One-Pot Analytical Pipeline for Efficient and Sensitive Proteomic Analysis of Extracellular Vesicles

Abstract

Extracellular vesicle (EV) proteomics emerges as an effective tool for discovering potential biomarkers for disease diagnosis, monitoring, and therapeutics. However, the current workflow of mass spectrometry-based EV proteome analysis is not fully compatible in a clinical setting due to inefficient EV isolation methods and a tedious sample preparation process. To streamline and improve the efficiency of EV proteome analysis, here we introduce a one-pot analytical pipeline integrating a robust EV isolation approach, EV total recovery and purification (EVtrap), with in situ protein sample preparation, to detect urinary EV proteome. By incorporating solvent-driven protein capture and fast on-bead digestion, the one-pot pipeline enabled the whole EV proteome analysis to be completed within one day. In comparison with the existing workflow, the one-pot pipeline was able to obtain better peptide yield and identify the equivalent number of unique EV proteins from 1 mL of urine. Finally, we applied the one-pot pipeline to profile proteomes in urinary EVs of bladder cancer patients. A total of 2774 unique proteins were identified in 53 urine samples using a 15 min gradient library-free data-independent acquisition method. Taken altogether, our novel one-pot analytical pipeline demonstrated its potential for routine and robust EV proteomics in biomedical applications.

Graphical Abstract

INTRODUCTION

Over the past decade, extracellular vesicles (EVs) have emerged as important mediators of cell-to-cell communication, homeostasis, disease pathogenesis, and other biological functions.^1–3 Depending on the physiological conditions, EVs carry a unique set of cargoes, including proteins,⁴ nucleic acids,⁵ and metabolites⁶ that can reflect molecular processes in the cell of origin, further highlighting their potential use for biomarker discovery and disease diagnosis.^7–9 In fact, EV-based clinical detection offers several advantageous features. The nature of EVs that they are released into biofluids, such as urine, plasma, and saliva, is beneficial for developing liquid biopsies that are minimally or non-invasive compared to tissue biopsies.^10,11 Detection of EV cargoes in biofluids can also favor continuous sampling and longitudinal monitoring in a clinical setting.^12,13 Moreover, a growing number of studies in recent years have shown that EV markers could be identified even before disease progression and utilized for the detection of early-stage cancers.^14–17 In addition, since EVs are membrane-enclosed structures, the protein or nucleic acid contents can be protected from degradation while circulating in biofluids.^18–20 Because of these advantages, a detailed understanding of EV cargoes and their relationships with diseases is crucial in preclinical and clinical studies to develop EV-based diagnostic tools. Most importantly, as personalized medicine becomes a more common practice, molecular profiling of EVs can provide valuable information for more effective diagnostic and therapeutic strategies that may reduce the need for surgical intervention.^21,22

Recently, mass spectrometry (MS)-based proteomic analysis has successfully demonstrated its powerful capability to profile EV proteins and discover potential biomarkers for diseases such as cancers, neurodegenerative conditions, and cardiovascular diseases.^20,23–25 Among different strategies, bottom-up proteomics is the most widely used approach for the identification and quantification of EV proteins, and the typical workflow mainly involves several steps, including the isolation of EVs from biofluids, extraction of EV proteins, proteolytic digestion of extracted proteins, and peptide cleanup before the MS run.^26–28 Although various protocols and technologies are available for obtaining a more comprehensive EV protein profile, the tedious sample preparation and lengthy processing time are still major concerns and seriously limit EV proteomic analysis in clinical applications.^29,30 For instance, isolation of EVs using differential ultracentrifugation (DUC) normally requires several hours depending on the sample condition, such as complexity or viscosity, which negatively influences the feasibility of this method in a clinical setting.^31,32 Other drawbacks of DUC, like low recovery and loss of EV subpopulations, have also been discussed.^33–35 Besides EV isolation, the sample processing for proteomic analysis usually involves overnight enzymatic digestion, sample cleanup (e.g., desalting), and several sample concentration steps, leading to decreased throughput and sample loss.³⁶ Therefore, a simpler and more efficient EV proteomic workflow is needed.

