Human Urinary Exosomes as Innate Immune Effectors

Abstract

Exosomes are small extracellular vesicles, approximately 50 nm in diameter, derived from the endocytic pathway and released by a variety of cell types. Recent data indicate a spectrum of exosomal functions, including RNA transfer, antigen presentation, modulation of apoptosis, and shedding of obsolete protein. Exosomes derived from all nephron segments are also present in human urine, where their function is unknown. Although one report suggested in vitro uptake of exosomes by renal cortical collecting duct cells, most studies of human urinary exosomes have focused on biomarker discovery rather than exosome function. Here, we report results from in-depth proteomic analyses and EM showing that normal human urinary exosomes are significantly enriched for innate immune proteins that include antimicrobial proteins and peptides and bacterial and viral receptors. Urinary exosomes, but not the prevalent soluble urinary protein uromodulin (Tamm–Horsfall protein), potently inhibited growth of pathogenic and commensal Escherichia coli and induced bacterial lysis. Bacterial killing depended on exosome structural integrity and occurred optimally at the acidic pH typical of urine from omnivorous humans. Thus, exosomes are innate immune effectors that contribute to host defense within the urinary tract.

Exosomes form as intraluminal vesicles of multivesicular bodies (MVBs), contain membrane and cytoplasmic proteins, have a cytoplasmic-side inward membrane orientation, and are released intact into the extracellular space (Figure 1A). First described in maturing ovine reticulocytes,¹ exosomes are released by many cell types² and have been conventionally regarded as a vehicle for shedding obsolete protein. However, emerging evidence has revealed a variety of exosomal functions, including the intercellular transfer of membrane receptors³ and RNA,^4–6 induction of immunity,⁷ antigen presentation,⁸ modulation of bone mineralization,⁹ and antiapoptotic responses.¹⁰

Figure 1: Vesicles isolated from human urine are consistent with exosomes. (A) Exosomes are derived from the endocytic pathway (1–4) forming through invagination of the limiting membrane of the MVB (3). They are released into the urinary space from renal tubular epithelial cells through fusion of the MVB with the apical plasma membrane (4). (B and C) Exosomal isolation was confirmed by the identification by negative stain EM of nanovesicles (black arrows; characteristic mean 50-nm size distribution. (D) Uromodulin (white streaks in B and dark in C; open arrows in B and C) cofractionated with exosomes but was confirmed to be extraexosomal (5 nm gold-labeled; white arrows in C). (E) Western blot confirmed the presence of known exosomal constituents in vesicle preparations but did not confirm them in precipitated protein from exosome-depleted urine. (F) Immuno-EM with 5 (TSG101 and CD63) or 15 nm (AQP2) gold particle-labeled antibodies showed vesicular residency of known exosomal constituents (arrows). Vesicles were nonpermeabilized; thus, positive staining with an anti-CD63 antibody directed against an extracellular epitope indicated the cytoplasmic side inward membrane orientation characteristic of exosomes. EDUP, exosome-depleted urine protein; HKM, human kidney membrane; MW, molecular weight; TSG101, tumour susceptibility gene 101.

Nanovesicles were first shown in human urine by Kanno and colleagues¹¹ and subsequently, were shown to represent exosomes.¹² Consistent with a renal tubular epithelial origin, renal tubular epithelial cells contain MVBs at the apical surface, and urine exosomes contain apical membrane proteins from every cell type along the nephron.^13,14 The array of functions ascribed to exosomes in other tissues has kindled recent interest in the functional significance of urinary exosomes. Hogan et al.¹³ suggested interaction of exosome-like vesicles with primary cilia of renal epithelial cells, and Street and coworkers¹⁵ showed in vitro uptake of exosomes by a renal cortical collecting duct cell line, leading to speculation that exosomes may provide intrarenal proximal-to-distal transapical renal tubular epithelial signaling through RNA transfer. However, most studies on urine exosomes to date have focused on biomarker discovery, resulting in the publication of several urine exosome protein compendia.^{12,13,16–21}

Published reports of the urinary exosomal proteome have limited value in illuminating the potential functions of urinary exosomes for several reasons. First, protein identification by mass spectrometry (MS) has, until recently, yielded results with unacceptably low reproducibility and high false-positive rates,^22,23 and previous reports are not free of these limitations. Second, studies have often aimed to maximize the number of protein identifications and hence, biomarker candidates rather than applying or reporting rigorous protein identification thresholds. Third, most have relied on pooled samples from up to six donors and have not reported interindividual variability or reproducibility.

