Tumor-derived Exosomes and Antitumor Immunity

Exosomes are small extracellular vesicles (sEVs) derived from the endocytic cell compartment and produced by all living cells (1). EVs were discovered in early 1980s by pioneering work of several investigators (2), and Pan and Johnstone (3) are credited with providing the first electron microscope images of EVs produced by sheep reticulocytes. At the time, EV secretion was thought to be necessary to remove cellular waste. Today, EVs are emerging as an intercellular communication system (4). During the span of the last 30 y, it has become clear that EVs are indispensable for conveying messages between cells in bacteria, plants, animals, and humans (4, 5). Presumably, evolution has preserved the vesicular system because of its key importance for species survival. The realizations by the scientific community of the existence of EVs and the role that EVs play in cellular crosstalk have led to the intense worldwide investigation of their characteristics, cellular interactions, and functions. Today, much is known about the genetic/molecular content of EVs and their interactions with recipient cells, including their ability to transfer “targeted” information to a wide variety of cells in different tissues (6). Also, EVs are ubiquitous components of all body fluids, circulate freely across all organ barriers, and deliver signals to all systems, including the immune system (7, 8). The signals that EVs carry and deliver reflect the functional status of cells that produce them and thus change as these cells experience physiological alterations (e.g., differentiation, aging) or pathological changes (e.g., infection, cancer). This means that molecular profiles of EVs and their signaling repertoire are different in health and disease (9), potentially providing EVs with the attribute of serving as disease biomarkers.

This review considers the role EVs play in modulating the immune system in cancer as an example of cellular/molecular/genetic signaling that leads to major reprogramming of the tumor microenvironment (TME). The focus is on tumor-derived exosomes (TEX), a subset of sEVs released by tumor cells, and on effects TEX exert on immune cells with profound consequences for cancer escape from the host immune system, disease progression, and response to therapy.

The relationship between cancer and the host immune system has been extensively investigated, and the understanding of genetic, epigenetic, molecular, and physiological consequences of mutual interactions that evolve and change as the tumor develops is of critical importance. Evidence for the immune system involvement in all stages of cancer development and progression/regression is indisputable, and genetic/molecular mechanisms regulating these processes are emerging (10, 11). The immune system guards against and eliminates dangerous non-self in the form of exogenous invaders (e.g., infectious agents) or endogenous signals (e.g., mutated proteins). However, when at some point the immune system fails to protect, the intricate balance between developing cancer and the host immune system changes in favor of the tumor (11). When and how exactly this change occurs remain the main subjects of current research, but overwhelming evidence indicates that it corresponds to acquisition by the tumor of the ability to suppress antitumor immune responses (12, 13). Gradual but insidious immune suppression specific for immune activity against developing cancer is a hallmark of cancer, and numerous mechanisms that cancers use for suppression of antitumor immunity have been identified and reviewed (12). These include the following: 1) soluble factors such as IL-10, TGF-β, IL-35, IDO, arginase, galectin 9, adenosine, COX-2, PGE₂, and oxygen radicals; 2) inhibitory receptors/ligands (e.g., PD-1/PD-L1); 3) immune regulatory cells (e.g., regulatory T cells [Tregs], myeloid-derived suppressor cells; 4) metabolic checkpoints (e.g., glucose deprivation); 5) MHC class I downregulation/loss or inactivation of β₂-microglobulin on tumor cells; and 6) TEX (14). The latter represent the latest addition to the growing list of agents that tumors produce to silence the immune cells capable of harming/eliminating expanding tumor cells. Research into the role that TEX play in restraining immune cells from harming the tumor has provided some insights into mechanisms that TEX use to subvert antitumor immunity.

