A multiplex quantitative proteomics strategy for protein biomarker studies in urinary exosomes

Urinary exosomes have received considerable attention as a potential biomarker source for the diagnosis of renal diseases. Notwithstanding, their use in protein biomarker research is hampered by the lack of efficient methods for vesicle isolation, lysis, and protein quantification. Here we report an improved ultracentrifugation-based method that facilitates the solubilization and removal of major impurities associated with urinary exosomes. A double-cushion sucrose/D2O centrifugation step was used after a two-step differential centrifugation to separate exosomes from the heavier vesicles. After the removal of uromodulin, 378 and 79 unique proteins were identified, respectively, in low- and high-density fractions. Comparison of our data with two previously published data sets helped to define proteins commonly found in urinary exosomes. Lysis, protein extraction, and in-solution digestion of exosomes were then optimized for MudPIT application. More than a hundred exosomal proteins were quantified by four-plex iTRAQ analysis of single and pooled samples from two different age groups. For healthy men, six proteins (TSN1, PODXL, IDHC, PPAP, ACBP, and ANXA5) showed significant expression differences between exosome pools of those aged 25–50 and 50–70 years old. Thus, exosomes isolated by our method provide the basis for the development of robust quantitative methods for protein biomarker research. Urinary exosomes have received considerable attention as a potential biomarker source for the diagnosis of renal diseases. Notwithstanding, their use in protein biomarker research is hampered by the lack of efficient methods for vesicle isolation, lysis, and protein quantification. Here we report an improved ultracentrifugation-based method that facilitates the solubilization and removal of major impurities associated with urinary exosomes. A double-cushion sucrose/D2O centrifugation step was used after a two-step differential centrifugation to separate exosomes from the heavier vesicles. After the removal of uromodulin, 378 and 79 unique proteins were identified, respectively, in low- and high-density fractions. Comparison of our data with two previously published data sets helped to define proteins commonly found in urinary exosomes. Lysis, protein extraction, and in-solution digestion of exosomes were then optimized for MudPIT application. More than a hundred exosomal proteins were quantified by four-plex iTRAQ analysis of single and pooled samples from two different age groups. For healthy men, six proteins (TSN1, PODXL, IDHC, PPAP, ACBP, and ANXA5) showed significant expression differences between exosome pools of those aged 25–50 and 50–70 years old. Thus, exosomes isolated by our method provide the basis for the development of robust quantitative methods for protein biomarker research. Single-protein biomarkers support the molecular diagnosis and medical management of various disorders in clinical practice today. Urine is one of the most attractive biomarker sources for large-scale noninvasive clinical screening programs. Sample variability, complexity, and the wide dynamic concentration range of urinary proteins, however, present a significant analytical challenge for biomarker discovery.1.Court M. Selevsek N. Matondo M. et al.Toward a standardized urine proteome analysis methodology.Proteomics. 2011; 11: 1160-1171Crossref PubMed Scopus (51) Google Scholar Therefore, the number of high-throughput high-confidence urinary protein biomarkers identified by proteomic approaches currently is still somewhat limited. Exosomes are extracellular membrane-bound nanovesicles originating from the intraluminal vesicles of multivesicular bodies secreted by diverse cell types under both normal and pathological conditions.2.Simpson R.J. Lim J.W.E. Moritz R.L. et al.Exosomes: proteomic insights and diagnostic potential.Expert Rev Proteomics. 2009; 6: 267-283Crossref PubMed Scopus (831) Google Scholar Despite their role in immune system modulation,3.Li X.-B. Zhang Z.-R. Schluesener H.J. et al.Role of exosomes in immune regulation.J Cell Mol Med. 2006; 10: 364-375Crossref PubMed Scopus (113) Google Scholar the biological role of exosomes remained elusive until Lötvall's group demonstrated that exosomes can transfer genetic information from one cell to another.4.Valadi H. Ekstrom K. Bossios A. et al.Exosome-mediated transfer of mRNAs and microRNAs is a novel mechanism of genetic exchange between cells.Nat Cell Biol. 2007; 9: 654-659Crossref PubMed Scopus (8922) Google Scholar Exosomes were shown to transport proteins, mRNAs, and microRNAs and to modulate immune reaction, angiogenesis, cell proliferation, and tumor cell invasion. Consequently, their use in biomarker research as a source of disease-relevant cargo proteins shows great promise. Exosomes have been isolated from various biofluids including urine.5.Keller S. Ridinger J. Rupp A.