Article

Proteomic Analysis of Urinary Microvesicles and Exosomes in Medullary Sponge Disease and Autosomal Dominant Polycystic

Maurizio Bruschi,1 Simona Granata,2 Laura Santucci,1 Giovanni Candiano,1 Antonia Fabris,2 Nadia Antonucci,2 Andrea Petretto,3 Martina Bartolucci,3 Genny Del Zotto ,4 Francesca Antonini,4 Gian Marco Ghiggeri ,5 Antonio Lupo,2 Giovanni Gambaro,6 and Gianluigi Zaza 2 1Division of , Dialysis, Abstract and Transplantation, Background and objectives Microvesicles and exosomes are involved in the pathogenesis of autosomal dominant Laboratory of polycystic kidney disease. However, it is unclear whether they also contribute to , a Molecular sporadic kidney malformation featuring cysts, , and recurrent kidney stones. We addressed this Nephrology, 3Laboratory of Mass knowledge gap by comparative proteomic analysis. Spectrometry—Core Facilities, Design, setting, participants, & measurements The protein content of microvesicles and exosomes isolated from 4Department of the urine of 15 patients with medullary sponge kidney and 15 patients with autosomal dominant polycystic kidney Research and Diagnostics, and disease was determined by mass spectrometryfollowedby weightedgenecoexpression network analysis,support 5 fi Division of vector machine learning, and partial least squares discriminant analysis to compare the pro les and select the Nephrology, Dialysis most discriminative proteins. The proteomic data were verified by ELISA. and Transplantation, Istituto di Ricovero e Results A total of 2950 proteins were isolated from microvesicles and exosomes, including 1579 (54%) identified in Cura a Carattere fi Scientifico, Istituto all samples but only 178 (6%) and 88 (3%) speci c for medullary sponge kidney microvesicles and exosomes, and Giannina Gaslini, 183 (6%) and 98 (3%) specific for autosomal dominant polycystic kidney disease microvesicles and exosomes, Genoa, Italy; 2Renal respectively. The weighted gene coexpression network analysis revealed ten modules comprising proteins with Unit, Department of similar expression profiles. Support vector machine learning and partial least squares discriminant analysis Medicine, University identified 34 proteins that were highly discriminative between the diseases. Among these, CD133 was upregulated HospitalofVerona, Verona, Italy; and in exosomes from autosomal dominant polycystic kidney disease and validated by ELISA. 6Division of Nephrology and Conclusions Our data indicate a different proteomic profile of urinary microvesicles and exosomes in patients Dialysis, School of with medullary sponge kidney compared with patients with autosomal dominant polycystic kidney disease. The Medicine, Columbus- fi Gemelli University urine proteomic pro le of patients with autosomal dominant polycystic kidney disease was enriched of proteins Hospital Catholic involved in cell proliferation and matrix remodeling. Instead, proteins identified in patients with medullary University, Rome, Italy sponge kidney were associated with parenchymal calcium deposition/nephrolithiasis and systemic metabolic derangements associated with stones formation and bone mineralization defects. Correspondence: Prof. CJASN 14: 834–843, 2019. doi: https://doi.org/10.2215/CJN.12191018 Gianluigi Zaza, Renal Unit, Department of Medicine, University HospitalofVerona, Introduction The hypothesis that extracellular vesicles are present Piazzale A Stefani 1, fi 37126 Verona, Italy. Extracellular vesicles, such as microvesicles (diameter in human urine (8) was con rmed by the proteomic Email: gianluigi. of 100–1000 nm) and exosomes (diameter of 30–100 identification of membrane proteins in a pellet isolated [email protected] nm), are membrane-enclosed particles released by by the ultracentrifugation of urine samples (9). Such most cells under normal and pathologic conditions urinary extracellular vesicles contain cell-specific (1–5). Microvesicles are shed directly from the plasma marker proteins from every segment of the nephron membrane, whereas exosomes are formed by the (9,10), and they offer a source of potentially valuable fusion of intracellular multivesicular bodies (also urinary biomarkers (10). The intrinsic characteristics of known as late endosomes) with the plasma membrane, extracellular vesicles also suggest that they may play leading to the release of their vesicular contents into an important role in kidney development and kidney the extracellular space. These vesicles can mobilize a disease. Accordingly, extracellular vesicles seem to be large number of biologic factors, including receptors, involved in the mechanism of cystogenesis in autoso- other proteins, nucleic acids, and lipids, thus shuttling mal polycystic kidney disease, a common hereditary information to other cells (6). The transfer of RNA and kidney disorder with a prevalence of 0.1%–0.25%. miRNA can reprogram recipient cells and modify their Autosomal polycystic kidney disease gives rise to phenotype (7). predominantly kidney symptoms, including cysts that

834 Copyright © 2019 by the American Society of Nephrology www.cjasn.org Vol 14 June, 2019 CJASN 14: 834–843, June, 2019 Urinary Proteome Analysis Differentiated MSK versus ADPKD, Bruschi et al. 835