TEX are sEVs formed via invagination of the outer membrane of intracellular endosomal organelles called multivesicular bodies located in the cytosol of tumor cells (15). Tumors produce a large variety of vesicles and particles ranging in size from <30 nm (exomeres), 30–150 nm (exosomes or sEVs), 200–500 nm (microvesicles [MVs]), to >1000 nm (apoptotic bodies) (16). These vesicles differ from each other by size and also by their biogenesis: whereas sEVs originate in the endosomal cell compartment, MVs are formed by pinching off from the surface of cellular membrane. Thus, the molecular contents of sEVs and MVs differ, and they can be distinguished by the presence of the endocytic markers (e.g., ALIX, TSG101) in sEVs and their absence in MVs. Instead, MVs carry various membrane markers on the surface, such as integrin β₁ and cytosolic proteins (e.g., calnexin) in the lumen (17). Exosomes are also produced by nonmalignant cells, and in physiologically normal conditions they mediate intercellular crosstalk. Compared to healthy cells, tumors produce and release into extracellular space sEVs in significantly higher numbers (18). Thus, sEV numbers in body fluids of cancer patients significantly exceed those found in healthy donors (HDs) (19). In cancer patients, TEX represent a proportion of all circulating EVs, and the TEX fraction is dependent on the rate of TEX production and the rate of TEX removal by the host reticuloendothelial system (20). In general, numbers of circulating TEX are higher in advanced metastatic disease than in early disease or disease responding to therapy (21).

TEX differ from circulating sEVs produced by nonmalignant cells by the content of their cargos (proteins, lipids, glycans, nucleic acids) and by TEX functional attributes (22). Fig. 1 illustrates the rich molecular/genetic surface and luminal content of a “representative” TEX with a proviso that this content changes in TEX produced by cells of different tumor types. TEX carry the components of the major immunosuppressive pathways that the parent tumor cells use and, similar to their parent tumor cells, induce immune cell dysfunction (23). TEX transfer their components, including mRNA, microRNA, DNA, signaling proteins, enzymes, transcription factors, and cytokines to recipient cells, thereby inducing phenotypic and functional changes in these cells (24). When immune cells are recipients of TEX, the changes usually but not invariably lead to suppression of cellular activities and translate into immune dysfunction (24). Thus, TEX represent a unique category of sEVs delegated by the tumor to downregulate antitumor immunity. Although various tumors may use distinct mechanisms for inducing immune suppression, the quantity and quality of TEX that tumors release seem to correlate with the degree of immune dysfunction and disease progression existing in the tumor-bearing host (25, 26). Because the immune cell dysfunction in cancer is emerging as a biomarker of response to therapy and of outcome (27, 28), it follows that TEX carrying an excess of immunosuppressive cargos could potentially serve as biomarkers of existing immune dysfunction and of disease outcome. Thus, TEX and mechanisms that drive their protumor activities and interfere with immune cell antitumor functions have been of primary interest in cancer research.

The plasma of cancer patients contains a heterogeneous mix of sEVs derived not only from tumor cells but also from various tissue-resident and circulating nonmalignant cells, including immune cells (29, 30). Because of this heterogeneity, the intercellular crosstalk mediated by TEX cannot be distinguished from that mediated sEVs produced by nonmalignant cells without isolation of these sEV subsets and examination of their molecular content. TEX represent a variable, often only small proportion of circulating sEVs in cancer plasma, and their isolation and separation from sEVs produced by nonmalignant cells require special approaches. We and others have used Ab-based immune capture for positive or negative selection of TEX starting with sEVs isolated from plasma by ultrafiltration and size exclusion chromatography (31). Others use different methods for sEV isolation from body fluids, such as several hours-long ultracentrifugation at 100,000 × g followed by high-resolution iodixanol density gradient fractionation (32) or technologies based on biosensing microfluidic devices (33). Currently, the major barrier in studies of EVs is the lack of a uniformly accepted isolation procedure that is readily translatable to clinical applications.