-K. et al.Body fluid derived exosomes as a novel template for clinical diagnostics.J Trans Med. 2011; 9: 86Crossref PubMed Scopus (526) Google Scholar Application of proteomics to urinary exosomes secreted by the urinary tract epithelial cells has recently been reviewed.6.van Balkom B.W.M. Pisitkun T. Verhaar M.C. et al.Exosomes and the kidney: prospects for diagnosis and therapy of renal diseases.Kidney Int. 2011; 80: 1138-1145Abstract Full Text Full Text PDF PubMed Scopus (163) Google Scholar,7.Moon P.-G. You S. Lee J.-E. et al.Urinary exosomes and proteomics.Mass Spectrom Rev. 2011; 30: 1185-1202Crossref PubMed Scopus (72) Google Scholar The limited progress of urinary exosomes in biomarker discovery can be explained by the high and variable concentration of uromodulin, also referred to as Tamm–Horsfall glycoprotein, and the lack of protocols enabling urinary exosomal proteins for quantitative proteomics. Uromodulin assembles into intracellular filaments forming three-dimensional matrix. This filament network traps exosomes and prevents their efficient isolation and purification by the traditional methods.8.Fernandez-Llama P. Khositseth S. Gonzales P.A. et al.Tamm-Horsfall protein and urinary exosome isolation.Kidney Int. 2010; 77: 736-742Abstract Full Text Full Text PDF PubMed Scopus (217) Google Scholar,9.Rood I.M. Deegens J.K.J. Merchant M.L. et al.Comparison of three methods for isolation of urinary microvesicles to identify biomarkers of nephrotic syndrome.Kidney Int. 2010; 78: 810-816Abstract Full Text Full Text PDF PubMed Scopus (194) Google Scholar To overcome this problem, recently, dithiothreitol was used to reduce the intermolecular disulfide bonds of uromodulin.8.Fernandez-Llama P. Khositseth S. Gonzales P.A. et al.Tamm-Horsfall protein and urinary exosome isolation.Kidney Int. 2010; 77: 736-742Abstract Full Text Full Text PDF PubMed Scopus (217) Google Scholar,10.Pisitkun T. Shen R.-F. Knepper M.A. Identification and proteomic profiling of exosomes in human urine.Proc Natl Acad Sci USA. 2004; 101: 13368-13373Crossref PubMed Scopus (1630) Google Scholar Treatment with dithiothreitol results in a somewhat higher yield of urinary exosomes; however, it does not seem to work in all cases. This prompted us to revise and improve the current isolation method. We have set up new protocols for the isolation/purification and also for the lysis and subsequent solubilization of membrane proteins to meet the need of protein biomarker discovery platform. Quantitative proteomics based on iTRAQ labeling and MudPIT was applied to demonstrate the feasibility of multiplex quantitative approach. For the iTRAQ experiment, four samples were prepared (Table 1) with the aim of detecting differences in the proteomes of urinary exosomes of (i) an individual and their age-matched pooled control, (ii) individuals of two different age groups, and (iii) two different exosome preparations (conventional single-cushion and new double-cushion) in a single multiplex analysis.Table 1The four urinary exosome samples analyzed using iTRAQ labeling and MudPIT quantitative proteomicsSample numberExosome preparation methodAge (years)Number of individualsSample volume per individual (ml)Sampling methodLabel1aSamples 1 and 4 were prepared from the same pool of individual urine samples.Double cushion25–452050PoolediTRAQ-1142Double cushion50–702050PoolediTRAQ-1153Double cushion4311000SingleiTRAQ-1164aSamples 1 and 4 were prepared from the same pool of individual urine samples.Single cushionbUrinary vesicles prepared by the single-cushion method using phosphate-buffered saline buffer.25–452050PoolediTRAQ-117a Samples 1 and 4 were prepared from the same pool of individual urine samples.b Urinary vesicles prepared by the single-cushion method using phosphate-buffered saline buffer. Open table in a new tab Current urinary exosome isolation and purification methods are based on ultracentrifugation or filtration. In the two-step differential centrifugation protocol,10.Pisitkun T. Shen R.-F. Knepper M.A. Identification and proteomic profiling of exosomes in human urine.Proc Natl Acad Sci USA. 2004; 101: 13368-13373Crossref PubMed Scopus (1630) Google Scholar first a low-velocity sequential centrifugation is performed to remove cells and cellular debris. The second step is ultracentrifugation of the supernatant at 100,000–200,000g velocity to sediment urinary vesicles. Vesicles isolated by differential centrifugation are contaminated by abundant urinary proteins (uromodulin, albumin, etc.) and are heterogeneous in size. To isolate more homogeneous vesicle population at a higher purity grade, the crude preparation is further processed by sucrose gradient or sucrose cushion centrifugation. The latter uses a small density cushion typically composed of 1mol/l sucrose in deuterium oxide (D2O) for the separation of exosomes by the formation of a micro-gradient.2.Simpson R.J. Lim J.W.E. Moritz R.L. et al.Exosomes: proteomic insights and diagnostic potential.Expert Rev Proteomics. 