Here, we sought, for the first time, to ascribe functionality to human urinary exosomes. We initially performed rigorous, conservative tandem MS analysis of separate human urinary exosomal samples to allow enrichment scoring²⁴ to elucidation of urine exosomal function.

Results

Exosomal samples were obtained from 10 healthy volunteers (five men and five women) ages 23–36 years. Nine subjects were Caucasian, and one man was Mauritian. Samples prepared from 455±28 ml second morning void contained 0.54±0.18 μg protein/ml urine. Exosomal integrity was confirmed by the electron microscopy (EM) demonstration of 54.5±14-nm vesicles (Figure 1, B and D), the identification of known exosomal markers TSG101, enolase-1, CD63, podocin, and aquaporin-2 by Western blot and immunogold EM (Figure 1, E and F), and the confirmation of a cytoplasmic side inward membrane orientation (Figure 1F).

We applied a number of methodological approaches to overcome limitations of previous studies. (1) Samples were not pooled but analyzed separately. (2) MS was performed without and with a uromodulin exclusion list²⁵ on a high-sensitivity instrument. (3) Data were analyzed using stringent and novel bioinformatics approaches²⁶ (Supplemental Material). (4) All ambiguous peptides were excluded unless matched only to products of a single gene. From 50 μg exosomal protein per subject, we identified 601 unique proteins by MS, with a median (interquartile range [IQR]) P value of 1.17×10⁻¹³ (4.08×10⁻⁴) (Supplemental Table 1). Importantly, the use of a fixed uromodulin exclusion list²⁵ and a posterior error-derived protein type I error estimator termed espresso (Supplemental Material) unmasked proteins that would not otherwise have been evident (Figure 2A). The complete proteome included the known exosomal markers TSG101, CD14, and CD59. However, 307 (51%) had not previously been identified in exosomes from any source compared with the EXOCARTA database.²⁷ There was only minimal overlap with the soluble urinary proteome (Supplemental Figure 1).

Figure 2: The human urinary exosomal proteome. (A) Using conventional methods (Mascot scoring and no uromodulin exclusion), only 237 proteins would have been evident (green). The application of a uromodulin exclusion list (purple) to conventional methods and the use of espresso (cyan) greatly increased protein identifications, which was further amplified by their combination. (B) Cellular localization of 601 proteins identified from exosomal pellets by tandem MS. A significant proportion consisted of cytoplasmic and membrane proteins or constituents of the endocytic pathway or vesicles. (C) Heat map showing the overlap between subjects for all proteins identified from two or more peptides (yellow) or one peptide (red) or not seen (black). (D) Most proteins were identified in the minority of subjects, consistent with the stochastic nature of MS. The x axis represents the number of subjects in whom each protein was observed; 47 proteins were observed in all 10 subjects, showing the value of analyzing multiple samples separately. (E) Principal protein groupings were consistent with previous reports of exosomes but also included those exosomes with an innate immune role.

As previously reported for exosomes,¹² the cellular origin of the majority of identified proteins was in vesicles or the endocytic vesicular pathway, cytoplasm, plasma membrane, and nucleus (Figure 2B). Apical membrane proteins from all nephron segments were detected, including aminopeptidase N, carbonic anhydrase II and IV, chloride intracellular channel 1, Cubilin, Dipeptidase 1, calbindin-D_28k, and Vacuolar ATPase 6 V₀ subunit C. Most proteins were identified from fewer than one half of samples; 47 (8%) proteins were identified in all 10 samples (Figure 2, C and D), and 97 (16%) proteins were identified in at least 8 of 10 samples. Each of the shared proteins was identified from 6 (IQR=2–24) peptides, indicating a set of consistently identifiable urinary exosomal constituents (Figure 2E). This core exosomal proteome included a large number of proteins involved in the endocytic pathway, MVB formation, and exosomal biogenesis and cytoskeletal proteins required for exosomal structural integrity. A considerable number was involved in transcription and RNA processing, suggesting a potential role in the targeting of RNA to exosomes,⁴ or involved in innate immunity and the response to infection. Uromodulin was present in all 10 samples, consistent with previous reports, but confirmed as extraexosomal by immunogold EM (Figure 1D). All MS data have been deposited with the ProteomeXchange consortium repository (http://proteomecentral.proteomexchange.org; reference PXD000117).

