Characterization of exosome-like vesicles released from human tracheobronchial ciliated epithelium: a possible role in innate defense

Abstract

Airway mucus forms the structural basis of the local innate immune defense mechanism. It is an integrated, active, viscoelastic gel matrix evolved to protect the exposed lung from physical, chemical, and pathological erosion. Exosomes are biologically active vesicles secreted by different cell types including epithelial, hematopoietic, and some tumor cells. They are also present in some biological fluids such as serum, urine, breast milk, and bronchoalveolar lavage fluid. In this study, we demonstrate for the first time that exosome-like vesicles with antiviral properties are present in human tracheobronchial epithelial (HTBE) cell culture secretions. These vesicles have been isolated by differential centrifugation and are characterized further by mass spectrometry, flow cytometry, immunoblotting, electron microscopy, and light-scattering methods. HTBE vesicles exhibited characteristic exosomal size (30–100 nm) and morphology (cup-shaped) with a buoyant density in sucrose of 1.12–1.18 g/ml. Biochemical characterization further revealed typical surface, cytoskeletal, and cytoplasmic proteins characteristic of exosomes, including the multivesicular and late endosomal membrane markers Tsg101 and CD63. The presence of RNA was also observed. The epithelial mucins MUC1, MUC4, and MUC16 also contributed to the vesicles’ structure. Notably, α-2,6-linked sialic acid was associated with these mucin molecules and subsequent functional analysis showed that these vesicles have a neutralizing effect on human influenza virus, which is known to bind sialic acid. Taken together, these findings suggest that airway epithelial cells release exosome-like vesicles and that these structures may be involved in diverse physiological processes in airway biology, including innate mucosal defense.— Kesimer, M., Scull, M., Brighton, B., DeMaria, G., Burns, K., O’Neal, W., Pickles, R. J., Sheehan, J. K. Characterization of exosome-like vesicles released from human tracheobronchial ciliated epithelium: a possible role in innate defense.

Biological surfaces exposed to the external environment, such as the human respiratory tract, are often protected by mucus, the first component of the local innate defense system. Mucus consists not only of glycosylated mucin proteins, but also a network of enzymes and scavenging proteins that adhere to, immobilize, destroy, and/or remove a range of foreign bodies, toxins, and pathogens that might otherwise pose a risk to the host. Indeed, many of the proteins associated with innate defense in the respiratory tract are found within mucus gels (1, 2). In addition to mucus, the surfaces of the airway epithelial cells are decorated by a dense glycoconjugate layer called the glycocalyx. This layer consists mainly of large glycoproteins, called epithelial mucins, as well as other glycoproteins, glycolipids, and proteoglycans. Sialic acid residues appear as terminal sugars attached to the epithelial surface glycoconjugates such as proteoglycans, glycolipids, mucins, and other glycoproteins and are known to play a crucial role in the pathogenesis of certain pathogens, as well as many physiological processes. However, since glycocalyx layers are often underrepresented on cell lines, the interactions of these layers present in the respiratory tract in vivo, with pathogens such as respiratory viruses, are not well defined.

Cells shed their surface membranes as microvesicles in normal and pathological conditions such as shear and oxidative stress, hypoxia, osmotic shock, and pathogenic insult (3,4,5). Microvesicles are also released from cells via endosomal vesicle/multivesicular body (MVB) pathways by fusion with the plasma membrane (6). These vesicles, termed exosomes, and are smaller and more homogeneous in size (30–100 nm) than membrane-shed vesicles (100–1000 nm). Exosomes are secreted by different cell types, including epithelial cells (7, 8), hematopoietic cells (reticulocytes; ref. 9), dendritic cells (10), B cells (11), T cells (12), mast cells (13), platelets (4), microglia (14), and some tumor cells (15, 16). In addition, exosomes have been isolated and characterized from different biological fluids, such as urine (17), BAL fluid (18), serum (19), and human breast milk (20). The molecular organization of these structures depends on the source from which they are derived. Apoptotic cells have also been shown to shed membrane particles (21), although no proteomic or electron microscopy (EM) evidence shows whether these particles have a vesicular structure. Thery et al. (10) later demonstrated that exosomes derived from dendritic cells are clearly distinct from apoptotic cell vesicles by their appearance, protein composition, and floatation on sucrose gradients. Their observations indicate that vesicles from apoptotic cells contain abundant amounts of histones, float at high sucrose density (1.24–1.28 g/ml), are larger and heterogeneous in size, and are strongly stained by uranyl acetate in EM (10). Furthermore, exosome release is constitutive and is not dependent on apoptotic processes (8).

Although the function of exosomes remains largely unknown, previous studies suggest that exosomes may be involved in a broad range of biological processes, including stimulation of the immune system and modulation of selected cellular activities (see reviews, refs. 22,23,24,25). Additional work has demonstrated that mouse and human mast cell-derived exosomes contain both mRNA and microRNA, which can be delivered to and function within another cell (26). This novel finding suggests another function to the exosomes, i.e., intercellular communication.

In this study, we identify and characterize exosome-like vesicles isolated from the apical secretions (mucus) of an in vitro model of the human ciliated airway epithelium. The HTBE exosome-like vesicles described here represent highly organized structures, possibly arising from different epithelial cell types found in these specialized cultures, that contain RNA, and a proportion have mucin coats composed of known airway tethered mucins. The identification of α-2,6-linked sialic acid on the vesicle surface, and the known role for this sialic acid residue in mediating influenza virus infection, led us to test whether vesicles may interact with the influenza virus. We show that human airway-derived exosome-like vesicles can neutralize human influenza virus infection, which suggests a potential role for exosomes in mucosal innate defense.

