|Home | About | Journals | Submit | Contact Us | Français|
Urinary exosomes have been proposed as starting material for discovery of protein biomarkers of kidney disease. Current protocols for their isolation use a two-step differential centrifugation process. Due to their low density, exosomes are expected to remain in the low-speed (17,000 × g) supernatant and to sediment only when the sample is spun at high speed (200,000 × g). Analysis using western blots and electron microscopy found that urinary exosomes are also present in the low-speed pellet entrapped by polymeric Tamm-Horsfall protein thus diminishing the procedures reproducibility. Here we show that addition of dithiothreitol to the low-speed pellet disrupted the polymeric network presumably by reduction of disulfide bonds linking the monomers. This modification shifted the exosomal proteins from the low-to the high-speed pellet. Also, by shifting the Tamm-Horsfall protein to the high-speed pellet, the use of dithiothreitol makes it feasible to use Tamm-Horsfall protein to normalize excretion rates of exosomal proteins in spot urines. We tested this by western blot, and found that there was a high degree of correlation between exosomal proteins and Tamm-Horsfall protein in the high-speed pellet. Since the yield of exosomes by differential centrifugation can be increased by chemical reduction, Tamm-Horsfall protein may be a suitable normalizing variable for urinary exosome studies when quantitative urine collections are not practical.
Exosomes are small (20–100 nm) membrane vesicles that originate as the internal vesicles of multivesicular bodies in many cell types. They are released from the multivesicular body lumen into the extracellular environment (including plasma and urine) on fusion of the outer membrane of the multivesicular bodies with the plasma membrane. Since the initial description of exosomes in urine by Pisitkun et al. in 2004,1 there have been several studies focusing on the physiological and pathophysiological significance of these vesicles in urine.2–7 Urinary exosomes contain proteins that are characteristic of every renal tubule epithelial cell type, as well as podocytes and transitional epithelia from the urinary collecting system.1 Therefore urinary exosomes provide a suitable starting material for biomarker discovery relevant to a variety of disease processes. One barrier to success with this approach is the presence of large amounts of Tamm-Horsfall protein (THP) in urine.8
Tamm-Horsfall protein, also known as uromodulin, is the most abundant protein in urine under physiological conditions. It is a glycosylphosphatidylinositol-linked membrane protein synthesized exclusively in the thick ascending limb of Henle’s loop. It has a zona pellucida (ZP) domain of approximately 260 amino acids including 8–10 conserved cysteine residues. THP is targeted to the plasma membrane and is secreted into the urine through an extracellular proteolytic cleavage, allowing delivery into the tubule fluid. THP is found in the urine as a high-molecular-weight polymer assembled into filaments or matrices. The ZP domain presumably functions as a polymerization module.9–11 The function of THP has been investigated but there is no strong consensus regarding its role in normal kidney physiology.12 A study in THP gene-deficient mice suggested that THP may regulate the transporters in Henle’s loop.13
The isolation of urinary exosomes usually shows a two-step differential centrifugation process.1 In an initial low-speed spin (17,000 × g), high-density membranes and poly-merized THP are normally removed. Owing to the low density of the exosomes, they are expected to remain in the supernatant from this spin. In the second spin at 200,000 × g, urinary exosomes are normally found in the pellet.1 The large polymeric THP network has the potential to trap exosomes in the 17,000 × g pellet, thus preventing efficient and reproducible isolation in the subsequent 200,000 × g spin. In this work we show a substantial quantity of urinary exosomes precipitates by entrapment by polymeric THP in the 17,000 × g spin. We show that the entrapment can be eliminated by chemical reduction of disulfide bonds with dithiothreitol (DTT), thereby depolymerizing the THP. In addition, we also discuss the lack of a standard method for normalizing protein measurements in urinary exosomes and propose the use of THP as a normalizing variable.
Figure 1 shows a Coomassie-stained gel of the 17,000 and 200,000 × g pellets from pooled normal human urine (three subjects) with and without addition of the reducing agent DTT at the time of 17,000 × g centrifugation. The large band centered at about 92 kDa is known to represent THP monomer.8 As can be seen, a substantial amount of THP was present in the 17,000 × g pellet without reduction and most of it was moved to the 200,000 × g fraction when the sample was reduced (Figure 1).
To explain the findings in Figure 1, we carried out electron microscopy of the 17,000 × g pellets prepared from one subject (Figure 2) either without (panels a and b) or with (panel c) DTT treatment. As shown in Figure 2a and b, the THP formed long polymeric filaments that associated laterally to form ropelike structures. The complex THP network was seen to contain small (20–100 nm) vesicles compatible with exosomes (Figure 2b). When treated with DTT, the THP meshwork was dissipated (Figure 2c). Figure 2d shows a few short convoluted THP oligomers present in the 200,000 × g pellet incubated with DTT at the time of the 17,000 × g centrifugation.