2009; 6: 267-283Crossref PubMed Scopus (831) Google Scholar Although sucrose gradient and cushion centrifugations allow a better separation of vesicles of different sizes, it does not solve the problem of the co-purifying uromodulin.11.Hogan M.C. Manganelli L. Woollard J.R. et al.Characterization of PKD protein-positive exosome-like vesicles.J Am Soc Nephrol. 2009; 20: 278-288Crossref PubMed Scopus (240) Google Scholar We aimed to find suitable conditions for the solubilization of uromodulin aggregates in order to shift uromodulin from the vesicle-containing layer into the soluble protein fraction during the cushion centrifugation. We have found that 20mmol/l Tris, pH 8.6, buffer facilitates the solubilization of filaments and keeps uromodulin in solution. To improve the separation of exosomes from the heavier vesicles and/or membrane fragments, the single cushion was replaced by a double cushion. In this step, the crude exosome pellet solubilized in the Tris buffer is centrifuged on a double layer composed of 1 and 2mol/l sucrose in D2O. Qualitative transmission electron microscopy image of the 1-mol/l fraction shows small (30–50nm in diameter) intact membrane vesicles with the typical morphology of exosomes (Figure 1b). Fewer filaments and more homogeneous vesicle distributions can be observed in this preparation than in vesicles obtained by the conventional single-cushion method (Figure 1a). In the 2-mol/l fraction, larger urinary vesicles and aggregates were detected (Figure 1c). Starting with 1000ml urine, the process yields vesicles containing 100μg of proteins in the 1-mol/l fraction and 10μg in the 2-mol/l fraction. SDS-polyacrylamide gel electrophoresis (PAGE) and western blot analyses performed at different steps of the process show how the amount of co-purifying uromodulin is reducing at each step (Figure 2a, lanes 3–6). By comparing the SDS-PAGE and western blot images of the 1-mol/l fraction (Figure 2, lane 6) with the urinary vesicles prepared by the conventional differential centrifugation (Figure 2, lane 3) and single-cushion (Figure 2, lane 8) methods, clear improvements in protein pattern (Figure 2a) and quantity of exosome markers (Figure 2b) were seen. Vesicles were purified five times from the same biological sample (pooled samples of 10 healthy male donors) with good reproducibility (Figure 3). Different biological replicates resulted in similar pattern too (Figure 3). We also purified urinary vesicles from patients having micro- and macro-proteinuria and found very similar SDS-PAGE pattern to that of healthy people (data not shown). Interestingly, the two vesicle fractions (1 and 2mol/l) show a very different SDS-PAGE pattern (Figure 2), which was further investigated in a proteomic study.Figure 2SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and western blot (WB) analyses showing the different stages of urinary exosome isolation/purification process. (a) SDS-PAGE and (b) WB images. L1, total urine; L2, exosome-depleted urine; L3, crude exosome after differential centrifugation; L4, 15,000g pellet; L5, 15,000g supernatant; L6, exosomes purified in the 1-mol/l sucrose fraction by double-cushion method using 20mmol/l Tris, pH 8.6; L7, urinary vesicles purified in the 2-mol/l sucrose fraction by the double-cushion method; L8, exosomes purified by the single-cushion method using phosphate-buffered saline buffer; M, protein molecular mass market. Western blots were probed with antibodies against the following market proteins: Na-K-Cl cotransporter isoform 2 (NKCC2), apoptosis-linked gene-2-interacting protein X (ALIX), Nap/Hp exchanger 3 (NHE3), tumor susceptibility gene 101 (TSG101), β-ACTIN, aquaporin 2 (AQP2), and CD9.View Large Image Figure ViewerDownload (PPT)Figure 3SDS-polyacrylamide gel electrophoresis (PAGE) image shows a good analytical reproducibility of the double-cushion sample preparation method of urinary exosomes using a pooled sample of healthy volunteers in five consecutive experiments (left, lanes A1–A5). Exosomes isolated/purified from two different pooled samples of healthy volunteers show (right, lanes B1 and B2) very similar SDS-PAGE patterns too. Lane M: protein molecular mass marker.View Large Image Figure ViewerDownload (PPT) The 1- and 2-mol/l fractions obtained by double-cushion centrifugation were subjected to in-gel and in-solution digestion–based proteomic analyses. Proteins identified by at least two unique peptides and P<0.05 significance are reported in Supplementary Tables online. In all, 378 and 79 unique proteins were identified in the 1- and 2-mol/l fractions, respectively. Proteins in the main SDS-PAGE bands of the 2-mol/l fraction (semenogelin 1, semenogelin 2, olfactomedin, PSCA, and PPAP) previously have been identified in seminal prostasomes.12.Utleg A.G. Yi E.C. Xie T. et al.Proteomic analysis of human prostasomes.The Prostate. 