MATERIALS AND METHODS

Cell culture

Primary airway epithelial cells were isolated by the University of North Carolina (UNC) Cystic Fibrosis Center Tissue Culture Core and expanded on plastic to generate passage 1 cells before being plated at a density of 600,000 cells/well on permeable Transwell-Col (T-Col; 24 mm diameter) supports (27). Human tracheobronchial epithelial cultures were generated by provision of an air–liquid interface for 4–6 wk to form well-differentiated, polarized cultures that resemble in vivo pseudo-stratified mucociliary epithelium (28). Mucus secretions were obtained by performing two sequential 1 ml PBS washes on the apical surface of the cultures. Each wash was collected following a 30 min incubation at 37°C. Culture washings obtained from 10 individual cultures were pooled and centrifuged at 3000 g for 10 min to remove shed cells. Washings were subsequently subjected to isopycnic centrifugation and/or differential sedimentation as described below.

Isolation of vesicles by differential sedimentation

Previously published isolation protocols (7, 10, 17, 26, 29) in other systems that used differential sedimentation were slightly modified to isolate exosomes from HTBE secretions, which contain complex protein content (2) and are viscous in nature. Briefly, pooled HTBE secretions were diluted 1:1 with PBS, centrifuged at 3000 g and 10,000 g to eliminate cell debris and other particles, and subsequently pelleted at 65,000 g. The pellet was then washed with PBS and pelleted again at 100,000 g. This washing procedure was repeated to remove any protein or mucin contaminants, which are abundant in the HTBE secretions. Isolated vesicles were resuspended in PBS and filtered twice through 0.22-μm filters to eliminate impurities such as membrane pieces shed from the airway epithelium, i.e., microvilli and ciliated surfaces.

Sample preparation for MS analysis

The filtered vesicle pellet was resuspended in 50 mM ammonium bicarbonate with a mass spectrometry (MS)-compatible detergent (RapiGest; Waters, Milford, MA, USA) at the final concentration of 0.1 w/v. Samples were reduced with 10 mM dithiothreitol (DTT) for 2–3 h at 37°C. Free thiols were alkylated with 30 mM iodoacetamide for 1 h in the dark at ambient temperature. Reduced and alkylated samples were then digested with proteomic-grade trypsin (0.5 μg; Sigma, St. Louis, MO, USA) for 18 h at 37°C. The resulting peptides were dried down 10 times by volume, using a Heto vacuum concentrator (Heto-Holten, Allered, Denmark), mixed 1:1 with 1% formic acid, and subjected to nano-LC-ESI-MS/MS analysis.

MS

Digested samples (2 μl) were introduced into a Waters Q-Tof micro mass spectrometer via a Waters CapLC system, which was configured with a PepMap™ C18 (LC Packing, 300 μm ID × 5 mm) preconcentration column (Dionex, Bannockburn, IL, USA) in series with an Atlantis® dC18 NanoEase™ (75 m × 150 mm) nanoscale analytical column (Waters). Samples were separated on the column using a gradient of 5% acetonitrile in 0.1% formic acid to 60% acetonitrile in 0.1% formic acid over 45 min. All data were acquired using Masslynx 4.0 software (Waters). All analyses were repeated twice for each sample, and peptides identified in the first run were excluded from the second analysis. The processed data were searched against the National Center for Biotechnology Information (NCBI; Bethesda, MD, USA) nonredundant protein database (version 20070805; 2,739,666 sequences) and Swiss-Prot (Release 48.7; 190,255 sequences) using an in-house MASCOT (Matrix Science, London, UK) search engine (Version 2.0). Parameters used for the MASCOT search were Taxonomy human: 0.2 Da mass accuracy for parent ions and 0.3 Da accuracy for fragment ions. One missed cleavage was allowed, and carbomidomethyl-Cys and methionine oxidation were used as fixed and variable modifications, respectively. If a peptide was assigned to multiple isoforms or multiple entries of a protein, we specify only the major form of the protein unless a specific peptide points to a region of the protein, which exists only in one of the isoforms.

CsCl density gradient of vesicle extract

The purified vesicle preparation was treated with 4 M GuHCl plus detergent (0.1% triton), and CsCl was added to the sample to make the density 1.35 g/ml. Isopycnic density-gradient centrifugation was performed for 60 h at 100,000 rpm on a Beckman TL-100 benchtop ultracentrifuge using TLA-100.2, 10 × 1 ml rotor (Beckman Coulter, Fullerton, CA, USA). The sample was emptied as 0.1 ml fractions from the top of the tube, and the fractions were then subjected to agarose gel electrophoresis as described below.