The immune capture strategy we have used for the separation of TEX from other sEVs in cancer plasma is useful for EV studies in a preclinical setting (25, 34). The two-step positive selection for TEX isolation utilizes streptavidin beads and biotin-labeled Abs that recognize tumor-specific Ags on the TEX surface (34). It yields a highly enriched fraction of TEX captured on beads and sEVs derived from nonmalignant cells (non-TEX) in solution. Although the recovery of TEX is generally excellent, TEX adherence to beads may create difficulties with their subsequent functional characterization, especially in assays measuring enzymatic activities. The immune capture strategy can only be successful when Abs specific for a highly overexpressed tumor-associated Ags (TAAs) are available, such as, for example, the anti–chondroitin sulfate peptidoglycan-4 (CSPG4) epitope-specific Abs we used for isolation of melanoma cell–derived sEVs from patients’ plasma (25, 34). The mAb specific for an epitope of CSPG4, also known as high molecular weight melanoma-associated Ag, was developed and extensively evaluated by Ferrone and colleagues (35). It recognizes the Ag expressed on tumor cells and TEX but not on any other nonmalignant cell in the body except pericytes (36). It allows for effective separation of TEX from non-TEX in melanoma and other cancers overexpressing CSPG4 (36). The major limitation of positive TEX selection by immune capture is the paucity of Abs specific for tumor-specific Ags, as Abs specific for mutated Ags are only applicable in individual cases. The following commercially available Abs recognizing Ags overexpressed by tumors have been used for TEX selection with a variable success: Abs to epidermal growth factor receptor (EGFR), EGFR-2 (HER2), EGFRvIII, epithelial cell adhesion molecule (EpCAM), hyaluronic acid receptor (CD44v3), glypican-3 (GPC-3), prostate specific membrane Ag (PSMA), heat shock protein 70 (HSP70), and CD24, among others. Positive TEX selection can also be performed using a mix of Abs specific for the TAAs overexpressed on the tumor cell surface and carried by TEX (37). An alternative separation approach involves Ab-based negative selection, where exosomes derived from nonmalignant cells (e.g., T cells) are captured by the relevant anti-CD3 Abs, thereby enriching the CD3⁻ exosome fraction in TEX (38). This strategy allows for recovery and studies of both CD3⁺ and CD3⁻ exosome fractions in the same plasma sample (39). In cancer plasma, the CD3⁻ sEV fraction comprises TEX and non-TEX at various patient-specific ratios (38). Of note, separation from plasma of CD3⁺ sEVs, which are products of T lymphocytes and mimic contents of parental T cells, allows for indirect assessments of molecular and functional characteristics of producer T cells in addition to enabling studies of the TEX-enriched CD3⁻ sEV fraction in the same plasma specimen.

Upon isolation from body fluids or supernatants of cultured tumor cells, TEX undergo analysis to confirm their endocytic origin and define their phenotypic, molecular, and functional characteristics. Tetraspanins (CD9, CD63, CD81) are used as pan-exosome markers, and ALIX, TSG-101, syntenin, flagellin in the absence of cytosolic calnexin, and gp94 serve to confirm the TEX endosomal origin (31). The TEX surface phenotype is evaluated by on-bead flow cytometry that determines fluorescence intensity relative to isotype controls or by single vesicle flow cytometry, provided an instrument enabling gating on nanovesicles is available. The protein cargo is evaluated by Western blots or proteomic analyses and functional characteristics in coincubation assays with various recipient target cells. In vivo assays using tumor-bearing mice injected with sEVs are also performed as previously described (40). The International Society for Extracellular Vesicles has published a set of recommendations to guide EV characterization (41), which in the absence of a uniformly accepted isolation procedure represents an invaluable resource for investigators.

Much of what is known about TEX comes from studies with vesicles recovered from supernatants of cultured tumor cells, which contain only tumor cell–derived EVs and are the best source of TEX. As TEX can be readily labeled with membrane-binding dyes or adenoviral vectors, their engagements with a wide variety of recipient cells can be modeled in vitro or in vivo. Each tumor cell produces EVs with a unique phenotype/genotype, and the examination of all TEX present in culture provides the sum of characteristics for TEX originating from a given tumor type. Phenotypic analyses provide profiles of proteins carried on the vesicle surface membrane, and mild fixation to ensure permeability of the membrane allows for evaluations of the vesicle content in the lumen. Nucleic acid, lipid, or glycan extractions from TEX using protocols commonly used for cells work well with EVs, albeit only when adequate numbers of vesicles are available for analysis. Characterization of the TEX cargo often extends to specific molecular pathways or the pathway components of interest (e.g., enzymes), specific signaling molecules, or transcription factors (42–44). TEX functional attributes can be studied in coincubation experiments with recipient cells using assays that measure TEX cellular uptake, internalization, and various functional changes that TEX induce in recipient cells (i.e., differentiation, migration, proliferation, secretion, death). These studies performed in numerous laboratories worldwide have led to the conclusion that TEX deliver inhibitory signals to immune cells and protumor signals to various tissue cells (45, 46).