2003; 56: 150-161Crossref PubMed Scopus (211) Google Scholar It was therefore reasonable to deduce that the 2-mol/l fraction contains vesicles heavier than exosomes, such as urinary-secreted prostasomes. Of 79 proteins, 15 proteins, including MASP2 and MUC6 prostasomal proteins, are present only in the 2-mol/l fraction, whereas the remaining (64 proteins) were found to be common in the two sets. The identified proteins were functionally annotated using the open-source protein annotation tool, STRAP.13.Bhatia V.N. Perlman D.H. Costello C.E. et al.Software tool for researching annotations of proteins: open-source protein annotation software with data visualization.Anal Chem. 2009; 81: 9819-9823Crossref PubMed Scopus (191) Google Scholar The systems biology modeling, which included gene ontology and canonical genetic pathway analysis, shows that urinary exosomes are particularly enriched in proteins involved in biological regulation and binding (Figure 4). Indeed, 33 different Ras-related Rab proteins, known regulators of different steps of membrane traffic, including vesicle formation and movement, were identified. Different components of endosomal sorting complexes required for transport (ESCRT), such as ESCRT-I (VPS28, VPS37B, and VPS37D), ESCRT-II (VPS25), ESCRT-III (CHMP1B, CHMP2A, CHMP2B, CHMP4B, and CHMP5), and VPS4 (VPS4A, VPS4B, and VTA1), with a known role in exosome biogenesis were also identified. Ten and four proteins from the protein families of Annexins and Copines, respectively, with known calcium-dependent phospholipid-binding properties were identified too. Another binding protein family highly represented in our data set (by 13 proteins identified) is the guanine nucleotide–binding proteins (G-proteins) involved as modulators and transducers in various transmembrane signaling systems. In addition, CD9, CD38, CD55, CD59, and CD63 antigens present on the surface of exosomes and known to participate in immune regulation were identified. These findings are in line with the key biological functions of exosomes discovered recently in cell–cell communication and in the modulation of different cellular mechanisms including the immune system. Therefore, the proteomes of urinary vesicles obtained by this improved isolation and purification process (Supplementary Tables online) may provide a promising source for robust and reproducible biomarker discovery studies. Download .pdf (.07 MB) Help with pdf files Supplementary Table 2 A detailed comparison was performed between the proteins identified in the 1-mol/l fraction of the double-cushion ultracentrifugation and two previously published data sets using differential centrifugation10.Pisitkun T. Shen R.-F. Knepper M.A. Identification and proteomic profiling of exosomes in human urine.Proc Natl Acad Sci USA. 2004; 101: 13368-13373Crossref PubMed Scopus (1630) Google Scholar,14.Gonzales P.A. Pisitkun T. Hoffert J.D. et al.Large-scale proteomics and phosphoproteomics of urinary exosomes.J Am Soc Nephrol. 2009; 20: 363-379Crossref PubMed Scopus (559) Google Scholar and microfiltration15.Merchant L.M. Powel D.W. Wilkey D.W. et al.Microfiltration isolation of human urinary exosomes for characterization by MS.Proteomics Clin Appl. 2010; 4: 84-96Crossref PubMed Scopus (163) Google Scholar in combination with mass spectrometry (MS)-based protein identification (Supplementary Table S1B online). For this comparison, only proteins identified with at least two unique peptides were considered. So far, the largest data set comes from the work of Knepper's group,10.Pisitkun T. Shen R.-F. Knepper M.A. Identification and proteomic profiling of exosomes in human urine.Proc Natl Acad Sci USA. 2004; 101: 13368-13373Crossref PubMed Scopus (1630) Google Scholar,14.Gonzales P.A. Pisitkun T. Hoffert J.D. et al.Large-scale proteomics and phosphoproteomics of urinary exosomes.J Am Soc Nephrol. 