progressively disrupt the kidney parenchyma, leading to autosomal dominant polycystic kidney disease was de- interstitial fibrosis, cellular infiltration, and the loss of pendent on the revised Ravine criteria (20). The study functional nephrons. was carried out in accordance with the Declaration of The proteomic analysis of urinary exosome-like vesicles Helsinki and approved by the institutional ethical board (particularly those containing polycystin) revealed ap- of the University Hospital of Verona (Verona, Italy; proximately 500 autosomal dominant polycystic kidney code 1312CESC) and the Independent Ethics Committee disease–associated proteins, many with signaling func- (Comitato Etico Regione Liguria) on October 14, 2014 tions (11). Furthermore, the quantitative proteomic analysis (study number 408REG2014). of urinary extracellular vesicles from patients affected by a complete spectrum of chronic kidney functional damage Isolation of Microvesicles and Exosomes highlighted 30 proteins strongly associated with the autoso- Second morning urine samples were obtained from mal dominant polycystic kidney disease phenotype, includ- patients and healthy donors. Extracellular vesicles were ing periplakin, envoplakin, villin-1, and complement C3 (12). isolated by centrifugation. Briefly, aliquots of 16 ml were In contrast to the wealth of information available for centrifuged at 16,0003g for 30 minutes at 16°C to remove autosomal dominant polycystic kidney disease, little is cells, debris, and organelles, such as mitochondria. To known about the role of extracellular vesicles in the onset obtain the microvesicle fraction, the supernatant was of medullary sponge kidney, a sporadic cystic kidney centrifuged at 22,0003g for 120 minutes at 16°C (21). The malformation that involves nephrocalcinosis and recurrent microvesicle pellet was rinsed in PBS and centrifuged again kidney stones (13). The detailed analysis of extracellular at 22,0003g; this rinse/centrifugation cycle was carried out vesicles could provide insight into the pathogenesis of this five times in total to obtain a clean microvesicle fraction. The rare disease. Despite sporadic genetic associations (14,15) 3g – supernatant was then centrifuged at 100,000 for 120 and the dysregulation of a few biologic factors (16 18), the minutes at 16°C to pellet the exosomes. The pellet was systemic and kidney biologic/cellular network underlying resuspended in 1 ml 0.25 M sucrose, loaded on a 1-ml this disease is poorly characterized, and its relationship with 30% sucrose cushion, and centrifuged at 100,0003g for other cystic diseases is unclear. 120 minutes at 16°C. The pellet was rinsed in PBS and To address this knowledge gap, we carried out a com- centrifuged again at 100,0003g for 10 minutes at 4°C, and prehensive comparative proteomic analysis of urinary this rinse/centrifugation cycle was carried out five times microvesicles and exosomes to identify differences between in total to obtain a clean exosome fraction. For each assay, medullary sponge kidney and autosomal dominant poly- we have performed the same purification procedure. Each in terms of the mechanism of cysto- pellet fraction was stored at 280°C until use. The size and genesis and identify putative diagnostic biomarkers that purity of microvesicles and exosomes isolated by ultracen- distinguish these diseases. In fact, at the moment, no trifugation were confirmed by dynamic light scattering, diagnostic biomarkers are available for both diseases. whereas the antigen profile of exosomes and microvesicles Although some urinary biomarkers for autosomal dominant was performed by Western blot as described in Supple- polycystic kidney disease (NGAL, M-CSF, and MCP-1) (19) mental Material. have been proposed, none of them have been used in clinical practice (19). Additionally, most of them are only effective in the advanced stage of the disease. Identification of Mass Spectrometry both diseases at early stages could help clinicians start The samples were processed by the in-StageTip method prevention, diet adjustment, and for selected patients, with two poly(styrene divinylbenzene) reverse phase m pharmacologic treatment. Finally, they could potentiate sulfonate disks (22). Each pellet was solubilized in 25 l diagnostic accuracy for medullary sponge kidney (this 2% sodium deoxycholate, 10 mM Tris(2-carboxyethyl) disease is often undiagnosed and confused with other cause phosphine, 40 mM chloroacetamide, and 100 mM Tris of nephrocalcinosis or papillary ductal plugging), minimize (pH 8.5). Microvesicles or exosomes were lysed, reduced, patients’ radiation and/or nephrotoxic contrast media expo- and alkylated in a single step, and then, they were loaded sure from medical imaging (e.g., intravenous urography and into the StageTip. The lysates were diluted with 25 mM Tris m CT urography), and reduce underdiagnosis of noncontrast (pH 8.5) containing 1 g of trypsin. The samples were fi m fl CT scans. acidi ed with 100 l 1% (vol/vol) tri uoroacetic acid and washed three times with 0.2% (vol/vol) trifluoroacetic acid. The proteins were eluted in 60 ml 5% (vol/vol) ammonium Materials and Methods hydroxide containing 80% (vol/vol) acetonitrile. Detailed Patients descriptions of mass spectrometry instrumentation, data The study included 15 adult patients with autosomal analysis, and biologic validation with homemade ELISA are dominant polycystic kidney disease and 15 adult patients reported in Supplemental Material. with medullary sponge kidney matched for age, sex, and geographical origin as well as a cohort of 17 healthy donors Statistical Analyses matched for age and sex (Table 1, Supplemental Figure 1). After normalization using the Normalyzer R-package The patients were followed up by the Renal Unit at the with the LOESS-G method (23), mass spectrometry data Department of Medicine, University Hospital of Verona were analyzed by unsupervised hierarchical clustering (Verona, Italy), and they were enrolled after providing using multidimensional scaling with k means and Spear- informed consent. Medullary sponge kidney diagnosis was man correlation to identify outliers and the dissimilar- performed as previously reported (15). The diagnosis of ity between samples. The normalized expression profiles 836 CJASN

Table 1. Baseline characteristics of the study participants

Medullary Sponge Autosomal Dominant Polycystic Variable Healthy Controlsa Kidney Kidney Disease

Age, yr 266426652768 Sex (men/women) 6/9 7/8 8/9 eGFR, ml/min per 1.73 m2 132615 133612 13968 Plasma calcium, mg/dl 9.560.3 9.460.4 9.460.3 Plasma phosphate, mg/dl 3.160.5 2.960.5 2.860.5 Plasma sodium, mmol/L 140621396213864 Plasma potassium, mmol/L 3.860.6 3.960.2 3.960.1 Proteinuria, g/24 h 0.0860.06 0.0760.07 0.0460.09 Urine volume, ml/d 17866212 17506581 17426526 Systolic BP, mm Hg 119641186711765 Diastolic BP, mm Hg 746576667564

Values are expressed as mean6SD. P values were determined by ANOVA except for sex, which was determined by Fisher exact test. aIncluded only in the flow cytometry analysis.