Using enrichment scoring (ES),²⁴ we found, in addition to the expected enrichment for cytoskeletal proteins (ES=10.45) and proteins involved in the endocytic pathway and vesicle formation (ES=8.06), a very significant enrichment for proteins involved in immunity and host defense (ES=3.27, P<0.001). The MS-derived type I error estimates for proteins falling into this grouping were consistently small (Figure 3A), indicating high significance. These 29 proteins (Figure 3B, Table 1) included archetypal antimicrobial proteins and peptides and fell into two categories: those proteins with known bacteriostatic (such as mucin-1, fibronectin, and CD14) or bactericidal (e.g., lysozyme C, calprotectin [S100A8/A9], and dermcidin) roles and those proteins that function as microbial receptors or bind to bacterial surface molecules.^28–33 Importantly, 28 of 29 immune proteins have known expression in kidney (Table 1). We confirmed the presence and exosomal residency of a representative group of these innate immune exosomal proteins, including lysozyme C, dermcidin, mucin-1, calprotectin, and myeloperoxidase, by Western blot and immunogold EM (Figure 3C, Supplemental Figure 9). The EM appearances differ markedly from the considerably larger and more heterogeneous neutrophil granule,³⁴ where some of these proteins are also found.

Figure 3: Exosomes are enriched for innate immune proteins. (A) The exosomal proteome was significantly enriched for proteins with a known role in host defense. Proteins falling into this category (red) were identified with a high degree of statistical confidence compared with all nonimmune proteins called. (B) The immune group included bactericidal and bacteriostatic proteins as well as those proteins known to function as bacterial or viral receptors. (C) Western blot confirmed the presence of a representative group of proteins identified by MS, including myeloperoxidase (MPO), mucin-1 (MUC1), dermcidin (DCD), calprotectin (S100A8/A9 heterodimer), and lysozyme C (LYZ). Positive controls represent human kidney membrane, except DCD (purified DCD) and calprotectin (neutrophil lysate). Immunogold EM shows the decoration of vesicles with 5-nm gold particles.

Table 1: Identified exosomal proteins with a known innate immune role

Because of this enrichment, we next tested the hypothesis that exosomes may inhibit growth of or kill Escherichia coli, the organism responsible for up to 90% of urinary tract infections (UTIs) in humans.^35–37 First, BL21, a laboratory strain of E. coli, was incubated with exosomes, suspension buffer alone, or purified human uromodulin, because uromodulin has itself been proposed as an antimicrobial urinary defense protein³⁸ and it cofractionates with exosomes during ultracentrifugation.^25,39 Initial colony counting⁴⁰ showed significant reduction in bacteria after incubation with exosomes (with or without uromodulin) but not uromodulin alone. However, this method is unable to assess real-time growth. To achieve this result, we transformed BL21 with the luxCDABE operon⁴¹ (BL21-lux) to constitutively express luciferase; transfected organisms spontaneously omit 490-nm light. Compared with buffer alone, exosomes from each of four healthy volunteers significantly inhibited growth of BL21-lux, whereas growth was not altered by uromodulin (Figure 4). There was only minor interindividual variation (Supplemental Figure 2). The effect of exosomes on bacterial growth occurred rapidly, becoming apparent within 30 minutes of coincubation (Supplemental Figure 3). Complete inhibition of bacterial growth was dependent on exosomal structural integrity (Figure 4A, Supplemental Figure 4).

