Classification, Functions, and Clinical Relevance of Extracellular Vesicles

1. Classic Pathway of Exosome Formation.

The “classic pathway” of exosome formation is by far the best studied and involves the formation of intraluminal vesicles within MVEs (Fig. 2). In turn, MVEs can fuse with either lysosomes for cargo degradation or with the plasma membrane to secrete the intraluminal vesicles, which are then released as exosomes.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Pathways of exosome formation. Cells release exosomes via two mechanisms. The classic pathway (left) involves the formation of intraluminal vesicles (ILVs) within MVEs. In turn, the membrane of MVE fuses with the plasma membrane, resulting in the release of ILVs. When secreted, ILVs are called exosomes. Alternatively, the direct pathway (right) involves the release of vesicles, indistinguishable from exosomes, directly from the plasma membrane.

Different intracellular sorting pathways exist in directing proteins toward intraluminal vesicles predestined for either degradation or secretion, thus implicating the existence of different types of MVEs. Redundant transmembrane receptors are sorted to intraluminal vesicles predestined for lysosomal degradation after ubiquitination, a post-translational modification that is executed by ESCRT (Babst, 2005; Michelet et al., 2010). On the other hand, there is no compelling evidence that the ESCRT is involved in the sorting of transmembrane receptors to intraluminal vesicles that are predestined to become secreted as exosomes (Théry et al., 2001; Buschow et al., 2005, 2009; Trajkovic et al., 2008; Simons and Raposo, 2009; Tamai et al., 2010; Bobrie et al., 2011). Although several ESCRT proteins (Théry et al., 2001) and ubiquitinated proteins (Buschow et al., 2005) are present in exosomes, the ubiquitinated proteins are soluble proteins and not transmembrane proteins, suggesting that ubiquitination may have occurred in the cytosol rather than by the ESCRT. Likewise, the sorting of the proteolipid protein to intraluminal vesicles predestined to become secreted as exosomes is independent from the ESCRT but depends on the sphingolipid ceramide (Marsh and van Meer, 2008; Trajkovic et al., 2008). Further evidence for two pathways comes from studies showing that lysobisphosphatidic acid induces the formation of intraluminal vesicles predestined for lysosomal degradation but does not affect the formation of exosomes (Matsuo et al., 2004). This is confirmed by studies in which cholesterol is labeled with perfringolysin O, which reveals perfringolysin O-positive and -negative MVEs in B cells, of which only the cholesterol-containing MVEs fuse with the plasma membrane resulting in the release of exosomes (Möbius et al., 2002). In addition, both epidermal growth factor and the epidermal growth factor receptor (EGFR) travel to exosomes via MVEs not containing the lipids bis(monoacylglycero)phosphate/lysobisphosphatidic acid, whereas the bis(monoacylglycero)phosphate/lysobisphosphatidic acid-containing vesicles are degraded by lysosomes (White et al., 2006). Thus, clearly two intracellular pathways of MVE sorting exist.

Cytosolic domains of proteins (Theos et al., 2006) or lipid domains enriched in the tetraspanins CD9 or CD63 are thought to play a role in the sorting of transmembrane proteins toward intraluminal vesicles (Buschow et al., 2009; Bobrie et al., 2011). Several different types of small GTPases from the Rab family play a role in the intracellular trafficking of MVEs toward either the plasma membrane (Savina et al., 2005; Hsu et al., 2010; Ostrowski et al., 2010) or to lysosomes for degradation (Stenmark, 2009), as well as cytosolic calcium levels (Savina et al., 2003; Fauré et al., 2006; Krämer-Albers et al., 2007), citron kinase, (Loomis et al., 2006), and a still unidentified combination of soluble N-ethylmaleimide-sensitive factor attachment protein receptors, which is involved in the final fusion of the MVE membrane with the plasma membrane (Rao et al., 2004).

2. Direct Pathway of Exosome Formation.

Next to the “classic pathway” of exosome biogenesis, there is a second and much more immediate route of exosome formation (Fig. 2). T cells and erythroleukemia cell lines release exosomes directly from their plasma membrane, both spontaneously as well as upon expression of HIV Gag or Nef, or after cross-linking of surface receptors (Booth et al., 2006; Fang et al., 2007; Lenassi et al., 2010). These vesicles are indistinguishable from exosomes formed by the classic endosomal pathway because they are enriched in classic exosome markers such as CD63 and CD81 and have a similar diameter and density. The extent to which such exosomes are also released from other cells or in vivo (e.g., in biological fluids) is unknown.

1. Immune Suppression.

By transporting ligands and receptors (Fig. 3), exosomes can orchestrate cell growth and development, and modulate the immune system. Activated T cells and peripheral blood mononuclear cells release vesicles exposing Fas ligand (FasL), a death receptor ligand, suggesting that these vesicles play a role in cell death during immune regulation (Martínez-Lorenzo et al., 1999). These vesicles are likely to be exosomes because FasL is stored in MVEs in T cells (Monleón et al., 2001). The sorting of FasL to MVEs and exosomes is regulated by phosphorylation and monoubiquitination (Zuccato et al., 2007), and the secretion of FasL-exposing exosomes is mediated by diacylglycerol kinase-α (Alonso et al., 2005, 2011). Intraperitoneal injection of FasL-exposing exosomes triggers apoptosis of macrophages in vivo (Hohlbaum et al., 2001).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Immune suppression. Cells, tumor cells, and EBV-infected cells release vesicles exposing proteins (LMP-1) or death receptor ligands (FasL, NKG2D) that induce T-cell apoptosis.

The ability of vesicles to modulate the immune response is likely to play a role in normal growth and development. For instance, the invading trophoblast is a semiallograft and should in theory be rejected by the mother as “foreign” material. During normal pregnancy, the trophoblast evades the maternal immune system by killing activated T cells that have been sensitized to paternal alloantigens. First-trimester trophoblast cells release exosomes exposing FasL, which are capable of inducing Fas-mediated T-cell death, suggesting that exosomes contribute to this immune privilege (Abrahams et al., 2004; Frängsmyr et al., 2005). Because sera from pregnant women exhibit higher levels of FasL-associated with exosomes at term compared with preterm, it is thought that immune suppression by exosomes may contribute to normal pregnancy and delivery (Taylor et al., 2006). Placenta-derived exosomes also expose several ligands of the natural killer (NK) cell receptor NKG2D, UL-16 binding proteins 1–5, and major histocompatibility complex (MHC) class I chain-related protein, which all down-regulate NKG2D on NK cells, CD8⁺ cells, and γδ T cells, thereby all potentially contributing to fetal immune escape (Hedlund et al., 2009).