Gel (agarose/SDS-PAGE) electrophoresis, immunoblotting

The vesicle pellet was resolubilized in 8 M urea reduction buffer (25 mM tris; 2% SDS, pH 8.0; 5 mM DTT), boiled for 5 min, and then separated on a precast 4–20% SDS-PAGE gel (Pierce, Rockford, IL, USA) using an Amersham Biosciences Hoefer se 260 electrophoresis apparatus (Amersham, Piscataway, NJ, USA). For immunoblotting, gels were equilibrated in transfer buffer [250 mM Tris, 20% methanol (v/v), 200 mM glycine, 0.1 SDS (w/v), pH 8.0] 15 min prior to transfer onto nitrocellulose membranes. Transfer was performed via semidry electrotransfer with an Amersham Biosciences Multiphore II NovaBlot unit, according to manufacturer’s instructions. After transfer, the membranes were probed with antibodies specific for the following: MUC1 (B2729, gift from Fujirebio Diagnostics Inc., Malvern, PA, USA, which reacts with glycosylated MUC1 and recognizes a region of the MUC1 core protein; and ab37435, Abcam, Cambridge, MA, USA, which targets cytoplasmic tail); EBP 50 (ab3452; Abcam), CD133 (ab16518; Abcam), annexin II (ZO14; Invitrogen, Carlsbad, CA, USA), CD63 (215-020; Ancell, Alexis Biochemicals, San Diego, CA, USA), and Tsg101 (T5951, Sigma). To see charge dispersion of mucins, agarose gel electrophoresis was performed in 1% (w/v) agarose gels as described previously (30). After electrophoresis, the molecules were Western blotted onto nitrocellulose membranes by vacuum transfer and then probed for mucins with antibodies specific for MUC1 (B2729), MUC4 (ab60720; Abcam), and MUC16 (OC125; Cell Marque, Rocklin, CA, USA). Immunodetection was performed using infrared dye-labeled secondary antibodies and visualized using a Li-Cor Odyssey infrared detection system according to the manufacturer’s protocol (Li-Cor Biosciences, Lincoln, NE, USA).

Light scattering and refractive index analysis of the vesicle preparations

Preparations were chromatographed on a Pharmacia Sephacryl S1000 column (15×2.5 cm; Amersham) eluted with 0.2 M NaCl at a flow rate of 400 μl/min. The column effluent was passed through an inline Dawn DSP laser photometer coupled to a Wyatt/Optilab 903 inferometric refractometer (both Wyatt Technologies, Santa Barbara, CA, USA) to measure light scattering and absolute sample concentration, respectively. Light scattering measurements were taken continuously at 18 angles between 15° and 151°; the captured data were integrated and analyzed with the Astra software provided with the Dawn photometer.

Sucrose gradient of the vesicles

A linear sucrose gradient (0.25–2.0 M) in 0.2 M HEPES (pH 7.4; adjusted with NaOH) was created using gradient maker into 12 ml ultraclear SW 40 centrifuge tubes. The vesicle pellet was resuspended in 0.5 ml of 0.125 M sucrose in HEPES and layered onto the gradient. The gradients were centrifuged overnight (∼16 h) at 210,000 g at 4°C, in an SW-40 swinging bucket rotor. Fractions (1 ml) were unloaded from the top, and the densities of the fractions were determined by refractive index measurements. Fractions were then diluted by adding 10 ml of 0.2 M HEPES buffer and recentrifuged to pellet vesicles at 100,000 g, 1 h. The pellets were resuspended with 8 M urea/2 M thiourea/2% SDS extraction buffer, then subjected to either electrophoresis, or antibody analysis by slot blotting.

Immuno-EM

EM and immuno-EM analysis of vesicles were performed as described previously (7, 29). Briefly, the vesicle pellet was resuspended in 4% PFA and deposited onto formvar/carbon-coated EM grids. The vesicle-coated grids were washed twice with PBS (3 min each), twice with PBS/50 mM glycine, and finally with PBS/0.5% BSA (10 min); stained with 2% uranyl acetate; and then viewed for transmission EM (TEM) using a Zeiss EM900 (Carl Zeiss, Oberkochen, Germany). For the immuno-gold labeling with antibodies, blocked grids were transferred to a drop of the antibody in PBS/0.5% BSA and incubated for 30 min. The grids were then washed with PBS/0.5% BSA 5 × 3 min, incubated with gold-labeled secondary antibody in PBS/0.5% BSA for 30 min,and then washed 5 × 3 min in 100 μl drops of PBS/0.5% BSA. The grids were stained with 2% uranyl acetate and then viewed for TEM using a Zeiss EM900.

Flow cytometry

Flow cytometry analysis of the HTBE vesicle beads was performed as described previously (29). Briefly, the final vesicle preparation was incubated with 4-μm-diameter aldehyde/sulfate beads (Interfacial Dynamics, Eugene, OR, USA). After adsorbing the vesicles, the beads were subjected to glycine, followed by the addition of BSA (0.5% in PBS) to saturate any remaining free binding sites on the beads. HTBE vesicles were stained with antibodies against MUC1, MUC4, MUC16, keratan sulfate (5D4; Seikagaku Kogyo, Tokyo, Japan), MHC class I (Joan-1; NeoMarkers, Lab Vision, Fremont, CA, USA) and class II (LN3; NeoMarkers, Lab Vision), CD59, CD133, CD63 (215-020; Ancell, Alexis Biochemicals), and lectins specific for α-2,6- and 2,3-linked sialic acid (SNA and MAA, respectively; Vector Lab, Burlingame, CA, USA). FITC-labeled anti-mouse IgG (Sigma) and streptavidin-labeled (Vector) or Alexa Fluor® 647-labeled (Invitrogen) anti-rabbit secondary antibodies were used. C-Myc (9E10) IgG1 and Ron (C-20) rabbit polyclonal IgG antibodies (Santa Cruz Biotechnologies, Santa Cruz, CA, USA) were used as isotype-matched controls. Samples were analyzed by a CyAn ADP flow cytometer (Beckman Coulter), and the data were analyzed using Summit 4.3 software (Dako Colorado, Inc., Fort Collins, CO, USA). A minimum of 1 × 10⁴ beads per sample was examined.

