The potential role of tumor-derived exosomes in diagnosis, prognosis and response to therapy in cancer

Abstract

Introduction:

Small extracellular vesicles (sEV) produced by tumors and called TEX mediate communication and regulate the tumor microenvironment. As a “liquid tumor biopsy” and with the ability to induce pro-tumor reprogramming, TEX offer a promising approach to monitoring cancer progression or response to therapy.

Areas Covered:

TEX isolation from body fluids and separation by immunoaffinity capture from other EVs enables TEX molecular and functional characterization in vitro and in vivo. TEX carry membrane-bound PD-L1 and a rich cargo of other proteins and nucleic acids that reflect the tumor content and activity. TEX transfer this cargo to recipient cells, activating various molecular pathways and inducing pro-tumor transcriptional changes. TEX may interfere with immune therapies, and TEX plasma levels correlate with patients’ responses to therapy. TEX induce local and systemic alterations in immune cells which may have a prognostic value.

Expert opinion:

TEX have a special advantage as potential cancer biomarkers. Their cargo emerges as a correlate of developing or progressing malignant disease; their phenotype mimics that of the tumor; and their functional reprogramming of immune cells provides a reading of the patients’ immune status prior and post immunotherapy. Validation of TEX and T-cell-derived sEV as cancer biomarkers is an impending future task.

1.0. Introduction

Over the last decade, an increasing number of studies have provided evidence that tumors develop diverse and powerful defensive strategies to protect themselves from the host responses [1]. This phenomenon is referred to as ‘tumor immune escape,’ and it has been recognized as a major mechanism responsible for resistance of tumors to immune therapies and for the lack of success in generating long-lasting clinical responses to vaccines or immune checkpoint inhibitors (ICIs) in cancer patients [2,3]. Immune escape of tumors is best illustrated by in vivo studies in tumor models, where the delivery of immunogenic stimuli to immune-competent animals promotes the development of tumor antigen-reactive T-lymphocytes but ultimately fails to provide effective anti-tumor immunity and instead induces tolerance and enhances tumor progression [4]. Tumors escape the host immune system by creating a highly immunosuppressive tumor microenvironment (TME), which favors tumor progression. The establishment of the TME involves a gradual reorganization of the tumor stroma, numerous alterations in cytokine profiles with enrichment in immunosuppressive factors (TGF-β, IL-10), induction and recruitment of immunosuppressive cells, such as regulatory T cells (Tregs) or myeloid-derived suppressor cells (MDSC) into tumors, and targeted elimination or functional paralysis of tumor-specific cytotoxic T cells (CTL) [5,6]. Exosomes released by cancer cells called tumor-derived exosomes or TEX are emerging as one of the key players of tumor-mediated immune suppression in the TME and in the periphery in patients with cancer [7].

The goal of this review is to examine the role of TEX in cancer progression and to provide a critical analysis of recent scientific findings that report on mechanisms used by TEX to directly and/or indirectly modulate functions of immune cells in the TME and in the peripheral tissues of patients with cancer. We will address the impact TEX appear to exert on patients’ responses to immunotherapies. We will also consider the potential of TEX as components of liquid tumor biopsies and will evaluate the current status of TEX as non-invasive cancer biomarkers. Finally, focusing on the latest discoveries, we will discuss the intercellular cross-talk that TEX engage in, orchestrating communication between tumor cells, tumor-surrounding non-malignant cells in the stroma and local or circulating immune cells in tumor-bearing hosts.

2.0. Definition of TEX and TEX origin

TEX are produced and released into the extracellular space by tumor cells. Much of what we know about TEX comes from studies of murine and human tumor cell lines, where all cells produce small extracellular vesicles (sEV) that are TEX, and which can be readily recovered from culture supernatants for analysis [8]. The nomenclature of EVs has not been finalized, and the International Society for Extracellular Vesicles (ISEV) has issued and updated guidelines for studies and classification of EVs [9]. The guidelines call for reporting the subcellular origin, size, biochemical composition and cell/tissue origin of EVs to establish the working definition for various EV subpopulations. Studies of TEX from supernatants of cultured tumor cells provided data informing about TEX biogenesis, molecular and genetic content of TEX and their in vitro and in vivo functions. Among various EVs produced by tumor cells, exosomes represent the smallest subpopulation of EVs with a diameter of 30–150 nm and the endocytic biogenesis that is distinct from that of larger EVs, such as microvesicles (MVs: 200–1,000 nm) or apoptotic bodies (>1,000nm). Exosome biogenesis has been recently reviewed in depth [10,11] and only a short overview is presented here (Figure 1). It begins with the inward invagination of the parent cell plasma membrane followed by a vesicle closure to form early endosomes. Then, intraluminal vesicles (ILVs) are formed within the lumen of the early endosomes through numerous inward membrane invaginations, leading to the formation of mature, ILV-containing multivesicular bodies (MVBs). A fraction of endosomal proteins and other cellular contents not directed to the lysosomal pathway are sorted into ILVs during the MVBs formation. The endosomal sorting complex required for transport (ESCRT) is the key regulator of early endosomes maturation, MVB formation and cargo sorting [12]. It consists of four protein complexes (ESCRT-0, ESCRT-I, ESCCRT-II and ESCRT-III), and together with an ESCRT-independent pathway that mainly recruits lipids, is responsible for sorting and directing the intraluminal cargo to the maturing MVB. Upon maturation, the MVB filled with ILVs fuses with the cellular plasma membrane, resulting in the release of the mature ILVs as sEVs, also called exosomes, into the extracellular space (Figure 1). Proteins of the Rab family of GTPases play a key role in the exosome biogenesis and in exosome release into the extracellular environment. In contrast to exosomes, MVs are formed by “budding” of the plasma membrane. While TEX isolated from tumor cell supernatants, as discussed below, are largely small EVs of endocytic origin, they might also contain various proportions of MVs, depending on the method employed for their isolation. In this review, sEVs (50–150 nm) originating from MVBs of tumor cells that carry endocytic markers and the tumor-associated molecular cargo are referred to as TEX.

Figure 1. Biogenesis of TEX and uptake by a recipient cell.

Endocytosis of proteins expressed on the surface of the parent tumor cell and their processing in the endosomes leads to formation of intraluminal vesicles (ILVs) by inverse invagination of the late endosome membrane. Multivesicular bodies (MVBs) filled with ILVs are formed by fusion of late endosomes. Note that topography of the proteins in vesicles resembles that of the parent cell surface. Upon MVB fusion with the cell membrane, vesicles are released into extracellular space. They deliver their cargo to the recipient cell by one or more of several mechanisms which may include phagocytosis, clathrin-dependent endocytosis, lipid raft-mediated uptake, caveolin-mediated internalization, direct fusion or receptor-ligand binding followed by uptake. Surface-delivered signals and activation of intracellular molecular pathways by the internalized molecules lead to phenotypic and functional reprogramming of the recipient cell.