The surface membrane of TEX is decorated by a variety of receptor/ligands, adhesion molecules, opsonins, MHC proteins, and nucleoproteins (Fig. 1). The somewhat unexpected simultaneous presence on the TEX surface of numerous stimulatory proteins (CD40, CD40L, OX40, OX40L, CD80, CD86) together with inhibitory death receptor ligands (PD-L1, CTLA-4, TRAIL, FasL) suggests that TEX could perform regulatory functions depending on the microenvironment in which they operate. Because the TEX surface phenotype recapitulates that of the surface membrane in parent tumor cells (9), signaling mediated by TEX resembles that of their producer tumor cells. Importantly, signals that TEX carry are highly amplified, as they are distributed among millions of TEX released by tumor cells. We estimate that plasma of melanoma patients may contain as many as 10¹² sEVs/ml, of which TEX represent a variable proportion, depending on the disease stage/activity (22). TEX carry and deliver these signals throughout the body, although recent data indicate that the lung and liver are preferential targets of sEVs (47). Invariably, TEX interact with circulating and tissue-bound immune cells, and the image of these interactions that comes to mind has a T cell surrounded by a huge excess of sEVs simultaneously delivering multiple signals that result in an accelerated response. The latter is likely to be a sum of conflicting inhibitory/stimulatory signals that the recipient T cell must sort out and respond to. Clearly, the response that the T cell makes will determine its fate. In the TME, an excess of TEX carrying immunoinhibitory signals shifts the balance to suppression or cell death.

The coterie of proteins decorating the surface membrane of TEX (Fig. 1) determines, enables, and presumably regulates TEX initial engagements with recipient immune cells, including a selection of recipient cell, uptake/internalization, and translation of the delivered signals into actionable transcriptional or translational changes (48). Although it has been suggested that TEX may be directed by the tumor to a specific cellular address, evidence in support of such preselected delivery is scarce. Perhaps the data showing preferential engagement of TEX with activated CD8⁺ effector T cells culminating in apoptosis and much less effective signaling with activated CD4⁺ T cells and NK cells could be considered as a preferential “target cell selection” (49). However, the molecular basis of this preferential elimination by TEX of activated CD8⁺ T cells remains unclear. In contrast, molecular mechanisms involved in TEX uptake by different types of immune or nonimmune recipient cells are defined and may involve phagocytosis, membrane fusion, opsonization, endocytosis, pinocytosis, and/or surface receptor/ligand signaling (50). TEX are rich in integrins, especially α_Vβ₃ integrin, which facilitate their entry into recipient cells, promoting metastasis (51). The time of TEX entry varies broadly, and it is determined by a recipient cell type. Specifically, the entry of EVs into phagocytic cells such as dendritic cells (DCs) or macrophages is rapid (few minutes), takes longer in T cells (minutes to hours) depending on the T cell activation, or is variable, ranging from minutes to hours in malignant or nonmalignant epithelial cells. Once internalized, EVs deliver their cargo and simultaneously activate several intracellular molecular pathways, resulting in phenotypic and functional reprogramming of the recipient cells (45). The molecular signals driving EV-mediated reprogramming in various types of recipient cells are under intense current investigation.