2009; 20: 363-379Crossref PubMed Scopus (559) Google Scholar publicly available in the Urinary Exosome Database (http://dir.nhlbi.nih.gov/papers/lkem/exosome/). The data set is built on two gel-based, large-scale proteomic studies using two different types of ion-trap instruments for peptide analysis. The data set includes proteins identified in pooled samples of healthy volunteers (male and female) of different age. It reports 531 proteins that have been identified with at least two unique peptides. Merchant et al.,15.Merchant L.M. Powel D.W. Wilkey D.W. et al.Microfiltration isolation of human urinary exosomes for characterization by MS.Proteomics Clin Appl. 2010; 4: 84-96Crossref PubMed Scopus (163) Google Scholar on the other hand, identified 94 proteins by isolating exosomes using microfiltration and in-solution digestion–based shotgun method from a pooled urine sample. Figure 5 shows the Venn diagram of the comparison (Supplementary Table S1B online). Out of the 378 proteins identified in our work (Supplementary Table S1A online), 216 and 48 proteins were found to overlap with Gonzales14.Gonzales P.A. Pisitkun T. Hoffert J.D. et al.Large-scale proteomics and phosphoproteomics of urinary exosomes.J Am Soc Nephrol. 2009; 20: 363-379Crossref PubMed Scopus (559) Google Scholar and Merchant15.Merchant L.M. Powel D.W. Wilkey D.W. et al.Microfiltration isolation of human urinary exosomes for characterization by MS.Proteomics Clin Appl. 2010; 4: 84-96Crossref PubMed Scopus (163) Google Scholar data sets, respectively. It should be noted that, in spite of the variability of the biological sample and difference in methodologies, the overlap with Gonzales's data set is relatively high (46%). In total, there are 247 proteins that have been identified by at least two of the three methods. These we considered as the common and most readily identifiable proteins of urinary exosomes, and we called them as the 'the core urinary exosome proteome'. The core protein set was cross-checked against a data set containing 1416 entries from exosome studies conducted on human cell lines and biological fluids (except urine) extracted from the ExoCarta database16.Mathivanan S. Simpson R.J. ExoCarta: a compendium of exosomal proteins and RNA.Proteomics. 2009; 9: 4997-5000Crossref PubMed Scopus (667) Google Scholar (Supplementary Table S1C online). Out of the 247 core proteins, 171 proteins (69%) were found to be identified in biological sources other than urine, whereas 76 proteins (31%) were specific of urine. Some of these can be contaminants (such as uromodulin), whereas others (such as membranes of solute carrier family 12) could be true exosomal proteins specific to the epidermal cells lining the urinary tract. In a similar way, we compared those proteins that have been identified by only one of the three methods. Proteins identified only by this work overlap with a higher percentage (50%, 78 out of 156) with the human exosomal proteins listed in ExoCarta (Supplementary Table S1C online) compared with the data sets of Gonzales (41%) and Merchant (38%). This is in accordance with the transmission electron microscopy (Figure 1), SDS-PAGE (Figure 2a), western blot (Figure 2b), and STRAP (Figure 4) analysis, and indicates a higher content of true exosomal proteins and lower contamination in sample prepared by the double-cushion method. Download .pdf (.26 MB) Help with pdf files Supplementary Table 1B Download .pdf (.29 MB) Help with pdf files Supplementary Table 1A Download .pdf (.9 MB) Help with pdf files Supplementary Table 1C In-solution digestion–based quantitative strategy is being increasingly applied to the discovery phase of protein biomarker research in clinical proteomics over the tedious gel-based analysis. In urinary exosome research, however, it is hampered by inefficient lysis and solubilization of highly resistant lipid vesicles, resulting in low protein yield and altered protein composition. Here we show that using an acid-cleavable detergent (RapiGest) can greatly facilitate vesicle lysis, protein solubilization, and proteolysis enhancement for downstream liquid chromatography–mass spectrometry (LC-MS/MS) analysis of the proteins of urinary vesicles. The detergent is cleaved and easily removed after digestion by solid-phase extraction. Figure 6 shows the experimental setup used in this work. The iTRAQ labeling method was used to measure protein relative abundances. Table 1 displays the samples for iTRAQ run, and Figure 7 shows the SDS-PAGE images of the samples before the labeling reaction. Samples were prepared in parallel according to the single- and double-cushion protocols from a pooled urine sample of 20 healthy male donors in the age group of 25–45 years. Effects of age and sample pooling on the relative protein quantities were monitored in the same experiment. A total of 114 proteins were quantified, and the measured weighted median protein ratios are reported in Table 2. A twofold change was considered to be biologically significant in this analysis. The results confirm that the isolation/purification method does highly influence the protein abundances (Table 2, ratio 117/114). A total of 37 exosomal proteins show a significantly decreased level in the sample