Figure 4: Exosomes inhibit growth of E. coli. (A) All exosomal additions were made at 25 μg/ml. All curves represent mean±SEM; y axes show luminescence (relative light units×10⁶), and x axes show time in hours. Growth curves are left-censored at the onset of detectable bacterial growth and arranged by growth medium (columns) and organism (rows). All data represent at least three replicates per condition. Data for BL21 represent triplicates for each of four biologic replicates. For all strains of E. coli tested, highly significant inhibition of growth in either medium occurred with the addition of exosomes compared with controls (P values shown). With the exception of UPEC in LB, where uromodulin achieved some growth inhibition compared with control (P=0.04), bacterial growth did not significantly differ with uromodulin compared with control in any experiment. Intact exosomes inhibited growth of BL21 compared with lysed exosomes (shown in green; P<0.001); lysed exosomes induced some growth inhibition compared with buffer only (P<0.001). As expected, given the protein- and carbohydrate-rich nature of LB medium compared with urine, all organisms grew more vigorously in LB. For Nissle, UTI89, and E. coli O6:H1 (CFT073) grown in urine, the number of viable organisms at peak bacterial growth (dashed lines labeled A–C) was additionally assessed by conventional colony counting (corresponding bar graphs in B). Addition of exosomes to donor urine resulted in significant reductions in the number of CFU for Nissle (P=0.009), UTI89 (P=0.05), and CFT073 (P=0.003). Colony counts are shown as CFU per milliliter×10⁶.

To assess clinical relevance, we first transformed a uropathogenic E. coli (UPEC) strain, obtained from an infected patient, to express the luxCDABE cluster (UPEC-lux). Despite more vigorous growth, UPEC-lux growth was still almost arrested by exosomes from healthy volunteers (Figure 4A). We repeated these experiments with both BL21-lux and UPEC-lux using exosome-depleted urine as growth medium. Exosomes were equally effective against both organisms, although as expected, their overall growth was slower in urine. We next sought to determine whether exosomes were effective against well characterized, luxCDABE-transformed, commensal (Nissle) or standard uropathogenic (UTI89 and CFT073) model strains of E. coli. In both exosome-depleted donor urine and Luria-Bertani (LB) media, we observed consistent, significant growth inhibition of all three organisms by exosomes but not by uromodulin alone (Figure 4A). Colony counting at peak bacterial growth confirmed that the observed differences in luminescence were attributable to significant differences in the number of viable organisms (Figure 4B).

The effect of coincubation with exosomes on bacterial integrity was next evaluated by scanning EM. UPEC-lux incubated with exosomes or control and freeze-dried after 5 or 15 minutes showed clear evidence of exosome-induced bacterial lysis (Figure 5). Organisms incubated with exosomes for 5 minutes showed an increased proportion of lysed phenotypes (12%) compared with control samples (1.8%, P=0.003), increasing to 56% versus 2% after 15 minutes (Figure 5, A and C), consistent with the rapid inhibition of BL21-lux growth observed earlier (Supplemental Figure 3).

Figure 5: Exosomes induce lysis of E. coli. (A) Incubation of UPEC with exosomes (Exos) showed rapid induction of bacterial lysis. An increase in lytic phenotypes was detected as early as 5 minutes after incubation (P=0.003); after 15 minutes, more than 50% of UPEC had undergone lysis (P<0.001). (B–D) Intact organisms (B and D, white arrow) and whole organisms that had lost integrity (C and D, black arrows) were counted, but lysed bacterial fragments (D, open arrows) were not.

All these bacterial growth experiments were performed at pH 5.5–6.0, typical of omnivorous human urinary pH; exosomes were, however, also effective at pH 6.5 and pH 7.0 (Supplemental Figures 5 and 6). Furthermore, we noted that 0.5 μg/ml exosomes were effective against a starting UPEC-lux concentration (representing 500 relative light units) of >100 CFU/ml (Supplemental Figure 8), confirming our observations to be highly physiologically relevant.

Discussion

Urinary exosomes released by renal tubular epithelia are present in human urine, where their function is unknown. Here, we examined the normal human urinary exosomal proteome through in-depth MS and showed significant enrichment for known innate immune proteins. EM studies confirmed the exosomal residency of antibacterial proteins and peptides. Urinary exosomes from healthy volunteers inhibited the growth of different uropathogenic and commensal E. coli strains and induced bacterial lysis, and these effects were dependent on the structural integrity of urinary exosomes. These findings suggest that urinary exosomes function within the renal tract as innate immune effectors.