Isolation and detection of RNA

We followed procedures described by Valadi et al. (26) to determine whether RNA is a structural part of HTBE exosome-like vesicles. In brief, RNA and proteins were isolated using Trizol (Invitrogen) and RNeasy mini kits (Qiagen, Valencia, CA, USA) according to the manufacture’s instructions. Detection of RNA was performed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The cDNA was synthesized using reverse transcriptase (Invitrogen) in the presence of radioactive α-³²P-CTP, 10 mCi/ml (Amersham), according to the manufacturer’s instructions. Total cDNA was subjected to a 0.8% agarose gel electrophoresis. The gel was subsequently transferred to a Hybond N⁺ membrane (Amersham) overnight, then visualized using film autoradiography exposure. To confirm that the RNA was present only inside the vesicles, the HTBE exosome-like vesicles were treated with 0.4 μg/μl RNase (Roche, Nutley, NJ, USA) for 10 min at 37°C prior to isolation.

Inhibition assay for influenza virus-vesicle interaction

To determine whether HTBE exosome-like vesicles neutralized influenza A virus, we incubated recombinant H3N2 human influenza virus A/Victoria/3/75 with purified vesicle preparations and performed standard plaque assays on Madin Darby canine kidney (MDCK) cell monolayers. Briefly, influenza A virus stocks were diluted from 10⁷ to 10³ plaque-forming units (PFU)/ml in PBS. Each viral dilution (100 μl) was then mixed with 100 μl of either vesicle preparation (equivalent to 10 μg vesicles), sialidase pretreated exosomes, PBS + BSA (50 μg/ml), or PBS alone, then incubated at room temperature for 20 min. Each mixture was then serially diluted (10-fold dilutions) in serum-free DMEM and applied to MDCK cell monolayers for 1 h at 37°C. Inoculum was then removed, and the cells were treated with 2% agar-based overlay medium. Three days following overlay, MDCK cell monolayers were fixed and stained with crystal violet in order to visualize and quantitate plaques. Sialidase (neuraminidase from Streptococcus; Seikagaku Kogyo) treatment of the vesicles was performed for 30 min at 37°C according to manufacturer’s recommendations.

RESULTS

Initial identification of HTBE vesicles

The existence of vesicular structures in HTBE culture secretions was originally suggested by immunofluorescent and EM studies investigating tethered mucin MUC1 localization in human tracheobronchial epithelial cultures. These data revealed that MUC1, though mainly located on microvilli, was also strongly associated with punctuate supramolecular structures secreted into the mucus (Fig. 1a). EM studies designed to identify these molecular structures suggested the presence of vesicular structures of varying sizes that were closely associated with the cilia and the microvilli (Fig. 1b).

Figure 1.

Sections of HTBE cell cultures. a) Transverse section of an HTBE cell culture stained with monoclonal B2729 to MUC1. Thin band of staining at the interface (white arrowhead) highlights the presence of MUC1 in the glycocalyx surrounding the microvilli, but MUC1 expression is not strongly associated with the surrounding cilia. Clearly, MUC1 also releases into the mucus above the cilia, where it appears in both dispersed cloud-like structures or in highly particulate or punctuate structures (white circle). Small solid arrow denotes extent of cilia; larger arrow indicates overlayered mucus (20–30 μm). Schematic (right) represents height of the cell, microvilli and cilia. b) Section through the cell culture with mucus layer, fixed with 1% OsO₄ in FC-72 and stained with 2% uranyl acetate/lead citrate, then viewed by EM. Figure indicates the presence of different-sized vesicles (arrowhead) in the periciliary region of the cells, together with cilia (solid arrows) and microvilli (broken arrows). Scale bars = 5 μm (a); 250 nm (b).

Isolation and characterization

To isolate and further characterize the vesicular structures observed in Fig. 1, we utilized a classic exosome isolation protocol that exploits their differential sedimentation properties (see Materials and Methods). Electron microscopic analysis of the vesicles isolated by sedimentation indicated spherical, membrane encapsulated particles. A typical electron micrograph of stained vesicles (Fig. 2) illustrates diameters varying between 30–90 nm. In general, the appearance of the vesicles was similar to those reported from other systems (7, 8, 18, 20, 31, 32) in which the vesicles have been described as being exosomes. A dense staining around some of the vesicles indicates the presence of a filamentous halo of charged material ∼10–30 nm in thickness surrounding the vesicles.

Figure 2.

Electron microscopic morphological analysis of vesicles. Isolated HTBE vesicles were fixed and deposited onto formvar/carbon-coated copper grids, as described in Materials and Methods, and stained with 2% uranyl acetate. Vesicles were then viewed for TEM using Zeiss EM900. Vesicles display a cup-shaped morphology with an average 50–100 nm size characteristic of exosomes. Dense staining around some vesicles indicates possible carbohydrate decoration ∼10–30 nm surrounding vesicles (arrows). Arrowheads indicate vesicles that show no such dense staining. Inset: immunogold labeling of MUC1 mucin indicates that some vesicles, especially those showing intense staining, carry that molecule. Scale bar = 100 nm.