of the proteins were then used to construct the coexpression Furthermore, about 40% of the proteins found in extra- network using the weighted gene coexpression network cellular vesicles were associated with one or both kidney analysis package in R (24). Additionally, to identify the hub diseases: 95% were found in the medullary sponge kidney proteins of modules that maximize the discrimination be- samples, and 100% were found in the autosomal dominant tween the selected clinical traits, we applied a nonparametric polycystic kidney disease samples (Figure 1, B and C). The Mann–Whitney U test, machine learning methods (such as cellular origins of the proteins in the exosomes were very nonlinear support vector machine learning), and partial similar in the medullary sponge kidney and autosomal least squares discriminant analysis. A complete and de- dominant polycystic kidney disease samples, with 18% of tailed description of the data analysis has been reported in proteins originating from membranes, 32% originating from Supplemental Material. the cytoplasm, 10% originating from the nucleus, and 39% originating from other organelles (Supplemental Figure 3). Similar results were observed for the microvesicle proteins, Results with 34% originating from membranes, 26% originating Characterization of Exosomes and Microvesicles from the cytoplasm, 8% originating from the nucleus, and The size and purity of microvesicles and exosomes 32% originating from other organelles. fi isolated by ultracentrifugation were con rmed by dynamic The significant overlap among the groups of proteins fi light scattering, revealing a Gaussian distribution pro le foundineachsamplewasconfirmed by constructing a 6 6 with peak means at 1000 65 or 90 5 nm, respectively, the two-dimensional scatter plot of the multidimensional typical sizes for microvesicles or exosomes, respectively scaling analysis (Supplemental Figure 4). No samples were (Supplemental Figure 2, A and B). There was no difference excluded during the quality check performed by nonhierar- in size between the microvesicles and exosomes isolated chical clustering (Supplemental Figure 5). We used weighted from patients with medullary sponge kidney and patients gene coexpression network analysis to identify proteins with autosomal dominant polycystic kidney disease. associated with each type of extracellular vesicle and disease, Western blot analysis revealed that the exosomes were revealing a total of ten modules comprising proteins with positive for CD63 and CD81 but not CD45, whereas the similar expression profiles. To distinguish between mod- fi microvesicles showed the opposite antigen pro le (Sup- ules, we chose an arbitrary color for each module (Figure plemental Figure 2C). 2A). The number of proteins included in each module ranged from 44 (gray) to 930 (turquoise). The gray, brown, Protein Composition of Exosomes and Microvesicles pink, and blue modules showed closer relationships with The protein composition of exosomes and microvesicles the medullary sponge kidney, autosomal dominant poly- from the urine of patients with medullary sponge kidney cystic kidney disease, microvesicle, and exosome groups, and patients with autosomal dominant polycystic kidney respectively (Figure 2B). disease was determined by mass spectrometry. We iden- Next, we applied the Mann–Whitney U test to identify tified 2950 proteins in total, 1579 (54%) of which were the proteins that best distinguish the type of disease in the present in all four sample types. Among the medullary microvesicles or exosomes (Figure 3, A and B) and the type sponge kidney samples, only 178 (6%) and 88 (3%) proteins of extracellular vesicle in the medullary sponge kidney or were exclusively found in the exosomes and microvesicles, autosomal dominant polycystic kidney disease samples respectively. Similarly, among the autosomal dominant (Figure 3, C and D). This revealed a total of 255 discrim- polycystic kidney disease samples, only 183 (6%) and 98 inatory proteins, 50 that distinguished between medullary (3%) proteins were exclusively found in the exosomes and sponge kidney and autosomal dominant polycystic kidney microvesicles, respectively (Figure 1A); .60% of all of the disease microvesicles, 90 that distinguished between med- extracellular vesicle proteins that we identified were present ullary sponge kidney and autosomal dominant polycystic in exosomes, and .80% were present in microvesicles. kidney disease exosomes, 150 that distinguished between CJASN 14: 834–843, June, 2019 Urinary Proteome Analysis Differentiated MSK versus ADPKD, Bruschi et al. 837

A ADPKDEx MSKMv (2375) (2020)

183 88 (6.2%) (3%) 32 145 (1.1%) 48 (4.9%) (1.6%) 178 88 74 98 (6%) (3%) (2%) (3.3%)

1579 59 (53.5%) 82 (2%) (2.8%) 52 189 (1.8%) (6.4%) 55 (1.9%) MSKEx ADPKDMv (2345) (2177)

B C Exocarta Exocarta (1649) Associated to (1649) Associated to Kidney diseases Kidney diseases Associated to (1172) (1172) MSKMv Associated to ADPKDMv MSK (22) (2020) PKD (112) (2177)

153 134 3 Vesiclepedis 32 Vesiclepedis (1948) 2 (1948) 1 29 21 27 19 21 9 35 16 40 36 11 2 10 1 36 16 1 2 1 2 2 17 5 100 35 79 3 822 88 24 23 50 795 105 33 71 518 89 45 546 110

2 1 4 8 3 5 9

336 6 7 225 289 87 244

MSKEx ADPKDEx (2345) (2372)

Figure 1. | Venn diagram of total proteins detected in exosomes and microvesicles from the urine of patients with medullary sponge kidney (MSK) and patientswith autosomal dominantpolycystickidney disease (ADPKD)identified by massspectrometry.(A) TheVenndiagramshows common and exclusive proteins in MSK and ADPKD. The numbers represent the distinct proteins in the overlapping and nonoverlapping areas. (B and C) The numbers represent the distinct proteins in the overlapping and nonoverlapping areas. The data were extracted from the Exocarta, Vesiclepedia, UniProt, Open Target, DisGeNET, and Atlas databases. The majority of the proteins identified in extracellular vesicles correspond to proteins already described as components of exosomes or microvesicles or associated with kidney disease (about 40%). We found that 95% and 100% of the proteins were associated with MSK and ADPKD, respectively. PKD, polycystic kidney disease. exosomes and microvesicles in the autosomal dominant corresponding expression profiles (Figure 4A) and polycystic kidney disease samples, and 62 that distin- prepared a graphical representation for their cluster guished between exosomes and microvesicles in the med- separation (Figure 4B). ullary sponge kidney samples (Supplemental Table 1, The diversity of expression profiles among the proteins Supplemental Figures 6 and 7). Support vector machine in this core panel indicated their association with different learning and partial least squares discriminant analysis functions, and therefore, GO analysis of functional anno- were then used to highlight the proteins that maximize the tations was used to build a scatter plot of enriched gene – discrimination between different sample types, revealing signatures on the y axis and log10 P values on the x axis a core panel of 34 proteins that allowed us to distinguish (Supplemental Figure 8). The size of scatters is proportional to the four conditions with an accuracy of 100% (Figure 4, A the number of proteins associated with each biologic pro- and B). After Z-score analysis, we built a heat map of the cess. After Z-score analysis, we built a heat map showing the 838 CJASN