Although many of the innate immune proteins identified in our analysis had been previously identified in urinary exosomal analyses (Supplemental Table 1),^{12,13,16–21} enrichment for this protein grouping has not previously been reported. However, enrichment analysis relies on the reliability of the background proteome. Our analysis of separate urinary exosomal preparations, along with key methodological advances, has revealed a number of proteins identified from urinary exosomes for the first time. Furthermore, rather than place emphasis on maximizing the number of protein identifications, we made every attempt to ensure the robustness of those proteins ultimately included in this proteome. These key aspects of our study enabled the clear identification of enrichment for innate immune proteins (confirmed by direct methods) as well as the expected enrichment for those proteins with a cytoskeletal role, involved in MVB or exosomal biogenesis, or resident on the luminal surface of the nephron.

UTIs represent the most common bacterial infection in humans, with an estimated 150 million UTIs per annum globally.^42,43 Most of these UTIs are caused by E. coli.^35,36,44 The anatomic location of the distal urethra results in its continuous exposure to large numbers of bacteria. Nevertheless, the urinary tract is usually sterile above the urethral meatus, indicating the existence of highly effective innate immune mechanisms within the renal tract. To date, these mechanisms remain incompletely understood. For infection to occur, uropathogenic organisms must overcome a variety of physical and immune obstacles and ascend the urinary tract. First, organisms must adhere to uroplakin-covered urothelium, which is limited by the shear flow of urine and the glycoprotein uromodulin, which impairs the ability of bacteria to adhere and facilitates their expulsion during voiding.⁴⁵ Second, invading organisms are exposed to free radicals and soluble antimicrobial proteins and peptides present within urine, although the soluble concentrations of these proteins and peptides seem too low to be directly bactericidal.^46,47 To evade these defenses, organisms typically invade epithelial cells, where they proliferate intracellularly before egressing back into the urinary space, a process that may, in turn, trigger acute inflammatory responses through Toll-like receptor activation. Our findings indicate that, in addition, this defensive array includes exosomes, which must be evaded or overwhelmed for UTI to occur. Constitutively released by renal tubular epithelia, a continuous stream of exosomes acts as an innate immune sentinel within the urinary tract. Because exosomes contain molecules at once attractive and lethal to bacteria, they provide vehicles for the efficient, targeted distal delivery of antimicrobial molecules and confer on renal epithelia the ability to effect distant bacterial killing. Furthermore, they may function as decoys in limiting interaction of bacterial adhesion molecules with epithelial surfaces.

Our study has several strengths. First, this study is the first comprehensive report of in-depth proteomic analysis of individual, rather than pooled, urinary exosomal samples. Second, we report that a urinary exosomal proteome was constituted by statistically robust and reproducibly identifiable protein identifications. Third, we show a clinically relevant effect of urinary exosomes on the most common human urinary pathogen E. coli using several methodologies. However, our findings should be interpreted within the limitations of the study. It is not possible to infer from our data which exosomal proteins represent the key effectors, because targeted deletion of individual exosomal proteins has not yet been possible. More than one third of innate immune proteins were consistently identified (≥8 of 10 samples), and this subgroup may be central. However, we cannot exclude the possibility that other less consistently identified members of this proteome or indeed, unidentified molecules are important. Furthermore, even with contemporary high-sensitivity mass spectrometers, mitigation against the masking effect of abundant proteins is incomplete. Lower abundance proteins may, therefore, have gone undetected or may not have been consistently seen. In contrast to the situation in vivo, where exosomes would be continually replenished, our in vitro experiments used a single exosomal dose; however, the strength of the observed effect further supports its physiologic relevance. Finally, uromodulin cofractionates with urinary exosomes during isolation; therefore, we have not directly assessed the effects of uromodulin-free exosomal preparations, because the exosomal population would be highly compromised.

Our findings add to an expanding body of evidence that exosomes are biologically active in a wide variety of tissues.^4,10,48–50 They are consistent with a report of protection against viral respiratory pathogens by respiratory epithelium-derived exosomes,⁵¹ in which Kesimer and colleagues⁵¹ showed inhibition of influenza A virus infection of Madin–Darby canine kidney cells by respiratory epithelium-derived exosomes in vitro. The demonstration that urinary exosomes are able to induce bacterial lysis and inhibit growth at bacterial concentrations that are highly physiologically relevant makes identifying factors that modulate exosomal release and constitution key priorities for future work and may reveal potential therapeutic targets for the treatment of UTIs.