Tumor cells produce large quantities of TEX that circulate freely in all body fluids [13]. The molecular content of TEX resembles that of the parent tumor cells, including the topography of proteins on the surface of TEX (Figure 1). It has been argued that since TEX mimic the content of the parent tumor cells, they could serve as readily accessible, non-invasive “liquid tumor biopsies.” TEX carry a rich membrane-associated cargo of immunosuppressive and immunostimulatory proteins, MHC class I and II antigens, tetraspanins (e.g., CD9, CD63, CD81, CD82), chaperones (e.g., heat‐shock protein (HSP) 70, HSP90), adhesion molecules (e.g., integrins), tumor-associated antigens (TAAs), lipids, and glycolipids. In their lumen, TEX have a large portfolio of cellular proteins, enzymes, soluble factors and nucleic acids, including miRNAs [14]. The next critical step is the delivery of the molecular cargo TEX carry to recipient cells. TEX are known to be able to interact with their recipient cells by direct signaling via receptor-ligand binding using cognate receptors or ligands expressed on surfaces of targeted cells (Figure 1). TEX can also be taken up by recipient cells through various other mechanisms, such as lipid raft‐, caveolae‐, and clathrin‐dependent endocytosis, phagocytosis or micropinocytosis, and deliver their content, including nucleic acids, such as mRNA and micro-RNA, to recipient cells [15]. It has been suggested that components of the TEX cargo determine the tissue‐ and organ‐specificity of released TEX. In other words, TEX produced by parent tumor cells are provided with an address of their ultimate destination upon release into extracellular space. The existence and exact molecular content of this putative address system remain speculative and are still under investigation. TEX are rich in integrins, for example, and different integrins in their surface membrane have been shown to regulate TEX adhesion to specific cell types or to extracellular matrix (ECM) components in particular organs [16]. In any case, TEX deliver molecular messages from the tumor to nonmalignant cells, including immune cells, present in the TME and in the periphery [17]. Figure 2 illustrates interactions of TEX with various cells residing in the TME and effects of these interactions on the tumor promotion. In addition to these juxtracrine interactions with locally-situated cells, TEX also mediate autocrine effects, further enhancing tumor growth and metastasis [7]. TEX-mediated effects can be direct, such as, e.g., functional suppression of immune cells by engaging inhibitory receptors on their cell surface, or can be indirect, involving functional reprogramming of recipient cells which subsequently orchestrate promotion of the tumor. Whether direct or indirect, results of TEX signaling or uptake involve dramatic transcriptional changes of multiple genes in recipient cells towards the phenotype that supports tumor growth and its survival [18,19]. Based on these data, TEX are considered to be promising biomarkers of cancer diagnosis and progression as well as potentially important surrogates of tumor-induced immune suppression in cancer and as biomarkers of responses to immune therapies.

Figure 2. The schematic view of effects TEX exert on cells in the TME.

TEX can mediate direct effects or indirect effects, where the cells reprogrammed by TEX induce changes in phenotypic and functional characteristics of their neighbors in the tumor. TEX also mediate autocrine effects promoting growth of autologous tumor as well as paracrine effects promoting tumor metastasis to sites distant from the primary tumor.

3.0. Methods for TEX isolation

The current challenge in the field is the establishment of the ‘gold standard’ for the isolation of TEX that can provide high recovery of morphologically intact, purified and functional vesicles of endocytic origin and that is reproducible as well as applicable to processing of clinical samples. The field has progressed from investigation of TEX in supernatants of cultured tumor cell lines to analysis of TEX in complex biological fluids, especially plasma and urine [20]. Reproducibility of TEX isolation methods is especially critical when serial samples of biological fluids are to be tested. Exosomes in biological fluids are heterogenous in size, origin and composition. Exosomes may share physicochemical properties with other EVs, in which case co-isolation of different types of EVs is likely to occur. The co-isolation of exosomes with liposomes, chylomicrons or MVs and the presence of contaminants (e.g., lipoproteins or protein aggregates) is a common barrier to the success of currently practiced isolation techniques.

3.1. Isolation by differential ultracentrifugation

Differential ultracentrifugation (diff. UC) has been the most commonly used method for isolating exosomes, especially from cell culture media [21]. This method was used in the initial studies reporting on the exosome isolation from the supernatant of reticulocytes in 1987 [22]. Diff UC sediments exosomes based on their size, using forces from 100,000 to 200,000xg. To remove debris and larger vesicles, several differential centrifugation steps at lower gravitational forces precede the final exosome-pelleting step [23]. However, exosomes isolated by diff UC from biological fluids are often of low purity, because MVs, lipoproteins and protein aggregates of similar sizes, density and mass might be co-pelleted. The method also has several other drawbacks, including low reproducibility, low throughput due to time and equipment required for UC, low RNA-yields and exosome aggregation or even damage, all of which are not compatible with clinical utilization [24,25]. Nonetheless, this method has been favored by researchers who use conditioned cell culture media as an exosome source, since it allows for large volume preparation and, aside from a requirement for an ultracentrifuge, is inexpensive and relatively simple.

As an improvement of the diff. UC method, another type of centrifugation commonly used for the separation of blood cells has been adapted for exosome isolation, i.e., the density-gradient UC. Here, a continuous or stepwise density gradient is built up in isolation medium prior to centrifugation, and the EVs are separated according to their density within the gradient. Usually, protein aggregates concentrate at the bottom of the centrifugation tube, while exosomes are recovered in the layer of medium with density between 1.10 and 1.18 gmL⁻¹. As the isolation gradient, a single or double sucrose or iodixanol cushion may be used. In a comparative study, Paolini et al showed that density-gradient UC provided exosomes with higher purity and well preserved biological functions relative to the classical UC or a one-step precipitation kit [26]. However, although density-gradient UC can effectively separate exosomes from common contaminants like protein aggregates, it cannot separate EVs with buoyant density similar to that of exosomes, such as small MVs. To solve this problem, moving-zone density-gradient centrifugation (also known as rate zonal centrifugation) may be used, which separates particles by size and density [27]. Although both density-gradient isolation methods are an improvement over the classical diff. UC, they also have disadvantages, including a limited input volume, high sample loss, potential exosome damage and loss of biological functions.

3.2. Isolation by other methods

Exosome isolation from human plasma or serum is especially challenging, because they not only contain mixtures of vesicles derived from different cell types but also components such as lipoproteins, apoptotic bodies, plasma proteins such as immunoglobulins and albumin, chylomicrons etc., which should be removed during the isolation process. It has been shown that in human plasma, lipoproteins are present at several orders of magnitude higher concentrations than EVs and cannot be successfully separated by traditional one-step physical-base purification methods [28,29]. Karimi et al has proposed a two-step isolation procedure combining density cushion separation followed by size-exclusion chromatography, which separated plasma EVs from lipoproteins and plasma proteins with an efficiency that allowed mass-spectroscopic analysis of the isolated EVs [30].

In the last 10 years, a size-based separation technique termed size-exclusion chromatography (SEC), widely applied for the high-resolution separation of macromolecules, has been adapted for exosome isolation and is currently available as commercial isolation kits. SEC is especially well suited for exosome isolation from biological fluids, since it removes most, but not all, contaminating proteins. The advantages of SEC are a minimal volume requirement and minimal sample loss, which are important for processing of limited samples such as cerebrospinal fluid (CSF). Also, SEC is a high throughput, relatively rapid method with excellent reproducibility when using small (0.5–1.0ml) sample volumes and a high yield of morphologically intact, biologically active exosomes [31,32]. We have adapted this method in our laboratory for isolation of total plasma exosomes from cancer patients’ plasma for functional studies and therapeutic applications [33]. However, SEC is not perfect, as it does not provide a successful separation of vesicles from lipoproteins and protein aggregates of a similar size. To remove lipoproteins, an additional step of ultrafiltration is often used [34,35].

Numerus new and increasingly sophisticated techniques are being introduced almost daily for isolation of EVs or EV subsets [36]. As these technological advances are being introduced and evaluated, the field remains uncertain as to the optimal EV isolation strategy. The choice of the appropriate isolation method depends on the vesicle source and the intended downstream analysis of the isolated vesicles and in all cases should consider vesicle integrity, recovery and purity [37]. Therefore, no specific isolation separation method that is suitable for all applications is currently available and may never be available. In practice, a combined application of two or more techniques offers the best strategy for TEX isolation (for example, SEC combined with immunoaffinity capture or density gradient centrifugation and filtration). A tailor-made choice of the isolation method for a given TEX source may have to be devised and may require the isolation of total exosomes from a body fluid followed by capture that is based on the availability of one or a panel of antibodies with specificity for antigens overexpressed in parent tumor cells and carried by TEX.