The ability of exosomes to mediate cell killing at a distance is also used by tumor cells to evade the immune system. FasL is present in MVEs of melanoma cells and a prostate cancer cell line, and FasL-exposing vesicles from these tumor cells kill Jurkat T cells in vitro. Therefore, tumor-derived vesicles may provide an important front-line defense mechanism by inhibiting the homing and antitumor activity of immunocompetent cells (Andreola et al., 2002; Martínez-Lorenzo et al., 2004; Abusamra et al., 2005). In addition, epithelial ovarian cancer cells from ascites release FasL-exposing microvesicles that are capable of inducing T-cell apoptosis (Abrahams et al., 2003). Likewise, FasL-exposing microvesicles isolated from sera of patients with oral squamous cell carcinoma induce T-cell apoptosis, and in these patients, there is a correlation between the microvesicle-associated levels of FasL and tumor burden, suggesting that vesicle-mediated immune suppression (indirectly) promotes tumor growth and development in vivo (Kim et al., 2005a). FasL is not the only mediator of tumor-induced suppression of the immune system, because exosomes from human tumor cell lines and mouse mammary tumor cells inhibit (interleukin-2-induced) proliferation of NK cells, illustrating that tumor-derived exosomes suppress the cytotoxic response to tumor cells (Liu et al., 2006).

The loss of NKG2D, an activating receptor for NK, NKT, CD8⁺, and γδ⁺ T cells, in cancer is also one of the key mechanisms of immune evasion. Exosomes from cancer cell lines or isolated from pleural effusion of patients with mesothelioma down-regulate NKG2D expression of NK cells and CD8⁺ cells via a transforming growth factor (TGF)-β-dependent mechanism, indicating that NKG2D may be one of the targets for exosome-mediated immune evasion (Clayton et al., 2008). In addition, the high incidence of relapse and fatal outcome in many blood malignancies such as leukemia and lymphoma has been associated with the expression of several NKG2D ligands on exosomes (Hedlund et al., 2011). Alternatively, differentiation of myeloid cells to myeloid-derived suppressor cells, immature myeloid cells that have the ability to suppress T-cell activation and thereby promote tumor progression, induce the expression of TGF-β and prostacyclin E2 within tumor-derived exosomes (Xiang et al., 2009; Liu et al., 2010). In another study, tumor-derived exosomes from mouse cell lines were shown to inhibit immune surveillance by binding to myeloid-derived suppressor cells, thereby increasing the immune suppressive activity of these cells. This interaction was mediated by the exposure of HSP72 on exosomes. Further evidence that tumor-derived exosomes contribute to immune suppression in vivo comes from the observation that inhibition of exosome production by a drug to treat high blood pressure, dimethyl amiloride, enhances the antitumor efficacy of the chemotherapeutic drug cyclophosphamide in mouse tumor models (Chalmin et al., 2010). Finally, local immune responses are inhibited by high levels of extracellular adenosine, which prevent T-cell activation. Exosomes from several cancer cell lines expose potent ATP and 5′AMP-phosphohydrolytic activity, thereby contributing to locally high concentration of adenosine. Exosomes from pleural effusion of patients with mesothelioma contained 20% of the total ATP-hydrolytic activity, indicating that exosomes may contribute to inhibition of T cells in the tumor environment by this mechanism (Clayton et al., 2011). Taken together, exosomes are likely to orchestrate the efficacy of the immune system by a whole set of different mechanisms important not only for normal development but also for tumor development.

Besides healthy and tumor cells, viruses and parasites exploit the ability of exosomes to kill immune cells at a distance to evade the immune system. The latent membrane protein-1 (LMP-1) of Epstein-Barr virus (EBV), which is exposed on exosomes from EBV-infected cells, inhibits the proliferation of peripheral blood mononuclear cells. This mechanism is thought to be relevant in development of EBV-associated tumors, such as nasopharyngeal carcinoma and Hodgkin's disease, by allowing tumor cells to evade the immune system (Flanagan et al., 2003). Cells infected with the intracellular parasite Leishmania donovani secrete exosomes that contain and deliver parasite proteins to host target cells, and suppress the immune response by preventing the activation of human monocytes (Silverman et al., 2010a; Silverman et al., 2010b). Exosomes from plasma of mice immunized to keyhole limpet hemocyanin specifically suppress a keyhole limpet hemocyanin delayed-type hypersensitivity inflammatory response, which is partially dependent on exosomal FasL (Kim et al., 2007b).

Although considerable evidence thus suggests that cell-derived vesicles are involved in immune suppression, there are also studies showing that exosomes may exhibit opposing activity or do not contribute to suppression. For example, exosomes from synovial fibroblasts of rheumatoid arthritis patients expose TNF- α, which delays T-cell activation-induced cell death and thus contributes to apoptosis resistance (Zhang et al., 2006). In addition, soluble factors but not the microvesicles from patients with non–small-cell lung carcinoma (NSCLC) were shown to induce T-cell apoptosis (Prado-Garcia et al., 2008). Furthermore, exosomes from human pancreatic tumor cells inhibit tumor progression by counteracting constitutively active survival pathways, thereby inducing apoptosis of tumor cells rather than apoptosis of immune cells (Ristorcelli et al., 2008, 2009). In sum, the exact role and contribution of microvesicles or exosomes to immune suppression depends on the models studied and is likely to be complex.

2. Antigen Presentation.

Vesicles are used by cells not only to suppress the immune system but also to present antigens (Fig. 4). For example, intestinal epithelial cells endocytose dietary proteins, and serum obtained from mice fed with a particular protein can suppress hypersensitivity reactions when this serum is administered to other mice before exposure to this protein. Processing of the antigen in the gut is required to achieve oral (immune) tolerance via serum transfer. Although the true nature of the “tolerogen” in serum is unknown, exosomes are likely candidate tolerogens, because human and mouse intestinal epithelial cells release exosomes exposing MHC class I and MHC class II molecules at both the apical and basolateral sides, suggesting a possible involvement of exosomes in the transcellular transport of antigens from the lumen of the gut to immune cells (van Niel et al., 2001; Van Niel et al., 2003). To test the ability of vesicles to deliver “transcellular information” to the immune system in vivo, mice received peritoneal injection with exosomes from epithelial cells that had been incubated with pepsin/trypsin ovalbumin hydrolysate (hOVA) to mimic luminal digestion. Exosomes from intestinal epithelial cells treated with hOVA and interferon-γ, which is added to increase the secretion of exosomes (Van Niel et al., 2003), expose abundant MHC-class II/OVA complexes. These exosomes, however, induce a specific humoral immune response in vivo, suggesting that these exosomes trigger an immunogenic rather than a tolerogenic response (Van Niel et al., 2003). Because these exosomes preferentially interact with dendritic cells (DCs), they can strongly facilitate antigen presentation to T cells, thereby providing luminal antigenic information to local immune cells and thus facilitating immune surveillance at mucosal surfaces (Mallegol et al., 2007).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Antigen presentation. Cells, tumor cells, and infected cells all release vesicles that can be phagocytosed by DCs. DCs induce an immune response but also release antigen-exposing vesicles that present antigens to other DCs, thereby enhancing the immune response.