Our group has previously introduced the extracellular vesicle total recovery and purification (EVtrap) technique that is based on chemical affinity to capture EVs onto magnetic beads modified with both hydrophilic and lipophilic functional groups.³⁷ EVtrap has demonstrated successful recovery of EVs at a much higher yield with fewer contaminants from various types of biofluid samples compared to DUC and several other existing isolation methods.^25,37,38 Moreover, the whole EV isolation procedure using EVtrap beads can be accomplished within one hour without any expensive equipment, making it more compatible with clinical applications. By taking advantage of the EVtrap approach for EV isolation, we aim at integrating the EV isolation with the downstream sample preparation for EV proteomic analysis.

Recently, multiple attempts have been made to improve the efficiency and throughput of sample preparation for proteomic analysis. For example, the utilization of photocleavable surfactant, Azo, to assist in sample preparation has demonstrated effective extraction of proteins from EVs.³⁹ This one-pot Azo-based method also enabled rapid trypsin digestion, thereby increasing the overall throughput of proteomic analysis. Another promising method, single-pot solid-phase-enhanced sample preparation (SP3), has been introduced and applied in recent proteomic studies.^40,41 By using carboXylate-modified paramagnetic beads and solvent-driven protein capture, the original SP3 protocol showed unbiased recovery of proteins and peptides and good compatibility with a wide range of common chemicals used in proteomics sample preparation (e.g., detergents, chaotropes, or salts).⁴⁰ Without the need for complicated handling procedures, SP3 provides a simple and rapid sample preparation that outperforms other existing protocols. Many paramagnetic beads with different surface chemistries (e.g., Dynabeads, Sera-Mag beads, and MagReSyn beads) have been utilized on the SP3 platform.^41,42 One recently introduced approach, Mag-Net, has also demonstrated the feasibility of using strong-anion exchange magnetic beads for the SP3 procedure and analysis of plasma EV proteome.⁴³ To achieve streamlined and efficient EV proteomic analysis, a sample process platform that can perform a complete EV proteomics workflow, including both EV isolation and MS-based proteomics sample processing, is essential.

In this study, we report a one-pot analytical pipeline for high-throughput proteomic analysis of EVs isolated from urine samples. Urinary EVs were captured onto the EVtrap beads, and the EV-bound beads were directly subjected to lysis and the subsequent acetonitrile-driven protein capture for on-bead enzymatic digestion. EV peptides were then extracted without the need for desalting and analyzed by LC–MS/MS. By utilizing the one-pot pipeline, we detected ∼1000 unique EV proteins from 1 mL of urine using a short 15 min LC gradient and direct data-independent acquisition (directDIA) analysis. Importantly, the whole procedure from EV isolation to MS analysis can be completed in ∼8 h, demonstrating superior efficiency compared to the existing workflow that normally takes 2 days. We have also applied the one-pot pipeline to investigate the difference in EV proteome between the urine samples from healthy individuals and bladder cancer patients. Several bladder cancer-associated protein markers were successfully identified and showed significant changes in abundance in patients versus healthy controls. The results from this study demonstrated the potential of this simple and efficient one-pot EV proteomics pipeline for clinical applications.

EXPERIMENTAL SECTION

Experimental details on DUC, the existing EVtrap-based proteomics pipeline, Western blot, and silver staining are included in the Supporting Information.

Sample Collection

For characterization and comparison of the one-pot analytical pipeline, urine samples were collected from healthy individuals after informed consent. For the bladder cancer experiment, urine samples were collected from 37 patients and 16 healthy individuals under approval from Purdue University Human Research Protection Program and Indiana University Human Subjects Office Institutional Review Boards. All collected urine samples were further centrifuged at 2500g for 10 min to remove cell debris, apoptotic bodies, and other large particles. The supernatant was collected and stored at −80 °C for the experiments.