Emerging data suggest that in cancer, TEX drive tumor-mediated reprogramming of all cells, impacting functions of immune as well as nonimmune cells with profound consequences for tumor immune escape (24, 45, 52). TEX transfer horizontal information between different nonmalignant cells or other cancer cells within the TME (46). This information transfer results in extensive metabolic reprogramming of recipient cells (52). Focusing on T lymphocytes, we and others have reported that in melanoma (and other cancers) activated CD8⁺ effector T cells interacting with TEX undergo apoptosis that could not be blocked with Abs to death receptor ligands carried on the surface of the interacting partners (49). Recently, we investigated expression levels of various signaling receptors (PD-1, CTLA-4, Fas, CD39, CD73, CD40, OX40, TNFR, DR4, DR5) and ligands (PD-L1, CD80, FasL, TRAIL, CD40L, OX40L) on the surface of activated recipient immune cells just prior to their interaction with TEX in coincubation assays and TEX-induced apoptosis. We observed relative sensitivity of activated CD8⁺ T cells and resistance of B cells and NK cells to TEX-induced apoptosis and speculated that this might be due to differences in expression levels of receptors/ligands on the surface of the recipient cells (49). Indeed, expression levels of the apoptosis-relevant proteins as well as costimulatory molecules differed in various immune cells: CD8⁺ T cells expressed high levels of PD-L1, Fas, CD39, and CD73 and somewhat lower levels of OX40 and Ox40L; B cells expressed lower levels of inhibitory receptors/ligands relative to CD8⁺ T cells and showed enrichment in CD40, OX40, and CD80 costimulatory proteins; and NK cells expressed moderate PD-L1 levels and high levels of Fas and TNFR2, but relatively low levels of the other death receptors/ligands (49). A similar screen of immunoregulatory proteins carried on the TEX surface revealed moderate expression levels of all suppressive proteins and a variable enrichment in costimulatory proteins, the finding of which is consistent with the ability of TEX to either suppress or activate recipient cells. Fig. 2 presents an image of T cells interacting with TEX, with each carrying at least one immunoregulatory protein on its surface, although in reality, each TEX likely carries multiple signaling proteins. TEX bind to complementary receptors/ligands on the T cell surface and induce signals that may be either stimulatory or inhibitory. The cumulative effect of this signaling by multiple TEX results in immune suppression or activation depending on the ratio of suppressive/stimulatory proteins in the TEX cargo. In the TME, inhibitory signals prevail, and the recipient T cell undergoes apoptosis, which is strictly dependent on the TEX numbers and composition of molecular cargo in TEX-engaging T cells.

The TEX–T cell engagement illustrated in Fig. 2 is based on the concept that death receptor/ligand signaling determines the fate of recipient T cells via extrinsic apoptosis. Surprisingly, however, blocking of death receptor/ligand signaling with neutralizing Abs specific for Fas, PD-1, CTLA-4, TRAIL, CD39, CD73, or with TGF-β inhibitors failed to prevent TEX-induced T cell apoptosis (49). Only ∼10% of apoptotic activity relative to isotype controls was measured when a mix of all neutralizing reagents was used. Furthermore, when an attempt was made to block entry of PKH26-labeled TEX into recipient T cells using protein kinase or heat at 80°C or inhibitors of vesicle uptake, such as Dynasore, Pitstop, or cytochalasin D, apoptosis of recipient T cells was reduced but not blocked. Interestingly, none of the inhibitors used was able to completely block TEX entry into T cells or the resulting apoptosis. Considering a possibility that TEX entering T cells delivered caspases or agents promoting caspase activity, we pretreated recipient T cells with the pan-caspase inhibitor Z-VAD-FMK, and neither TEX entry into T cells nor T cell apoptosis was inhibited (49). In aggregate, these studies led to the conclusion that TEX entering T cell cytosol induce intrinsic apoptosis, which is associated with downregulation in expression of survival proteins Bcl-2, Bcl-x_L, and cFLIP and with a mitochondrial injury (49). The latter resulted in leakage from mitochondria and accumulation in the cytosol of cytochrome c and Smac, as shown by Western blots of the mitochondrial and cytosol fractions isolated from recipient T cells coincubated with TEX (49). Recent studies suggest that TEX entering into a T cell create physical stress and deliver extraneous materials, including nucleic acids, to the cytosol, culminating in molecular alterations known as stress-associated molecular patterns (53, 54). Our unpublished data show that upregulated expression of stress proteins (PERK, BIP, CHOP, ATF4, or IRE1) and the unfolded protein response are manifestations of TEX entry into T cells and drive their apoptosis. Because mitochondria are highly sensitive to cellular stress, it appears that TEX entering a T cell alter mitochondrial integrity and functions, resulting in unavoidable, relentless T cell death that cannot be stopped (49). We showed that this mechanism is generally used by TEX produced by different types of tumor cells but not by sEVs derived from nonmalignant cells used as controls in the coincubation experiments. The nature of signals delivered by TEX that initiate mitochondrial dysfunction in T cells is under current investigation.