In a mouse model of allergic asthma, serum or isolated exosomes from serum of donor mice fed with ovalbumin were transferred to mice followed by intranasal exposure to ovalbumin. Mice receiving serum or isolated exosomes from ovalbumin-fed mice showed tolerance to ovalbumin exposure, because these mice had a reduced concentration of airway eosinophils, lower serum levels of total IgE and ovalbumin-specific IgE, and an increased concentration of activated T cells with a regulatory phenotypethan control animals. Thus, exosomes have the ability to prevent allergic sensitization (Almqvist et al., 2008).

Exosomes, isolated from conditioned medium of B-cell lines originating from patients with birch pollen allergy and loaded with T-cell-activating peptides from a major birch allergen, induce a dose-dependent T-cell proliferation and increase the production of interleukin 5 and 13 by T cells, indicating that B-cell exosomes may contribute to and worsen allergic immune responses (Admyre et al., 2007). In contrast, exosomes from BAL fluid of mice exposed to an olive pollen allergen inhibit the IgE response, cytokine production, and airway inflammation, all signs of allergy, upon exposure to the allergen, showing that these exosomes induce tolerance rather than allergic immune responses (Prado et al., 2008).

The initiation of T-cell-mediated antitumor immune responses requires uptake, processing, and presentation of tumor antigens by DCs. Exosomes from mouse tumor cells transfer tumor antigens to DCs in vitro. The binding of these intercellular adhesion molecule-1 (ICAM-1)-exposing exosomes to DCs, which express lymphocyte function-associated antigen-1 (LFA-1) and macrophage 1-antigen as the two major ligands of ICAM-1, is mediated only by LFA-1. Because both mature (CD8⁺) and immature (CD8⁻) DCs expose LFA-1, it has been suggested that the interaction between ICAM-1 and LFA-1 plays a role in the exosome-mediated transfer of antigens to DCs (Segura et al., 2007). After binding and uptake of tumor-derived exosomes, DCs are capable of inducing CD8⁺ T cell-mediated antitumor effects on established mouse tumors in vivo, indicating that tumor-derived exosomes can present tumor antigens to elicit an antitumor immune response (Wolfers et al., 2001).

After uptake of antigens, DCs also secrete exosomes that expose both MHC complexes and T-cell costimulatory molecules. Although an antigen-specific CD4⁺ T cell activation is induced by injection of antigen- or peptide-exposing exosomes from DCs in vivo, this response only occurs in the presence of mature DCs in vitro, suggesting that the exchange of functional peptide-MHC complexes between (subsets of) DCs by DC-derived exosomes may be a prerequisite to induce an efficient and antigen-specific T-cell response in vivo. As such, this exosome-mediated mechanism may amplify the initiation of the primary adaptive immune response (Théry et al., 2002). Indeed, DC-derived exosomes transfer functional MHC class I/peptide complexes to other DCs, but the presence of mature DCs is essential to efficiently prime cytotoxic T cells in vivo (André et al., 2004).

Mature DCs express MHC class II to present peptide antigens to T cells. Immature DCs, however, have only low surface levels of MHC class II because the peptide-loaded MHC class II complexes are ubiquitinated and predestined for lysosomal degradation. Activation of immature DCs inhibits ubiquitination of the MHC class II-antigen complexes, which results in an increased cell surface exposure of these complexes. In turn, exosomes exposing MHC class II-antigen complexes are secreted from antigen-loaded DCs when these cells contact antigen-specific CD4⁺ T cells. This secretion is preceded by accumulation of intraluminal vesicles exposing both MHC class II and CD9 in MVEs. The sorting of MHC class II molecules to these intraluminal vesicles requires incorporation into CD9-containing detergent-resistant membrane domains but is independent of ubiquitination, indicating that different intracellular pathways exist for trafficking of MVEs predestined for either lysosomal degradation or secretion (Buschow et al., 2009). It should be pointed out that exosomes from DCs may also be internalized by epithelial cells, which will then produce and release cytokines and chemokines. Thus, exosomes from DCs play a role not only in adaptive immunity but also in innate immunity (Obregon et al., 2009).

Exosomes may also present antigens from microorganisms and allergens. Toxoplasma gondii can cause severe sequelae in fetuses of mothers who acquire infection during pregnancy as well as life-threatening neuropathy in immunocompromised patients. Exosomes from a mouse DC cell line incubated with T. gondii-derived antigens trigger a strong systemic humoral immune response in vivo and conferred good protection against infection, suggesting that DC-derived exosomes can be used for immunoprophylaxis (Aline et al., 2004; Beauvillain et al., 2007; Beauvillain et al., 2009). Exosomes produced by DCs loaded with protozoan parasite Leishmania major antigens provide an effective vaccine against cutaneous leishmaniasis (Schnitzer et al., 2010), and infection with Mycobacterium tuberculosis primes macrophages to release increased numbers of exosomes and microvesicles exposingM. tuberculosis peptide-MHC class II complexes and HSP70 that enhance antimicrobial T-cell responses (Anand et al., 2010; Ramachandra et al., 2010). Finally, intramuscular immunization of chickens with exosomes from DCs incubated with Eimeria tenella (avian coccidiosis) antigens induces protective immunity against avian coccidiosis (Del Cacho et al., 2011). In these examples, presentation of antigens by exosomes induces immune tolerance rather than immune responses. Taken together, presentation of antigens by exosomes affects the immune response at various levels in vitro as well as in vivo. Nevertheless, although exosomes are involved in presentation of specific antigens, they may also play other roles in host defense, for instance, exosomes from human tracheobronchial epithelial cells expose epithelial mucins that immobilize human influenza virus by binding viral α-2,6-linked sialic acid (Kesimer et al., 2009).