A B 1.0 MEyellow -0.067 0.067 -0.081 0.081 (0.6) (0.6) (0.5) (0.5) 1

MEblue -0.51 0.51 -0.5 0.5 0.8 (4e-05) (4e-05) (4e-05) (4e-05)

MEblack -0.18 0.18 0.3 -0.3 (0.2) (0.2) (0.02) (0.02) 0.5 0.6 MEgreen -0.19 0.19 0.02 -0.02 (0.1) (0.1) (0.9) (0.9) Height -0.29 0.29 0.16 -0.16 MEturquoise 0.4 (0.03) (0.03) (0.2) (0.2) 0 -0.15 0.15 0.8 -0.8 MEpink (0.2) (0.2) (1e-14) (1e-14) 0.2 -0.42 0.46 0.55 -0.55 MEred (7e-04) (2e-04) (6e-06) (6e-06)

-0.55 0.55 0.18 -0.18 -0.5 MEbrown (6e-06) (6e-06) (0.2) (0.2) Dynamic Tree Cut: -0.51 0.51 -0.43 0.43 MEmagenta (4e-05) (4e-05) (6e-04) (6e-04)

Merged 0.55 -0.55 -0.29 0.29 MEgray -1 Dynamic: (6e-06) (6e-06) (0.02) (0.02)

MSK ADPKD Mv Ex

Figure 2. | Module identification and clinical trait relationship. (A) Dendrogram of all proteins identified in the extracellular vesicles of patients with medullary sponge kidney (MSK) and patients with autosomal dominant polycystic kidney disease (ADPKD) clustered on the basis of a dissimilarity measure with topological overlap matrix (TOM) (1-TOM). (B) Heat map of the relationships between module eigengenes and the trait indicator of samples. Module-trait weighted relationships and their P values (in parentheses) between the identified modules and trait indicator. The color scale on the right shows module-trait relationship from 21 (blue) to one (red), where blue represents a perfect negative correlation and red represents a perfect positive correlation. Mv, microvesicles; Ex, exosomes; ME, module eigengenes. expression profiles of the enriched biochemical pathways ROC analysis were 0.98 (95% CI, 0.94 to 1) and P,0.001 (Figure 4C). Interestingly, this revealed that proteins in- (patients with autosomal dominant polycystic kidney dis- volved in cell migration/adhesion were over-represented in ease versus healthy controls), 0.82 (95% CI, 0.67 to 0.97) and the microvesicles of patients with polycystic kidney disease, P=0.003 (patients with autosomal dominant polycystic whereas those involved in the regulation of the epithelial kidney disease versus patients with medullary sponge cell differentiation were over-represented in the exosomes kidney), and 0.70 (95% CI, 0.51 to 0.89) and P=0.05 (patients of patients with autosomal dominant polycystic kidney with medullary sponge kidney versus healthy controls) disease. (Figure 4E). The cutoff, sensitivity, specificity, and likelihood ratio are reported in Supplemental Table 2. ELISA for CD133 in Exosomes-Validated Proteomics A homemade ELISA for urinary CD133 was performed in exosomes from all patients and healthy controls to Discussion validate proteomic data. We found that CD133 was highly Microvesicles and exosomes are known to be involved in expressed in patients with autosomal dominant polycystic the pathogenesis of several chronic kidney disorders, but kidney disease compared with patients with medullary few studies have focused on their role in kidney cystic sponge kidney and healthy controls (Figure 4D). The diseases (9,11,25,26), and their potential involvement in medians (interquartile ranges [IQRs]) were 1.04 (IQR, medullary sponge kidney disease has not been addressed. 0.54–1.68), 0.4 (IQR, 0.22–0.76), and 0.28 (IQR, 0.16–0.34) In this study, we used mass spectrometry to identify the for patients with autosomal dominant polycystic kidney protein content of microvesicles and exosomes to gain disease, patients with medullary sponge kidney, and insight into medullary sponge kidney–related cystogenesis healthy controls, respectively, and P values were P,0.001 and its similarities and differences compared with autoso- for Kruskal–Wallis test analysis. Also, ROC analysis re- mal dominant polycystic kidney disease. By applying a vealed that the expression of CD133 in urinary exosomes layered statistical analysis approach, we found 34 core can discriminate patients with autosomal dominant poly- proteins that distinguished the microvesicles and exo- cystic kidney disease from healthy subjects and patients somes of medullary sponge kidney and autosomal dom- with medullary sponge kidney. The areas under the curve, inant polycystic kidney disease. Interestingly, most of the 95% confidence intervals (95% CIs), and P values of these proteins were assigned to a small number of specific CJASN 14: 834–843, June, 2019 Urinary Proteome Analysis Differentiated MSK versus ADPKD, Bruschi et al. 839

A B MSKMv ADPKDMv MSKEx ADPKDEx 16 10

14 8 12

10 6 P-value 8 P-value 10 10 6 4 -Log -Log 4 2 2

0 0 -9 -6 -3 0369 -9-6-30369

Log2 Fold Change Log2 Fold Change

C D ADPKDMv ADPKDEx MSKMv MSKEx 18 14 16 12 14 10 12 10 8 P-value P-value 10 8 10 6 -Log 6 -Log 4 4 2 2 0 0 -9 -6 -3 0 3 6 9 -9 -6 -3 0369