Interestingly, B cells and NK cells are less sensitive to TEX-mediated intrinsic apoptosis, possibly because differential expression of receptor proteins on their surfaces leads to activation of alternative molecular pathways in these recipient cells. Recent data suggest that TEX from various tumor cells cause NK cell dysfunction through a variety of mechanisms. NK cells were reported to downregulate expression of NKG2D and to reduce perforin, granzyme B, and IFN-γ production, resulting in a loss of cytolytic activity after coincubation with TEX (55). In B cells, TEX were shown to activate the adenosinergic pathway, leading to excessive production of immunoinhibitory adenosine and failure to proliferate (56, 57).

DCs, which are functionally indispensable for the development of specific antitumor immunity, were shown to be impaired in maturation and functions after incubation with TEX obtained from cancer plasma or supernatants of tumor cell lines (21, 58, 59). Coculture of human DCs and cultured melanoma cells was shown to result in decreased expression of the Ag-presenting machinery components and impaired functions of DC (60). Similarly, TEX from tumor cell line supernatants decreased expression levels of the Ag-presenting machinery components involved in Ag processing, specifically TAP1, in human immature DCs (21), whereas in matured human DCs, TEX downregulated expression levels of costimulatory surface proteins and the HLA complex, presumably interfering with Ag presentation to cognate T cells. Thus, TEX are negative modulators of DC differentiation into competent APCs.

M0 macrophages that are generated from monocytes using M-CSF or GM-CSF and coincubated with TEX are readily polarized to proinflammatory M1 or tumor-like M2 (tumor-associated macrophages). TEX, signaling directly via surface-associated proteins, are as effective in mediating M0 macrophage polarization into M1 and M2 as are cytokines (e.g., IFN-γ) or other commonly used polarizing agents, including LPS (61). They have been shown to induce and regulate M2 differentiation via the engagement of various molecular pathways, for example, downregulation of PTEN expression and activation of the PI3K/AKT signaling pathway (62, 63). TEX-induced expression of PD-L1 has been observed in macrophages and monocytes (64). These results show that TEX can directly interact with macrophages, interfering with their differentiation and enhancing acquisition of the metastasis-promoting, tumor-associated macrophage–like phenotype (65).

Mechanisms of information transfer by TEX may also involve indirect TEX interactions with the recipient immune cells that are mediated via Tregs and/or myeloid-derived suppressor cells (66–68). TEX drive proliferation and differentiation of these regulatory immune cells and enhance their suppressive functions (69). Specifically, TEX upregulate expression levels of surface ectonucleotides, CD39 and CD73, and adenosine production by Tregs, converting them into highly effective immunosuppressive cells (68).

It is becoming increasingly clear that TEX also reprogram functions of nonimmune cells in the TME. For example, evidence is emerging that cancer-associated fibroblasts produce sEVs that mediate cancer-associated fibroblast–tumor cell interactions and drive tumor progression via hyperactivation of TGF-β signaling in recipient tumor cells (70). Similarly, sEVs in the TME were reported to drive epithelial–mesenchymal transition and cancer metastasis (71). A major consequence of TEX-driven reprogramming of immune and nonimmune cells in the TME is acquisition by these recipient cells of the ability to produce and release a new crop of sEVs that mimic the protumor attributes of TEX (i.e., are immunosuppressive) and at the same time phenotypically and functionally resemble the parental T cells. This is best illustrated by TEX reprogramming of tumor-infiltrating T cells. Because tumors produce excessive numbers of TEX that effectively target and reprogram tumor-infiltrating T cells, the TME becomes rapidly filled with CD3⁺ sEVs mediating immunosuppressive protumor activities, and the vicious cycle of immune cell dysfunction and death initiated by TEX continues as long as the tumor delivers TEX and reprogrammed T cells produce CD3⁺ sEVs. Mechanistically, reprogramming of T cells (and other immune or tissue cells) is accomplished by TEX through delivery to the recipient cell surface of signals, which can be transient (“touch-and-go”) or sustained, and which initiate transcriptional activity in the responding cells. Upon entry, TEX disrobe and transfer their content of proteins, enzymes, cytokines, and nucleic acids (mRNA, microRNA, DNA) from the tumor to the recipient cells. Thus, both surface and laminar components of TEX take part in reprogramming. An excessive TEX production by the tumor (think billions of copies) ensures that these transcriptional and subsequent translational activities are amplified by the enhanced vesiculation of the reprogrammed recipient cells.