prepared by the single-cushion method. In addition, a 20-fold increase in uromodulin was detected. On comparing the two different age groups (Table 2, ratio 115/114), significant differences were found in six proteins. Interestingly, altered expressions of five out of these six proteins (TSN1, PODXL, IDHC, PPAP, and ANXA5) have been implicated in cancer and cancer prognosis. In all, 12 downregulated and 9 upregulated proteins were found to be differently expressed when a pooled sample of 20 individuals was compared with that of a single person. Most prominently, higher levels of Annexin-1 (5.3-fold), PSCA (4.8-fold), and Aquaporin-2 (2.9-fold), and lower level of PDXL (7.8-fold), were measured in the single sample. These observations indicate that while pooling individual samples certainly reduces biological variation but can potentially hide biological significance.Figure 7Comparison of the four urinary exosome samples (Table 1) prepared for the multiplex quantitative proteomic workflow (Figure 6). SDS-polyacrylamide gel electrophoresis (a) and western blot (b) analyses of the four exosome samples (Table 1) prepared for the quantitative iTRAQ analysis. Samples 1–3 were prepared by the double-cushion method using 20mmol/l Tris, pH 8.6, and sample 4 was prepared by the single-cushion method using phosphate-buffered saline buffer (conventional preparation). L1, exosomes isolated from a pooled sample of 20 healthy male volunteers in the age group of 25–50 years; L2, exosomes isolated from a pooled sample of 10 healthy male volunteers in the age group of 50–70 years; L3, exosomes isolated from a pooled sample of a male individual aged 43 years; L4, exosomes isolated by the single-cushion method from a pooled sample of 20 healthy male volunteers in the age group of 25–50 years; Lane M, protein molecular mass marker. Western blots (b) were probed with antibodies against the following marker proteins: Na–K–Cl cotransporter isoform 2 (NKCC2), apoptosis-linked gene-2-interacting protein X (ALIX), Na+/H+ exchanger 3 (NHE3), tumor susceptibility gene 101 (TSG101), β-ACTIN, aquaporin 2 (AQP2), and CD9.View Large Image Figure ViewerDownload (PPT)Table 2Weighted median ratios of urinary exosomal proteins quantified in the 4-plex iTRAQ experiment using MudPIT quantitative proteomicsTable 2Weighted median ratios of urinary exosomal proteins quantified in the 4-plex iTRAQ experiment using MudPIT quantitative proteomics The objective of this study was to solve the problem of co-purifying urinary proteins in the preparation of urinary exosomes and to set up protocols that allow the MudPIT-based strategy for the quantification of exosomal proteins. The major barrier for protein biomarker studies through urinary exosomes is the presence of highly abundant urinary proteins forming polymeric networks and entrapping the vesicles. As a first step, we developed a method for the efficient isolation and purification of urinary exosomes based on impurity solubilization and sucrose double-cushion centrifugation. We show that uromodulin and other cytoplasmic filament contaminations can be efficiently removed by the method (Figures 1 and 2), whereas the vesicles remain intact (Figure 1). The method enables the separation of exosomes from larger vesicles, such as prostasomes (Figure 1). Next, we set up a new protocol for the lysis of exosomes and subsequent solubilization of membrane proteins by using a commercial acid-cleavable detergent. This step was found to be fundamental to successfully interface urinary exosome preparation method with current in-solution digestion-based quantitative proteomics. Urinary vesicles recovered from the two cushions were analyzed by gel-based and, for the first time by, in-solution digestion–based proteomics. Proteomic analyses identify fewer proteins in the new preparations than previous analysis in earlier preparations (Supplementary Tables online). This is partly due to the removal of contaminating urinary proteins and partly to the use of stringent criteria we applied for protein identification. In fact, the majority of the proteins present in the exosome database today have been identified based on a single peptide and needs to be revised. It is demonstrated that the methodological improvements in the preparation of urinary exosomes facilitate in-solution digestion, iTRAQ labeling, and MudPIT-based quantitative proteomics (Figure 6). In a preliminary study, the effects of age and sample pooling on protein expression have been studied (Figure 7). We found significant difference in six exosomal protein expressions when comparing two different age groups of healthy individuals (Table 2). On the other hand, preliminary data suggest that the sample pooling possibly compromises the results of the analysis. Second morning urine samples (up to 200ml) were collected from healthy male volunteers of two age groups: 25–50 and 50–70 years. Protease inhibitors, sodium azide, phenylmethylsulfonyl fluoride, and leupeptin (AppliChem, Darmstadt, Germany) were prepared and added to the samples according to Pisitkun et al.10.Pisitkun T. Shen R.