Log2 Fold Change Log2 Fold Change

Figure 3. | Volcano plots of univariate statistical analysis as applied to urinary extracellular vesicle samples. The plots are on the basis of the 2 fold change (log2) and the P-value ( log10) of all proteins identified in (A) Mv from MSK and ADPKD; (B) Ex from MSK and ADPKD; (C) Mv and Ex from ADPKD; (D) Mv and Ex from MSK. Red circles indicate proteins related to the selected clinical trait with statistically significant changes between the clinical traits selected in this study. ADPKD, autosomal dominant polycystic kidney disease; MSK, medullary sponge kidney; Mv, microvesicles; Ex, exosomes. functions, including the regulation of epithelial cell dif- confirm this hypothesis. Accordingly, the kidney pro- ferentiation, kidney development, cell migration, cell genitor cells in human kidney papillary loops of Henle can adhesion, carbohydrate metabolism, and extracellular differentiate into both neural-like and epithelial-like line- matrix organization. ages as well as producing tubules (30). An abundant One of the core proteins was prominin 1 (CD133), a population of CD133+ cells was also shown to be present pentaspan transmembrane glycoprotein that localizes to in the cystic wall and kidney tubules of patients with membrane protrusions and is often expressed on adult autosomal dominant polycystic kidney disease (31). The stem/progenitor kidney cells, where it is thought to role of these cells is not yet clear, but it would be maintain stem cell properties by suppressing differenti- interesting to evaluate more patients with autosomal ation. The high-level expression of prominin 1 is associ- dominant polycystic kidney disease at different disease ated with several types of cancer (27–29). This protein stages (from asymptomatic to the late disease stage) and was more abundant in the exosomes of patients with clarify whether CD133+ (and CD24+) cells are associated autosomal dominant polycystic kidney disease, reflecting with a better or worse prognosis. the attempted tissue repair in response to the aberrant rate of We also found that the cellular repressor of E1A stimulated proliferation and apoptosis, which would require kidney genes 1 (CREG1), a factor that interacts with the IGF2 progenitor cells. The upregulation of other proteins in- receptor to regulate cell growth, was more abundant in volved in cell migration/adhesion, such as Cadherin 4, or autosomal dominant polycystic kidney disease. This protein the epithelial cell differentiation, such as CREG1, seems to may facilitate stem cell differentiation and activity, which 840 CJASN

A C MSK ADPKD ADPKD MSK ADPKD ADPKD MSK Mv – + MSK

–+ Ex Ex Mv Mv Mv Ex Ex

CARMIL3 ZSCAN32 Regulation of cell adhesion SPP1 OLR1 EEF1G MATN2 Regulation of cell migration SEMG2 PRG2 GUCA2B DPT Cell adhesion ENOSF1 CLSTN3 DAG1 Regulation of epithelial VPS4A cell differentiation involved PROM1 CREG1 in kidney development ANKFY1 FLRT3 Extracellular matrix B3GNT8 organization DNAJB6 COL14A1 VWA7 Carbohydrate metabolism ANKRD18B NAAA CPAMD8 FAT4 MAN2A2 ZFHX3 LRRC40 MAL ITIH5 PAM TSPAN9 hPEPT1-RF

B D E 20 2.5 100 MSKEx ADPKD 15 Ex MSKMv 2.0 80 ADPKD 10 Mv

5 1.5 60

0 1.0 40 -5 Sensitivity (%) CTR vs MSK Component 2 (19.5%)

Optical Density (RU/ml) AUC=0.70 P=0.047 0.5 -10 CTR vs ADPKD 20 AUC=0.98 P<0.0001 -15 0.0 MSK vs ADPKD AUC=0.82 P=0.003 -20 0 -20 -15 -10 -5 0 5 10 15 20 25 CTR MSK ADPKD 0 20 40 60 80 100 Component 1 (27%) 100-Specificity (%)

Figure 4. | Proteins and gene ontology annotation that maximize the discrimination among all conditions. (A) Heat map of 34 core proteins identified through the combined use of univariate statistical analysis, support vector machine learning, and partial least squares discriminant analysis. In the heat map, each row represents a protein, and eachcolumn corresponds to a condition. Normalized Z scores of protein abundance are depicted by a pseudocolor scale, with red indicating positive expression, white indicating equal expression, and blue indicating negative expression compared with each protein value, whereas the dendrogram displays the outcome of unsupervised hierarchical clustering analysis, placing similar proteome profile values near to each other. (B) Two-dimensional scatter plot of multidimensional scaling analysis of exosome (solid symbols) and microvesicle (open symbols) of medullary sponge kidney (MSK; red triangles) and autosomal dominant polycystic kidney disease (ADPKD; black squares) samples using the above 34 highlighted proteins. Ellipses indicates 95% confidence intervals. Visual inspection of the dendrogram and heat map shows the ability of these proteins to clearly distinguish between the different conditions. (C) The heat map shows biologic process enrichment for different extracellular vesicle samples. In the heat map, each row represents a protein, and each column Cont. CJASN 14: 834–843, June, 2019 Urinary Proteome Analysis Differentiated MSK versus ADPKD, Bruschi et al. 841