3.3. Immunoaffinity capture of TEX

Exosomes isolated from cancer patient’s plasma by SEC or other method are a mix of vesicles derived from cancer cells (TEX) and non-cancer cells (non-TEX). Thus, it becomes necessary to separate TEX from non-TEX to evaluate the role of TEX as cancer biomarkers. On their surface, TEX carry a unique set of membrane-embedded molecules, which mimic those present in the parent tumor cells [38]. Using antibodies specific for these TAAs, TEX could be separated from non-tumor-derived vesicles by immunoaffinity capture. We have developed a two-step method to separate TEX from non-TEX, combining an initial purification and concentration by mini-SEC with immunoaffinity-based capture of TEX, using antibodies specific for TAAs present only in the tumor and TEX but not expressed on normal cells or non-TEX [39]. This method is described in a greater detail in Sections 7.1 and 7.2 of this review. We and others also use microarrays or chips coated with antibodies specific for antigens overexpressed by tumor cells and carried by TEX these cells produce. Since Abs specific for tumor antigens are rarely available, a mix of several Abs recognizing antigens known to be overexpressed and essential for the tumor wellbeing are “painted” on the surface of a solid support [40] and used for TEX capture from body fluids. Proteins carried on the surface of the immunocaptured TEX are then identified and quantified by staining with labeled detection mAbs followed by confocal imaging or flow-based detection [41].

3.4. Antigen detection on TEX surface

Flow cytometry has become a useful tool for detection of antigens carried by immunocaptured TEX. Flow cytometers equipped with high-sensitivity lasers able to detect nanoparticles have become available for exosome phenotyping but in their absence, on-bead flow cytometry can be utilized as described [42,43]. Using flow and pre-titered fluorochrome-labeled detection Abs, we are able to obtain semi-quantitative individual assessments based on relative flow intensity (RFI) values for each targeted protein carried by TEX [44]. The Ab-based amplification increases sensitivity of detection even beyond that of mass spectrometry to enable detection of exosome antigens present in femtomolar quantities. In addition, the method makes it possible to measure double-or multiple-labeled exosomes and to evaluate intraluminal content of vesicles after permeabilization.

The immunocapture methods can be readily upscaled, so that quantities of the isolated exosome fractions are adequate for either the protein or miRNA detection or both. The detection steps (i.e., profiling) can be targeted to specific cargo components. The approach is limited by the availability of tumor antigen-specific Abs. To overcome this limitation, mixtures of Abs recognizing TAAs highly overexpressed on tumor cells, such as EPCAM, EGFR or CSPG4, are used for enrichment for TEX and have been used for the construction of microarrays for exosome capture from body fluids [45–48]. The immune based capture and characterization of TEX obtained from plasma or other biofluids of patients offers a unique opportunity for future molecular profiling of the TEX cargo in relation to clinical and clinicopathological data as well as disease activity in patients with different tumor types. Preliminary data suggest that separation of TEX from total plasma exosomes is a highly promising platform for the development of biomarkers for monitoring cancer diagnosis, progression and response to therapy [42–45].

4.0. Functional impact of TEX on cells in the TME and in the periphery

TEX isolated from supernatants of cultured tumor cells or body fluids are expected to be functionally competent. Literature suggests that TEX can interact with a wide variety of recipient cell types in vitro or in vivo, when injected into experimental animals, and exert profound effects on functions of recipient cells [49]. This lack of selectivity in TEX uptake and interactions with various cellular targets is best appreciated upon examining the TME. It appears that functions of all components of the TME are influenced by TEX, as indicated in Figure 2. Further, TEX-induced changes in the TME clearly benefit the tumor by promoting tumor growth and metastasis, enhancing blood vessel growth, altering structural elements of the extracellular matrix (ECM) and inducing immune suppression in the infiltrating hematopoietic cells. This large-scale reprogramming of the TME by TEX begins early in the process of carcinogenesis, e.g., at the pre-malignant lesion stage, as shown by Razzo et al. [50], and continues throughout tumor progression and metastasis [51]. While it has been accepted in oncology that the growing tumor alters its environment, molecular mechanisms responsible for implementation of the tumor-induced alterations have remained obscure and have been variously attributed to genetic changes, signaling events, soluble factors, metabolic alterations, cytokine cascades or some/all of the above (reviewed in [52,53]).

Current data indicate that TEX, serving as tumor messengers capable of reaching and addressing all cells in the TME, are responsible for the TME reprogramming and for promoting epithelial-to-mesenchymal transition [54]. The TEX attributes, such as rapid intercellular distribution, surface interactions via receptor-ligand signaling, efficient uptake by endocytosis or phagocytosis, all attest to the TEX ability to deliver membrane-protected, biologically-active genetic and molecular elements to various non-malignant and malignant cells in the TME and beyond. Table 1, presents a selected list of functional changes that have been shown to be mediated by TEX in the TME of solid tumors and indicates the cell types in the TME that are known to be affected by TEX. Not surprisingly, the changes TEX induce extend to autologous tumor cells, stem cells, fibroblasts, mesenchymal stromal cells, fibroblasts, endothelial cells and tumor-infiltrating immune cells. Notably, mechanisms responsible for TEX induced changes in the TME vary but involve either signaling by proteins or transcriptional activation by RNA species carried by TEX and delivered to recipient cells [12]..

Table 1.

Selected examples of effects TEX induce in functions of various recipient cells^a

TEX origin	Responder cells	Responsible cargo	Functions induced	Reference
HNSCC	HNSCC cell	CD39/CD73	Adenosine production	[142]
	Immune cells	FasL, TGF-β, PD-L1	Immune suppression	[57]
	HUVEC	VEGF, other proteins	tube formation; vessel growth	[51]
	Fibroblasts	miR196a	cisplatin resistance	[143]
Melanoma	Melanoma cell	transfer of MET	pro-metastatic phenotype	[144]
	Immune cells	FasL, TRAIL, PD-L1	Immune suppression	[67,126]
	HUVEC (indirect)	SOC1/JAK2/STAT3 signaling	pro-angiogenic switch in CAFs	[145]
	Fibroblasts	lncRNA Gm26809	reprogrammed to CAFs	[146]
Breast Ca cells	Breast Ca cell	RISC-associated miRs	proliferation, oncogenic transformation	[147]
	Immune cells	STAT5 signaling inhibited	suppression of T cell killing	[148]
	HUVEC	Heparan sulfate	increased HPSE2, angiogenesis	[149]
	Fibroblasts	miR-105	CAF activation, tumor growth	[150]
Ovarian Ca	OvCa cell	various TAA	pre-metastatic niche formation	[151]
	Immune cells	Arginase-1	suppression of immune cells	[83]
	HUVEC	ATF2, MTA1	angiogenesis	[152]
	Fibroblasts	TGF-β	CA differentiation, tumor promotion	[153]
Colorectal Ca	CRC cell	Wnt4	migration, invasion, metastatic phenotype	[154]
	Immune cells	miR-1246	macrophages become pro-tumor	[155]
	HUVEC	HIPK2	promotes tube formation	[156]
	Fibroblasts	miR-1249–5p	CAF expansion	[157]
AML	AML cell	miR-150, −155	suppress hematopoiesis	[158]
	Immune cells	TGF-β, FasL, MICA/B	immune suppression	[93]
	HUVEC	VEGF, VEGFR mRNA	migration, tube formation	[159]
	Fibroblasts		leukemia cell survival	[160]

^aAbbreviations: HNSCC, head and neck squamous cell carcinoma; AML, acute myelogenous leukemia; HUVEC, human vascular endothelial cells; CAF, cancer-associated fibroblasts.