3. Transfer of Signaling Components.

Cell-derived vesicles also play a role in the intercellular exchange of signaling components. Treatment of microvesicles from platelets and erythrocytes with nonpancreatic secretory phospholipase A₂, which is inactive toward whole cells, generates lysophosphatidic acid, which in turn triggers platelet aggregation (Fourcade et al., 1995). Because secretory phospholipase A₂ and lysophosphatidic acid-containing microvesicles are both present in synovial fluid of inflamed joints, the formation of lysophosphatidic acid is likely to occur in vivo (Fourcade et al., 1995). Likewise, treatment of microparticles from activated platelets with secretory phospholipase A₂ produces intravesicular arachidonic acid, which is transferred and then metabolized by platelets to thromboxane A₂ by cyclooxygenase-1, and by endothelial cells to prostacylin by cyclooxygenase-2 (Barry et al., 1998). The ability of platelet microparticles to increase adhesion of monocytes to endothelial cells is due to the presence of intravesicular arachidonic acid (Barry et al., 1998). Vesicles can evidently transfer lipids between cells for further metabolization, and thus lipid transfer may play a role in atherosclerosis and inflammation.

An elegant example of vesicle-mediated transfer of other signaling elements comes from studies on progesterone-induced Ca²⁺ signals, which play a key role in sperm motility. The progesterone-induced Ca²⁺ signals require fusion with prostasomes, which transfer progesterone receptors, cyclic adenosine diphosphoribose-synthesizing enzymes, ryanodine receptors, and other Ca²⁺-signaling tools to the sperm. Inhibition of cyclic adenosine diphosphoribose-synthesizing enzymes or depletion of ryanodine receptors from prostasomes reduces sperm motility and fertilization rates, showing that sperm motility and fertilization may depend on the acquisition of Ca²⁺-signaling tools from prostasomes (Park et al., 2011; Ren, 2011).

Another example comes from the Wnt-signaling pathway in DCs. The Wnt pathway plays an important role in both normal development and cancer. Wnt signaling activity, however, is suppressed by secretion of β-catenin in exosomes, revealing that exosomal packaging and release of cytosolic proteins can affect cellular signaling pathways (Chairoungdua et al., 2010). Exosomes can also contain proteins involved in the apoptotic signaling pathway. For instance, survivin, an inhibitor of caspase activation that promotes proliferation, survival and tumor cell invasion, is also secreted by exosomes (Khan et al., 2011). Thus, it is clear that cells can exchange functional signaling components between cells.

4. Inflammation.

Cell-derived vesicles can trigger the production of tissue factor (TF) and proinflammatory cytokines by activating target cells (Fig. 5). For instance, microparticles from N-formyl-Met-Leu-Phe-stimulated human polymorphonuclear leukocytes (PMNs) activate endothelial cells by increasing tyrosine phosphorylation of the 46-kDa c-Jun NH₂-terminal kinase (JNK) 1. This activation results in expression and production of TF and interleukin-6 (Mesri and Altieri, 1998; Mesri and Altieri, 1999). Because these vesicles, called “ectosomes,” are released at sites of inflammation in vitro and in vivo and expose the complement receptor 1 required for binding to opsonized bacteria, these vesicles have been suggested to represent an “ecto-organelle” designed to focus antimicrobial activity onto opsonized surfaces (Hess et al., 1999). On the other hand, the effects of microparticles from PMNs may be strongly cell-type-dependent, because human macrophages are inhibited upon incubation with PMN-derived microparticles, resulting in production of TGF-β and inhibition of the inflammatory response of macrophages to zymosan or lipopolysaccharide (LPS). This inhibition requires binding of microparticles to macrophages but is independent from their phagocytosis (Gasser and Schifferli, 2004; Eken et al., 2010). Furthermore, PMN-derived microparticles contain the anti-inflammatory protein annexin-1, and intravenous delivery of these microparticles blocks recruitment of PMN in an inflammatory model in an annexin-1-dependent mechanism in vivo, suggesting that PMN-derived microparticles may have both pro- and anti-inflammatory properties at sites of inflammation (Dalli et al., 2008).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Inflammation. Cells can transfer inflammatory mediators, including interleukin (IL)-1β, platelet-activating factor (PAF), and TNF receptor-1 (TNFR-1), to cells. Activated cells express and produce proinflammatory cytokines, TF, monocyte chemotactic protein (MCP) 1 and 2, and ICAM-1, which all support inflammation. Vesicles can also induce cellular differentiation of monocytes into macrophages, thereby inducing insulin resistance.

Cell-derived vesicles have the ability to affect inflammation by transferring transport receptors and inflammatory mediators. Microvesicles or exosomes are a major pathway for fast secretion of interleukin 1β, which lacks a signal peptide (MacKenzie et al., 2001; Wilson et al., 2004; Bianco et al., 2005). Interleukin 1β was recently shown to be one of the key mediators within microparticles from monocytes, together with other components of the inflammasome, responsible for activation of endothelial cells (Wang et al., 2011). Likewise, mouse macrophages, stimulated with ATP, release exosomes with entrapped interleukin-1β, caspase-1 and other components of the inflammasome (Qu et al., 2007). In addition, microvesicles from P2X7 receptor-stimulated mature DCs contain interleukin-1β (Pizzirani et al., 2007). Furthermore, microparticles contain platelet-activating factor (Watanabe et al., 2003) and can expose the soluble (non–cell-bound) but full-length 55-kDa form of TNF receptor 1, which is also found in human serum and lung epithelial lining fluid (Hawari et al., 2004).

Proteins are not the only biologically active component of vesicles that play a role in inflammatory responses, because the lipid fraction of microvesicles activates macrophage Toll like receptor (TLR) 4. Because the ability of these microparticles to activate TLR4 is impaired by preincubation with an inhibitor of phospholipase D, which does not affect the release of microvesicles but inhibits hydrolysis of phosphatidylcholine to phosphatidic acid, this further supports the lipid nature of the active component (Thomas and Salter, 2010). In addition, exosomes from DCs and macrophages, and microvesicles from human plasma, contain functional enzymes for synthesis of leukotrienes, which are potent lipid inflammatory mediators (Esser et al., 2010).

The ability of vesicles to modulate the inflammatory response is not limited to blood. Autologous vesicles from synovial fluid and microparticles from T cells, monocytes, and platelets trigger the production and release of interleukins 6 and 8, matrix metalloproteases, monocyte-chemotactic proteins 1 and 2, vascular endothelial growth factor (VEGF), and ICAM-1 by synovial fibroblasts, indicating that these vesicles enhance the destructive activity of these fibroblasts in rheumatoid arthritis (Berckmans et al., 2005; Distler et al., 2005; Beyer and Pisetsky, 2010; Boilard et al., 2010; Distler and Distler, 2010). One of the mechanisms by which microparticles activate synovial fibroblasts is transfer of arachidonic acid from leukocytes to synovial fibroblasts via microparticles (Jüngel et al., 2007). Furthermore, synovial microparticles expose TF and are extremely procoagulant, which could explain the presence of fibrin deposition within inflamed joints (Berckmans et al., 2002). In addition, microparticles from human airway epithelial cells stimulate the production of proinflammatory mediators, thereby possibly enhancing the inflammatory airway response (Cerri et al., 2006).