One-Pot EV Proteomic Pipeline

Urinary EVs were isolated by the EVtrap approach based on the previous study with some modifications.³⁷ The EVtrap beads (Tymora Analytical) were added to the urine samples at a 1:50 (v/v) ratio, and the samples were incubated by end-over-end rotation for 30 min. After the incubation, the supernatants were removed using a magnetic separator rack. The beads were then washed with washing buffer followed by PBS and 50 μM triethylamine (TEA, Millipore-Sigma). The lysis solution containing 12 mM sodium deoxycholate, 12 mM sodium lauroyl sarcosinate, 100 mM triethylammonium bicarbonate buffer, 10 mM TCEP, 40 mM CAA, and protease inhibitor (Thermo Fisher Scientific) was directly added to the beads, and the samples were incubated for 10 min at 95 °C with shaking at 1200 rpm. This step solubilized the EVs and denatured, reduced, and alkylated the EV proteins. After the samples were cooled down to room temperature for 5 min, acetonitrile was added to the sample to a final concentration of 70% (v/v). The supernatants were removed, and the beads were further washed with 70% acetonitrile three times. To perform a phase transfer surfactant-aided enzymatic digestion,⁴⁴ the beads were reconstituted in the digestion solution containing 2.4 mM sodium deoxycholate, 2.4 mM sodium lauroyl sarcosinate, 50 mM triethylammonium bicarbonate buffer, and Lys-C (Wako) at a 1:100 (wt/wt) enzyme-to-protein ratio for 1 h at 37 °C with shaking at 1200 rpm. Trypsin was added to a final 1:50 (wt/wt) enzyme-to-protein ratio for 3 h at 37 °C with shaking at 1200 rpm. The samples were acidified with trifluoroacetic acid (TFA) to a final concentration of 1% (v/v). Ethyl acetate solution was added to the samples at a 1:1 ratio to remove the surfactants. The mixture was vortexed for 2 min and centrifuged at 20,000g for 3 min to obtain aqueous and organic phases. The organic phase (top layer) was removed, and the aqueous phase was collected. This surfactant removal step was repeated one more time. Acetonitrile was added to the aqueous phase to a final concentration of 60% (v/v). Peptide-containing supernatants were collected, and the peptide concentration was determined using the Pierce Quantitative Colorimetric Peptide Assay (Thermo Scientific) according to the manufacturer’s instructions. Peptide samples were dried in a vacuum centrifuge for LC–MS/MS analysis.

LC–MS/MS Analysis

Dried peptide samples were dissolved in 0.05% TFA with 2% (v/v) ACN. A total of 10 μL (0.5 μg) of each sample was injected into an Ultimate 3000 nano UHPLC system (Thermo Fisher Scientific, Waltham, MA, USA). Peptides were captured on a 2 cm Acclaim PepMap trap column and separated on a heated 50 cm Acclaim PepMap column (Thermo Fisher Scientific) containing C18 resin. The mobile phase buffer includes 0.1% formic acid in HPLC grade water (buffer A) with an eluting buffer containing 0.1% formic acid in 80% (v/v) acetonitrile (buffer B) run with a linear 15 min gradient of 5–45% buffer B at a flow rate of 300 nL/min. HPLC was coupled online with a Q-Exactive HF-X mass spectrometer (Thermo Fisher Scientific). The mass spectrometer was run in data-independent mode, in which a full-scan MS (polarity: positive; scan range: 390 to 1010 m/z with a resolution of 120,000; automatic gain control target (AGC): 3E6, maximum injection time: 45 ms; spectrum data type: centroid) was followed by MS/MS with 8.0 m/z staggered-isolation windows schemes (polarity: positive; 15,000 resolution; normalized collision energy: 25%; AGC - 3E6, maximum injection time: 22 ms; loop count: 75; spectrum data type: centroid).⁴⁵

DirectDIA Data Analysis for Identification and Quantification

The signal extraction and quantitation of the DIA data were performed in Spectronaut (Biognosys, v15), utilizing a directDIA (library-free) standard setting with several modifications. For Pulsar Search, specific digest types with trypsin/P enzymes, seven minimal peptide length, 52 maximum peptide length, a maximum of two missed cleavages, carbamidomethyl at cysteine as fixed modification, acetyl protein N-term and oxidation at methionine as variable modifications, and five as maximum variable modifications were used. The FDR at PSM, peptide, and protein group were set to 0.01. DIA analysis used dynamic retention time prediction with local regression calibration. Meanwhile, the identification was analyzed with mutated decoy generation and dynamic size at 0.1 fractions of library size. The quantitation was performed at the MS2 level by enabling cross-run normalization. Qvalue sparse and no imputing were used as data filtering.

