FormalPara Key Points

Exosome processing and role in cancer formation and progression is varied between cells and tissue states.

Exosomes relate to diagnostic, prognostic, and therapeutic potential on many levels.

Several promising approaches to using exosomes for biomarkers, delivery vehicles, and therapeutics are changing the landscape of cancer diagnosis and treatment.

1 Introduction

In this review, we discuss exosomes as related to their role in cancer. In vitro and animal studies have demonstrated that many cell types communicate via intercellular signaling using soluble factors and secreted extracellular vesicles (EVs) [1,2,3]. While they are secreted by all cells, cancer cells release significantly more EVs than normal cells, and these can be found in most bodily fluids, including urine, ascites, serum, and plasma [4,5,6,7,8]. Exosomes are a type of EV that delivers cargo containing bioactive lipids, cytokines, growth factors, receptors, transcription factors, DNA, non-coding regulatory RNAs, and messenger RNAs (mRNAs) into recipient cells. This cargo can have profound effects on recipient cells and reflect the cell of origin. Intercellular communication by exosomes contributes to the regulatory signaling of both normal and pathological processes, including cancer [1, 3, 9, 10]. Exosomes have a common morphology, size distribution, and specific marker expression, regardless of tissue origin. Although there are common exosome characteristics, the markers and cargo they carry differ with the original tissue, cell, and cell state. Additionally, the cargo content of exosomes can reflect the origin of the cell and may even indicate pre-metastatic niche formation. With these considerations, exosomes have been recognized as potential diagnostic and prognostic markers for various diseases [11, 12] as well as the assessment of treatment efficiency. Eventually, the data from exosome studies may be used to discover new ways to detect and eliminate malignant precursor cells.

Exosomes were first described as EVs released by tumor cell lines [13]. Since then, they have been described as small, round, and cup-shaped vesicles that collect in multivesicular bodies (MVBs) with size ranges reported from 20 to 200 nm. Exosome biogenesis involves inward budding at the plasma membrane which produces intracellular early endosomes [14]. These bud inward, forming small vesicles termed intraluminal vesicles (ILVs) within a MVB [14]. There are several processes that are believed to facilitate the biogenesis and secretion of exosomes (Fig. 1). These include the endosomal soring complex required for transport (ESCRT) pathway [15,16,17,18] and a ceramide-dependent pathway [16, 19, 20]. Their formation may be controlled by the interaction of syndecan heparin sulphate chains with syntenin-1 and ALG-2 (apoptosis-linked gene 2)-interacting protein X (Alix) for recruitment of the ESCRT machinery [21,22,23]. In addition, protein cargo sorting [24, 25] and release [26] involve the ESCRT complex [19, 27] and other associated proteins such as Alix [21, 28] and Tsg101 [29]. The four components of the ESCRT pathway (ESCRT-O, -I, -II, and -III) aid in the movement of ubiquitinylated proteins [16, 30, 31]. Biogenesis has also been shown to occur in an ESCRT-independent manner through ceramide, flotillin [21], phospholipase D2 (PLD2)/ADP ribosylation factor 6 (ARF6) [32], and CD81 [33]. The ceramide-dependent pathway depends on lipid raft modification to ceramide [19, 20], which ultimately prompts the formation of ILVs in the MVB [16].

Fig. 1
figure 1

Exosome biogenesis. a Inward budding of the plasma membrane creates early endosomes. b Invagination of the early endosome membrane results in ILVs. This may be accomplished by several mechanisms, including: c Lipid rafts may be converted to ceramide, prompting inward budding of the membrane. d Syntenin may bind to ARF6 through the action of PLD2, promoting inward budding. e Ubiquitinated proteins may bind to ESCRT O, which recruits ESCRT I, II, and III. Cooperative interaction of sydecan chains with syntenin, the CD63 tetraspanin, and Alix promotes invagination, followed by interaction of Alix with ESCRT III. f The endosome with the ILVs is referred to as a MVB. The ILVs that remain for secretion to the extracellular space are exosomes. Note that the process by which exosomes are created results in the same orientation of surface markers as the cell membrane. Alix ALG-2 (apoptosis-linked gene 2)-interacting protein X, ARF6 ADP ribosylation factor 6, ESCRT endosomal soring complex required for transport, ILVs intraluminal vesicles, MVB multivesicular body, PLD2 phospholipase D2