4.1. Reprogramming of non-malignant cells by TEX

Another interesting aspect of the recipient cell functional response to TEX is the up-regulation in the level of exosomes with pro-tumor functions that are produced by all TEX-reprogrammed recipient cells in the TME. Thus, not only TEX but also exosomes which are the by-products of all reprogrammed cells in the TME are contributing to tumor growth and metastasis, through creating a vicious cycle of exosome-mediated alterations that are necessary for establishing and driving the metastatic milieu. Figure 3 illustrates how glioblastoma-derived TEX establish such a vicious cycle by driving reprogramming of M0, M1 and M2 macrophages to pro-tumorigenic tumor-associated macrophages (TAMs). The TAMs proceed to produce and release an excess of exosomes carrying immunosuppressive and tumor growth-promoting proteins, including arginase-1 [55]. This series of molecular events involved a TEX-mediated conversion of normal cells (M0, M1) to metabolically reprogrammed cells (M2) actively generating pro-tumorigenic arginase-1⁺ exosomes [52]. This example suggests that TEX-reprogrammed normal cells producing an excess of arginase-1+ exosomes in the TME could potentially serve as biomarkers of tumor aggressiveness and progression.

Figure 3. Tumor-associated macrophages (TAMs) reprogrammed by glioblastoma-derived TEX are critical for the tumor progression.

An in vitro model resembling the crosstalk between macrophages and glioblastoma cells indicates that glioblastoma-derived exosomes reprogram naïve, M1 and M2 macrophages into TAMs characterized by elevated levels of Arginase1, CD206, IL-10, CTLA-4 and PDL-1. In turn, these TAMs, produce Arginase-1+ exosomes which mediate protumor activities. This reprogramming mechanism involves the macrophage-glioblastoma crosstalk and is an example of the role of exosomes in regulating immune cell-tumor cell communication. The figure was prepared by Dr. J.H. Azambuja based on results described in [55].

TEX also reprogram functions of various cells in the periphery. We and others have shown that the protein content as well as the specific proteins in the cargo of total exosomes from plasma of cancer patients can be used to differentiate cancer patients with early- vs. late-stage disease [43]. Total exosome protein levels in plasma are significantly elevated relative to those seen in healthy donors (HDs) [33,56], and these levels increase with disease progression or decrease when patients respond to therapy [42,43,57–60]. Thus, the exosome content and levels in plasma reflect disease activity and are also critical for monitoring of cancer patients’ responses to therapy. TEX are considered to be the key participants in establishing metastases, and TEX carry numerous inducers of the epithelial-mesenchymal transition (EMT) [61]. Current evidence indicates that TEX perform crucial roles in almost all steps of the metastatic cascade and provide signals that activate the EMT programme in epithelial cells, promoting their motility, invasion, intravasation into tissues and formation of the pre-metastatic niche [54,62]. To determine whether exosome numbers/content in plasma could forecast therapeutic responses, exosomes were isolated from plasma of patients with recurrent metastatic HNSCC receiving a palliative treatment with photodynamic therapy (PDT) [62]. Nine patients donated plasma 24h prior to (t1) and at 3 time points after PDT (t2= 24h, t3= wk1 and t3=wks 4–6). Plasma exosomes were isolated and co-incubated with epithelial target cells. All patients showed clinical improvements 4–6 wks after therapy. The patients’ exosome protein levels decreased dramatically in the course of response to therapy (Figure 4A). Exosomes obtained at t1/t2, when disease was evident, were enriched in N-Cadherin and TGF-β1; induced the mesenchymal phenotype upon being internalized by epithelial tumor cells, upregulated vimentin and transcripts for Snail, Twist, α-SMA, Slug and ZEB1 in these recipient cells and promoted proliferation, migration, and invasion in vitro of the recipient cells. These are all features associated with the EMT. In contrast, exosomes obtained at t3/t4 after PDT, when patients improved, carried E-cadherin, restored epithelial morphology and EpCAM expression in recipient tumor cells (Figure 4B), down regulated expression of mesenchymal genes and inhibited proliferation, migration and invasion of these tumor cells. Thus, plasma-derived exosomes perfectly recapitulated in vitro the conversion of tumor cells from the mesenchymal to epithelial phenotype in response to therapy. In addition, the exosomes served as non-invasive biomarkers of the reverse epithelial to mesenchymal transition after PDT [62]. This example illustrates the role of circulating exosomes in reprogramming of the metastatic microenvironment and suggests that therapy-induced alterations in exosome components are clinically relevant.

Figure 4. Characteristics of plasma exosomes of HNSCC patients with recurrent metastatic HNSCC responding to photodynamic therapy (PDT).

In A, significant decreases in total plasma exosome protein levels after therapy. In B, confocal images of cultured tumor cells coincubated with plasma exosomes harvested from HNSCC patients’ plasma pre- and post-therapy with PDT. Note a reversion to the epithelial phenotype in tumor cells co-incubated with t3/t4 exosomes. Reproduced with permission from M-N Theodoraki et al. [62].

4.2. Molecular cargos of TEX

The molecular cargo carried by TEX is enriched in prepackaged protein and RNA components, and currently, TEX cargos are being intensively interrogated to identify the molecular mechanisms used by TEX to induce phenotypic and functional alterations in recipient cells. The accumulating data point to simultaneous activation of numerous molecular pathways and to many miR-driven translational alterations in the recipient cells internalizing TEX, as indicated in the PDT example cited above. However, it remains unclear how TEX carrying the cargo presumably packaged and released by many individual tumor cells within a growing tumor manage to selectively deliver and activate in tandem various epithelial, endothelial, mesenchymal and hematopoietic cells within the TME. It is unknown whether all TEX that tumor cells release carry the same cargo of various molecular/genetic messages or whether various TEX subsets containing unique messages destined for different cells are produced by individual tumor cells. Evidence from studies of oncomeres, i.e., EVs that carry oncogenes, and of EVs that contain mutated genes corresponding to those in the parent tumor cells [63] suggests that TEX faithfully recapitulate contents of producer tumor cells. There are some indications that recipient cells may play a key role in TEX uptake and thus determine the type of message that is delivered. For example, it has been reported that T cells, in contrast to B cells, NK cells or macrophages, do not readily take up TEX, so that TEX-mediated alterations are a result of surface receptor-ligand signaling [19,64]. In contrast, phagocytic cells, such as macrophages, readily internalize and process TEX, which deliver their molecular cargo to the cell interior and induce transcriptional activation [55]. We have reported that TEX of melanoma patients carry the protein cargos that differ both qualitatively and quantitatively from those carried by non-TEX, and that melanoma patients’ TEX differ from exosomes of HDs functionally, with only the former mediating immune suppression in coincubation-type experiments [39].The data which support existence of phenotypic and functional differences between TEX and vesicles derived from non-malignant cells form a basis for the concept that TEX might serve as potential surrogates of the tumor. Since the TEX cargo appears to change with tumor progression and therapy [65–67], TEX emerge as a liquid biopsy potentially predictive of developing metastases or of response to cancer treatments. While still largely speculative, these characteristics of TEX are attracting much attention.

5.0. TEX impact on immune cells in the TME and periphery

TEX carry a diverse cargo of proteins, among them inhibitory receptors and ligands. To date, several mechanisms of immunosuppression mediated by TEX have been described, and include direct suppression of anti-tumor immune response by immune checkpoint or apoptotic signaling, dampening activity of innate immune cells, activation of suppressor cells and reprogramming non-malignant TME-resident cells towards immunosuppressive phenotypes.

TEX can promote expansion of myeloid suppressor cells (MDSCs) with enhanced immunosuppressive activity within the TME and in the periphery [49]. For example, TEX-associated Hsp72 [68], PGE₂ and TGF-β [69] and PD-L1 [70] are contributors to the expansion and activation of MDSCs. TEX containing TGF-β and PGE₂ were shown to skew the differentiation of myeloid precursor cells toward MDSCs in a MyD88-dependent manner [71]. Recently, in addition to TEX-associated proteins, an array of several miRNAs has been identified as elements responsible for TEX-mediated expansion/activation of MDSCs [72–74].