Molecular weight, size, and concentration analyses were performed on the large biomolecules present in the supernatants of the 10,000 and 100,000 g steps during sedimentation as well as on the resolubilized pellet of the vesicular preparation. These samples were analyzed by inline light-scattering measurements after gel chromatography on a Sephacryl S1000 column eluted with 0.2 M NaCl (Supplemental Fig. 1). The average M_r for the distribution of the S1000 void of the 10,000 g supernatant was 39 × 10⁷, and radius of gyration (Rg) was 242 nm with a concentration of 1200 μg/ml. The M_r of 100,000 g supernatant was 8.6 × 10⁷, with 154 nm Rg and 960 μg/ml concentration. The vesicle pellet had M_r of 3.3 × 10⁷, with 100 nm Rg and 250 μg protein per milliliter of original volume of HTBE secretion average concentration (Supplemental Fig. 1, inset table). These analyses confirmed that both their size and mass were consistent with exosome-like particles.

We next identified proteins associated with HTBE exosomes by digesting vesicle preparations with trypsin (see Materials and Methods) and subjected them to LC-MS/MS analysis. A subsequent MASCOT search identified ∼40 proteins, listed in Supplemental Table 1 according to their possible function. The most abundant proteins identified in HTBE secretions (2), including MUC5B mucin, LPLUNC1, and complement component C3, were not observed in the vesicle preparation, indicating that a pure vesicle preparation was obtained. The membrane-bound mucin MUC1 was robustly observed (5 peptides) in the preparations, while the detection of MUC4 and MUC16, although present, were preparation dependent. Our results reiterate the presence of numerous, previously described, exosomal proteins from other systems, such as annexins, tetraspanin, 14-3-3 proteins, and the chloride intracellular channel protein. Typical exosome-associated cytoskeletal proteins such as actin, ezrin, radixin, moesin, and tubulin were also identified, in addition to cytosolic enzymes and several calcium-binding proteins. However, some proteins identified in our analysis have not been identified previously in exosomes, such as MUC4, MUC16, brain acid soluble protein 1, calcyphosine, and calmyrin. Unlike the analyses of exosomes derived from some other systems (12, 16, 33,34,35), MHC class I and MHC class II proteins were not detected in HTBE exosome-like vesicles by MS analysis.

The presence of the membrane-bound mucins in our MS analysis corroborated our initial observation of MUC1 localized to punctuate supramolecular structures secreted into the mucus. To visualize further this particular mucin in association with purified HTBE exosomes, we utilized immunogold labeling techniques in combination with antibody against MUC1 (B2729). The detection of colloidal gold on the outer layer of some of the exosomes confirmed the presence of MUC1 (Fig. 2, inset). We next asked whether the mature glycosylated mucin was the dominant form in our exosome preparations, or whether we were detecting fragments from the membrane-bound C-terminal domain. High buoyant density in CsCl gradients is characteristic of mature glycosylated mucins. Thus, to determine whether the mucin proteins in our vesicle preparations were of the mature form, we dispersed vesicle preparations in 4 M GuHCl plus triton and centrifuged them as described in Materials and Methods. Western blotting of selected gradient fractions following agarose gel electrophoresis confirmed the enrichment of MUC1 (Fig. 3a), MUC4 (Fig. 3b), and MUC16 (Fig. 3c), in the dense fractions. All three mucins exhibited electrophoretic behavior typical of mature, densely glycosylated mucin molecules in agarose gel (30). Since sialic acid-bearing molecules (such as mucins) are key determinants for interactions between certain airway pathogens and the airway epithelial cell surfaces, we also blotted for α-2,6- and α-2,3-linked sialic acid. The presence of α-2,6-linked sialic acid residues on the mucins was confirmed in this experiment (Fig. 3d, e), although some α-2,6 was also detected in the lower density fractions, consistent with the presence of other glycoconjugates (e.g., glycolipids). Very little α-2,3-linked sialic acid reactivity with Maackia amurensis agglutinin (MAA) was detected (data not shown).

Figure 3.

Identification and characterization of epithelial mucins in the vesicle preparation. Vesicle pellets were solubilized in lysis buffer containing detergents (0.1% triton+0.1% Rapigest) in 6 M GuHcl and subjected to density-gradient centrifugation after reducing as described in Materials and Methods. a–d) Aliquots from selected (odd-numbered) fractions were subjected to 1% (w/v) agarose gel electrophoresis. After electrophoresis, the gel was transferred to nitrocellulose by vacuum blotting, prior to detection with MUC1 (B2729) (a), MUC4 (ab60720) (b), MUC16 (CA125) (c), and SNA lectin (α-2,6-linked sialic acid) (d). Arrowhead indicates position of the well. Also shown are migration positions of the low-charged glycoform mucins (solid circle) and the high-charged glycoform of the mucins, as well as other glycoproteins and immature mucin precursors (asterisk) and the dye front (df) where all the globular proteins and other small molecules migrate. e) Aliquots from selected (odd-numbered) fractions were also analyzed for SNA lectin (circles) and MUC1 reactivity (dotted lines) after slot blotting. The density of each fraction was determined by refractometry (triangles).

The presence of additional membrane (CD133, CD63), cytoskeletal (EBP50), and cytoplasmic (Annexin II, MUC1_cyto) proteins was confirmed by SDS-PAGE analysis of vesicle preparations (Fig. 4). Notably, the multivesicular body (MVB) marker Tsg101, which indicates an endosomal origin of the vesicles, was clearly identified by Western blot analysis (Fig. 4). Tetraspanin CD63 revealed a broad smeary band, which disappeared after treating with DTT and boiling (data not shown). The presence of the cytoplasmic domain of MUC1 (MUC1_cyto) as compared to the glycosylated domain is seen in Fig. 4, lanes 1 and 2. Clearly the treatment (SDS, DTT, and boiling) used separated these two regions of the molecule. A MUC1 antibody that recognizes only the cytoplasmic domain indicated the presence of considerable heterogeneity in size, which is perhaps consistent with different cleavage fragments and/or interactions with other proteins. The absence of MHC class I and II proteins in the MS analysis was also confirmed by Western blot analysis (data not shown).