Bioinformatics Analysis

The clinical sample data were analyzed using the Perseus software (version 2.0.7.0). The intensities of proteins were extracted from Spectronaut search results, and the missing values of intensities were replaced by a normal distribution with a downshift of 1.8 SDs and a width of 0.3 SDs. In the volcano plot, the significantly changed proteins were identified by the p-value and change fold. The vertical dotted line represents the difference of the ratio of 2 or 0.5 (log₂(ratio) = ±1), which is the strict criteria for the fold change threshold in our study; the horizontal dotted line represents a p-value of 0.05 (−log₁₀(0.05) = 1.30), which is significant based on a two-sample t-test with a permutation-based FDR cutoff 0.05 for each dataset. Hierarchical clustering analysis of significant proteins (FDR < 0.05, difference >2 or <0.5) was performed using Euclidean distance and average linkage. All visualizations of volcano plots, heatmap, density plots, and box plots were created using R (version 4.2.2). The KEGG pathway analysis and GO biological processes analysis were performed using ShinyGO (V0.77).⁴⁶

Data Availability

All mass spectrometry data have been deposited to the ProteomeXchange Consortium via the jPOST partner repository with the dataset identifier PXD041118.

RESULTS AND DISCUSSION

Development of a One-Pot EV Proteomics Pipeline

As EV isolation using EVtrap is based on functionalized magnetic beads, we examined whether we could use the EVtrap beads for both capturing EVs and in situ single-tube sample preparation for EV proteomic analysis (Figure 1). EVtrap beads with functionalized lipophilic and hydrophilic groups can bind to the lipid bilayer membranes of EVs, enabling effective isolation of EVs from biofluids. The EVs are then lysed on the beads using the phase-transfer surfactant protocol for efficient protein extraction.^47,48 The subsequent introduction of an organic solvent leads to the capture of proteins on the beads through either a hydrophilic interaction mechanism between the bead surface and proteins, or a recently proposed protein aggregation mechanism.^40,49 On-bead digestion of the captured proteins is directly performed in the same tube after removing contaminating agents by several washes. Finally, the digested peptides are recovered by adding an organic solvent while the proteolytic enzymes are left on the beads. This entire EV proteomics pipeline can be done in a one-pot manner and produces purified peptide samples that are ready for downstream LC–MS analysis.

Figure 1.

Schematic workflow of the one-pot EV proteomics pipeline.

We optimized several steps in the procedure to improve the EV isolation and protein/peptide recovery. In the first step, we improved the isolation specificity of EVs by removing more contaminating urine proteins. Since a high concentration (100–200 mM) of TEA was used to increase pH and elute EVs from the beads in the EVtrap approach, we reasoned that a mild concentration of TEA could be used to elute contaminating molecules with lower binding affinity compared to EVs. By testing the effect of various TEA concentrations on the elution of contaminants, 50 μM TEA was found to elute contaminating proteins, while EVs started being eluted out from 100 μM TEA (Figure S1A,B). Hence, we added a final washing step using 50 μM TEA. To examine the EV isolation efficiency, EV-bound beads after EVtrap were boiled in lithium dodecyl sulfate (LDS) lysis buffer, and the EV marker CD9 was detected by Western blotting. As shown in Figure 2A, the CD9 signal was much higher (>5-fold) in the EVs enriched by the EVtrap beads compared to the EVs isolated by DUC (Figure S2).

Figure 2.

Characterization of the one-pot EV proteomics pipeline by (A) detection of the CD9 EV marker using Western blotting and (B) silver-staining detection of total protein. Urine control: 100 μL (10%) of crude urine after 2500g centrifugation. DUC: differential ultracentrifugation of 1 mL of urine. EVs on beads: beads collected and boiled in LDS after EVtrap. 70% ACN binding: supernatant collected after adding 100% ACN into EV lysate to a final concentration of 70%. 70% ACN wash: supernatant collected during the washing steps using 70% ACN. Proteins on beads: beads collected and boiled in LDS after the washing steps. Peptide extraction: supernatant of peptide elution using 60% ACN. Enzymes on beads: beads collected and boiled in LDS after peptide elution.