• 5.1.Regulatory T cells (Treg) are up-regulated in cancer patients and contribute to tumor immune escape and progression [75]. Almost 10 years ago, we reported that TEX produced by ovarian tumor cells, but not exosomes derived from non-malignant cells, induced in vitro expansion of human Treg from primary activated CD4+ T cells [76]. Furthermore, TEX were shown to enhance suppressor functions of Tregs by upregulating expression levels of TGF-β1, FasL, IL-10, CTLA-4, granzyme B and perforin in these cells through TEX-associated TGF-β1 and IL-10. In another study, we have shown that TEX isolated from plasma of HNSCC patients contained ectonucleotidases (CD39/CD73) and induced adenosine production in CD4+CD39+ Treg and that the elevated content of CD39/CD73 in TEX reflected the presence of advanced disease in HNSCC patients [77]. The stimulatory effects of TEX on Treg in humans and mice were later confirmed by others [78].
• 5.2.Activated T lymphocytes seem to be especially vulnerable to the suppressive mechanisms exerted by TEX. Studies showed that TEX negatively regulate two key T-cell receptors: the T-cell receptor (TcR) and interleukin 2 receptor (IL-2R). Evidence indicates that TEX are responsible for low TCR/ CD3ζ zeta expression levels frequently seen in circulating T cells of cancer patients and in tumor-infiltrating T cells (TILs), especially in patients with advanced malignancies [79,80]. The loss of the CD3ζ chain leads to the blockade of T-cell proliferation and correlates with worse prognosis and shorter overall survival (OS), as reported for OvCa [80], HNC [81], and breast cancer [82]. The mechanism responsible for TEX-mediated down-regulation of CD3ζ in patients with OvCa has been recently identified [80]. TEX were reported to carry arginase-1 (ARG1) which acts as a metabolic immune checkpoint, depriving T-cells in the TME or in the circulation of the semi-essential amino acid, L-arginine, leading to CD3ζ downregulation and suppression of T-cell growth. Further, TEX were shown to deliver the active enzyme to the draining lymph nodes (LNs) in vivo, either directly by inhibiting T cell proliferation or, acting indirectly, by interfering with dendritic cell (DC) functions. High ARG1 expression in primary tumors or in in TEX and increased ARG1 activity in plasma correlated with worse prognosis in OvCa patients [83]. TEX also inhibit functions of IL‐2R on T cells [84] and reduced JAK expression and phosphorylation in activated T cells [85]. Since the integrity of the JAK pathway is essential for functions of IL-2, IL-7 and IL-15, the cytokines sharing the γ chain of the IL-2R, down-regulation of JAK activity by TEX is detrimental to T-cell proliferation [86]. It has also been shown that exosomes derived from a diverse set of cancer cell types stimulate apoptosis in CD8⁺ T cells, in a process that is dependent on the activity of ligands such as FasL, TRAIL or PD-L1 on the TEX surface [65,85,87–89]. TEX-mediated apoptosis is accompanied by caspase-3 cleavage, cytochrome c release, loss of mitochondrial membrane potential and DNA-fragmentation [90]. Thus, TEX induce apoptosis of activated CD8⁺ T cells by engaging extrinsic as well as intrinsic apoptotic cascades. Further, a dramatic, time-dependent AKT dephosphorylation and concomitant decreases in expression levels of anti-apoptotic proteins BCL-2, BCL-xL and MCL-1 accompanied by an increase in levels of pro-apoptotic BAX were observed in these cells during co-incubation with TEX [90]. Recently, the presence of PD-L1 on the TEX surface has gained attention as yet another contributor to TEX mediated immune suppression in cancer [91].
• 5.3.Dendritic cells (DCs) maturation and differentiation are inhibited by TEX, thus preventing antigen presentation and T-cell activation in the tumor-draining LNs. In coincubation experiments, TEX-treated DCs failed to induce CD4⁺ T-cell proliferation and activation, while promoting Treg differentiation. Further, it has been reported that TEX altered DC maturation and altered expression levels of surface markers and of antigen presenting machinery (APM) components [56]. In another study of OvCa-derived TEX, we showed that ARG1-containing TEX injected subcutaneously into mice, were detected in DCs isolated from the tumor-draining LNs. Ovalbumin (Ova)-specific CD8⁺ T cells isolated from those LNs failed to proliferate in response to Ova, in contrast to T cells from mice not injected with TEX. In vitro, DCs incubated with ARG1⁺ TEX failed to induce antigen-specific T cell proliferation. In mice bearing ARG1⁺ OvCa tumors, ARG1⁺ TEX were found in the peritoneal cavity and were taken up by up to 1% of peritoneal activated CD11c+ DCs. However, as shown in vivo and ex vivo experiments, upon internalization of ARG1⁺ TEX, DCs lost their stimulatory activity and acquired a suppressive phenotype [83].
• 5.4.NK cells are also susceptible to TEX-mediated suppression. In 2006 Liu et al showed that TEX produced by murine mammary carcinoma inhibited cytolytic activity of NK cells in vitro and ex vivo. TEX-treated mice had lower NK-cell numbers in lung and spleen, and these NK cells were functionally impaired [92]. Other experiments showed that TEX carrying TGF-β1 induced down-regulation of NKG2D receptor expression in human NK cells [66] and this mechanism of TEX-mediated NK-cell suppression was observed in patients with AML [93] or HNC [57]. Furthermore, IL-15 was able protected NK cells from the adverse effect of TEX. TGF-β-mediated activation of the SMAD-pathway triggering NK-cell functional deficiency was also reported for renal cell carcinoma [94]. The stress-inducible NKG2D ligands MICA/B and ULBP-1 and −2 are carried by TEX and act as a decoy, down-regulating the NKG2D-mediated cytotoxicity of NK cells in T- and B-cell leukemia/lymphoma [95] and other cancers as well as animal models of cancer [96–99].
• 5.5.Macrophages infiltrating the TME turn into tumor-associated macrophages (TAMs), expressing an M2-like phenotype that suppress the function of cytotoxic T-lymphocytes. In several studies it has been shown that TEX contribute to the TAM polarization towards the M2 phenotype [100–104], as also discussed above (Figure 3).
• 5.6.Circulating immune cells. Not only immune cells in the TME but also circulating immune cells in patients with cancer are a target for TEX. It should be noted, that coincubations of isolated immune cell subsets with TEX produced by tumor cell lines provided evidence and identified molecular mechanisms responsible for TEX-mediated suppression as well TEX-mediated activation of these cells [76,79,85]. Since TEX carry TAA and various co-stimulatory proteins, they are being evaluated as potential vaccines in mice as well as man [105,106]. However, vaccination with TEX in tumor-bearing hosts is bound to be difficult and may require engineering of TEX to introduce co-stimulatory factors. The enrichment of TEX in immunosuppressive proteins and the relative paucity of stimulatory proteins in TEX are reflected in the type of signaling they induce in responder cells. In cancer patients with compromised anti-tumor immunity stimulatory signals are rare. TEX delivering tumor antigens and simultaneously an excess of suppressive ligands are unlikely to promote antigen processing and presenting functions of dendritic cells in LNs, as discussed above. The LN-residing or circulating T cells, especially activated T cells, interacting with TEX are also reprogrammed to become dysfunctional, exhausted or prone to undergo apoptosis [107,108]. The TEX-induced reprogramming of the immune cell repertoire in cancer is profound, includes all types of immune cells, and is reflected by the profile of TEX these reprogrammed immune cells produce. The possibility for converting the suppressive to stimulatory TME by altering TEX signaling remains a goal of future immunotherapy.

Testing T cell-derived exosomes in cancer patients’ plasma for clues of the immune status in TEX-reprogrammed T cells has recently become possible [109]. Using immunoaffinity capture, it was possible to separate exosomes from plasma of HNSCC patients into the T cell-derived CD3(+) and CD3(−) fractions using anti-CD3 mAbs [39]. Surprisingly, in these experiments, T cell-derived CD3(+) exosomes represented up to 50% of total plasma exosomes in HNSCC [42]. The CD3(−) exosomes were largely CD44v3⁺, an indication they were derived from tumor cells, and they carried PD-L1 as well as other immunosuppressive ligands. These CD3(−) exosomes induced stronger ex vivo apoptosis of CD8+ T-cells and a more potent reprogramming of CD4⁺CD39⁺ T cells into adenosine producing suppressive Tregs than their CD3(+) counterparts derived from paired T cells. Reports indicate that CD3(+) exosomes derived from T cells in cancer patients are e enriched in immunosuppressive proteins and consistently inhibit functions of other T cells [58]. Separation of TEX from non-TEX and of CD3(+) from CD3(−) exosomes in plasma allows for simultaneous profiling of tumor-derived and T cell-derived vesicles in the same plasma specimens. This is an important advantage of a liquid biopsy with exosomes. Novel technologies, such as a proximity-dependent barcoding assay used to profile proteins of individual exosomes without their separation into subsets [110], are likely to further enhance the import of a liquid biopsy in the near future.