Peripheral blood of patients with preeclampsia contains elevated concentrations of placenta-derived vesicles compared with normal pregnancy. These vesicles, often called syncytiotrophoblast microparticles, bind to monocytes and the endothelium, thereby triggering the production of proinflammatory mediators. This suggests that syncytiotrophoblast microparticles contribute to the increased systemic inflammatory responsiveness in preeclampsia (Germain et al., 2007; Messerli et al., 2010; Southcombe et al., 2011).

Microparticles from human atherosclerotic plaques have been shown to mediate the functional transfer of ICAM-1 to endothelial cells, thereby promoting the adhesion of monocytes and trans-endothelial migration. The transfer of ICAM-1 from microparticles to endothelial cells results in increased phosphorylation of extracellular signal-regulated kinase 1/2 after ICAM-1 ligation (Rautou et al., 2011). Thus, plaque MPs may further facilitate atherosclerotic plaque progression.

Exosome-like vesicles released from adipose tissue are taken up by peripheral blood monocytes after intravenous injection, which then differentiate into macrophages producing increased levels of TNF-α and interleukin-6. Injection of these vesicles into wild-type C57BL/6 mice results in development of insulin resistance, whereas injection into TLR-4 knockout B6 mice results in much lower levels of glucose intolerance and insulin resistance, pointing to a role for TLR4 in the induction of inflammatory mediators by exosome-like vesicles from adipose tissue (Deng et al., 2009). Evidently, exosomes and other vesicles can influence the inflammatory response in several ways.

5. Tumor Growth, Metastasis, and Angiogenesis.

Exosomes can transfer the metastatic activity of highly metastatic Bl6–10 melanoma tumor cells to poorly metastatic F1 melanoma tumor cells in vitro. Mice develop lung metastatic colonies when F1 cells are injected together with exosomes from Bl6–10 exosomes, whereas mice injected solely with F1 cells develop no metastatic colonies (Hao et al., 2006). The ability of exosomes to promote metastasis and angiogenesis can be increased when exosomes are released under hypoxic conditions (i.e., conditions that many tumors encounter when growing) (Park et al., 2010).

One of the molecular mechanisms underlying the intercellular transfer of metastatic activity is the transfer of an oncogenic growth factor receptor or their ligand (Fig. 6). The truncated form of the EGFR vIII is transferred from glioma cancer cells by microvesicles to glioma cells lacking this receptor, in a mechanism involving detergent-resistant membrane domains (lipid rafts). The transferred receptor is fully functional and results in a full transfer of the oncogenic activity, including activation of the transforming signaling pathways (mitogen-activated protein kinase, AKT), changes in expression of EGFRvIII-regulated genes (VEGF, Bcl-x_L, p27), morphological transformation, and increase in anchorage-independent cell growth capacity (Al-Nedawi et al., 2008). On the other hand, human breast and colorectal cancer cells release exosomes containing full-length, signaling-competent EGFR ligands, and the invasive potential of colon cancer cell lines is related to the concentration of exosomes containing amphiregulin, one of the EGFR ligands, suggesting that ligand exchange contributes to cancer invasiveness and metastasis (Higginbotham et al., 2011).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Tumor growth, metastasis, and angiogenesis. Tumor cell-derived vesicles facilitate tumor development in many ways. The vesicles can transfer oncogenic activity by transferring growth factor receptors (EGFRvIII) or ligands to other tumor cells, vesicles can remove Fas-associated protein with death domain (FADD), so that tumor cells become less susceptible to apoptosis, and vesicles can transfer P-glycoprotein (P-gp), a multidrug transporter, between cells, thereby contributing to drug resistance. Tumor cell-derived vesicles also affect neighboring cells by transferring the Notch ligand Dll4 to endothelial cells, thereby inducing branch formation. Finally, vesicles can support and initiate coagulation, matrix degradation and angiogenesis.

Vesicles can also contribute to tumor growth and development via several other mechanisms. For instance, cancer cells can release vesicles containing the Fas-associated death domain, a key adaptor protein that transmits apoptotic signals and becomes lost in many cancer cells (Tourneur et al., 2008).

In addition, exosomes may protect tumor cells from entering or accumulation of antitumor drugs, thereby possibly contributing to (multi)drug resistance. For instance, exosomes from HER2-overexpressing breast cancer cell lines, or exosomes present in serum from patients with breast cancer, capture the humanized antibody trastuzumab, thereby reducing the effective concentration of this anticancer drug (Ciravolo et al., 2012). Vesicles may also contribute to drug resistance by exchanging drug transporters between cells. One of the most important drug transporters is P-glycoprotein, a transmembrane protein and member of the ATP binding cassette superfamily, which can be exchanged between cells via vesicles (Gong et al., 2012). P-glycoprotein, a key protein involved in multidrug resistance, is widely expressed at pharmacological interfaces such as the blood-brain barrier, blood-testis barrier, and along the gastrointestinal track (where the influx of carcinogens and harmful chemicals is prevented). Overexpression of P-glycoprotein by cancer cells correlates with anticancer drug failure in many types of cancer. There is also evidence that drugs are sorted, concentrated, and removed from cells via secretion of exosomes (e.g., cisplatin in drug-resistant ovarian carcinoma cells) (Safaei et al., 2005).

Vesicles can also promote vascular development, which is highly relevant for tumor growth. For instance, Notch signaling is an evolutionarily conserved pathway that plays an essential role in vascular development and angiogenesis. δ-like 4 (Dll4) is a Notch ligand that is up-regulated during angiogenesis by endothelial cells and regulates the differentiation between tip cells and stalk cells of the neovasculature. Dll4 is incorporated into exosomes of endothelial cells and tumor cells, and it can be transferred to endothelial cells, where it is incorporated into the plasma membrane. Once incorporated, Dll4 inhibits Notch signaling and induces the loss of Notch receptor, which results in a tip cell phenotype and increases branch formation of vessels (Sheldon et al., 2010).

There are other ways in which vesicles promote tumor growth, including the release of active matrix-degrading enzymes (Hakulinen et al., 2008), stimulation of angiogenesis (Taverna et al., 2011), and production of TF (Uno et al., 2007), but these mechanisms have been summarized excellently elsewhere (Zöller, 2009; Rak, 2010; Peinado et al., 2011). Please see also section III.A.7.

6. Morphogens.

The morphogen Wnt is important in development of the mature nervous system. Wnt is transported in vesicles across synapses in Drosophila melanogaster larval neuromuscular junctions (Korkut et al., 2009). However, the transport of morphogens is not limited to D. melanogaster; it also occurs in higher organisms. For instance, microvesicles from T cells or human blood contain functional Hedgehog proteins, which play a role in development, and these vesicles are capable of inducing differentiation of pluripotent erythroleukemic cells in vitro (Martínez et al., 2006).