To investigate the apparent entrapment of exosomes in more detail, we processed pooled urine (three subjects) with and without DTT and then carried out immunoblotting of the 17,000 and 200,000 × g fractions for two exosomal markers, Alix and TSG101, as well as two integral membrane proteins normally present in urinary exosomes, CD9 and aquaporin-2 (AQP2)1 (Figure 3). Because cooling is known to increase THP polymerization and because urine samples are normally stored either at 4 or −80 °C, we compared the same urine samples processed fresh (Figure 3a), stored at 4 °C for 3 days before processing (Figure 3b), and stored at −80 °C for 3 days before processing (Figure 3c). The signals for Alix, TSG101, and AQP2 in the 17,000 × g pellets were diminished by DTT reduction regardless of storage, whereas CD9 was not seen in the 17,000 × g pellet even without DTT. This finding is consistent with the conclusion above that there is significant entrapment of exosomes by the unreduced THP polymeric network. The absence of CD9 in the unreduced 17,000 × g fraction may suggest that some subpopulations of exosome-like vesicles escape entrapment.
In additional experiments, the 17,000 × g pellet was incubated with different DTT concentrations (50, 100, and 200 mg/ml), temperatures (37, 45, 50, 60, and 95 °C), and incubation times (2, 5, 10, and 15 min) (Supplementary Figures). The best enrichment of urinary exosome markers, without any protein aggregation or degradation, was observed with 200 mg/ml of DTT at 37 °C for 5–10 min. In each case, the DTT was added to the 17,000 × g pellet, the pellet was resuspended, incubated, and recentrifuged.
We have repeated the isolation of urinary exosomes with and without DTT five times, collecting urine from different normal subjects each time. We have observed the following: (1) the size of the 17,000 × g pellets is variable between different experiments, (2) the addition of DTT to the low-speed centrifugation pellet shifts the majority of THP from the low-speed pellet to the high-speed pellet, (3) exosomal markers are generally not detected in the low-speed pellet when it is incubated with DTT, and (4) urinary exosomal enrichment is variable between the different experiments but the highest signal in the 200,000 × g pellet is always realized when the 17,000 × g pellets are reduced with DTT.
To assess the yield of exosome-like particles with the DTT-reduction protocol, we counted vesicles using electron micrographs of the 200,000 × g pellets from six normal human subjects. The same fraction of the pellet was placed on electron microscopic grids and 20 electron micrographic fields were counted at × 20,000 magnification from each subject (see Supplementary Table 1 for the raw vesicle counting data). A sampling error of ± 9% was estimated for this procedure (Supplementary Table 1). Urinary ‘exosome-like particles’ were defined on the basis of the size (20–100 nm) and the shape (round) of the vesicles (Figure 4). Note that the DTT treatment has no obvious effect on morphology of the exosome-like particles. The minimal rate of excretion of these particles was calculated from the data giving a value of 4440 ± 1639 (s.d.) particles per min. Because, we do not know the efficiency of binding of the particles to the electron microscopy grids in these determinations, we consider this a ‘minimum’ value.
To determine whether the exosome marker protein abundance correlates with the counts of exosome-like particles, we carried out immunoblotting in the 200,000 × g fraction of the same samples used for the exosome counting described in the previous paragraph (n = 6) (Figure 5a), comparing the exosomal proteins Alix, CD9, TSG101, HSP70, and AQP2. These immunoblots were loaded with an amount of each sample calculated to give an equal number of exosome-like particles based on the quantitative electron microscopy described above. As can be seen, the band intensity values of all markers were highly variable among the normal human subjects. The most likely explanation for this is that the exosome-like particles varied from sample to sample in composition or ability to bind to the electron microscopy grid.
We asked whether THP may provide a reliable normalization variable to allow quantitative analysis in untimed collections of urine (‘spot’ urine samples). To address this question, we also carried out immunoblotting for THP in the same samples described in the last section (Figure 5a, bottom immunoblot). This allowed us to plot the signal from each exosomal protein against the THP signal from immunoblotting (Figure 5b). The correlation coefficients for each protein are also summarized in Figure 5b. As seen, the correlation was high with correlation coefficients ranging from 0.84 to 0.99. The highest correlation coefficients were for Alix and TSG101, which are components of the protein complexes that form pro-exosomes in multivesicular bodies. Thus, THP amount is representative of the amount of exosomal markers and can potentially be substituted as a normalizing variable in spot urines. This is significant because it obviates the need for immunoblotting of exosomal markers for normalization, as the amount of THP from immunoblots and the amount of THP deduced from densitometry of Coomassie-stained gels correlate very highly (Figure 6). Consequently, normalization can potentially be carried out on the basis of measurements of THP in Coomassie-stained gels.