Exosomes are secreted from the cell when the MVB fuses with the plasma membrane instead of fusing with a lysosome, releasing their contents extracellularly [14, 34, 35]. Exosomes may also be secreted by direct budding from the plasma membrane. Exosome release involves contributions from Rab proteins [23, 36,37,38], the actin cytoskeleton [39], the microtubule network [40], and cholesterol [41]. Importantly, Rab-mediated release has been shown to occur in a cell-specific manner [38, 42]. This process has been confirmed in antigen-presenting cells (APCs) [43, 44], epithelial cells [45], and tumor cells [46], among others. Due to the mechanism of intake and release, exosomes contain proteins from the plasma membrane and cytosol, as well as the extracellular domain of surface receptors from the cell [47], with very little protein from other organelles [48]. As a result, they bear both intraluminal and transmembrane proteins with the same orientation as the plasma membrane.

Exosomes enable intercellular communication without direct cell-to-cell contact. This communication can be antigen presentation, immune regulation, spread of infection, and initiation of altered signaling [49] through several mechanisms (Fig. 2). When juxtacrine signaling occurs, the exosomes bind a surface receptor on the recipient cell, initiating intracellular signaling. This signaling is markedly different from fusion and phagocytosis, which transfer the contents of the exosomes into the cell. In endocytosis, the entire exosome is transferred into the cell. The exosomes can also be internalized by endocytosis through several mechanisms reported to occur at acidic pH [41]. There are reports of endocytotic internalization of oligodendroglial and microglial exosomes by neurons [50, 51]. When direct fusion occurs, the contents of the exosomes are delivered to a recipient cell by fusion with the recipient cell membrane resulting in direct release of the cargo to the recipient cytoplasm. The binding of exosomes to the surface of recipient cells has been shown to be mediated by integrins, intercellular adhesion molecules [52], tetraspanins [53], specific glycoproteins [54], and heparan sulfate proteoglycans [55].

Fig. 2
figure 2

Mechanisms of exosome uptake by recipient cells. a The process by which exosomes are created results in the same orientation of surface markers as the cell membrane. b Exosome information is transferred to recipient cells through several mechanisms. Some cell types utilize multiple mechanisms. EBV Epstein-Barr virus, HIV human immunodeficiency virus, MHC myosin heavy chain, NK natural killer cells, Tregs regulatory T cells

The variety of mechanisms for the loading of proteins into exosomes, release mechanisms, and signal delivery routes are not necessarily contradictory. It points to the heterogeneity of exosome populations. In fact, the silencing of specific components of the exosome biogenesis machinery do not affect all exosomal markers in the same way. For example, breast cancer cells secrete exosomes with differing sizes, microRNA, and protein content [56]. In epithelial cells and a colon carcinoma cell line, exosomes released from the apical side of the cells are enriched in specific and different proteins from the basolateral side, even though both populations are identical on electron microscopy and typical exosome markers [57, 58].

