Published ahead of print on February 19, 2004, doi:10.1165/rcmb.2003-0238OC
© 2004 American Thoracic Society DOI: 10.1165/rcmb.2003-0238OC
Proteomic Analysis of Exosomes Isolated from Human Malignant Pleural EffusionsDepartment of Pulmonary Medicine, Department of Neurology, and Department of Clinical Genetics, Erasmus MC, Rotterdam, The Netherlands Address correspondence to: M. P. L. Bard, Department of Pulmonary Medicine, Erasmus Medical Centre H-Ee2253a, P.O. Box 1738, 3000 DR, Rotterdam, The Netherlands. E-mail: m.bard{at}erasmusmc.nl
Exosomes are membrane vesicles from endosomal origin secreted by various cells such as hematopoietic, epithelial, and tumor cells. Exosomes secreted by tumor cells contain specific antigens potentially useful for immunotherapeutic purposes. Our aim was to determine if exosomes are present in human cancerous pleural effusions and to identify their proteomic content. Exosomes were purified by sucrose gradient ultracentrifugation, and electron microscopy was used to check both concentration and purity of exosomes. Proteins were separated by one-dimensional sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and protein bands were identified by matrix-assisted laser desorption ionization time-of-flight mass spectrometry and Western blotting. Exosomes were present in pleural fluid obtained from patients suffering from mesothelioma (n = 4), lung cancer (n = 2), breast cancer (n = 2), and ovarian cancer (n = 1). As previously reported by others, antigen-presenting molecules, cytoskeletal proteins, and signal transduction–involved proteins were present. Proteins not previously reported were identified (SNX25, BTG1, PEDF, thrombospondin 2). Different types of immunoglobulins and complement factors were abundantly present in the sucrose fractions containing exosomes. Exosome-directed specificity of these immunoglobulins was not observed. In conclusion, sucrose gradient ultracentrifugation allows isolation of exosomes from malignant pleural effusions. However, pleural fluid proteins and especially immunoglobulins are coisolated and may hamper the use of exosomes isolated from malignant effusion for immunotherapy programs.
Abbreviations: B-cell translocation gene, BTG • dendritic cell, DC • electron microscopy, EM • immunoglobulin, Ig • heat shock protein, HSP • matrix-assisted laser desorption ionisation time-of-flight, MALDI-TOF • major histocompatibility complex, MHC • polyacrylamide gel electrophoresis, PAGE • pigment epithelium-derived factor, PEDF • sodium dodecyl sulfate, SDS • sorting nexin, SNX
Exosomes are membrane vesicles from endosomal origin that are secreted by various cells, e.g., hematopoietic, epithelial, and tumor cells. The function of exosomes is largely unknown, but some studies have shed some light on the effect of exosomes on immunity. Exosomes produced by intestinal epithelial cells have the ability to induce an antigen-specific tolerance (1). These "tolerosomes," which carry MHC class II molecules may be used by epithelial cells to induce tolerance to food antigens. In contrast, a tumor mice model study has demonstrated that exosomes secreted by tumor peptide-pulsed dendritic cells (DCs) have the ability to induce a specific anti-tumor response (2). The presence of tumor antigens on tumor-derived exosomes and their ability to be taken up by DCs has been demonstrated in an in vitro human study (3). These studies have demonstrated that exosomes from tumor cells and DCs can be implemented into cancer immunotherapy programs (4, 5). Little information is available on in vivo production and function of tumoral exosomes in humans. Recently, André and colleagues have reported the purification of exosomes from cancerous ascites fluid and demonstrated the induction of a specific anti-tumoral response by autologous exosome-loaded DCs (6). This study confirms that exosomes are secreted in human malignant ascites effusion and demonstrated that exosomes have the capacity to transfer specific tumoral antigens to DCs and to induce a specific anti-tumoral response. The protein composition of in vitro produced exosomes has been studied using Western blotting (2, 7), flow cytometry of exosomes-coated beads (8), and mass spectrometry (9–12). These studies have demonstrated that exosomes from different cellular origins share common groups of proteins (13). These groups are (i) proteins involved in antigen binding and presentation such as heat-shock proteins, MHC class I, and II proteins; (ii) proteins involved in intracellular membrane fusion and transport such as annexins and rab proteins; (iii) proteins involved in targeting and cell adhesion such as tetraspanins and integrin proteins; (iv) cytoskeletal proteins such as actin and tubulin; and (v) metabolic enzymes. These previous studies suggest that exosomes express a limited set of proteins and can be used as an antigen source for immunotherapeutic purpose especially in cancer therapy. The aim of our study was to determine if exosomes are present in pleural effusions of patients suffering from different cancer types involving the pleura and to analyze their protein composition.