Figure 4.

Western blot identification by SDS-PAGE of some proteins found in vesicles. Proteins were separated on a 4–20% gradient SDS-PAGE gel under reducing conditions. The gel was Western blotted onto nitrocellulose membranes and probed with antibodies against the extracellular domain of MUC1 (B2729) (lane 1), cytoplasmic MUC1 (lane 2), EBP50 (lane 3), CD133 (lane 4), Annexin II (lane 5), and Tsg101 (lane 6). Positions of molecular markers are indicated (kDa).

Investigation regarding surface expression of both mucins and additional proteins on the HTBE exosome-like vesicles was performed by flow cytometry analysis. Vesicles showed a very strong signal for α-2,6-linked sialic acid (detected by elderberry bark lectin binding) and MUC1. Strong signals for MUC4 and MUC16 mucins were detected, and keratan sulfate, CD59, and CD63 were also enriched in the vesicles, while CD133 reactivity was slightly positive (Fig. 5). We did not detect any MHC class I and class II molecule reactivity (Fig. 5).

Figure 5.

Flow cytometry analysis of surface membrane proteins of HTBE vesicles. Vesicles were embedded into 4-μm beads, then stained with a panel of specific monoclonal antibodies for MUC1, MUC4, MUC16, keratan sulfate (5D4), CD59, CD63, MHC class I and II, polyclonal antibody against CD133, and lectin specific for α-2,6-linked sialic acid (SNA). Specific secondary antibodies coupled with FITC (anti-mouse) and Alexa fluor 647 (anti-rabbit) were used for detection (filled black) and compared with isotype controls (filled gray). Results are shown as one representative experiment of six different vesicle preparations.

Sucrose isopycnic density centrifugation was used to assess the density and homogeneity of the vesicle preparation. The vesicle preparations were resuspended in 0.125 M sucrose in 0.2 M HEPES buffer, pH 7.4, then layered onto 0.25–2.0 M continuous sucrose gradient, and spun for 14 h (see Materials and Methods). The distributions of MUC1, CD63, and Tsg101 were analyzed by immunostaining after slot blotting onto nitrocellulose membranes (Fig. 6a) and distribution of β-actin and MUC1_cyto by Western blotting after SDS-PAGE (Fig. 6b). CD63, Tsg101, and actin reactivity revealed a broader distribution around fractions 6–9, at the corresponding densities of 1.12–1.18 g/ml. However, MUC1 reactivity showed a sharper distribution centered around fractions 8 and 9, with the buoyant densities of 1.16–1.18 g/ml. This is consistent with the densities of typical exosomal preparations from other systems (7, 20, 29).

Figure 6.

Sucrose isopycnic density centrifugation of vesicle preparation. Vesicle preparation was resuspended in 0.125 M sucrose in 0.2 M HEPES buffer, pH 7.4, then layered onto 0.25–2.0 M continuous sucrose gradient and spun for 14 h (see Materials and Methods). a) Distributions of MUC1, CD63, and Tsg101 were analyzed by immunostaining after slot blotting onto nitrocellulose membranes. Results are shown as means ± sd; n = 4. b) Distribution of β-actin and MUC1_cyto was analyzed by Western blotting after SDS-PAGE.

In a recent study, Valadi et al. (26) showed that exosomes derived from mouse and human mast cell lines contained both mRNA and miRNA, although DNA was absent. The HTBE-derived exosome-like vesicles used in this study were subjected to a similar analysis. Vesicles were treated with RNase, washed, and thereafter extracted and subjected to analysis as described in Materials and Methods, and the RNA was quantified by bioanalyzer (Fig. 7b). The RNA analysis of vesicle pellet without RNase treatment is also shown as Fig. 7a. The main exosomal RNA population detected was around 200–400 nucleotides. No ribosomal RNA (18S- and 28S-rRNA) was detected. The analysis of the cDNA after conversion of the exosomal RNA revealed cDNA synthesis capability (Fig. 7c).

Figure 7.

RNA detection and transcribed cDNA analysis of HTBE exosomes. a, b) RNA measurement of exosomal pellet (a) and RNase-pretreated exosomal pellet (b). HTBE exosomes contain a substantial amount RNA, most within 200–400 nt, with no ribosomal RNA. c) To test whether the exosomal RNA is sufficient to support cDNA synthesis, mRNA was converted to cDNA using polydT primed reverse transcriptase synthesis in the presence of radioactive dCTP. Product was separated on a 0.8% agarose gel and visualized by film autoradiography.