Concise Methods

Exosomal Isolation

Exosomes were isolated from 10 healthy volunteer urine samples as previously described.²⁵ The five men and five women ages 23–36 years were on no regular medication and had not consumed antibiotics or other medication within the previous 1 month. Current UTI was excluded. Briefly, subjects urinated directly into a container with protease inhibitors, including PMSF (500 μl 0.5 M solution), leupeptin (450 μg), and sodium azide (15 ml 100 mM solution). Urine was centrifuged within 30 minutes of collection (Beckman AVANTI J26-XP centrifuge; JA-17 fixed angle rotor; polyallomer 50-ml centrifuge bottles) for 20 minutes at 17,000×g. The supernatant was passed through a sterile 0.22-µm filter and ultracentrifuged (Beckman Optim L-100 XP VAC Ultracentrifuge; Ti45 fixed-angle titanium rotor; Beckman 70-ml polycarbonate ultracentrifuge bottles) for 135 minutes at 235,000×g and 4°C. Each ultracentrifugation pellet was suspended in 50 µl suspension buffer (250 mM sucrose and 10 mM triethanolamine [pH 7.6]) and pooled with the other pellets from the same urine sample. To avoid confounding effects on bacterial growth, urine samples for these experiments were collected without protease inhibitors or sodium azide, triethanolamine was removed from the suspension buffer, and exosomal pellets were washed one time by resuspension and ultracentrifugation.

Exosome-Depleted Urinary Protein

Samples were prepared as described for exosomal isolation above. The supernatant from the 235,000×g centrifugation was retained after harvesting exosomal pellets. This supernatant contained the nonexosomal soluble urinary protein. For MS, protein was precipitated from these supernatants (typically 360 ml per subject) by the addition of ammonium acetate and acetone (Supplemental Material). Pellets were resuspended with Laemmli sample buffer.

Sample Preparation for MS

Protein from resuspended exosomal pellets was concentrated by precipitation and quantified using a Bradford protein binding colorimetric assay (Bio-Rad). Protein pellets were stored at −80°C until use.

For MS, protein pellets were suspended in Laemmli sample buffer and incubated at 95°C; 50 µg protein from each sample was fragmented on a 4%/12% SDS–polyacrylamide gel. After staining, each gel track was separated into 28 equal sections, which were processed individually for the remainder of the workflow. After destaining, proteins were reduced and alkylated in gel, washed with NH₄HCO₃, and dehydrated in acetonitrile, and proteins were digested with modified trypsin.

Liquid chromatography–MS/MS was performed using an Eksigent NanoLC-1D Plus (Eksigent Technologies) HPLC system and an LTQ Orbitrap Mass Spectrometer (Thermo Fisher Scientific). Peptides were separated by reverse-phase chromatography (Dionex). Peptides were loaded onto a 5-cm C18 precolumn (300 μm inner diameter; LC Packings) from the autosampler. Peptides were eluted onto the analytical column using gradients for solvent A (water+0.1% formic acid) and B (acetonitrile+0.1% formic acid) of 5%–50% B over 40 minutes. A New-Objective nanospray source was used for electrospray ionization. m/z Values of eluting ions were measured in the Orbitrap mass analyzer with a mass range of 350–1600, and the resolution was set at 7500.

MS Data Processing

Peptides from each gel segment were run two times. All segments were run with dynamic exclusion. From these runs, a fixed exclusion list was generated for the abundant protein uromodulin and superimposed on a dynamic exclusion list as described elsewhere.²⁵ Data from these two sets of runs were combined. MS data were processed using the SEQUEST Bioworks Browser (version 3.3.1 SP1; Thermo Fisher Scientific) to generate MS/MS peak lists. Combined peak list files were submitted to the MASCOT search algorithm (version 2.2.1; Matrix Science) and searched against the IPI-Human Database, version 4.3. Spectra were rescored using MASCOT-Percolator, a machine learning tool that minimizes false discoveries and incorporates target decoy searching.²⁶ Protein identification required two or more unique peptides, with a false discovery rate of 0.1. Single peptide identifications were included if the MASCOT-percolator posterior error probability was <0.01. In addition, we included proteins based on an in-house protein type I error estimator termed espresso (Supplemental Material). The cellular location of proteins was evaluated by searching the Ensembl gene identifier for each protein against the Human Protein Atlas Database (http://www.proteinatlas.org). Renal expression was evaluated using the bioGPS Gene Portal System (http://www.biogps.org). Comparison of exosomal proteins with previous reports was made using the Exocarta exosomal protein database.²⁷ Enrichment scoring was performed using the DAVID bioinformatics tool.²⁴ This method provides a measure (by tests of proportions) of whether functional categories are overrepresented within a gene list compared with what is expected from stochastic sampling of the entire human gene set.