Patients Patients with a histologically proven malignancy were asked to participate in the study. All patients were more than 18 yr old and signed informed consent. Indication for pleural effusion evacuation was in most cases exertional dyspnea relief. Clinical characteristics of the patients (n = 9) are summarized in Table 1. Primary tumor sites were mesothelioma (n = 4), non–small cell lung carcinoma (n = 2), adenocarcinoma of the breast (n = 2), and ovarian adenocarcinoma (n = 1). All patients with mesothelioma proved to be nonoperable due to mediastinal involvement or low performance status at presentation. A pleural involvement was the presenting symptom of the two patients with lung cancer and the patient with ovarian cancer. The two patients with breast cancer were initially treated with surgery and locoregional radiotherapy, 6 and 9 yr after diagnosis multiple metastases appeared in these two patients.
Pleural Fluid The pleural fluid evacuation procedure was standard. After a local anesthesia (Lidocaine 1%), a metallic needle was introduced in the pleural cavity through an intercostal space. Pleural fluid was gently aspirated and used immediately for exosome isolation. All fluids were nonhemorrhagic and proved to be exudates. Biochemical characteristics of the pleural fluids are summarized in Table 2. High level of albumin, immunoglobulin, and complement were observed in pleural effusion (Table 2). The microscopical analysis after May-Giemsa staining showed that lymphocytes and tumor cells were the most abundant present cells. Neutrophils, eosinophils, and mesothelial cells were occasionally observed. In all cases, no infectious agent was observed and bacterial cultures were negative.
Sucrose Gradient Ultracentrifugation Exosome isolation was performed as previously described (6). In brief, 200–400 ml of pleural fluid was sequentially centrifuged at 300 x g for 10 min, 2,000 x g for 20 min, and 10,000 x g for 30 min to pellet cells and debris. Exosomes were then pelleted at 64,000 x g for 90 min. Pellet was resuspended in 2 ml 0.32M sucrose containing 10 µl 5 µg/ml phospholipid analog fluorescein-DHPE (N-(fluorescein-5-thiocarbamoyl)1,2-dihexadecanoyl-sn-glycero-3-phosphaethanolamine, triethylammonium salt; F-362: Molecular Probes, Leiden, The Netherlands). Fluorescein-DHPE was used to label the phospholipid bi-layers of both cell and exosome membranes. The fluorescein labeled pellet was layered on a continuous sucrose density gradient (0.32–2.5 M sucrose, 20 mM HEPES, pH 7.2) and centrifuged overnight at 100,000 x g (SW41 rotor; Beckman Instruments [Fullerton, CA]). Two hundred–microliter fractions were collected from the top (low density) to the bottom of the tube (high density) of the sucrose gradient. For each fraction, fluorescence was measured in a fluorescence microplate reader (Bio-Rad Benchmark, Hemel Hempstead, UK; emission 519 nm, excitation at 495 nm). Fluorescent fractions from the gradient were harvested, diluted in phosphate-buffered saline, and centrifuged at 100,000 x g for 60 min (SW60 rotor; Beckman Instruments). Pellets were resuspended in phosphate-buffered saline, aliquoted and stored at –80°C. The final quantification of exosomal proteins was measured by CBQCA kit according to the manufacturer's recommendations (Molecular Probes).
Electron Microscopy Indirect immunogold labeling of exosomes was performed with a goat anti-human immunoglobulin Fc antibody coupled to 10-nm gold particles (Aurion, Wageningen, The Netherlands).
One-Dimensional Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis
Matrix-Assisted Laser Desorption Ionization Time-of-Flight Analysis
Western Blot Analysis
Sucrose Gradient Ultracentrifugation After overnight centrifugation of fluorescent-labeled pleural sample, various yellow-white layers appeared in the sucrose gradient. In all patients, electron microscopic analysis of these fluorescent fractions confirmed the presence of round shaped homogeneous membrane vesicles which fulfilled the exosome definition. As illustrated in Figure 1, some variations in both shape and diameter of exosomes could be observed between the patients. For the same patient, the concentration of exosomes could vary between different fluorescent fractions and were occasionally mixed with cell membrane fragments (data not shown).