The objective of this study was to solve the problem of inconsistent isolation of urinary exosomes. We proposed that exosomes were being variably entrapped by polymeric networks of Tamm-Horsfall protein and showed such entrapment by electron microscopy. Our experiments showed that dissolution of the THP polymeric network through the use of the reducing agent DTT markedly reduced the amount of exosomal marker proteins (TSG101 and Alix) in the low-speed fraction from centrifugation of urine samples. Therefore, the exosomes were more completely retrieved in the subsequent high-speed fraction. We recommend therefore that the standard protocol for exosome isolation by differential centrifugation be modified to use DTT treatment of the resuspended low-speed pellet with a repeat of the low-speed (17,000 × g) spin. Figure 7 represents our current protocol based on the findings of this study. In addition we include a written protocol in the Supplementary Materials.
Tamm-Horsfall protein is related to the ZP family of proteins. ZP proteins form a polymeric meshwork that creates a barrier around a fertilized egg to prevent polyspermy. The polymerization is through disulfide linkages between ZP protein monomers resulting in a linear polymer. The pattern of distribution of cysteines in THP proteins is similar to that seen in ZP proteins. The success of DTT treatment in disrupting THP polymerization strongly supports the notion that disulfide linkages are also responsible for THP polymerization. THP was found to form ropelike structures by lateral aggregation of the THP monomeric chains (Figure 2). The physical basis for the lateral association is presently unknown.
In addition, we looked in additional detail about urine storage for exosome isolation. Our previous study concluded that freezing samples at −20 °C markedly reduced the retrieval of exosomes from urine samples14 but did not explain the loss. Here we compared storage at 4 and −80 °C with the yield using fresh urine samples. The immunoblotting did not show a substantial loss of exosomal marker proteins at either temperature. However, there was a small amount of retention of exosomal marker proteins in the 17,000 × g pellet even with DTT, suggesting that the mechanism of this small retention may depend on a process other than entrapment by THP networks (Figure 3). Thus, analysis of fresh samples seems to be superior. However, freezing at −80 °C appears to be a reasonable choice if long-term storage is necessary. Other than the temperature, the ionic strength and pH of urine can also contribute to THP aggregation.15 However, this study did not investigate the role of these factors on THP aggregation and exosome isolation. Thus, future studies are needed to address this issue experimentally.
Biomarker discovery and exploitation depend on development of methods for quantifying urinary excretion rates of proteins. However, the timed collections needed for the calculation of such rates are not always practical in a clinic setting. Thus, a normalization variable is needed to substitute for time in calculations of relative excretion rate of exosomal proteins. (Because water excretion is highly variable, comparison of absolute protein concentrations in urine is never a reasonable alternative.) One alternative for a normalizing variable would be exosomal markers such as Alix or TSG101. Such measurements, however, add to the complexity of the overall determination. Here we show that the level of THP in the 200,000 × g pellet correlates very highly with the levels of Alix or TSG101 (Figure 5). Hence, THP may be a suitable normalizing variable, especially as its level is easily assessed by densitometry of Coomassie-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS—PAGE) and does not require an immunoassay. THP is secreted by the thick ascending limb of Henle’s loop and, although its secretion is unlikely to be invariant among subjects, it is more likely to be relatively constant in the same subject. The advantage of THP over measurement of creatinine or some other substance filtered at the glomerulus is that it can be measured in the same high-speed differential centrifugation fraction that is used to assay potential biomarkers. Hence, factors that would cause THP to vary either physiologically or technically (in the isolation process) would be common to THP determination and the determination of any other protein in that fraction. Thus, we would expect that THP normalization would give a smaller systematic error than normalization by creatinine. In this study, we investigated the potential use of THP as a normalizing variable for estimation of relative excretion rates of exosomal proteins only in normal human subjects. Further studies are required to address the validity of THP as a normalizing variable in subjects with renal disease.
Urine was collected from 10 male healthy subjects (ages 22–43; NIDDK Clinical Research Protocol no. 00-DK-0107). Twelve-hour collections were used for the exosome counting by electron microscopy and in normalization experiments. The first morning void urine was used for the remainder of the experiments. The urine sample was collected in a sterile container added with a protease inhibitor mixture (volume per 50 ml of urine: 1.67 ml of 100 mM NaN3, 2.5 ml of 10 mM AEBSF, and 50 μl of 2 mM leupeptin). For the first morning 50 ml of void urine was collected and the protease inhibitor mixture was added immediately after collection. For the 12-h urine collection, 59 ml of the protease inhibitor mixture was added to a container before collection (based on the assumption that a normal human produces 600–800 ml of urine in 12 h). Except where explicitly noted, urine was processed immediately after collecting without storage.