7. Genetic Information.

Bacteria exchange genetic information via outer membrane vesicles (Yaron et al., 2000). In 2007, it was described that eukaryotic cells can exchange functional genetic information by vesicles as well (Fig. 7). Exosomes from mouse and human mast cell lines as well as exosomes from primary bone marrow-derived mouse mast cells were shown to contain mRNA and miRNA (Valadi et al., 2007). The mRNA of approximately 1300 genes was detectable, many of which were below the detection limit in the cytoplasm of the donor cells. In vitro translation showed that mRNA could be translated, and transfer of mouse exosomal RNA to human mast cells induces expression of mouse proteins. Thus, exosomes have the capacity to facilitate the intercellular exchange of functional genetic information. Exosomes from T cells transfer miRNA to antigen-presenting cells (Mittelbrunn et al., 2011), and exosomes from DCs contain different miRNAs depending on the maturation stage of the DCs. Fusion of these exosomes with acceptor DCs results in the transfer of functional miRNAs,as shown by the reduced expression of target mRNA. Thus, intercellular exchange of miRNAs by vesicles can affect post-transcriptional regulation (Montecalvo et al., 2012). In addition, exosomes from glioblastoma cells transfer functional mRNA, miRNA, and angiogenic proteins to brain microvascular endothelial cells, thus stimulating tubule formation and proliferation of glioma cells (Skog et al., 2008). Several mRNAs and miRNAs characteristic of gliomas were detected in some but not all sera of patients with glioblastoma, suggesting that exosomes contain diagnostic information. Likewise, microvesicles from a colorectal cancer cell line contain mRNA, including 27 mRNAs encoding cell-cycle related proteins. These exosomes triggered proliferation of endothelial cells, again confirming that vesicles can induce angiogenesis and thus promote tumor growth and metastasis (Hong et al., 2009).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Genetic information. Vesicles from normal cells, tumor cells, and infected cells release vesicles containing functional genetic information that can be transferred to recipient cells.

Exosomes from astrocytes and glioblastoma cells also contain mitochondrial DNA (Guescini et al., 2010), and microvesicles from several tumors and tumor cell lines have been shown to contain elevated levels of specific coding as well as noncoding RNA and DNA, mutated and amplified oncogene sequences, and transposable elements (Balaj et al., 2011). In addition, serum-derived exosomes from tumor-bearing mice contain the amplified oncogene c-myc, suggesting that the genetic information in tumor-derived vesicles can potentially be a unique tumor biomarker (Balaj et al., 2011). It is to be expected that vesicles in biological fluids will contain genetic information, and therefore it is not surprising that exosomes from, for example, human saliva, plasma, and milk all contain detectable levels of RNA, including mRNA (Lässer et al., 2011).

Pathogens have “hijacked” cell-derived vesicles as a transport vehicle to exchange genetic information between cells without the risk of being recognized by the immune system. Exosomes from nasopharyngeal carcinoma cells harboring latent EBV contain viral miRNAs, some of which are enriched compared with intracellular levels. This increased concentration of viral miRNAs may help to manipulate the tumor microenvironment to influence the growth of neighboring cells (Gourzones et al., 2010; Pegtel et al., 2010; Meckes et al., 2010). In addition, human herpes virus 4 or EBV encodes miRNAs that can be transferred from infected to uninfected cells by exosomes. Again, transferred miRNAs are functional because they down-regulate target gene expression in recipient cells (Pegtel et al., 2010). The miRNA-mediated gene silencing is a potentially important mechanism of intercellular communication. Several components of RNA-induced silencing complexes are present in exosomes, suggesting that this complex may be involved in the sorting of miRNAs into exosomes (Pegtel et al., 2011). Taken together, the role of vesicles in the exchange of functional genetic information between cells is not limited to prokaryotes but may also contribute to physiological and pathological processes in eukaryotes.

8. Prions.

Prion diseases are infectious and fatal neurodegenerative disorders due to accumulation of, for example, abnormally folded prion protein (PrP) scrapie (PrP^sc) in the central nervous system. PrP^sc catalyzes the further conversion of normal cellular PrP into PrP^sc. Both forms of prion protein, PrP and PrP^sc, have been found in exosomes. Because exosomes containing PrP^sc are infectious, they are likely to contribute to the spreading of prions (Fevrier et al., 2004; Vella et al., 2007). Likewise, a fraction of β-amyloid peptides is associated with exosomes in Alzheimer's disease. Because exosomal proteins accumulate in plaques of patient brains, exosomes are thought to play a role in the pathogenesis of Alzheimer's disease (Rajendran et al., 2006).

9. Viruses.

Viruses use vesicles for infection and survival (Fig. 8). Microparticles transfer receptors that are essential for entry of HIV to cells lacking these receptors. For instance, transfer of the chemokine receptors CCR5 and CXCR4 makes recipient cells susceptible to entry of HIV (Mack et al., 2000; Rozmyslowicz et al., 2003). In addition, uptake of HIV by cells results in camouflage of the virus. Capture of HIV by immature DCs produces viral particles that expose characteristic exosomal proteins as CD63 and CD9, and the presence of such proteins is thought to help the virus particles to escape from recognition by the immune system. These viral particles are more infectious to CD4⁺ T cells than cell-free viral particles (Wiley and Gummuluru, 2006). This camouflage can be further improved by coating with host-derived glycoproteins, thereby making viral particles even more closely resembling T-cell-derived vesicles (Krishnamoorthy et al., 2009). Because virus particles and cell-derived vesicles are of similar size and have many similarities in composition, this has caused major problems with isolation and purification of the virus particles (Bess et al., 1997; Aupeix et al., 1997; Coren et al., 2008; Park and He, 2010).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

Viruses. Cells infected with HIV release viral particles exposing exosomal (glyco)proteins, thus escaping from detection by the immune system. Vesicles from HIV-infected cells can transfer chemokine receptors (CCR5, CXCR4) to cells lacking these receptors, thereby facilitating the entry of HIV in these cells. In addition, vesicles from HIV-infected cells expose Nef, which triggers CD4⁺ T cell apoptosis. Vesicles contain also antiviral activity, in particular cytidine deaminase.

One of the earliest and most abundantly expressed viral proteins is Nef. Nef is present in the serum of patients with HIV and interacts with CXCR4, thereby inducing apoptosis and depletion of CD4⁺ T cells, which is one of the hallmarks of AIDS (Lenassi et al., 2010; Raymond et al., 2011; Shelton et al., 2012). Nef stimulates its own export via secretion of exosomes or CD45-exposing microvesicles. In addition, LMP-1, the major oncogene of EBV, is detectable in serum of infected patients. LMP-1 inhibits T cell proliferation and NK cytotoxicity, although other proteins, such as galectin 9, may be involved as well, thereby facilitating escape from the immune system (Meckes and Raab-Traub, 2011).