After the lysis of EVs using sodium deoxycholate, organic solvent acetonitrile was added to retain EV proteins on the EVtrap beads. We investigated different concentrations of acetonitrile for optimal binding performance and found that proteins can completely be retained on the EVtrap beads only at acetonitrile concentrations higher than 70% (Figure S3). This result is consistent with the previous studies showing that the binding of proteins to the beads in SP3 is driven by organic solvent-induced protein aggregation.⁴⁹ Therefore, 70% acetonitrile was chosen for retaining proteins, while detergents and other reagents including denaturing, reducing, and alkylation chemicals were effectively removed.⁴⁰ The discarded supernatants from the binding and washing steps and the beads after these steps were also collected, boiled in LDS buffer, and detected by Western blotting and silver staining (Figure 2A,B). The results showed that no proteins were detected in the two supernatants and nearly all urinary EV proteins were retained on the beads, indicating the effectiveness of protein recovery.

We further evaluated the possibility of a shorter time of Lys-C/trypsin sequential digestion directly on EVtrap beads and found that a total of 4 h Lys-C/trypsin incubation was sufficient to completely digest all proteins as the standard overnight incubation (Figure S4A). A similar digestion efficiency between the short and overnight digestion was further verified by observing the comparable missed cleavage rates, with ∼12% of peptides containing more than one missed cleavage (Figure S4B). The feasibility to use the short digestion time can greatly benefit the throughput of EV proteome analysis since digestion is usually the most time-consuming step in traditional proteomics sample preparation.

To recover peptides off the EVtrap beads, we aimed to obtain a peptide sample that does not need further cleanup steps before LC–MS analyses or downstream processes such as phosphopeptide enrichment. By testing different concentrations of acetonitrile, we found that 60% acetonitrile was able to retain Lys-C and trypsin on the beads while recovering EV peptides into the solution, resulting in overall high protein identification numbers (Figure S5A,B). There were no intact EV protein signals in the 60% acetonitrile supernatant and the beads portion after elution, again demonstrating the effectiveness of the shorter digestion time (Figure 2B). The eluted peptide samples were directly for LC–MS analysis without the additional desalting step.

One-Pot EV Proteomics Pipeline Matches or Outperforms the Existing In-Solution Digestion-Based Pipeline

In the existing in-solution digestion-based pipeline using the EVtrap approach (Figure 3A),^25,38,45 captured EVs need to be eluted first from the beads and dried for further processing. Following overnight digestion, the detergents need to be removed, and the large volume of the recovered diluted sample takes several hours to concentrate. To remove salts and other small molecules in the sample, a peptide desalting step is normally required before MS analysis. All the abovementioned steps result in a longer processing time needed, and the whole pipeline usually takes ∼2 days to obtain the EV proteome profile. On the other hand, the one-pot EV proteomics pipeline described here omits these time-consuming steps and only requires normal working hours (∼8 h) of processing time and MS analysis in total within a single day.

Figure 3.

Comparison of one-pot EV proteomics and existing EV proteomics pipelines. (A) Timeline comparison of the two procedures. (B) Total number of identified unique proteins in triplicates. (C) Overlap of proteins identified in both pipelines and top 100 EV markers in the Exocarta database. (D) Density plots of physical properties (molecular weight, isoelectric point, and GRAVY index) of identified unique peptides from each pipeline.

To understand the performance of the one-pot EV proteomics versus the existing procedures, we used 1 mL of urine as the starting material for both procedures and determined peptide concentrations using a colorimetric peptide assay (Figure S6). A higher total peptide amount (∼1.5-fold) obtained by the one-pot pipeline indicated the lower sample loss during the process, which could be attributed to the absence of sample transfer steps and an additional peptide desalting step. This favorable characteristic of the one-pot method may provide potential benefits for the analyses of post-translational modifications (PTMs), as a significant peptide amount is typically necessary for enrichment procedures to effectively analyze these low-abundant molecules.

Next, a final peptide amount of 0.5 μg from each pipeline was analyzed by LC–MS using a short 15 min LC gradient, and a library-free directDIA analysis was then performed for identification (1% FDR) and label-free quantification. We identified >5000 unique peptides from ∼1100 unique proteins using the existing pipeline, while the one-pot EV proteomics pipeline enabled the identification of >4000 unique peptides from ∼1000 unique proteins (Figure 3B, Tables S1 and S2). Quantitative reproducibility was assessed by the coefficients of variation (CV) of precursor and protein group intensities across three replicates (Figure S7). The median precursor CV and median protein CV in both pipelines were ∼10% and ∼7%, respectively, which indicates the good repeatability of the sample preparation procedures and the short gradient DIA method. A high degree of overlap in the identified proteins was observed between the two pipelines (Figure 3C). Moreover, there was no discernible bias in the physicochemical properties of the peptides from each pipeline, which was confirmed by analyzing the molecular mass, isoelectric point, and grand average hydropathy (GRAVY) of the peptides (Figure 3D). Interestingly, the peptides obtained by 60% acetonitrile extraction in the one-pot pipeline exhibit comparable properties to those processed by the desalting step in the existing pipeline. To specifically investigate the EV proteomes obtained by the two pipelines, we further compared our results with the ExoCarta database.⁵⁰ Among the top 100 EV markers and proteins, both pipelines identified ∼75 out of them (Figure 3C). The quantitative data generated from directDIA also demonstrated similar intensity levels for the EV markers (Table S3).