Urine (96 ml) was used for centrifugation. Urine samples were placed in a centrifuge (Beckman L8-70M ultracentrifuge; Beckman Coulter, Fullerton, CA) and centrifuged at 17,000 × g for 10 min at 37 °C (13,600 r.p.m.). The supernatant was saved and the 17,000 × g pellets were resuspended in an isolation solution (250 mM sucrose, 10 mM triethanolamine (pH 7.6)) followed by incubation with either DTT (final concentration of 200 mg/ml; MP Biomedicals, Solon, OH) or vehicle at 37 °C for 5–10 min. During the DTT or vehicle incubation, samples were vortexed every 2 min. The temperature, incubation time, and DTT concentration were chosen on the basis of preliminary experiments (see Supplementary Figures). The incubated suspensions were then transferred to clean centrifuge tubes (Beckman polycarbonate catalog no. 355630) and more isolation solution was added to a final volume of 8 ml. The samples were centrifuged again at 17,000 × g for 10 min at 37 °C. The two supernatants from the 17,000 × g spins were pooled and ultracentrifuged at 200,000 × g for 1 h at 37 °C. Pellets were solubilized in 1.5% SDS and 50 mM Tris-HCl (pH 6.8). These samples were used for immunoblotting as described below. For the electron microcopy of urinary exosomes, pellets obtained from the 200,000 × g spin from each subject were resuspended in 1 × phosphate-buffered saline.
Vesicle suspensions for electron microscopy were further diluted 1:10 with phosphate-buffered saline and 5 μl suspension was applied to glow-discharged formvar-carbon films on copper 300 mesh grids (catalog no. 71150; Electron Microscopy Sciences, Hatfield, PA). The adsorbed exosomes were negatively stained with 1% aqueous uranyl acetate. The samples were examined with a JEM 1200EX electron microscope (JEOL) equipped with an AMT XR60B digital camera (Advanced Microscope Technologies).
Exosomes from six normal subjects were used for this procedure. Five fields located in the center, upper right and left, and lower right and left of the grid were selected for viewing the exosomes from each subject. Four electron micrographs in each selected field were randomly taken at × 20,000 magnification. Under this magnification the exosomes were clearly identifiable by their size, between 20 and 100 nm, and their round shape.1 In brief, 20 electron micrographs were taken for each subject, therefore a total of 120 electron micrographs were taken for the entire procedure. The rate of excretion of ‘exosome-like particles’ was calculated based on the collection times for each sample, the dilution volumes at various stages of the procedure, and the areas of the electron microscopy field counted.
Immunoblotting was carried out as described previously.1 Briefly, urinary exosome pellets were solubilized in Laemmli sample buffer (1.5% SDS, 6% glycerol, and 10 mM Tris-HCl (pH 6.8)). Gel loading of urinary exosome was based on the same percentage of the pellet’s final volume or on the estimated number of exosome-like vesicles counted as described previously. Proteins were separated by one-dimensional SDS–PAGE and were transferred to nitrocellulose membranes, which were blocked and probed with antigen-specific primary antibodies. The blots were incubated with species-specific fluorescent secondary antibodies (Alexa 688) and were visualized using the Odyssey Infrared Imaging System (Li-Cor, Lincoln, NE).
Antibodies acquired from commercial suppliers were TSG101 (ab83; Abcam, Cambridge, MA), Alix (AIP1; BD Biosciences), CD9 (C-4; Santa Cruz Biotechnology), HSP 70 (StressMarq, Victoria, Canada), Tamm-Horsfall protein (H-135; Santa Cruz Biotechnology). We also used a rabbit antibody to AQP2 produced in our laboratory.16
Table S1. Exosome counting.
Figure S1. Coomassie blue-stained SDS–PAGE gels of 17,000 × g pellets before and after the treatments with different concentrations of DTT for 5 min at 37 °C.
Figure S2. Aquaporin-2 (AQP2) immunoblots of 17,000 and 200,000 × g pellets after incubation of the 17,000 × g pellets with 200 mg/ml DTT at different temperatures and incubation times.
Figure S3. Alix and aquaporin-2 immunoblots of 17,000 and 200,000 × g pellets after incubation of the 17,000 × g pellets with 200mg/ml DTT at 37 °C for 5, 10, or 15 min.
This study was supported by the Intramural Budget of the NHLBI (Project Z01-HL001285). PFLL was supported by Spanish Society of Nephrology and the Catalan Society of Nephrology. We thank the NHLBI Electron Microscopy Core Facility (M Daniels, Director) for outstanding support.
All the authors declared no competing interests.