Viruses also have other strategies to escape from the immune system. For instance, cells infected with hepatitis B virus, pox viruses, or herpes simplex virus, release empty viral particles exposing viral glycoproteins in concentrations up to 10,000 fold higher than that of the infectious particles, suggesting that these vesicles may act as potential decoys to distract the immune system (Meckes and Raab-Traub, 2011).

Viruses may use the ability of exosomes to kill immune cells at a distance. LMP-1, which is exposed on exosomes and in microvesicles from EBV-infected cells, inhibits proliferation of peripheral blood mononuclear cells. This could be relevant in EBV-associated tumors such as nasopharyngeal carcinoma and Hodgkin's disease, because it allows tumor cells to escape from immune eradication (Flanagan et al., 2003). The underlying sorting mechanism of LMP-1 to exosomes is unknown, but the presence of LMP-1 in exosomes is independent from ubiquitination or detergent-resistant membrane domains (Verweij et al., 2011). Expression of LMP-1 also changes the composition of exosomes by increasing the concentrations of important signaling molecules in cancer, including phosphatidylinositol 3-kinase, EGFR, and fibroblast growth factor-2, suggesting that EBV 1) manipulates and uses the exosomal pathway for tumor growth and development and 2) protects the exosomes from degradation (Ceccarelli et al., 2007; Meckes et al., 2010; Verweij et al., 2011). The uptake of exosomes from EBV-infected cells by epithelial cells indeed results in activation of growth-stimulating signaling pathways, showing that transfer of LMP-1 and other signaling molecules can manipulate the growth characteristics of neighboring cells (Meckes and Raab-Traub, 2011). Exosomes from an EBV-infected B cell line have been shown to bind preferentially to B cells in blood, whereas exosomes from milk, DCs, or noninfected B cells bind to monocytes. The interaction between exosomes from EBV-infected B cells and B cells is mediated by CD21 on the B cells and glycoprotein 350 on the exosomes. It has been hypothesized that exosome-based vaccines can be developed that interfere with this interaction (Vallhov et al., 2011).

Because retroviruses and exosomes have many similarities, including lipid composition and the presence of host-cell proteins, retroviruses have been proposed to have hijacked the pathway of exosome biogenesis for their production to escape from immune detection (i.e., the Trojan exosome hypothesis) (Gould et al., 2003). Although the finding that an inhibitory domain of the HIV Gag protein interferes with the sorting of both viral and exosomal proteins, such as CD63, supports this hypothesis (Gan and Gould, 2011), inhibition of exosome release by blocking ceramide synthesis does not interfere with the release of HIV, suggesting the existence of different routes (Brügger et al., 2006; Jolly and Sattentau, 2007; Sato et al., 2008; Izquierdo-Useros et al., 2009; Meckes and Raab-Traub, 2011).

Exosomes also have anti-HIV-1 activity (Tumne et al., 2009), and the human cytidine deaminase APOBEC3G, which is a part of a cellular defense system against HIV-1 and other retroviruses, is secreted in exosomes and is actually the major exosomal component explaining the anti-HIV-1 activity of exosomes (Khatua et al., 2009, 2010). Furthermore, vesicles from virus-infected cells can present antigens to activate the immune system (Walker et al., 2009; Meckes and Raab-Traub, 2011).

1. Cancer.

Exosomes from antigen-presenting cells, DCs, express functional MHC class I and II molecules and T-cell costimulatory molecules. Because DCs incubated with tumor peptides secrete exosomes that are capable of priming specific cytotoxic T cells to eradicate or suppress growth of established murine tumors in vivo (Fig. 11), exosomes have the potential to be used as vaccines to suppress tumor growth (Zitvogel et al., 1998; Andre et al., 2002). When these DC-derived exosomes were characterized, several proteins were discovered that might explain the ability of exosomes to induce immune responses. The most striking proteins that were discovered are mild fat globule-epidermal growth factor VIII (MFG-E8), also known as lactadherin, and the 73-kDa cytosolic heat shock cognate protein. Lactadherin can target exosomes to antigen-presenting cells as macrophages and DCs, whereas heat shock cognate protein 73 triggers antitumor immune responses in vivo (Théry et al., 1999). Because DCs are potent vaccine carriers but are short-lived and sensitive to cytotoxic T-cell-mediated elimination in vivo, antigen presentation by exosomes is considered a promising tool. In particular, because functional antigen-presenting exosomes could be recovered from draining lymph nodes in vivo after DCs had disappeared, illustrating that antigen presentation after delivery of DC vaccines persists for longer than expected and indicating that exosomes may have played a previously unrecognized role in presentation of antigens after DC vaccination (Luketic et al., 2007).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 11.

Cancer therapy. Tumor cells and tumor cell-derived vesicles activate DCs, which in turn release vesicles that activate other DCs. This activation amplifies the cellular (T-cell)-mediated antitumor response. Vesicles from DCs can also induce a humoral (B-cell)-mediated antitumor response. hsc, heat shock cognate protein.

Protocols have been developed and optimized to ensure a reliable manufacturing process to harvest DC-derived exosomes (Lamparski et al., 2002; Le Pecq, 2005; Johansson et al., 2008; Viaud et al., 2011) and several strategies have been explored to improve the ability of exosomes to function as immunotherapeutic antitumor agents, including loading of exosomes with oligonucleotides and TLR-3 and TLR-9 ligands (Chaput et al., 2004; Adams et al., 2005), or the priming of DCs with exosomes from tumor cells exposed to heat stress. The heat treatment leads to increased exposure of MHC class I molecules, chemokines, and HSPs, the last of these acting as molecular chaperones having potent adjuvant activity in the induction of antigen-specific T-cell responses and therefore increased immunogenicity (Dai et al., 2005; Chen et al., 2006, 2011). Other protocols include administration of the exosomes after pretreatment with immune-potentiating doses of cyclophosphamide (Taieb et al., 2006), priming of DCs with exosomes from genetically modified tumor cells expressing interleukin-2 or interleukin-12 (Yang et al., 2007; Zhang et al., 2010), or overexpression of fusion proteins consisting of tumor-associated antigens and the C1C2 domain of lactadherin (Hartman et al., 2011; Rountree et al., 2011). Furthermore, exosomes from mature DCs are more efficient in eradicating established tumors than exosomes from immature DCs in vivo, because exosomes from mature DCs express higher levels of MHC complexes and T-cell costimulatory molecules (Hao et al., 2007).