Inhibition of human influenza A virus infection of MDCK cells

Based on our observations that vesicles bear sialic acids, we hypothesized that HTBE exosome-like vesicles may interact with and potentially neutralize respiratory pathogens that utilize sialic acid to facilitate infection of target cells. To this end, we used human influenza A virus, which binds preferentially to α-2,6-linked sialic acid, as a biological probe to assess pathogen interaction with HTBE exosomes. Human influenza A virus was combined with vesicles and then plaqued on MDCK cells to determine changes in infectivity by reduction of input viral titer. Titer of human influenza virus, A/Victoria/3/75, was reduced after incubation with vesicles but not PBS alone (data not shown) or PBS + BSA (to control for the amount of protein present in the exosome preparation), which suggests that this virus was neutralized by binding to vesicles (Fig. 8a). Indeed, vesicles completely inhibited up to 10² PFU of virus and reduced higher initial viral doses by 85–99% (Fig. 8b). To determine whether this vesicle-mediated inhibition of viral infection was sialic acid dependent, we pretreated vesicles with an exogenous neuraminidase with broad-spectrum sialidase activity to remove sialic acid before combining them with virus. As shown in Fig. 8a, output titers for virus mixed with PBS + BSA and those mixed with neuraminidase-treated vesicles were the same, indicating that no neutralization occurred when the sialic acid was removed from the vesicles (Fig. 8a).

Figure 8.

Neutralizing effect of exosomes on influenza virus (A/Victoria/3/75) infectivity. a) Bar diagram compares analysis of raw data from plaque assay. Exosome-treated influenza viruses (patterned bars) grow 2–3 orders of magnitude less than PBS + BSA-treated viruses (black bars). No neutralizing effect was observed on neuraminidase-pretreated exosomes (gray bars). b) Bar diagram indicates 85–100% inhibition effect of exosomes on influenza virus, depending on virus dilution. Percentage inhibition is calculated as [(viral titer_PBS+virus) – (viral titer_{exosome+virus})]/viral titer_PBS+virus) × 100. Data show means ± sd of 4 independent experiments.

DISCUSSION

In this study we demonstrated that human tracheobronchial epithelial cells release vesicles into the apical secretions. We have also provided evidence that confirms the vesicles are exosome-like rather than apoptotic in character, i.e., they are excreted into airway secretion by fusion of the multivesicular bodies with the apical plasma membrane of the airway surface epithelial cells. The following evidence supports this assertion: HTBE vesicles display typical characteristics of documented exosomes, including size (30–100 nm), shape (cup-shaped) (7, 10, 17, 36, 37), and buoyant density in sucrose (1.12–1.18 gm/ml) (7, 29). They have characteristic exosomal surface and cytoplasmic and cytoskeletal proteins, i.e., CD59 (38), MUC1 (17, 20), tetraspanin CD63, TSG101, actin, ezrin, and radixin, as well as membrane fusion proteins, e.g., annexins (14, 26, 37, 39). These features, combined with the absence of DNA and related proteins such as histones and other indicators of apoptotic cell derived vesicles (10), are all consistent with this conclusion.

Proteomic and flow cytometric analyses revealed that HTBE vesicles expressed the complement inhibitory protein CD59, which suggests a possible role of vesicles in complement regulation. We found no detectable expression of Na/K ATPase and transferrin receptor (CD71) as markers of the basolateral membranes. Also, the endoplasmic reticulum (ER) marker calnexin was not found, which suggests that the vesicles were not derived from the ER. In addition, biochemical analysis revealed the presence of some membrane transport and channel proteins, such as chloride intracellular channel protein 1, the sodium-dependent phosphate transporter, and the amino acid transporter ATB0⁺. Still, one of the crucial membrane transport proteins expressed on the airway epithelial apical surfaces, cystic fibrosis transmembrane conductance regulator (CFTR) (40), was not detected by proteomic or other biochemical analyses, such as Western blot and flow cytometry (data not shown). Some membrane fusion proteins, such as Annexin A1 and AII, major components of fusogenic endosomal vesicles (41), and Rab GDP dissociation inhibitor beta, were present in the proteomic analysis, indicating that they originated from endosomal compartments. Several cytoskeletal-related proteins, such as tubulin and actin and its interacting/organizing proteins (i.e., cofilin, ezrin, radixin, and EBP50), were clearly identified. The presence of enzymes involved in intracellular metabolism (i.e., dehydrogenases, kinases, transferases, and enolase) inside the vesicular lumen suggest their inward invagination during intracellular formation. as previously identified (7, 14, 37). Identification of ubiquitin in the HTBE vesicle preparations suggested that some ubiquitinated proteins were present in the vesicles, as previously noted in B-cell- and dendritic cell line-derived exosomes (42), possibly derived from cytoplasmic proteins incorporated into the MVB pathway. Moreover, the presence of the late endosomal membrane protein CD63, a commonly used marker of exosomes (26) and ESCRT-1 (endosomal sorting complexes required for transport-1) subunit protein Tsg101, which mediates receptor sorting into MVBs, strongly suggest that the vesicles described here are consistent with fusion of endosomal MVBs with the apical membrane. These vesicles can be contrasted with the membrane-derived microvesicles, which are released from the surface membranes during membrane blebbing and are significantly bigger in size (100–1000 μm) (3) than the vesicles described here. Taken together, we propose that the HTBE vesicles described here should be considered as HTBE exosomes.

Van Niel et al. (7) demonstrated exosome-like vesicles derived from the intestinal epithelial cell (IEC) line, which were similar to previously described exosomes derived from antigen representing cells such as B lymphocytes and dendritic cells. The HTBE exosomes shown here have similar shape, size, and buoyant density and have some common proteins with IEC-derived exosomes. HTBE exosomes differ from IEC-derived exosomes in that they carry no antigen-presenting MHC class I and II molecules. However, the HTBE exosomes described here present different types of surface molecules, such as MUC1 and other membrane mucins, MUC4 and MUC16.