One-Dimensional Gel Separation Proteins from fractions containing only exosomes (without membrane fragments) were separated under reducing condition on 7.5% SDS-PAGE. For all patients, it appeared that the patterns of bands on the SDS gel were grossly similar (Figure 2).
In three representative patients (Patients 1, 2, and 3)—suffering from breast cancer, lung cancer, and mesothelioma, respectively—one fraction containing no exosomes (no), one fraction containing a mix of exosomes and cell membrane fragments (mix), and one fraction containing only exosomes (ex) were separated on 7.5% SDS-PAGE and stained by colloidal blue (Figure 3). For each three patients, common bands were observed in fractions with or without exosomes.
MALDI-TOF Analysis According to the common band pattern observed on the 1D gel for all patients, the MALDI-TOF analysis was performed in three representative patients (Patients 1, 2, and 3) suffering from breast cancer, lung cancer, and mesothelioma, respectively. This MALDI-TOF analysis was performed for fractions of the sucrose gradient containing only exosomes without membrane fragments (Figure 4). The results of this analysis are summarized in Tables 3, 4, and 5 for Patients 1, 2, and 3, respectively. Both protein names and accession numbers are related to MSDB databank. The calculated molecular mass was deduced from the mass fingerprint analysis, the apparent molecular mass was measured on the SDS-PAGE gels using molecular weight standard proteins. A facultative modification during the peptide mass fingerprint analysis was the presence of oxidative methionine. The number of peptides matched and the coverage percentage were deduced from the comparison between the identified peptides and the whole primary protein structure. The last column corresponded to the score given by Matrix Science Mascot UK software analysis, which was significant (P < 0.05) when higher than 61. For each protein digest, all significant results were given, and in case of a not significant score only the top-scored protein was indicated.
Surprisingly, an important group of identified proteins was composed of peptides originating from immunoglobulin (Ig) light and heavy chains (IgM, IgG1, IgG3 heavy chains, and Ig kappa light chain) and from various complement factors (C1q, C1r, C4a, H). These proteins corresponded to the large and most intense bands and could be detected in all patients. These immunoglobulins and complement proteins corresponded also to the common bands observed in the sucrose fractions containing no exosomes (Figure 3). Moreover, immunogold-labeled electron microscopy analysis showed that the exosomes were not labeled with the monoclonal antibody to human immunoglobulin Fc component in any patient (data not shown). We identified proteins already reported to be present in exosomes derived from tumor cells or antigen-presenting cells. MHC class I molecules were identified in exosomes from Patient 1 with a peptide coverage of 15% but a nonsignificant Matrix Science Mascot UK score. The cytoskeletal protein actin was identified in Patients 1 and 2. Some proteins involved in signal transduction as G protein and protein kinase were identified in Patients 1 and 2, respectively. In addition, we identified proteins that have not been previously described. Some were related to intracellular membrane trafficking proteins such as sorting nexin (SNX25) protein. Some were related to cell growth and differentiation such as B-cell translocation gene 1 (BTG1) protein and pigment epithelium-derived factor (PEDF), both overexpressed in malignant processes, suggesting a role in tumoral exosome biogenesis. Others were related to extracellular matrix organization and cell–matrix interaction such as bamacan (basement membrane-chondroitin sulfate proteoglycan) protein and thrombospondin-2.