-F. Knepper M.A. Identification and proteomic profiling of exosomes in human urine.Proc Natl Acad Sci USA. 2004; 101: 13368-13373Crossref PubMed Scopus (1630) Google Scholar and Zhou et al.17.Zhou H. Yuen P.S.T. Pisitkun T. et al.Collection, storage, preservation, and normalization of human urinary exosomes for biomarker discovery.Kidney Int. 2006; 69: 1471-1476Abstract Full Text Full Text PDF PubMed Scopus (443) Google Scholar Samples were stored at -80°C until processed. Urine samples were thawed, extensively vortexed, and subjected to iterative sequential centrifugations at 400, 800, and 15,000g velocity for 20min at each step at 4°C to remove whole cells, large membrane fragments, and other debris. The supernatant was centrifuged at 200,000g for 1h at 4°C using a Beckman 70 Ti rotor (Beckman Coulter, Fullerton, CA), and urinary vesicle containing pellet (crude exosome) was recovered. Crude exosome pellet was resuspended in 28ml phosphate-buffered saline (PBS; 150mmol/l NaCl at pH 7.2), under-layered by a single cushion composed of 1mol/l sucrose prepared in PBS/D2O, and centrifuged at 110,000g for 3h at 4°C using a Beckman SW 32 Ti rotor. The vesicles captured within the sucrose layer were collected, washed twice in PBS buffer, and centrifuged at 110,000g for 90min at 4°C using a SW 32 Ti rotor. The pellet was resuspended in 100–150μl PBS buffer. The crude exosome pellet was resuspended in 28ml of 20mmol/l Tris, pH 8.6, solubilization buffer and centrifuged at 15,000g for 20min at 4°C. The supernatant was subjected to cushion ultracentrifugation using a double layer. For this step, exosome sample was under-layered without disturbing the interface in the ultracentrifuge tube by 5.2ml 1mol/l sucrose and 3.5ml 2mol/l sucrose prepared in 20mmol/l Tris pH 8.6/D2O and centrifuged at 110,000g for 3h at 4°C using a SW 32 Ti rotor. The vesicles captured in the 1- and 2-mol/l cushions were collected, washed twice in solubilization buffer, and centrifuged at 110,000g for 90min, at 4°C using a SW 32 Ti rotor. Pellets were resuspended in 100–150μl solubilization buffer. Protein concentrations were determined by the micro-bicinchoninic acid protein assay (Pierce/Thermo Scientific, Rockford, IL). Samples were stored at -80°C until use. Urinary vesicles (2–3μg) were mixed at a ratio of 1:1 with 4% paraformaldehyde and applied onto glow-discharged carbon-coated 200-mesh copper grids. The adsorbed exosomes were stained with freshly prepared 2.0% aqueous uranyl acetate and embedded in a mixture of uranyl acetate (0.4%) and methyl cellulose (0.13%). The samples were examined using a JEM-1011 transmission electron microscope (JEOL, Tokyo, Japan) at 100kV. Protein samples (20μg) were electrophoretically separated on a precast Novex 4–12% Bis-Tris NuPAGE gel using MOPS running buffer (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions and stained with colloidal Coomassie blue. Immunoblot (western blot) analysis of proteins was performed on proteins separated by SDS-PAGE. The stained gel was destained in water, soaked in 1% SDS (pH 8.5) for 30min, and washed twice in water and once using the NuPAGE Transfer Buffer (Invitrogen) for 10min. Proteins were transferred onto polyvinylidene difluoride membranes (Invitrogen), blocked according to the manufacturer's instruction, and probed with mouse monoclonal antibodies: tumor susceptibility gene 101 (TSG101), apoptosis-linked gene-2-interacting protein X (ALIX), β-actin (β-ACTIN), Na+/H+ exchanger 3 (NHE3), CD9 (Santa Cruz Biotechnology, Santa Cruz, CA), and rabbit polyclonal antibodies aquaporin 2 (AQP2) and Na–K–Cl cotransporter isoform 2 (NKCC2; gift from G Capasso). The blots were incubated with alkaline phosphatase–conjugated secondary antibodies, and the immune complexes were detected using a Western Light chemi-luminescent detection system (Applied Biosystems, Bedford, MA) according to the manufacturer's protocol. Entire gel lanes were cut into 1.5-mm bands, proteins were reduced, alkylated, in-gel trypsin digested, and extracted as described earlier.18.Shevchenko A. Tomas H. Havlis J. et al.In-gel digestion for mass spectrometric characterization of proteins and proteomes.Nat Protoc. 