Methods to isolate exosomes are a subject of constant debate [59,60,61]. There are many ways to isolate from any number of tissues and cells. The most widely used and reported technique is ultracentrifugation (UC) [4, 61,62,63,64,65]. It has been reported that UC accounts for 81% of all exosome isolation techniques [66]. UC involves spinning suspensions at very high speeds with or without density gradients. Density gradient UC separates exosomes based on their size and mass compared to the medium. This requires large, vacuum-fitted, cooled ultracentrifuges that need much maintenance and user training. UC suffers from high exosome loss [67] and at high velocities (> 100,000g), proteins aggregate and exosomes may clump [5, 59, 68]. Additionally, the viscosity of plasma and presence of lipoproteins, which have similar density and diameter to exosomes, co-isolate with the exosome component [59, 62, 63, 69, 70]. Isolation may also be accomplished using polymer-based precipitation of the exosomes by binding water surrounding the particles [71, 72] or immuno-affinity capture with antibodies against consensus exosome markers [73, 74]. Exosomes cannot be identified by using standard flow cytometry because they are too small to be distinguished from junk signals. But, the existence of markers on the surface of exosomes can be used to sort them [75] if attached to beads with consensus exosomal cell surface markers, such as CD63 and CD81 for distinction. The exosomes bound to beads are then probed with fluorescent antibodies against the same or other consensus exosomal cell surface markers. The distribution of these markers is varied amongst exosome populations from various cell types and tissue states. Microfluidics passes solution with exosomes through small channels on microchips that are treated with antibody [76, 77]. Ultrafiltration utilizes filter-based fractionation at specific speeds and membrane porosity to capture exosomes and other EVs [67, 78]. Commercially available kits have been developed using these techniques [79]. However, these kits have been criticized for low purity, contamination by other microvesicle types, contamination by proteins, loss of exosomes, and undesirable effects on the exosomes themselves [80]. Size exclusion chromatography (SEC) has been shown to isolate exosomes from blood plasma with no significant impurities [62, 67, 81,82,83]. The principle of SEC is separation based on size differences [84, 85]. SEC has been used to isolate vesicles from sera, ascites, and saliva, and was shown to effectively separate vesicles from proteins [62, 63, 86,87,88,89,90]. Böing et al. [62] showed that there is no risk of lipoprotein complex formation or vesicle aggregation as compared to differential centrifugation. With respect to the undesirable effects on the exosomes themselves [80], compared to UC, SEC results in a good recovery of vesicles with almost complete removal of contaminants [4, 62].

An enormous amount of information can be acquired by analysis of exosome contents. These analyses can include differential expression of genomes, transcriptomes, microRNA, lipid and proteome profiles, in addition to more detailed analysis of post-transcriptional or post-translational modifications. The analysis of exosome cargo is accomplished through numerous methodologies, including mass spectroscopy for complete protein screening [91,92,93], the more specific Western blot, next-generation sequencing for complete transcriptome screening [94,95,96], reverse transcription polymerase chain reaction (RT-PCR) for more specific RNA questions, and Nanostring for panels of RNA and microRNA probes. Data from such studies has led to the discovery of various exosome sources involved in cancer establishment and progression. Some of these sources are described in this review.

1.1 Exosomes from Cancer-Associated Fibroblasts

To avoid necrosis as tumor size increases, vascularization at the tumor begins, which establishes the entry point for metastatic cells [97]. Many of the cancer progression processes are influenced by cancer-associated fibroblasts (CAFs) within the tumor. CAFs stimulate new angiogenesis by secreting stromal-derived factor-1 [98], recruiting endothelial progenitor cells [99] and stimulating them to form tube-like structures [100]. It has been demonstrated that CAFs and tumor cells jointly invade blood vessels and implant themselves in metastatic sites [101]. Tumor-derived exosomes (TDEs) can convert mesenchymal stem cells into CAFs [102]. In breast cancer, exosomes from CAFs cross-talk with cancer cells to enhance migration [103] and stimulate epithelial to mesenchymal transition (EMT) [98] by secreting exosome-processed transforming growth factor (TGF)-1 [98]. The TGF-1 effect on stroma differentiation is abrogated by blocking of exosomes, even if the TGF-1 levels in the tumor stroma remain constant [98]. The CAFs also use enzymes (matrix metalloproteinases) to create tunnels through the extracellular matrix that the tumor cells can utilize [104]. It has been shown in sarcoma that exosome matrix metalloproteinases stimulate metastasis [105] and are prognostic for invasiveness in head and neck cancer [106]. In addition, exosomal CD81 (a common exosomal marker) was shown to work with CAFs to promote cancer cell mobility in breast cancer [103].