It is well documented that T cells in the tumor-bearing hosts frequently have compromised immune competence, specifically lacking the ability to arrest cancer progression (72, 73). This is largely a result of TEX-mediated reprogramming of effector T cells. The latter, including T lymphocytes isolated from tumor tissues or body fluids (i.e., T lymphocytes present in the TME), are phenotypically and functionally different from T cells in HDs (22). They produce CD3⁺ sEVs, which, similar to TEX, have an immunosuppressive profile and mediate immune suppression (26). Thus, the isolation from cancer plasma and examination of T cell–derived CD3⁺ sEVs, which reciprocate the protein profile of parent T cells, provides key information about the frequency as well as the phenotypic and functional status of the patients’ T cells, potentially serving as a “liquid T cell biopsy” (22). The emerging data confirm this hypothesis and provide the rationale for monitoring CD3⁺ sEVs in lieu of T lymphocytes in cancer body fluids (D.H. Zandberg et al., submitted for publication). Because changes in the frequency, phenotype, and functions of CD3⁺ sEVs mimic those of the producer T cells, CD3⁺ sEVs can serve as T cell surrogates, and the degree of immune dysfunction that cancer induces in the patients’ T lymphocytes can be determined using CD3⁺ sEVs instead of T lymphocytes. In our hands, the CD3⁺ sEV frequency in pretherapy plasma correlated with progression-free survival at p < 0.005 and overall survival at p < 0.007 in 24 patients with recurrent/metastatic head and neck squamous cell carcinoma treated with the standard of care immunotherapy (IT) (pembrolizumab) (39). The study illustrates the potential of CD3⁺ sEV frequency in the patients’ plasma as a predictive biomarker of favorable response to IT. In the future, phenotyping of circulating CD3⁺ sEVs in a few drops of plasma could replace immune monitoring of T lymphocytes for measuring their immune activity or exhaustion.

TEX are present in large but variable numbers in all body fluids of cancer-bearing hosts. When TEX encounter immune cells, they deliver information that is processed by recipient cells and leads to suppression of the recipient cell activities, culminating in cell death. In HDs, circulating sEVs present in lower numbers and carrying different molecular cargos do not induce significant T cell apoptosis (48). The negative signaling by TEX contributes to dysfunction or exhaustion of effector T cells, suggesting that TEX are one of many manifestations of tumor-induced immune suppression that characterize cancer progression. In this context, TEX likely play an important role in cancer development, progression, and, potentially, response to ITs. The ability of TEX to induce extrinsic and intrinsic apoptosis in activated T cells or dysfunctions in NK cells, B cells, DCs, and macrophages suggests that TEX might contribute to resistance of tumors to oncologic therapies, as previously suggested (74). Because current anti-cancer IT is based on the principle that an efficient blockade of negative signaling by the tumor is necessary for achieving favorable therapeutic response, the abundance of immunosuppressive TEX in body fluids of cancer patients, especially those with advanced disease, is likely to counterbalance restorative effects of IT. The prediction would be that at elevated frequencies, TEX, which avidly sequester and remove anti–PD-1 therapeutic Abs and simultaneously eliminate/disarm newly rejuvenated, activated effector T cells (75), attenuate therapeutic benefits of IT. Adoptively transferred activated T cells, NK cells, or engineered CAR T cells might be especially vulnerable to TEX-mediated apoptosis. For example, the therapeutic failure of adoptive IT with activated NK cells in patients with acute myelogenous leukemia was linked to accumulations prior to IT of highly immunosuppressive TEX in the patients’ peripheral blood (76). Preliminary data confirm concentration-dependent susceptibility of CAR T cells to TEX-mediated lysis in vitro (author’s unpublished data), and TEX equipped with a targeted tumor Ag and PD-L1 were reported to preferentially target cognate CAR T cells, inhibiting their proliferation, migration, and function (77). These studies serve as examples of the TEX-negative role in cancer IT.