2007; 1: 2856-2860Crossref Scopus (3545) Google Scholar The peptides were analyzed by nanoflow reversed-phase liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) as described for the iTRAQ experiment, except that the length of the gradient was 17min. Urinary exosome sample was solubilized in 0.8% RapiGest SF (Waters, Milford, MA) and lysed by three freeze–thaw cycles in liquid nitrogen under sonication. Proteins (100μg) were reduced, alkylated, digested with trypsin, and labeled with iTRAQ reagents (Applied Biosystems, Foster City, CA) according to the manufacturer's instruction with the following minor modifications. After the labeling reaction, the four samples were pooled and 10μl of 20% (v/v) trifluoroacetic acid was added to cleave RapiGest. Samples were vortexed, incubated at 37°C for 1h, and centrifuged. The supernatant was vacuum-dried, resolubilized in 5% (v/v) methanol, and peptides were purified by solid-phase extraction using Oasis column (Waters) according to manufacturer's instructions. iTRAQ-labeled peptides were separated by two-dimensional liquid chromatography using SCX chromatography in the first dimension and reversed-phase chromatography in the second dimension. SCX was performed using a Polysulfoethyl-Asp column, 1.0mm ID × 15cm length (LCPackings, Sunnyvale, CA), with the following conditions: 95% solvent A (20% acetonitrile, 0.05% formic acid) and 5% solvent B (20% acetonitrile, 0.05% formic acid, 500mmol/l KCl) for 3min, solvent B ramped up to 90% for 40min and maintained for 7min at 100%. Over a period of 55min with a flow rate of 40μl/min, 47 fractions were collected, dried, and analyzed in the second dimension using a nanoflow LC, Ultimate 3000 (Dionex, Sunnyvale, CA). Samples of volume 20μl were loaded, purified, and concentrated on a reversed-phase monolithic pre-column, 200μm ID × 5mm length (LCPackings), at a flow rate of 25μl/min. Peptides were separated at a flow rate of 300nl/min on a PepSwift Monolithic column, 100μm ID × 5cm length (LCPackings), using the following gradient: (solvent C: 2% acetonitrile, 0.1% formic acid; solvent D: 98% acetonitrile, 0.1% formic acid) 5–50% D for 90min, 50–98% D in 6s and 98% D for 10min. Eluted peptides were analyzed in information-dependent acquisition mode using QSTAR Elite (Applied Biosystems) equipped with a nanoflow electrospray ion source. The Analyst QS 2.0. software (Applied Biosystems, Foster City, CA) was used with default parameters to generate and analyze peak lists extracted from information-dependent acquisition mass spectra. Mascot v.2.2 (Matrix Science, London, UK) was used to search data against SwissProt 2010_09 database (519348 sequences) using trypsin with one possible missed cleavage. Proteins identified by in-gel digestion proteomics, carbamidomethylation of cysteine, and oxidation of methionine were considered as fixed and variable modifications, respectively. An analysis of the false-positive rate of the protein identifications was performed by searching all tandem mass spectra from the nano-HPLC-ESI-MS/MS analyses against an in-house curated decoy SwissProt human protein database containing forward and reverse sequences. In addition, contaminants such as human keratins, porcine trypsin and so on were included in this database. The false-positive rate analysis resulted in the identification of 4397 unique peptides from the target database as compared with 71 peptides from the decoy database. On the basis of this analysis, we estimate the percentage of false-positives to be 1.61% for the present peptide data set. For quantitative analysis, iTRAQ modification at lysine residue and at the N termini of the peptide and carbamidomethylation of cysteines were set as fixed modifications. Oxidation of methionine and iTRAQ modification at tyrosine residues were set as variable modifications. Mass tolerance was set to 50p.p.m. for precursor and to 0.1Da for fragment ions. Mascot iTRAQ four-plex quantification method was used for peptide and protein quantification. The protein ratio was calculated as a weighted median ratio. Criteria for protein identification and quantification were as follows: a minimum of two unique peptides and P<0.05 significance threshold using MudPIT scoring. Open-source STRAP13.Bhatia V.N. Perlman D.H. Costello C.E. et al.Software tool for researching annotations of proteins: open-source protein annotation software with data visualization.Anal Chem. 2009; 81: 9819-9823Crossref PubMed Scopus (191) Google Scholar was used to obtain gene ontology (GO) terms associated with the urinary exosomal protein identified by proteomics. The study was supported by 'Ricercando 2011' from the Italian Society of Nephrology. The authors are grateful to Rosarita Tatè and Michele Cermola for the transmission electron microscopy analysis. Table S1. Urinary exosomal proteins identified in this work. Table S2. List of proteins identified by in-gel and in-solution digestion based proteomics in the 2M sucrose cushion of urinary vesicles. Supplementary material is linked to the online version of the paper at http://www.nature.com/ki