1.2 Platelets, Exosomes, and Cancer

Under normal circumstances, leukocytes and especially natural killer (NK) cells are found in close relation to metastatic cells in the bloodstream and eliminate these cells [107]. Metastatic cells need to escape this immune surveillance. Platelets adhere to metastatic cells to shield them from NK cell activity [108]. In addition, direct signaling between platelets and cancer cells induces an EMT-like transition and promotes metastasis [109]. The depletion of platelets in mice inhibits metastasis, and platelet reconstitution restores the metastasis [110]. Platelet-derived exosomes assist in the maintenance of cancer cells in the tumor and the blood circulation by also accompanying cancer cells in the bloodstream and enabling their survival [111] through interaction with endothelial cells, leukocytes, and cancer cells [112, 113]. Here, they assist tumor cell adhesion to the endothelium of the recipient vessels and enable extravasation of tumor cells from the bloodstream into the pre-metastatic niche [114], especially via P-selectin [113].

2 Tumor-Derived Exosomes

Normal human blood contains about 1.5 billion exosomes/mL [72]. It has been reported that there are more exosomes secreted from tumor tissue than healthy tissue [115, 116], as well as into patient plasma from the tumor tissue [116, 117]. In one study, this increase was not shown to further increase with tumor progression and was not related to the immune response to the tumor [115], but in other studies there is a correlation to tumor progression [117, 118]. It is possible that the increase is due to local stress within the tumor, such as hypoxia [119] and acidic pH [120]. For example, exosome release from melanoma cells is significantly higher when the medium is not buffered [121] and plasma exosome concentrations of patients with stage III–IV melanoma were shown to be significantly higher than in the plasma of lower stages or healthy patients [118]. Exosomes secreted directly from tumors are referred to as TDEs. These serve to mediate the immune response against a tumor, promote tumor progression, promote migration and engraftment, and affect the response of the tumors to treatment.

Briefly, some mechanisms by which TDEs mediate the immune response against a tumor is by inducing apoptosis of T cells, inhibiting dendritic cell differentiation, inhibiting NK cell cytotoxicity, and inducing myeloid suppressor cells and regulatory T cells [122]. Additionally, the pro-tumorigenic inflammatory response created by TDEs has been shown to support neo-angiogenesis, invasion, and matrix remodeling [123]. This relationship between TDEs and tumorigenic inflammation has been shown in pancreatic ductal adenocarcinoma [9] and liver cancer. TDEs affect the surrounding stromal EMT [124], migration [125], invasion, and engraftment in part by stimulating tumor-associated macrophages [126]. The Toll-like receptors on the surface of the macrophages interact with the TDEs, activating nuclear factor (NF)-κB in the macrophages [127]. For example, in breast cancer, an increased concentration of macrophages correlates with poor prognosis [126].

2.1 Role of Microenvironmental Factors

Since tumor cells are continuously subjected to various forms of stress, there are increased secretions of exosomes by cancer cells [128]. The acidic microenvironment is characteristic of the tumor microenvironment and is considered an important phenotype of malignant tumors [129, 130]. This acidic microenvironment stimulates release of exosomes and enhances their cell fusion capabilities by affecting the lipid composition [121]. Logozzi et al. [131] showed definitively in 2017 that the acidic environment of prostate cancer cells increases the release of exosomes from tumor tissue into patient plasma and this does not occur under buffered conditions. Furthermore, in 2018, Logozzi et al. [132] reported that acidity, as occurs in the tumor microenvironment, increases the release of homogenous exosomes in five different cancer cell lines. This was compared to buffered condition, which resulted in a smaller release of heterogenous exosomes. Clearly, environmental acidity is a key factor in TDE characteristics and release. Heat is a stressor that increases release of immunosuppressive exosomes from B cells in leukemia and lymphoma [133]. Heat may also lead to antitumor functions by inducing the release of exosomes with heat shock proteins (HSPs). Hypoxia is a stressor in the tumor that induces the secretion of exosomes containing proteins associated with cell migration, extracellular matrix degradation, growth hormone signaling, endocytosis, and vascular endothelial growth factor (VEGF) signaling [104, 134]. These signals alter the microenvironment and facilitate angiogenesis and metastasis. For example, malignant brain tumor glioblastoma cell growth is increased and more angiogenic when exposed to exosomes derived under hypoxic conditions, as compared to normoxia-derived exosomes [135]. Exosomes from hypoxic tumor cells can impair NK cell function by secretion of TGF-β1 [136]. This leads to activation of EMT, which plays an important role in metastasis [109]. These stress-induced changes can also provide protective signals influencing the response of distant cells. For example, during oxidative stress or starvation in neuronal development, oligodendrocytes secrete exosomes that are taken up by neurons, promoting neuronal viability [51]. During ischemia/reperfusion injury, mesenchymal stem cells and cardiac progenitor cells release exosomes that have a cardio-protective effect [137] and reduce inflammation [138]. Recognizing the response of tumors to stress and their subsequent exosome release is important for directing discovery of diagnostic and prognostic biomarkers.