6.0. TEX-associated PD-L1

One of the intensively investigated tumor escape mechanisms has been focused on expression and functions in exosomes of PD-L, one the checkpoint inhibitory ligands. PD-L1 is present in body fluids as a secreted soluble ligand, shed protein released from cell surface by ADAM10/ADAM17-mediated cleavage and as a component of EV membranes [111]. Soluble or membrane tethered PD-L1 engages PD-1 (CD247), an inhibitory receptor on T lymphocytes, thereby suppressing their anti-tumor activities. Tumor cells hijack this physiological pathway and use it to mitigate T-cell responses. Tumors express ligands for a variety of other inhibitory receptors, including CTLA-4, LAG-3 (lymphocyte-activation-gene-3), TIM-3 (T-cell immunoglobulin and mucin domain containing-3) or TIGIT (T-cell immunoreceptor with Ig and ITIM domains). Recent development of agents that block interactions between these receptors on immune cells and their respective ligands on tumor cells represents one of the most promising therapeutic approaches against tumor progression [112,113]. The administration of anti-PD-L1 or anti-PD-1 Abs to cancer patients has shown noticeable lasting clinical benefits in several different cancer types [114,115]. However, many cancer patients do not respond to immune checkpoint inhibition (ICI), and currently, it is not possible to predict who will or will not respond. The lack of biomarkers of response to the ICI has been a major barrier in immunooncology.

PD-1, the receptor for PD-L1, is expressed on most activated immune cells, including CD8⁺ and CD4⁺ T-cells, monocytes, NK cells, B cells and DCs. There are two PD-1 ligands, PD-L1 and PD-L2, which differ in their expression patterns. PD-L1 is constitutively expressed on most immune cells, whereas PD-L2 is expressed after induction on dendritic cells (DCs) and macrophages. As illustrated in Figure 5, the interaction between PD-1 and PD-L1 or PD-L2 leads to inactivation of the TCR signaling molecules, e.g., Zap70. The consequence is inhibition of cytokine secretion (IL-2, IFN-γ and TNF-α), blocking of proliferation and induction of T-cell apoptosis [116]. Tumor cells avoid the cytotoxic attack by anti-tumor T cells through upregulation of PD-L1 surface expression. PD-L1 expression on tumor cells and its presence in plasma has been correlated with poor prognosis [117]. However, in contrast to expectations, tumor-associated PD-L1 is not a reliable biomarker of patients’ response to immunotherapy with anti-PD-1 or anti-PD-L1 mAbs. Patients with PD-L1-positive tumors often do not respond to immune therapy and conversely, patients with PD-L1 negative tumors might experience a benefit [118]. Interestingly, it has been suggested that tumor cells with high levels of PD-L1 can package it into exosomes, which then distribute PD-L1 to the tumor-draining LNs, where it induces dysfunction of T cells and limits immunorestorative effects of anti-PD1 therapy [108]. It is now clear that PD-L1 detection in tumors by immunohistochemistry (IHC) is not reliable due to temporal and spatial tumor heterogeneity as well as packaging and re-distribution of PD-L1 into exosomes. Currently, efforts are being made to replace IHC by potentially more sensitive liquid biopsies, which would enable sampling of the tumor through TEX harvested form patients’ plasma pre, during and post therapy. In this scenario, the concept of TEX as a liquid biopsy for following tumor progression or response to therapy is very attractive. Thus, when the PD-L1 presence on human circulating exosomes was initially reported and linked to disease activity in cancer patients [43,57], numerous studies began a search for evidence that exosome-associated PD-L1, in addition to soluble PD-L1, could be a potential prognostic biomarker for tumor progression or a predictive biomarker for response to immunotherapy.

Figure 5. The role of PD-L1⁺TEX in suppression of CD8+T cell activation.

Activated CD8+ T cells in the TME upregulate expression of CD69 and PD-1 (left). PD-L1expressing tumor cells produce and release soluble PD-L1 (sPD-L1) and PD-L1 tethered to the TEX membrane (right). PD-L1⁺TEX preferentially bind to PD-1 on CD8⁺ T cell, and this signal downregulates expression of phosphorylated zeta chain-associated protein kinase 70 (ZAP70) thus interfering with the TCR activation. Downregulation of the TCR- associated zeta chain upon interaction with tumor cells or TEX has been previously reported [80]. This negative signaling induced by the PD-L1/PD-1 interaction results in blocking downstream nuclear translocation of nuclear factor of activated T cells (NFAT) and consequently suppresses T cell activities. The role of PD-L1⁺TEX versus sPD-L1 in this pathway is not clear, but PD-L1 tethered to the TEX membrane might deliver stronger negative signals than sPD-L1.

6.1. PD-L1 tethered to the TEX surface vs soluble PD-l

EVs carrying PD-L1 are produced by tumor cells and by many different types of non-malignant cells: immune cells, mesenchymal stem cells and other cells in the TME and in the periphery [119]. To date, most studies have been performed with PD-L1⁺ EVs from plasma of cancer patients and not with TEX. EVs or exosomes carrying PD-L1 have been detected in body fluids of patients with HNSCC, melanoma, prostate cancer, breast or lung cancer, glioblastoma and other cancer types [120]. Membrane-tethered PD-L1 carried by exosomes suppresses anti-tumor immune responses locally in the TME and systemically [121,122]. PD-L1⁺ exosomes preferentially target cytotoxic CD8+ T lymphocytes, inhibiting their functions by simultaneously activating different negatively signaling pathways [56]. PD-L1⁺ exosomes produced by a breast cancer cell line inhibited CD3/CD28-induced ERK phosphorylation and NFκB activation of T-cells in vitro and suppressed granzyme B activity of T cells found in the TME in vivo, reducing cytotoxic T-cell activity [122]. These PD-L1⁺ exosomes were able to transfer functional PD-L1 to PD-L1(−) tumor cells in a dose-dependent manner, restoring their growth in vivo. In patients with HNSCC, plasma-derived PD-L1⁺ exosomes served as correlates of the disease activity, tumor stage and lymph node involvement, while sPD-L1 levels did not correlate with clinicopathologic endpoints [43]. Other studies report a positive correlation of soluble PD-L1 levels with poor prognosis [123]. Interestingly, exosomes isolated from HNSCC patients’ plasma inhibited CD69 expression on human activated CD8⁺ T cells, and PD-L1 levels on exosomes correlated with their T-cell inhibitory activity [58]. The immunosuppressive impact of TEX carrying PD-L1 was confirmed by in vivo studies. Injections of PD-L1-positive TEX from murine or human HNSCC cell lines into mice with premalignant OSCC lesions induced accelerated tumor development and progression [124]. Moreover, in mice injected with TEX, the numbers of tumor infiltrating CD8⁺ and CD4⁺ T cells were significantly reduced. Similarly, Kim et al reported that exosomes containing PD-L1 inhibited IFN-γ production, induced apoptosis in Jurkat T-cells and enhanced tumor growth in vivo [89]. In the presence of PD-L1⁺ TEX, T-cells in the tumor draining LNs expressed markers of exhaustion and were unable to control tumor growth in mice with pancreatic cancer [108]. Genetic blocking of exosome biogenesis or PD-L1 deletion reversed the phenotype and strongly promoted proliferation, activation and cytotoxicity of T cells. Also, Chen at al reported that in patients with melanoma treated with anti-PD1 therapy, levels of PD-L1 in plasma-derived exosomes varied in the course of therapy, and the magnitude of increase in PD-L1+ exosomes early in therapy was predictive of poor response to the checkpoint blockade [67]. This association was attributed to the immunosuppressive role of exosomal PD-L1. In the same report, Chen et al demonstrated that in a syngeneic mouse model of melanoma, melanoma cell-derived PD-L1⁺ exosomes inhibited proliferation and cytotoxicity of CD8⁺ T cells and of tumor-specific OT-I T cells, simultaneously promoting tumor growth and decreasing infiltration of the tumor by CD8⁺ T-cells. Thus, murine TEX carrying PD-L1 were immunosuppressive, and blocking of PD-L1 activity with neutralizing mAbs restored immune competence of T cells and inhibited tumor growth [67].