was recently shown for the differentiation of embryonic example was SPP1 (osteopontin), a protein implicated in stem cells in cardiomyocytes, improving the integration of nephrolithiasis, a major clinical condition associated with stem cell–derived cardiomyocytes into recipient hearts (32). medullary sponge kidney (13). Osteopontin is intimately The exosomes sourced from our patients with autosomal involved in the regulation of both physiologic and path- dominant polycystic kidney disease not only contained ologic mineralization. In normal bone tissue, osteopontin higher levels of the proliferation regulator CREG1 but also, is expressed by osteoclasts and osteoblasts during bone proteins required for matrix remodeling (ITIH5) and the remodeling, and osteoclast-derived osteopontin inhibits regulation of salt secretion (GUCA2B or MAL). All of these the formation of hydroxyapatite during normal minerali- mechanisms are important for cyst formation and enlarge- zation (40). Osteopontin is also involved in kidney stone ment, which in autosomal dominant polycystic kidney formation (41). This protein is synthesized in the kidney disease, involve tubular cell proliferation, abnormalities in and secreted into the urine by epithelial cells, including the the extracellular matrix, and transepithelial fluid secretion loop of Henle, distal convoluted tubule, and papillary directed toward the cyst lumen. Because cysts are anatom- epithelium (42), inhibiting the nucleation, growth, and ically separated from their source tubule (33), the intracystic aggregation of calcium oxalate crystals (43) and the binding fluid does not originate from the glomerular filtrate, but of calcium oxalate crystals to kidney epithelial cells (44). rather, it originates from transepithelial fluid secretion (34). Osteopontin knockout mice are hyperoxaluric, leading to Autosomal dominant polycystic kidney disease is also the significant intratubular deposition of calcium oxalate, characterized by the disruption of the planar cell polarity whereas wild-type mice remove calcium oxalate effectively pathway, which is required for oriented cell division and (45). Therefore, the greater abundance of osteopontin in convergent extension to establish and maintain the struc- the microvesicles of our patients with medullary sponge ture of kidney tubules (35). We found that the FAT Atypical kidney could represent a defense mechanism against Cadherin 4 protein was more abundant in the exosomes of microcalcification, and it could, at least partially, explain patients with autosomal dominant polycystic kidney dis- the bone symptoms often observed in patients with this ease. The loss of this protein disrupts oriented cell division disease. Accordingly, 58% of patients with medullary and tubule elongation during kidney development, causing sponge kidney have a dual-energy x-ray absorptiometry tubule dilation (36). profile of osteopenia, and 14% have a profile of osteoporosis Notably, none of the proteins discussed above were unrelated to the common causes of bone demineralization, upregulated in medullary sponge kidney, showing a dif- particularly hyperparathyroidism and menopause (46). ferent mechanism of cystogenesis. The specificdiagnosisof Taken together, our results have shown for the first time medullary sponge kidney requires the anatomic feature of that the urinary microvesicles and exosomes of patients papillary precalyceal ectasias, sometimes associated with with autosomal dominant polycystic kidney disease and tiny medullary cysts, and such alterations can be unilateral patients with medullary sponge kidney have distinct or even limited to a portion of a single kidney medulla. proteomic profiles. The urine of patients with autosomal Unlike autosomal dominant polycystic kidney disease, the dominant polycystic kidney disease was enriched for tubular dilations and microcysts tend to be stable in terms proteins involved in cell proliferation and matrix remodeling, of size throughout life as if they formed at the same time as probably due to pathologic tissue remodeling prompting the kidneys. Taken together, our data confirmed earlier cystic development and enlargement. In contrast, the urine of reports indicating that medullary sponge kidney is an inborn patients with medullary sponge kidney revealed a proteome malformation similar to developmental disorders, such as indicative of a systemic biochemical imbalance that could congenital hemihypertrophy and Beckwith–Wiedemann syn- explain the predisposition of such patients to parenchy- drome, and kidney developmental anomalies, such as horse- mal calcium deposition/nephrolithiasis and extrarenal shoe kidney, unilateral kidney aplasia, and contralateral complications, including bone mineralization defects. congenital small kidney (37,38), with the absence of the Although small sample size and lack of independent sequence of events leading to cyst formation. Additionally, replication are major weaknesses of the study and addi- our data showing the abundance of proteins involved in cell tional research is required for validation, some of the proliferation and extracellular matrix remodeling in patients proteins (mainly CD133) that we identified could be suitable with autosomal dominant polycystic kidney disease could in in the future as diagnostic biomarkers that could help part explain why these patients are predisposed to the clinicians to distinguish between patients with medullary development of cancer, particularly kidney carcinoma (39). sponge kidney and patients with autosomal dominant poly- In contrast, only a few proteins were highly expressed in cystic kidney disease during the early stages of the disease, medullary sponge kidney, mainly in the microvesicles. One avoiding time-consuming and expensive clinical testing.

Figure 4. | Continued. corresponds to a condition. Normalized Z scores of protein abundance are depicted by a pseudocolor scale, with red indicating positive expression, white indicating equal expression, and blue indicating negative expression compared with each protein value, whereas the dendrogram displays the outcome of unsupervised hierarchical clustering analysis, placing similar proteome profile values near to each other. Visual inspection of the dendrogram and heat map shows the ability of these biologic processes to distinguish between the different types of extracellular vesicle in the MSK and ADPKD samples. (D) ELISA for CD133. Box plot showing the median and interquartile range value of the urinary exosome CD133 in all samples. CD133 was highly expressed in patients with ADPKD compared with patients with MSK and healthy controls. (E) ROC curve analysis revealed that the expression of CD133 in urinary exosomes can discriminate patients with ADPKD from healthy controls and patients with MSK. AUC, area under the curve. ROC, Receiver Operating Characteristic; Mv, microvesicles; Ex, exosomes; CTR, healthy controls. 842 CJASN