Overall, both the existing and one-pot EV proteomics pipelines appeared to provide similar proteome depth and quantification outcomes. However, the one-pot pipeline demonstrated clear advantages, such as a faster EV proteome profiling and simpler handling procedure, making it more suitable for clinical applications. We anticipate that this streamlined pipeline could benefit clinical proteomics by achieving a competitive turnaround time from receiving clinical samples to producing EV proteomic analysis results.

Bladder Cancer-Associated EV Proteins Can Be Identified Using the One-Pot EV Proteomics Pipeline

To demonstrate the applicability of the one-pot EV proteomics pipeline in clinical proteomics, we carried out a preliminary detection of potential urinary EV protein markers in bladder cancer. Bladder cancer is the 10th most common cancer in the world, and its incidence is still increasing worldwide.⁵¹ Cystoscopy is currently the standard test for diagnosis of bladder cancer; however, it is an invasive procedure and may lead to complications such as urethral perforation, hemorrhage, and infection.⁵² There have been a few attempts to analyze EVs as biomarkers that can provide an alternative source for non-invasive diagnostics.^53,54 To address the need for high-throughput clinical detection, we applied the one-pot EV proteomics pipeline to investigate the urinary EV proteomes from 37 patients already diagnosed with bladder cancer and 16 healthy individuals using 1 mL of urine from each. The clinical annotation of each sample is shown in Table S4. EVs were isolated from all urine samples using the EVtrap beads. The isolated EVs were then subjected to on-bead lysis, acetonitrile-driven protein capture, and on-bead digestion to obtain digested peptides. After peptide extraction, 0.5 μg of each peptide sample was injected into an Ultimate 3000 nanoLC coupled with a Q Exactive HF-X mass spectrometer. A short 15 min LC gradient and the data-independent mode were used for the acquisition of MS/MS spectra. Library-free direct-DIA was performed for identification and MS2-level quantification.

In the directDIA analysis result, 2774 unique proteins were identified and quantified across patients and healthy controls (Table S5, Figure S8). The volcano plot in Figure 4A shows the significantly upregulated and downregulated proteins in the patient group versus the control group with cutoff values of t-test permutation-based FDR (FDR = 0.05, −log₁₀(FDR) = 1.30) and difference in fold-change (fold-change = 2, log₂(fold-change) = ±1). The 38 statistically significant changing proteins in each patient and control group were further visualized in the heatmap (Figure 4B). Many of the significantly upregulated EV proteins have been reported to be associated with bladder cancer, and the quantification results of the representative proteins are shown in the box plots (Figure 4C). Actin-related protein 2/3 complex subunit 2 (ARPC2) is a member of the actin-related protein 2/3 (Arp2/3) family, which is involved in the regulation of actin polymerization and the formation of dendritic filopodia and lamellipodia.^55,56 Previous studies have found higher expression levels of ARPC2 in tissue specimens from bladder cancer patients, and a significant association between ARPC2 expression and various clinicopathological parameters, including tumor size, tumor multiplicity, tumor stage, tumor grade, and lymph node metastasis, was reported.^57,58 Likewise, most bladder cancer cells were found to exhibit higher protein levels of lactate dehydrogenase A (LDH-A).^59,60 The overexpression of LDH-A could contribute to the metabolic reprogramming in tumors and promote a glycolytic phenotype, which is a hallmark of cancer cells. Furthermore, LDH-A has been proposed as a cancer-selective therapeutic target for bladder cancers.⁵⁹ A high caspase 3 (CASP3) expression level has also been highlighted as a valuable indicator of early recurrence of invasive bladder cancer.^61,62

Figure 4.