Protein Confirmation and Antibodies

Proteins identified by MS were confirmed by Western blotting according to standard methods and immuno-EM (described below). For each protein, the same antibodies (Supplemental Material) were used for both techniques.

Bacterial Transformation

Bacteria were transformed by electroporation to express the luxCDABE operon (Bioware plasmid pXEN13), which results in ampicillin resistance and constitutive expression of luciferase.

Bacterial Growth Assays

Exosomal samples were collected from the same 10 healthy volunteers described earlier, but in contrast to samples prepared for MS, these samples were handled without protease inhibitors, triethanolamine, or sodium azide.

Bacterial Colony Counting

Colony counting was performed essentially as in Miles and Misra.⁴⁰ Briefly, serial dilutions of overnight bacterial cultures were spotted onto ampicillin-impregnated agar plates and incubated at 37°C for 18 hours. Colony counts were made from the highest bacterial concentration that yielded distinct colonies.

Bacterial Growth Curves

For each growth assay, negative controls comprised suspension buffer only or purified uromodulin (P135–1; SCIPAC) diluted in suspension buffer. Overnight bacterial cultures were diluted in LB broth to luminescence of 100–500 relative light units. At 4°C, 10 μl bacterial culture was placed in each well of a 96-well plate; 5 μg exosomal protein, uromodulin, or an equal volume of suspension buffer was added to each well. Volumes in all wells were standardized to 200 μl with LB or exosome-depleted urine as appropriate (Supplemental Figure 7). All samples were evaluated at least in triplicate. Plates were incubated at 37°C, and luminescence readings were obtained hourly until growth ceased. Exosomes were chemically lysed by sequential addition of ammonium acetate in methanol and acetone as described for protein precipitation above.

EM

Negative staining transmission EM with neutralized phosphotunstic acid or uranyl acetate on carbon film grids was performed with an FEI Tecnai G² electron microscope operated at 120 kV using an AMT XR30B digital camera. For immunolabeling, samples were placed on glow-discharged nickel grids for 30 seconds. Primary antibodies were applied for 15 minutes at room temperature. Secondary antibodies labeled with 5-, 10-, or 15-nm gold particles were applied for 15 minutes. Grids were blocked with PBS. For scanning EM, bacterial cultures were incubated with exosomes, and 10 μl were spotted onto glass coverslips, quench-frozen in propane-cooled liquid nitrogen, freeze-dried (Edwards Auto 306 Turbo), and gold-coated. Scanning EM was carried out with a Philips XL-30 FEG–scanning EM, and images were captured with an accelerating voltage of 5 kV.

Statistical Analyses

Data were analyzed with Stata SE, version 12.1. Data are presented as means±SDs or median (IQR) as appropriate. Parametric continuous variables were compared with the t test, and nonparametric continuous variables were compared with the Wilcoxon sign rank test. Bacterial growth was compared using response feature analysis with the area under the curve as the response feature, and comparisons were made by one-way ANOVA. Proportions were compared with Pearson’s chi-squared test of proportions.

Disclosures

None.

We thank M. Clatworthy and M. Berry for the generous donation of UTI89 E. coli and Ardeypharm GmbH for Nissle E. coli (Mutaflor).

This work was supported by an Action Medical Research Training Fellowship (to T.F.H.), Wellcome Trust Grant 088489 (to F.E.K.F.) and Strategic Award 079895 (to Cambridge Institute for Medical Research), the National Institute for Health Research Cambridge Biomedical Research Centre (CBRC), and the Biotechnology and Biological Sciences Research Council (BBSRC). T.F.H. received a Raymond and Beverley Sackler Research Studentship and is currently supported by the CBRC. P.D.C. was supported by BBSRC Research Studentship BB/D526088/1. L.G. is supported by a 7th Framework Programme of the European Union (262067-PRIME-XS).