7.0. TEX as potential prognostic biomarkers of response to immunotherapy and outcome

Since TEX, as reviewed above, directly or indirectly suppress anti-tumor functions of immune effector cells, it is expected that they may interfere with immune therapies. Emerging evidence indicates that TEX indeed substantially impact tumor as well as immune cell responses to immune therapies. Recently, Cordonnier et al. reported results of a prospective clinical trial in melanoma patients treated with anti-PD-1 mAbs, in which PD-L1⁺ exosomes (ExoPD-L1⁺) were monitored pre- and post-therapy to measure DeltaExoPD-L1 [125]. The PD-L1⁺ exosomes were captured from plasma using immobilized biotinylated anti-PD-1 mAbs, and ExoPD-L-1⁺ levels were measured by ELISA. The objective was to correlate ExoPD-L1⁺ levels with disease stage, response to immune checkpoint inhibitors (ICIs), overall survival (OS) and progression free survival (PFS). The study reported data that are in agreement with previously published results of Chen et al [67] and of other investigators [43,108,121], which in aggregate indicated that plasma levels of ExoPD-L1⁺ positively correlate with disease stage and progression and that variations in DeltaPD-L1 correlate with the tumor response to treatment.

While these results place PD-L1⁺ exosomes in a category of a promising biomarker of prognosis and response to ICIs in melanoma and other cancers, the origin of these PD-L1+ exosomes in plasma remains speculative, since all other studies discussed above, they were not validated TEX and might have originated from non-malignant PD-L1⁺ cells. The content of TEX in melanoma patients’ plasma is highly variable, ranging from 20–70% of total plasma exosomes [126], and it follows that a liquid tumor biopsy consiting of TEX, rather than all plasma exosomes, is likely to be more specific and more informative. To study the use TEX as a cancer biomarker, we considered it necessary to isolate TEX from plasma and separate them from melanoma patients exosomes produced by non-malignant cells.

7.1. Immune capture of TEX from plasma of melanoma patients.

Recently, the immune capture method mentioned above, was applied to isolation of TEX from plasma of patients with metastatic melanoma [39]. Using mAbs developed by Dr. Soldano Ferrone and specific for an epitope of Chondroitin Sulfate Proteoglycan-4 (CSPG4), it was possible to separate melanoma cell-derived TEX (MTEX) from other vesicles in patients’ plasma (Non-MTEX). The CSPG4 epitope used for production of these mAbs is expressed on >85% of melanoma cells but not on normal tissue cells. The mAb is not commercially available but can be obtained for research from Dr. Sodano Ferrone at Harvard. In the three-step method, the initial mini-SEC of pre-cleared plasma is followed by immune capture of MTEX and by antigen discovery using on-bead flow cytometry [127]. The SEC is required to remove the bulk of plasma “contaminants”, since previous experiments indicated that a direct addition of capture mAbs to pre-cleared plasma blocks the binding of mAbs to MTEX, presumably due to the presence of plasma proteins, compromising reproducibility of the method. After immune capture, Non-MTEX remaining in solution are recovered, placed on beads using biotinylated anti-CD63 Abs and their profile is evaluated by on-bead flow cytometry in parallel with MTEX. Quantitative phenotyping of the molecular cargos showed that melanoma-associated antigens (MAAs) such as TYRP2, Gp-100 or MelanA, are carried only by MTEX and are not detectable in NMTEX or in exosomes recovered from plasma of HDs. MTEX, in contrast to Non-MTEX, are highly enriched in immunosuppressive proteins and inhibit functions of various immune cells [126]. MTEX account for 23–66% of total exosomes, inhibit CD69 expression on activated T cells, induce apoptosis and suppress proliferation in CD8⁺ T cells and downregulate NKG2D expression in NK cells. In contrast, Non-MTEX, enriched in immunostimulatory proteins, are significantly less suppressive and tend to promote immune responses. Elevated MTEX/total exosome ratios and, surprisingly, Non-MTEX ability to induce apoptosis of CD8⁺ T cells emerge as positive correlates of the disease stage [126].

7.2. Immmune capture of TEX from plasma of patients with HNSCC

To confirm the utility of immune capture for TEX isolation from plasma, mAbs specific for CD44v3, a protein overexpressed in HNSCC cells, were also used. In this study, exosomes in plasma of HNSCC patients and of HDs were isolated and immunocaptured. CD44v3 levels on TEX in patients were higher than in HDs and correlated with the UICC stage and LN metastasis. Also, CD44v3⁺ exosomes were enriched in immunosuppressive proteins, including PD-L1, TGF-β, FasL or EGFR relative to CD44v3(−) exosomes, and this immunosuppressive cargo correlated with advanced disease [124]. Immune capture of TEX was also used in monitoring of patients with recurrent metastatic HNSCC treated with a combination of cetuximab, ipilimumab and IMRT in a phase I clinical trial. TEX and T cell-derived CD3⁺ exosomes from plasma were separated by immune capture and their molecular cargos were monitored pre, during and post therapy [59]. TEX were immunocaptured with a mix of Abs (anti-EGFR, -EpCAM, -MAGEA3, -CSPG4) painted on microarrays. T cell-derived exosomes were immunocaptured on beads coated with anti-CD3 mAbs. Importantly, TEX and T-cell derived circulating exosomes instead of immune cells were used for serial monitoring of patients. The objective was to evaluate the predictive value of the exosome molecular cargos for disease recurrence within 24mo after immunotherapy (IT). In HNSCC patients whose disease recurred within 24mo (n=5/18), total exosome protein, TEX/total exosome protein ratios, CD3⁺, CD3(−)PD-L1⁺ and CD3⁺CD15s⁺ (Treg) exosomes significantly increased relative to baseline levels. In patients who remained disease free (n=13/18), total exosome protein and TEX numbers decreased, CD3⁺ and CD3⁺CD15s⁺ exosomes stabilized, and CD3⁺CTLA4⁺ exosomes declined after ipilimumab. The results of this trial support the potential role of TEX and T-cell derived exosomes as non-invasive biomarkers of tumor as well as of T cell responses to IT in cancer [59].

7.3. TEX in patients’ plasma may interfere with immunotherapy

Despite undeniable success of immune therapy in oncologic diseases, many patients remain unresponsive to ICIs. For example, in a recent study, AML patients with relapsed refractory acute myelogenous leukemia (AML) treated with adoptive cell therapy (ACT) using expanded NK-92 cells were unresponsive to therapy. The patients were found to have high levels of blast-derived TEX in plasma collected prior to ACT. Exosomes isolated from pre-therapy plasma of these AML patients were co-incubated with NK-92 cells (NK-92 is a cell line) used for therapy. The exosomes isolated from patients’ plasma inhibited NK-92 cell functions and interfered with their anti-leukemia activity [128]. Further, the blockade of exosome-mediated suppression in part restored anti-leukemia functions of NK-92 cells. These findings link the presence and activity of blast-derived exosomes (TEX) with the resistance of AML blasts to ACT. This example provides an important lesson for other forms of ACT, where TEX in patients’ plasma could interfere with immune cells used for therapy and might limit expected therapeutic benefits of this form of therapy.