Recent data indicate that TEX also emerge as potentially highly specific and sensitive prognostic and predictive biomarkers of cancer progression and survival (78, 79). Two features of TEX account for their efficacy as biomarkers: 1) they carry proteins and transcriptional signatures that associate with the tumor clinicopathologic status, disease progression, and outcome (9, 25); and 2) they have the capability to reprogram phenotype and functions of targeted recipient cells, thereby revealing the degree of cancer-induced dysfunction that can be linked to disease activity, stage, progression, and response to therapy (24).

Furthermore, TEX carry a broad spectrum of immunosuppressive and immunostimulatory signals and can selectively upregulate functions of targeted immune cells (e.g., Tregs) or nonimmune cells (e.g., endothelial cells) to promote protumor functions. The upregulation of suppressor functions by Tregs (66) and promotion of angiogenesis in the case of endothelial cells (80) are well-documented examples of indirect protumor effects that TEX execute.

Because TEX directly and indirectly restrict antitumor immunity and promote resistance to cancer immunotherapy, strategies are being developed to reduce or eliminate TEX or to redirect their suppressive activities. TEX carry on their surface TAAs along with costimulatory proteins and could in theory be exploited as adjuvant-containing anticancer vaccines. The concept of TEX as vaccines for cancer originated in late 1990s (81) and was explored in tumor-bearing mice with some preliminary success using DC-derived exosomes (82) or from tumor cells (TEX) (83). However, early clinical trials of TEX-based therapeutic vaccines in patients with cancer demonstrated only moderate efficacy in enhancing antitumor immunity or overall patient survival (84, 85). Historically, overcoming tumor-induced immune suppression in patients with cancer has proven difficult, and overcoming negative TEX signaling, which predominates in the TME of cancer patients, is especially challenging. It would require 1) effective removal of circulating TEX prior to vaccination or immune therapy; 2) the simultaneous blockade of TEX production by tumors; 3) replacement of patients’ TEX with infusions of in vitro–generated TEX engineered to carry potent immunostimulatory signals and to selectively mediate suppression of endogenous inhibitory TEX. These strategies are currently being experimented with in tumor-bearing animals, but their application in the clinic is problematic at best. The therapeutic objective of removal/suppression of “bad” TEX without interference with functions of sEVs necessary for intercellular communication represents a major barrier. Nevertheless, numerous preclinical and more recent clinical studies are currently using engineered EVs for vaccination of mice or humans (2). Despite the existing challenges, there is hope that by introducing novel adjuvant strategies in combination with exosomes, antitumor efficacy of DC-derived exosome– or TEX-based vaccines will be increased to generate and sustain antitumor immunity in patients with cancer.

For many years, tumor immunity has been a “dark side” of the immune system. Remarkably adapted to efficiently deal with foreign or endogenous threats, the immune system fails to prevent or eliminate cancers. This is because tumors create a protective barrier that includes the production of a broad variety of factors targeting and disabling antitumor immune cells. Tumors have adapted the vesicular system used in health as a means of intercellular communication to specifically subvert antitumor immunity. TEX that tumors produce in great abundance mimic the contents of tumor cells and are equipped to export and deliver to immune cells the molecular/genetic signals that tumor utilizes for in situ protection. TEX distribute these signals near and far throughout the body and might be programmed by the tumor to target specific immune cell subsets. TEX also serve as an amplification system, as any one signal is distributed among billions of sEVs that carry and deliver it to immune cells that are activated, that is, armed to eliminate tumor cells. Thus, among the immunosuppressive factors that tumors produce, TEX are especially dangerous to effector immune cells, including those used for adoptive therapies or for T cells rejuvenating after IT. In cancer, the host immune system faced with the corrupted intercellular information transfer and insulted by an excess of TEX specifically targeting immune effector mechanisms is unable to cope. As we become aware of the mechanisms and magnitude of TEX protumor activities, the question arises of how to break the vicious cycle that tumors create and protect immune cells from TEX. Studies in progress aiming at disarming/redirecting TEX by genetic engineering or decreasing their production/release by tumor cells are ongoing (86, 87). The challenge is to protect the immune system without interfering with exosomal transfer of intercellular messages required for life. To meet the challenge, further understanding of the intricate balance existing between the immune system and the vesicular biology is necessary.

The author has no financial conflicts of interest.