It must be mentioned that exosomes from DCs not only induce a cell-mediated immune response but may also induce a humoral immune response. For example, exosomes from murine bone marrow-derived DCs incubated with intact diphtheria toxoid, or exosomes exposing a glycoconjugate that is cross-reactive with the capsular polysaccharide of Streptococcus pneumoniae type 14, induce specific antibody production in vivo (Colino and Snapper, 2006, 2007).

Instead of eradicating or suppressing the growth of already established tumors, also vaccination with exosomes has been tested for its efficacy to prevent tumor growth. Exosomes from plasmacytoma cells protected some 80% of the mice against a challenge with wild-type tumors. This protection is due to generation of specific cytotoxic T cells. The effects were not seen in severe combined immunodeficiency mice, and the immunity was tumor-specific (Altieri et al., 2004).

In several clinical trials, the efficacy of autologous exosome immunotherapy has been tested. In a phase I study, patients with advanced NSCLC with tumor expression of MAGE-A3 or -A4 were included. Patients underwent leukophoresis to produce DCs from isolated monocytes. Exosomes from DCs were isolated and loaded with a variety of MAGE peptides. Administration of exosomes was well tolerated; hardly any adverse effects were observed and several patients showed a long-term stability of disease, and activation of immune effector cells as NK cells and T cells in vivo was detectable in some patients (Morse et al., 2005). In a second study, exosomes incubated with MAGE 3 peptides were administered to patients with stage III or IV melanoma. Again, no adverse effects were observed, but also no MAGE 3-specific T-cell immune responses could be detected in peripheral blood, suggesting that adjuvant administration of exosomes may be of limited therapeutic relevance in this condition (Escudier et al., 2005). In both studies, however, monocyte-derived (immature) DCs were used, and treatment with interferon-γ has been shown to be a suitable maturation agent for these DCs. Exosomes from mature DCs produce exosomes that evoke a much stronger priming of the immune response in vivo than exosomes from immature DCs. At present, the ability of exosomes from mature DCs to trigger an antitumor immune response is tested in a phase II trial in patients with advanced NSCLC (Chaput and Théry, 2011).

Another approach is to use tumor-derived exosomes to raise an antitumor immune response. In a phase I trial, exosomes were isolated from ascites of patients with colorectal cancer. Patients were treated with exosomes plus granulocyte-macrophage colony-stimulating factor or only exosomes. Patients treated with the combination therapy showed beneficial tumor-specific antitumor cytotoxic T lymphocyte response (Dai et al., 2008).

Despite the fact that the different strategies used to explore the clinical use of exosomes as adjuvant therapy in antitumor immune therapy have not been as successful as hoped for, one has to bear in mind that the included patients all had advanced cancers and did not respond well to prior treatments. Therefore, the aim of these studies was not primarily to cure the patients but rather to prolong their survival and to stabilize disease after chemotherapy or radiation therapy (Théry et al., 2009).

2. Passing the Blood-Brain Barrier.

The therapeutic potential of RNA drugs requires efficient, tissue-specific, and nonimmunogenic delivery. Delivery of RNA interference to tissues at therapeutic effective doses, however, has proven difficult when using viruses,polycationic polyethylenimine-based nanoparticles, or liposomes(van den Boorn et al., 2011). Exosomes have been used to deliver short interfering RNA (siRNA) into the brain of mice in vivo. Exosomes were produced from autologous DCs, were genetically modified to express lysosomal-associated membrane protein-2b fused to a neuron-specific rabies virus glycoprotein peptide, and were loaded with glyceraldehyde 3-phosphate dehydrogenase siRNA. Intravenously injected rabies virus glycoprotein-targeted exosomes delivered glyceraldehyde 3-phosphate dehydrogenase siRNA specifically to neurons, microglia, and oligodendrocytes and induced a specific gene knockdown. The therapeutic potential of exosome-mediated siRNA delivery was shown by inhibition of more than 50% of expression and production of BACE1, a therapeutic target in Alzheimer's disease (Alvarez-Erviti et al., 2011). This targeted siRNA delivery is clinically very interesting because exosome-mediated drug delivery may overcome important obstacles in drug delivery, such as the blood-brain barrier, which is a major hindrance for treatment of neurological diseases (Lakhal and Wood, 2011). There is only one earlier study in which exosomes were used for drug delivery; in that study, exosomes were shown to transfer curcumin, an anti-inflammatory agent, to activated myeloid cells in vivo (Sun et al., 2010). In sum, because autologous exosomes can evade the immune system, are protected from complement deposition by exposing CD55 or CD59, and expose cell-type specific adhesion receptors, their use for drug delivery is promising.

3. Inflammation and Immune Response.

There are alternative strategies to use exosomes for clinical applications. For instance, exosomes from mouse DCs, genetically modified to express interleukin-4, FasL, or the tryptophan-degrading enzyme indoleamine 2,3-dioxygenase, have both immunosuppressive and anti-inflammatory properties when administered systemically in mouse models of experimental arthritis (Kim et al., 2006, 2007a; Bianco et al., 2009). The intra-articular administration of exosomes from DCs transduced with an adenovirus expressing interleukin-10 or from DCs treated with recombinant murine interleukin-10 suppresses delayed type hypersensitivity responses not only in the injected but also in the untreated contralateral joints and suppresses the onset of murine collagen-induced arthritis as well as the severity of arthritis, illustrating that exosomes can efficiently suppress inflammatory and autoimmune responses in vivo (Kim et al., 2005b).

Immature DCs release exosomes exposing lactadherin, which is required to opsonize apoptotic cells for phagocytosis. In an experimental sepsis model, decreased levels of lactadherin in spleen and blood were associated with impaired apoptotic clearance, and administration of exosomes promoted phagocytosis of apoptotic cells and reduced mortality, whereas lactadherin-deficient mice suffered from increased mortality. Thus, exosomes can attenuate the acute systemic inflammatory response in sepsis by enhancing the clearance of apoptotic cells and perhaps apoptotic vesicles via lactadherin interactions (Miksa et al., 2009).

4. Neovascularization.

Transplantation of human CD34⁺ stem cells to ischemic tissues reduces angina, improves exercise time, reduces amputation rates, and induces neovascularization in preclinical models. These beneficial effects of CD34⁺ stem cells on therapeutic angiogenesis, however, seem to be mediated, at least in part, by exosomes produced by these cells. Exosomes, isolated from CD34⁺ stem cell-conditioned medium, improved viability of endothelial cells and induced endothelial proliferation and tube formation in vitro and angiogenesis in various in vivo models. Thus, exosomes have an angiogenic activity in vitro and in vivo (Lai et al., 2010; Sahoo et al., 2011).