Acknowledgments 7. Valadi H, Ekstro¨m K, Bossios A, Sjo¨strand M, Lee JJ, Lo¨tvall JO: This study was performed (in part) in the Laboratorio Uni- Exosome-mediated transfer of mRNAs and microRNAs is a novel mechanism of genetic exchange between cells. Nat Cell Biol 9: versitariodiRicercaMedicaResearchCenter,UniversityofVerona. 654–659, 2007 This study was supported by Fondazione Cariverona call 2016 8. Thongboonkerd V, McLeish KR, Arthur JM, Klein JB: Proteomic (Principal Investigator Prof. Tagliaro) and Ministero Della Salute analysis of normal human urinary proteins isolated by acetone grant GR-2011-02350438. precipitation or ultracentrifugation. Kidney Int 62: 1461–1469, 2002 Disclosures 9. Pisitkun T, Shen RF, Knepper MA: Identification and proteomic profiling of exosomes in human urine. Proc Natl Acad Sci U S A Dr. Antonini, Dr. Antonucci, Dr. Bartolucci, Dr. Bruschi, Dr. 101: 13368–13373, 2004 Candiano, Dr. Del Zotto, Dr. Fabris, Dr. Gambaro, Dr. Granata, Dr. 10. Moon PG, You S, Lee JE, Hwang D, Baek MC: Urinary exosomes Ghiggeri, Dr. Lupo, Dr. Petretto, Dr. Santucci, and Dr. Zaza have and proteomics. Mass Spectrom Rev 30: 1185–1202, 2011 nothing to disclose. 11. Hogan MC, Manganelli L, Woollard JR, Masyuk AI, Masyuk TV, Tammachote R, Huang BQ, Leontovich AA, Beito TG, Madden BJ, Supplemental Material Charlesworth MC, Torres VE, LaRusso NF, Harris PC, Ward CJ: This article contains the following supplemental material online at Characterization of PKD protein-positive exosome-like vesicles. http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/ J Am Soc Nephrol 20: 278–288, 2009 CJN.12191018/-/DCSupplemental. 12. Salih M, Demmers JA, Bezstarosti K, Leonhard WN, Losekoot M, vanKooten C, Gansevoort RT,Peters DJ, Zietse R, Hoorn EJ; DIPAK Supplemental Material. Methods. Consortium: Proteomics of urinary vesicles links plakins and Supplemental Figure 1. Age and eGFR of all study participants complement to polycystic kidney disease. J Am Soc Nephrol 27: included in the study. 3079–3092, 2016 SupplementalFigure2. Characterizationof isolated exosomesand 13. Gambaro G, Danza FM, Fabris A: Medullary sponge kidney. Curr Opin Nephrol Hypertens 22: 421–426, 2013 microvesicles. 14. Fabris A, Lupo A, Ferraro PM, Anglani F, Pei Y, Danza FM, Supplemental Figure 3. Gene Ontology annotation of urinary Gambaro G: Familial clustering of medullary sponge kidney is extracellular vesicle proteins. autosomal dominant with reduced penetrance and variable ex- Supplemental Figure 4. Multidimensional scaling analysis of pressivity. Kidney Int 83: 272–277, 2013 extracellular vesicles from the urine of patients with medullary 15. Torregrossa R, Anglani F, Fabris A, Gozzini A, Tanini A, Del Prete D, Cristofaro R, Artifoni L, Abaterusso C, Marchionna N, Lupo A, sponge kidney (MSK) and patients with autosomal dominant D’Angelo A, Gambaro G: Identification of GDNF gene sequence polycystic kidney disease (ADPKD). variations in patients with medullary sponge kidney disease. Clin J Supplemental Figure 5. Sample clustering and trait indicators. Am Soc Nephrol 5: 1205–1210, 2010 Supplemental Figure 6. Venn diagram of statistically significant 16. Fabris A, Bruschi M, Santucci L, Candiano G, Granata S, Dalla Gassa A, Antonucci N, Petretto A, Ghiggeri GM, Gambaro G, differences in protein abundance in the different types of extra- Lupo A, Zaza G: Proteomic-based research strategy identified cellular vesicles from patients with medullary sponge kidney (MSK) laminin subunit alpha 2 as a potential urinary-specific biomarker or patients with autosomal dominant polycystic kidney disease for the medullary sponge kidney disease. Kidney Int 91: 459–468, (ADPKD). 2017 Supplemental Figure 7. Proteins network interaction. 17. Ria P,Fabris A, Dalla Gassa A, Zaza G, Lupo A, Gambaro G: New non-renal congenital disorders associated with medullary sponge Supplemental Figure 8. Gene ontology enrichment analysis for kidney (MSK) support the pathogenic role of GDNF and point to core discriminatory proteins in the extracellular vesicles of patients the diagnosis of MSK in recurrent stone formers. Urolithiasis 45: with medullary sponge kidney (MSK) and patients with autosomal 359–362, 2017 dominant polycystic kidney disease (ADPKD). 18. Fabris A, Ferraro PM, Comellato G, Caletti C, Fantin F, Zaza G, Zamboni M, Lupo A, Gambaro G: The relationship between Supplemental Table 1. List of all significant proteins identified calcium kidney stones, arterial stiffness and bone density: Un- using mass spectrometry. raveling the stone-bone-vessel liaison. J Nephrol 28: 549–555, Supplemental Table 2. ELISA cutoff, sensitivity, specificity, and 2015 likelihood ratio. 19. Kawano H, Muto S, Ohmoto Y, Iwata F, Fujiki H, Mori T, Yan L, Horie S: Exploring urinary biomarkers in autosomal dominant polycystic kidney disease. Clin Exp Nephrol 19: 968–973, 2015 References 20. Pei Y,Obaji J, Dupuis A, Paterson AD, Magistroni R, Dicks E, Parfrey 1. Heijnen HF, Schiel AE, Fijnheer R, Geuze HJ, Sixma JJ: Activated P, Cramer B, Coto E, Torra R, San Millan JL, Gibson R, Breuning M, platelets release two types of membrane vesicles: Microvesicles by PetersD,RavineD:Unified criteriaforultrasonographicdiagnosisof surface shedding and exosomes derived from exocytosis of mul- ADPKD. J Am Soc Nephrol 20: 205–212, 2009 tivesicular bodies and alpha-granules. Blood 94: 3791–3799, 21. Coumans FAW,Brisson AR, Buzas EI, Dignat-George F,Drees EEE, 1999 El-Andaloussi S, Emanueli C, Gasecka A, Hendrix A, Hill AF, 2. Ratajczak J, Wysoczynski M, Hayek F, Janowska-Wieczorek A, Lacroix R, Lee Y,van Leeuwen TG, Mackman N, Ma¨gerI, Nolan JP, Ratajczak MZ: Membrane-derived microvesicles: Important and van der Pol E, Pegtel DM, Sahoo S, Siljander PRM, Sturk G, de underappreciated mediators of cell-to-cell communication. Wever O, Nieuwland R: Methodological guidelines to study Leukemia 20: 1487–1495, 2006 extracellular vesicles. Circ Res 120: 1632–1648, 2017 3. Dear JW, Street JM, Bailey MA: Urinary exosomes: A reservoir for 22. Kulak NA, Pichler G, Paron I, Nagaraj N, Mann M: Minimal, en- biomarker discovery and potential mediators of intrarenal sig- capsulated proteomic-sample processing applied to copy-number nalling. Proteomics 13: 1572–1580, 2013 estimation in eukaryotic cells. Nat Methods 11: 319–324, 2014 4. Salih M, Zietse R, Hoorn EJ: Urinary extracellular vesicles and the 23. Chawade A, Alexandersson E, Levander F: Normalyzer: A tool for kidney: Biomarkers and beyond. Am J Physiol Renal Physiol 306: rapid evaluation of normalization methods for omics data sets. F1251–F1259, 2014 J Proteome Res 13: 3114–3120, 2014 5. van Balkom BW, Pisitkun T, Verhaar MC, Knepper MA: Exosomes 24. Langfelder P, Horvath S: WGCNA: An R package for weighted and the kidney: Prospects for diagnosis and therapy of renal correlation network analysis. BMC Bioinformatics 9: 559, 2008 diseases. Kidney Int 80: 1138–1145, 2011 25. Gonzales PA, Pisitkun T, Hoffert JD, Tchapyjnikov D, Star RA, 6. Mause SF, Weber C: Microparticles: Protagonists of a novel Kleta R, Wang NS, Knepper MA: Large-scale proteomics and communication network for intercellular information exchange. phosphoproteomics of urinary exosomes. J Am Soc Nephrol 20: Circ Res 107: 1047–1057, 2010 363–379, 2009 CJASN 14: 834–843, June, 2019 Urinary Proteome Analysis Differentiated MSK versus ADPKD, Bruschi et al. 843