Western Blot Analysis
We isolated exosomes from nine malignant pleural effusions by sucrose gradient ultracentrifugation and used MALDI-TOF mass spectrometry technology and Western blotting to identify their protein composition. We identified MHC class I and II proteins, heat shock proteins, cytoskeletal proteins, and signal transduction–involved proteins, all proteins that have already been reported as constituent of exosomes from other origin. We are the first to report the presence on exosomes of sorting nexin, BTG1, PEDF, bamacan, and thrombospondin 2. Surprisingly, the most abundantly present proteins were immunoglobulins and complement components. Exosomes present in cancerous pleural fluid may have various cellular origins. Cancerous pleural effusion is accompanied by a strong inflammatory response involving both cellular and humoral immunity. The most abundantly reported cells in the pleural effusion of our nine patients were lymphocytes and tumor cells. B-lymphocytes, T-lymphocytes, and antigen-presenting cells such as dendritic cells are important participants of the anti-tumoral immune response, and are all known to produce exosomes (7, 9, 11). Moreover, tumor cells are also able to produce exosomes (3, 6, 12). Although a lot of proteins are commonly present in exosomes secreted by various cell types (13), particular proteins are specifically enriched in exosomes secreted by specialized cells. For example, exosomes secreted by dendritic cells are rich in MHC class II and CD86 protein, both proteins involved in antigen presentation and stimulation of T cells (2). T cell receptors are abundantly present in exosomes secreted by T lymphocytes, suggesting a role of exosomes in intercellular communication (14). Tumor cells express MHC class I molecules and tumor markers (6, 12). Our results suggest that cancerous pleural fluid contains a combination of exosomes from various origins (mostly from lymphocytes and tumor cells). However, we are not able to determine the cellular source of these exosomes with currently available technology. As direct proof that mesothelioma cells can produce exosomes, we have found that cell lines derived from pleural fluid indeed produce exosomes (12). Immunoglobulin peptides were the most common proteins identified in the exosome-containing fractions from the malignant pleural effusions, corresponding to the most intense bands on the SDS-PAGE gel. IgM is the first antibody to be produced in a humoral immune response. IgG is the principal isotype found in blood and extracellular fluid. Both IgG1 and IgG3 can efficiently opsonize pathogens for engulfment by phagocytes and activate the complement system. Complement (C1q, C1r, C4a, and H factors) components were also frequently identified in exosomes from malignant pleural effusion. Both C1 and C4 complement factors are involved in the so-called classical complement pathway. Additionally tumor necrosis factor–stimulated gene 14 (TSG-14) protein is structurally related to C reactive protein and serum amyloid C component both involved in the first response to infections or tumor with the ability to activate complement (15). The importance of the high Ig and complement levels in our exosome preparations is threefold. First, Ig and complement determine and modify the anti-tumoral immune response. Presence of Ig, complement factors, and TSG-14 protein in pleural effusions could be related to the strong immune response occurring in the pleural space of patients with cancer. High concentration of Ig and complement has already been reported in malignant effusions (16–18) and was measured in the pleural fluid of our nine patients. Pleural Ig diffuse mostly from the bloodstream but can also originate from locally stimulated B-lymphocytes (19). Second, ultracentrifugation in sucrose gradient has been shown to be a reliable method for isolation of exosomes from cell culture supernatant (2, 13). This methodology has been recently used with success to isolate exosomes from malignant ascities (6). However, ultracentrifugation in sucrose gradient coisolates Ig and complement proteins. The presence of Ig in the sucrose gradient fractions that contain no exosomes and the absence of exosome labeling with an antibody targeted against Fc component of human Ig argues for the presence of free Ig and not for Ig bound to exosomes. Third, the potential presence of proteins not binding to exosomes in the exosome preparation had no functional consequences for an in vitro autologous cytolytic test as observed by André and colleagues. However, the use of exosome samples containing nonexosomal proteins may have deleterious consequences, especially in the case of heterologous cross-utilization. Induction of polyantigenic immune responses targeted to noncancer proteins could be not only responsible for a decrease of the anti-tumoral effect of the vaccine but could also induce an autoimmune response potentially dangerous to the patient. Use of exosomes as antigen source for cancer immunotherapy is promising. However, risk of presence of contaminating proteins, especially when exosomes are isolated from malignant effusion, must be taken into account before their in vivo use can be generalized. Antigen-presenting molecules such as MHC molecules (class I and II) and heat shock proteins were identified in exosomes isolated from cancerous pleural fluid. These molecules have been commonly identified in exosomes originating from various cells such as B cells, T cells, dendritic cells, and tumor cells (2, 3, 14). Moreover, in a recent report, André and colleagues have confirmed the presence of MHC class I molecules in malignant ascites-derived exosomes by electron microscopic immunostaining and Western blotting (6). Functions of these proteins in exosomes have been related to their capacity to transfer antigens