Application of one-pot pipeline in the detection of bladder cancer-associated proteins in urinary EVs. (A) Volcano plot representing the quantitative comparison of the urinary EV proteomes from bladder cancer patients and healthy individuals. (B) Heatmap of significantly upregulated and downregulated proteins in bladder cancer patients versus healthy controls. Hierarchical clustering analysis of significant proteins (FDR < 0.05, difference > 2 or < 0.5) was performed using Euclidean distance and average linkage. (C) Relative abundance data of the four representative bladder cancer-associated proteins showed significant upregulation (P < 0.05) in patients. All values are log2 conversions of normalized protein abundance as determined by LC–MS. The data points are not imputed. (D) Upregulated KEGG pathways and (E) biological processes in bladder cancer patients. All significantly upregulated proteins in bladder cancer patients were used as input for the enrichment analyses.

We further carried out Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis and gene ontology (GO) analysis of all significantly upregulated proteins in bladder cancer patients, and the result demonstrated that the pentose phosphate pathway (PPP) and the pentose biosynthetic process showed the highest fold enrichment (Figure 4D,E). Unsurprisingly, the PPP was upregulated in bladder cancer patients since the NADH generated from this pathway can help proliferate cells to meet their anabolic demands and survive under oxidative stress.⁶³ Previous proteomics, metabolomics, and genomics studies have all observed the upregulation of the PPP in bladder cancer.^58,64,65 Additionally, the rate-limiting enzyme of the PPP, glucose-6-phosphate dehydrogenase (G6PD), has been suggested as a poor prognostic factor in bladder cancer and a target of several drugs for inhibition of bladder cancer cell proliferation.^66,67 Consistent with these findings, a significantly higher abundance of GP6D in patient-derived EVs was also detected in our result (Figure 4C).

Our results demonstrated the feasibility of using the one-pot EV proteomics pipeline for the potential clinical detection of bladder cancer-associated proteins. While this study does not fully implement biomarkers for clinical use at this stage, it lays the foundation for future applications in identifying EV protein biomarkers that would benefit early cancer detection. To achieve this goal, a large-scale study involving thousands of samples and validation experiments will be necessary to refine the potential biomarkers. Such an extensive dataset will improve the robustness and accuracy of detection or diagnosis, ultimately bolstering the clinical applicability. Notably, most of the abovementioned studies detected the protein expression levels and discovered protein markers in bladder cancer using tissue specimens or cultured cells, which are invasive to obtain or may not be fully representative of physiological conditions. Instead, EV proteins from biofluids such as urine can be a more promising source for discovering cancer biomarkers and developing diagnostic tools. From this perspective, our one-pot EV proteomics pipeline can potentially make the processes of biomarker discovery and diagnostic testing more efficient.

By utilizing the one-pot EV proteomics pipeline, we were able to finish processing all 53 samples and MS analyses within three days in a typical biochemistry lab. The throughput was mainly limited by instrument time and manual sample processing. We believe that this streamlined pipeline can be easily adapted to automated processing using robotics,^68,69 which will further improve the throughput and reduce sample-handling errors. However, the centrifugation step for removing the surfactants from the digestion solution may impede the automation process. A thorough assessment of surfactant-free digestion is recommended to enhance automation feasibility. In addition, the throughput and proteome depth can also be increased by applying more advanced LC systems, mass spectrometry instruments, or data acquisition approaches.^70–73

CONCLUSIONS

In the present work, we successfully demonstrated the versatility of EVtrap beads for both the effective capture of urinary EVs and the isolation of EV peptides through a one-pot procedure. This single-tube EV proteomics pipeline was more time-efficient and simpler than the existing pipeline while still providing equivalent identification and quantification results. The application of this pipeline for the detection of urinary EV proteomes from bladder cancer patients further underscored its potential use for discovering cancer biomarkers and facilitating subsequent diagnosis. Our results also shed light on the potential of expanding this pipeline to analyze EV proteomes from other clinical biofluids, such as plasma, saliva, or lavage, which will be beneficial for the development of efficient liquid biopsy tests. With state-of-the-art MS and automation systems, we expect that this pipeline will open an avenue for a high-throughput, fast, and economical approach that could be standardized for largescale and longitudinal clinical proteomic analyses.