26. Hogan MC, Bakeberg JL, Gainullin VG, Irazabal MV, Harmon AJ, 37. Lambrianides AL, John DR: Medullary sponge disease in horse- Lieske JC, Charlesworth MC, Johnson KL, Madden BJ, Zenka RM, shoe kidney. Urology 29: 426–427, 1987 McCormick DJ, Sundsbak JL, Heyer CM, Torres VE, Harris PC, 38. Gambaro G, Fabris A, Citron L, Tosetto E, Anglani F, Bellan F, Ward CJ: Identification of biomarkers for PKD1 using urinary Conte M, Bonfante L, Lupo A, D’Angelo A: An unusual associ- exosomes. J Am Soc Nephrol 26: 1661–1670, 2015 ation of contralateral congenital small kidney, reduced renal 27. Weigmann A, Corbeil D, Hellwig A, Huttner WB: Prominin, a function and hyperparathyroidism in sponge kidney patients: On novel microvilli-specific polytopic membrane protein of the the track of the molecular basis. Nephrol Dial Transplant 20: apical surface of epithelial cells, is targeted to plasmalemmal 1042–1047, 2005 protrusions of non-epithelial cells. Proc Natl Acad Sci U S A 94: 39. YuTM,ChuangYW,YuMC,ChenCH,YangCK,HuangST,LinCL, 12425–12430, 1997 ShuKH,KaoCH:Riskofcancerinpatientswithpolycystickidney 28. Corbeil D, Ro¨per K, Hellwig A, Tavian M, Miraglia S, Watt SM, disease: A propensity-score matched analysis of a nationwide, Simmons PJ, Peault B, Buck DW, Huttner WB: The human AC133 population-based cohort study. Lancet Oncol 17: 1419–1425, hematopoietic stem cell antigen is also expressed in epithelial 2016 cells and targeted to plasma membrane protrusions. J Biol Chem 40. Hunter GK, Kyle CL, Goldberg HA: Modulation of crystal for- 275: 5512–5520, 2000 mation by bone phosphoproteins: Structural specificity of the 29. Florek M, Haase M, Marzesco AM, Freund D, Ehninger G, Huttner osteopontin-mediated inhibition of hydroxyapatite formation. WB, Corbeil D: Prominin-1/CD133, a neural and hematopoietic Biochem J 300: 723–728, 1994 stemcellmarker,isexpressedinadulthumandifferentiatedcellsand 41. Kleinman JG, Wesson JA, Hughes J: Osteopontin and calcium certain types of kidney cancer. Cell Tissue Res 319: 15–26, 2005 stone formation. Nephron, Physiol 98: 43–47, 2004 30. Ward HH, Romero E, Welford A, Pickett G, Bacallao R, Gattone 42. Giachelli CM, Pichler R, Lombardi D, Denhardt DT, Alpers CE, VH 2nd, Ness SA, Wandinger-Ness A, Roitbak T: Adult human Schwartz SM, Johnson RJ: Osteopontin expression in angiotensin CD133/1(+) kidney cells isolated from papilla integrate into de- II-induced tubulointerstitial nephritis. Kidney Int 45: 515–524, veloping kidney tubules. Biochim Biophys Acta 1812: 1994 1344–1357, 2011 43. Worcester EM, Beshensky AM: Osteopontin inhibits nucleation of 31. Lodi D, Ligabue G, Cavazzini F, Lupo V,Cappelli G, Magistroni R: calcium oxalate crystals. Ann N Y Acad Sci 760: 375–377, 1995 CD133 and CD24 expression in renal tissue of patients affected by 44. Wesson JA, Worcester E: Formation of hydrated calcium oxalates autosomal dominant polcystic kidney disease. Stem Cell Dis- in the presence of poly-L-aspartic acid. Scanning Microsc 10: covery 3: 211–217, 2013 415–424, 1996 32. Liu J, Qi Y, Li S, Hsu SC, Saadat S, Hsu J, Rahimi SA, Lee LY, Yan C, 45. Wesson JA, Johnson RJ, Mazzali M, Beshensky AM, Stietz S, Tian X, Han Y: CREG1 interacts with Sec8 to promote car- Giachelli C, Liaw L, Alpers CE, Couser WG, Kleinman JG, Hughes diomyogenic differentiation and cell-cell adhesion. Stem Cells J: Osteopontin is a critical inhibitor of calcium oxalate crystal 34: 2648–2660, 2016 formation and retention in renal tubules. J Am Soc Nephrol 14: 33. Grantham JJ, Geiser JL, Evan AP: Cyst formation and growth in 139–147, 2003 autosomal dominant polycystic kidney disease. Kidney Int 31: 46. Fabris A, Bernich P, Abaterusso C, Marchionna N, Canciani C, 1145–1152, 1987 Nouvenne A, Zamboni M, Lupo A, Gambaro G: Bone disease in 34. Terryn S, Ho A, Beauwens R, Devuyst O: Fluid transport and medullary sponge kidney and effect of potassium citrate treat- cystogenesis in autosomal dominant polycystic kidney disease. ment. Clin J Am Soc Nephrol 4: 1974–1979, 2009 Biochim Biophys Acta 1812: 1314–1321, 2011 35. Luyten A, Su X, Gondela S, Chen Y,Rompani S, Takakura A, Zhou Received: Accepted: J: Aberrant regulation of planar cell polarity in polycystic kidney October 15, 2018 March 7, 2019 disease. J Am Soc Nephrol 21: 1521–1532, 2010 36. Saburi S, Hester I, Fischer E, Pontoglio M, Eremina V, Gessler M, M. Bruschi, S.G., and L.S. contributed equally to this work. Quaggin SE, Harrison R, Mount R, McNeill H: Loss of Fat4 disrupts PCP signaling and oriented cell division and leads to cystic kidney Published online ahead of print. Publication date available at disease. Nat Genet 40: 1010–1015, 2008 www.cjasn.org.