to antigen presenting cells and to induce a specific immune response. Other proteins that we identified have already been reported as constituents of exosomes from other origins (e.g., actin, myosin, G protein, or protein kinase). We described several new proteins not previously described in exosomes by examining exosomes from malignant effusions. Sorting-nexin (SNX) family is a group of hydrophilic proteins implicated in the intracellular trafficking of proteins to various organelles. SNX1 has the capacity to bind membrane receptors such as epidermal growth factor receptor (EGRF), platelet derived growth factor or insulin (20). Moreover, overexpression of SNX1 induced an EGFR decrease on human cell surface suggesting that SNX1 plays a role in sorting EGFR to lysosomes for degradation (21). The presence of an endosomal trafficking protein such as SNX1 in exosomes, which are known to originate from the multivesicular late endosomal compartment, seems a quite likely possibility. Moreover, proteins involved in membrane intracellular transport such as annexins or rab proteins have been previously observed in exosomes secreted by dendritic cells (10). It could then be hypothesized that SNX1 plays a role in the intracellular membrane trafficking to the endosomal and subsequently to the exosomal compartment. Acidic ribosomal phosphoproteins from the 60S subunit of ribosomes, interact with elongation factors EF-1 and EF-2 and play an important role in the elongation step of protein synthesis. Elongation factor EF-1 has already been identified in B cell– and DC-derived exosomes (10, 11). However, various tumor cells like colon or hepatocellular carcinoma have an increased expression of acidic ribosomal phosphoprotein (22). The presence of translation-related proteins in exosomes may therefore also be explained by their high concentration in the cytosol of cancer cells. Identification of extracellular matrix organization– and cell matrix interaction–related proteins as bamacan (basement membrane-chondroitin sulfate proteoglycan) protein or thrombospondin-2 may be related to their high concentration in malignant pleural fluid. Indeed, an increased level of proteoglycans is a known event in epithelial tumors such as breast or lung carcinoma and overexpression of bamacan protein has been reported in cancer cells (23). Thrombospondin 2 is involved in the regulation of proliferation, adhesion, and migration of various cells, and has an inhibitory function of both angiogenesis and tumor growth (24, 25). Thrombospondin 2 interacts with cell surface receptors such as integrin or heparan sulfate proteoglycan (26, 27). A link between thrombospondin 2 expression and tumor progression has been reported in melanoma and breast carcinoma (28, 29). The binding of these extracellular matrix proteins on exosomes may be explained by the presence of matrix-binding proteins like integrins on the surface of exosomes, which have been reported previously (13). However the function of these extracellular matrix molecules in exosomes remains unknown. PEDF is a secreted protein that is expressed by various healthy and tumoral tissues. PEDF belongs to the serine protease inhibitor (SERPIN) gene family and is known to possess anti-angiogenic activities (30). Secretion of PEDF by Schwann cells induced differentiation of neuroblastoma tumor cells, which consecutively secrete PEDF, suggesting an anti-tumor feedback loop with the potential to limit tumor growth (31). The function of BTG1 (B-cell translocation gene 1) protein is incompletely known, but it may be involved in cell growth and differentiation and control of the cell cycle at a transcriptional level (32). Both PEDF and BTG1 proteins are then related to cell growth control, and their expression can be modified in cancerous cells, but their putative role in exosomes remains unclear. We could not identify any integrins, tetraspanins, or annexins in our exosome preparations by the technology used. These proteins have been isolated from exosomes derived from DC and tumor cells. The fact that we do not detect these proteins is probably caused by the relatively poor ability of the 1D SDS-PAGE gel to separate the different proteins. An extra initial step to separate the Ig and complement factors from the exosomes and the use of two-dimensional gel electrophoresis separation or other high resolving techniques may be a way to identify more proteins that are less profusely present. In conclusion, cancerous pleural effusion contain exosomes from various cellular origin. The proteomic analysis of pleural effusion-derived exosomes was marked by the identification of proteins that had not been identified in exosomes before and by a high concentration of antigen-antibody humoral immunity components. The latter might have important consequences both for in vitro and in vivo applications of exosomes that are isolated directly from body fluids.
M.P.L.B. received a research fellowship from the European Society of Medical Oncology. Received in original form June 20, 2003 Received in final form January 30, 2004
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
-cyano-4-hydroxy-trans-cinnamic acid (ACCA; Bruker Daltonics, Billerica, MA) in acetonitrile. From this sample-matrix solution 0.5 µl was pipetted onto a 400-µm 384-well anchorchip matrix-assisted laser desorption ionization (MALDI) plate (Bruker Daltonics, Bremen, Germany) and air-dried for 5 min. Mass spectra were acquired on a Biflex III (Bruker Daltonics) MALDI–time-of-flight (TOF) mass spectrometer equipped with a 337-nm nitrogen laser. A mass list of peptides was obtained for each protein digest with X-tof software and submitted to Matrix Science Mascot (London, UK) software using the most recent MSDB databank of the NCBI to identify the proteins. 
complex. J. Immunol. 168:3235–3241.