MUC1, MUC4, and MUC 16 are epithelial cell mucins that share a similar transmembrane binding mechanism and are believed to be heterodimeric in their C terminus due to intracellular cleavage mechanisms (see ref. 43 for review). However, they differ dramatically in the size of their glycosylated domains. These molecules have now been identified in a number of studies (1, 2, 44, 45) as being present in lung tissue, but knowledge of their cellular localization is less complete. MUC1 mucin has been identified previously as a part of the exosomes isolated from human breast milk (20) and urine (17). However, to our knowledge this is the first time that MUC4 and MUC16 mucins have been associated with any vesicular and/or exosomal preparations. Our proteomic, Western blot, and flow cytometry analyses also indicate that HTBE exosomes carry MUC1, MUC4, and MUC16 mucins; the detection of the MUC1 mucin is consistent, while MUC4 and MUC16 vary considerably between preparations. These molecules may control or contribute to their recognition properties. Isopycnic density gradient analysis in CsCl of the detergent-treated exosomes indicated that these mucins were found in the high-density region (1.40–1.55 g/ml), consistent with the presence of their heavily glycosylated domains. The cytoplasmic region of the MUC1 mucin, but not MUC4 nor MUC 16, was identified by MS and by immunoblotting after SDS-PAGE, suggesting that MUC1, at least, is present due to membrane tethering to the exosomes. The absence of peptides from the cytoplasmic region of MUC4 and MUC16, as detected by MS, may be due to the shorter cytoplasmic tails for these mucins, containing only a single tryptic peptide as compared with five available in MUC1. It has been proposed that MUC1 mucin plays a role in the regulation of T-cell responses (46) and a modulatory role during airway infection and inflammation caused by various pathogenic bacteria and viruses (47, 48). High expression levels of these mucins on many human carcinoma cells, such as breast, pancreatic, and cervical cancers, make them an attractive target for cancer immunotherapy. For instance, it has been demonstrated that engineered exosomes (to carry MUC1) attract an effective immunological response while suppressing tumor growth in an MHC-independent manner (49) and have been offered as a cell-free anticancer vaccine. It has also been demonstrated that both MUC4 (50, 51) and MUC16 (52) are associated with roles in tumor growth and metastasis and are target molecules for cancer immunotherapy (51, 53).

Both sucrose density gradient ultracentrifugation and EM analysis revealed distinct exosome populations, which suggests that they may arise from different cell types in the airway epithelia. Antibody analysis on surface proteins, and membrane mucins, after a sucrose density gradient, showed at least two different populations characterized by either the presence or absence of epithelial mucins. This observation is also consistent with the EM images that show dense uranyl acetate staining around some of the exosomes, indicating the presence of negative charges consistent with sulfated and sialylated oligosaccharides as potentially presented by mucins.

We observed that HTBE exosomes were highly enriched for α-2,6-linked sialic acid terminal sugars with little to no significant α-2,3-linked sialic acid. Our CsCl isopycnic density gradient centrifugation analysis of the detergent-treated exosomes revealed that α-2,6-linked sialic acid was enriched in a high-density (1.35–1.5 g/ml) region in which MUC1, 4, and 16 mucin immunoreactivity was also observed (Fig. 3). However, some immunoreactivity was detected in the low-density (1.2–1.25 gm/ml) and medium-density (1.25–1.30 gm/ml) regions, suggesting the presence of small molecules carrying α-2,6-linked sialic acid. This finding is consistent with the presence of molecules such as glycolipids and small glycoproteins, as was further confirmed by agarose gel electrophoresis (Fig. 3d, e).

The presence of sialic acid-rich mucins on HTBE exosome structures offered a potential function of exosomes in innate defense, given the known interactions of such moieties with many viruses and bacteria (54, 55). As an initial test of this hypothesis, we tested the effect of mixing exosomes with a viral pathogen, human influenza virus, known to attach to α-2,6-linked sialic acid. Inhibition experiments were performed using 10 μg/200 μl exosomes, which is one-fifth the actual concentration found in the HTBE cultures (with ∼600,000 cells). Indeed, these experiments revealed that exosomal preparations neutralized influenza virus infectivity by up to 85–99%. This high degree of inhibition was ablated when exosomes pretreated with neuraminidase (NA) were used, suggesting that the interaction between exosomes and the influenza virus was sialic acid dependent. Interestingly, viral inhibition also implied that the endogenous viral NA, important for cleaving progeny virions from the cell surface, could not cleave attachment of the virus from the exosome or that the relative rates of association/dissociation favored association. With respect to the relevance of HTBE exosomes in natural infection by human influenza viruses, we could surmise that these structures may have some ability to neutralize low, environmentally relevant exposures to these viruses. The inhibition shown in the present study implies that exosomes can interact with a relevant airway pathogen via sialic acid and that this interaction may impact the ability of this virus to infect target epithelial cells.

Knowledge of the mechanisms of release and functional biology of exosomes from different cell types is still in its infancy. We show here, for the first time, that exosome-like vesicles are present in the HTBE cell culture secretions and demonstrate that they express a range of surface protective proteins, including epithelial mucins. We also confirmed that they contain detectable amounts of RNA that can synthesize cDNA. Our functional analyses indicate that these structures may have a role in innate defense against respiratory pathogens that bind sialic acid, such as human influenza virus. Comparing our findings to the current exosome literature, we propose that these structures may be involved in diverse physiological processes in airway biology, including immunomodulation, inflammation, and bacterial and viral sequestration. The latter is a novel proposed function for exosomes and will lead to a greater understanding of the role of the exosomes in airway biology and innate defense.