In patients receiving Ab-based therapies, circulating TEX might interfere with IT. In cancer, Abs delivered as therapy usually target TAA and induce anti-tumor responses by antibody-dependent cell mediated cytotoxicity (ADCC). TEX present in body fluids carry the relevant TAAs and sequester delivered Abs, reducing ADCC. This type of interference with therapy by TEX has been demonstrated for widely used therapeutics, such as rituximab, which targets CD20 on B-cell lymphoma cells [129] or trastuzumab, which targets HER-2 on breast cancer cells [130,131]. Capello et al has shown that plasma exosomes of patients with pancreatic ductal adenocarcinoma (PDAC) utilize decoy-like functions by binding circulating autoantibodies in PDAC patients’ sera, thereby inhibiting complement-dependent cytotoxicity [132]. Thus, TEX may be critical determinants of tumor cell susceptibility to Ab therapies, allowing the tumor to escape from immune interventions [133].

Suppressive effects of TEX on therapeutic anti-tumor vaccines have also been noted. Often, such vaccines given to patients with advanced malignancy fail to rejuvenate anti-tumor responses and instead result in exhaustion and a loss of immune effector cells. Since vaccines induce re-activation of immune cells, and activated T, B or NK cells upregulate surface expression of receptors (TcR, IL-2R, PD-1, Fas, NKG2D, etc.), they become highly susceptible to TEX-mediated apoptosis. We suspect that a poor efficacy of many anti-tumor vaccines is due to accumulations in patients’ plasma of immunosuppressive TEX. Interestingly, accumulating plasma-derived exosomes might play a potential role as biomarkers of response to immunotherapy. Specifically, in patients with recurrent malignant glioma receiving a DC-based vaccine, exosomes isolated from plasma prior to and after the first vaccination were compared. Changes in total exosome protein levels and in levels of immunoregulatory proteins and in mRNA expression levels of selected immune-related genes were measured. These changes in exosomal proteins and mRNA reflected the response to therapy and immunologic activation in the vaccinated patients [134].

7.4. Correlative studies linking TEX with disease activity/progression

Long before it became possible to isolate TEX from patients’ plasma, we monitored total plasma exosome levels in various cohorts of cancer patients attempting to establish correlations between the exosome levels and/or cargos with disease activity, severity and progression [17]. These early studies suggested that plasma levels of exosomes enriched in FasL, PD-1, TGF-β or COX-2 had a prognostic value. Expression levels of these apoptosis-inducing molecules in TEX were correlated with the frequency of apoptosis-sensitive activated CD8⁺ T cells in the circulation of cancer patients. Subsequently, a significant correlation was established between spontaneous apoptosis of circulating CD8⁺ T cells observed in patients with cancer and the disease stage and prognosis [79,85,135]. In addition, Kim et al reported a significant positive association between the presence and levels of FasL on plasma-derived exosomes and the size of the tumor at the time of HNSCC diagnosis [136]. These correlative studies were later extended to hematopoietic malignancies. In AML, the composition of immunosuppressive factors, such as membrane-associated TGF-β1, in the exosome cargo was found to change with therapy. Alterations in levels of the TGF-β1 pro-peptide LAP and the mature form of TGF-β1 in exosomes isolated from AML patients’ plasma correlated with patients’ responses to chemotherapy [66,137].

The data suggesting that total or individual protein levels in TEX might correlate with cancer progression or responses to therapy have led to the engagement of extensive proteomic analyses of various EVs isolated from tumor cell supernatants and later of TEX in patients’ plasma [138–140]. The objective was the identification of protein signatures that might have a diagnostic or prognostic value. These studies have not been conclusive so far and are ongoing. Preliminary results of our studies investigating proteomic profiles of MTEX isolated by immune capture from plasma of melanoma patients, suggest that the proteomic approach to the analysis of MTEX cargos may be clinically relevant [139]. Numerous other studies have reported on the application of TEX to cancer diagnosis and therapy (Table 2). Many of the diagnostic and prognostic assessments of TEX are currently in clinical trials based on the prevailing notion that clinical applications of exosomes in general, and TEX in particular, represent an attractive tool for diagnostic and prognostic purposes [141]. Mostly, these studies feature nucleic acids, especially microRNAs, as promising biomarkers of early diagnosis and treatment of cancers [141]. The underlying tenet for all future clinical applications of TEX is that a more detailed examination of specific molecular components in exosomes might provide protein or miR signatures that will prove to be more accurate biomarkers of diagnosis, prognosis and response to therapy than any single soluble biomarker in plasma.

Table 2.

Examples of diagnostic/prognostic applications of tumor-derived exosomes^a

Disease target	Sample	Biomarker(s)	Clinical application	Reference
Bladder cancer	urine	eight lncRNAs	Predicts prognosis	[161]
		↑ PCAT1
Breast cancer	plasma	five miRNAs panel	Diagnosis	[162]
		↑ miR-122–5p in exosomes
Colorectal cancer	serum	↑ UCA1 in exosomes	Drug resistance	[163]
Hepatocellular carcinoma	serum	↑ miR-210	Microvessel density	[164]
	serum	↑ miR-224	Diagnosis/prognosis	[165]
Lung cancer	serum	six miRs ↑ miR-17–5p	Diagnosis	[166]
Melanoma	plasma	↑ PD-L1 in exosomes	Overall survival Progression-free survival	[125]
Oral SCC	plasma	↑ total exosome levels	Predicts prognosis	[167]
Ovarian cancer	serum	↑ miR-375 ↑ miR-1307	Diagnosis	[168]
Pancreatic ductal adenocarcinoma (resectable)	serum	↑ glypican-1 in exosomes	Predicts prognosis	[169]

^aAbbreviations: SCC, squamous cell carcinoma; PCAT1, prostate associated transcript1; UCA1, urothelial cancer associated 1.

8.0. Expert Opinion

In this review, we gathered and presented data supporting the role of TEX as promising biomarkers of diagnosis, prognosis and response to therapy in cancer. Emerging data from in vivo animal models and from ex vivo experiments with plasma-derived TEX or plasma exosomes enriched in TEX of patients with cancer point to the exosomal cargo as an important correlate of developing, progressive disease. Further, alterations in the patients’ immune status and the changes in molecular pathways the tumor experiences in response to therapy are reflected in the TEX protein and mRNA profiles. Especially, the levels of exosomal immunoinhibitory checkpoint molecules, such as PD-L1, appear to have significant impact on patients’ outcome and response to therapyse. Clearly, diagnostic, prognostic or predictive status of TEX in cancer requires further exploration. Pre-clinical in vitro, ex vivo and in vivo data of TEX produced by various tumor cell lines have provided insights into the mechanisms responsible for TEX attributes that make them eligible to serve as biomarkers of disease and/or responses to therapy. The fact that TEX phenotypic attributes can be linked with their functional reprogramming of target recipient cells endows TEX with a special advantage as biomarkers. In addition, the capability to correlate within the same plasma specimen functional characteristics of TEX and of T cell-derived EVs, provides exosomes with a double advantage of serving as biomarkers of the tumor and of the patients’ immune status at the time of blood draw. Both these features of plasma exosomes should be explored further, and the validation of the preliminary data need to be obtained in human clinical trials. Nevertheless, TEX and immune cell-derived exosomes are here to stay as potentially promising biomarkers in cancer and other diseases.

Article Highlights.

• Tumor-derived exosomes (TEX) originate from the endocytic compartment of tumor cells and mimic the molecular content of parental cells.

• TEX are present in all body fluids of cancer patients at levels that change with disease progression/regression.

• TEX are enriched in immunosuppressive proteins, including PD-L1, and suppress functions of immune effector cells in vitro and in vivo.

• Phenotypic and functional profiles of TEX have prognostic/diagnostic significance in cancer, but their role as “liquid biopsy” requires further clinical validation.