2.2 Cargo and Cancer

Exosomes, regardless of origin, contain proteins, lipids, and nucleic acids. The acquisition of cell membrane components and the specific sorting of cargo into exosomes are what distinguish exosomes as being from a particular cell, cell type, or cell state. These differences will distinguish the exosomes from one origin versus exosomes from another origin. The cargo of TDEs is varied and valuable to cancer progression. Consequently, understanding the cargo is important for research of tumor origination, as well as diagnostic, prognostic, and therapeutic uses. Since the exosomes contain lipids and receptors from the cell membrane, cytokines, growth factors, cytoplasmic receptors, transcription factors, non-coding regulatory RNAs, and mRNAs, they are good candidates for detection of the presence of abnormal cells using various body fluids. For example, significantly higher levels of macrophage migratory inhibition factor (MIF) were found in the serum exosomes of patients with pancreatic ductal adenocarcinoma with liver metastasis and progressive disease than in patients with full remission, suggesting MIF as prognostic marker in pancreatic carcinoma [9].

2.2.1 Proteins

Protein cargo of exosomes can be cytosolic, nuclear, transport-involved, adhesion-related, and plasma membrane-bound [48]. These can include cytokines, chemokines, growth factors, receptors, and hormones [139]. According to Exocarta [140], exosomes contain more than 41,860 different proteins. Commonly referenced proteins used to describe exosomes include integrins, tubulin and actin, sorting proteins Alix and Tsg101, major histocompatibility complex (MHC) class I (MHC-I) [141], HSP70, HSP90 [48], and tetraspanins CD9, CD63, CD81 and CD82. Differentially expressed protein cargo can be queried to indicate the status of the cells of origin. Transmembrane protein 256 in the urine has been shown to indicate prostate cancer with 94% sensitivity and 100% specificity [142]. Melo et al. [143] and Hu et al. [144] reported that Glypican-1 in serum exosomes can indicate tumor load/metastasis and stage, respectively, of pancreatic cancer patients with 100% sensitivity and specificity. Conversely, Yang et al. [145] reported much lower sensitivity and specificity for Glypican-1 in pancreatic cancer. Lai et al. [146] reported that, although their own previous studies show a relationship between Glypican-1 and pancreatic cancer, this marker alone is not sufficient and better results were seen using specific microRNAs. Hence, the use of Glypican-1 as a marker alone for pancreatic cancer is controversial. It is likely that more robust specificity will be achieved using a combination of proteins and microRNAs. Examples of other protein indicators found in exosomes are depicted in Table 1. Recent advances in proteomic analysis of exosomes by mass spectroscopy [147, 148] are sure to advance our understanding of exosome activity.

Table 1 Examples of published proteins found in tumor-derived exosomes

2.2.2 Lipids

Exosomes also contain lipids from membrane lipid rafts including ceramides, sphingolipids, cholesterol, and glycerophospholipids [48]. The localization of many cell membrane components to lipid rafts, which preferentially transfer to exosomes, allows selective omission of some signaling molecules that would otherwise lead to elimination of the exosomes within the extracellular space. Exosomal lipids have also been used for therapeutic targeting. For example, blocking ceramides to reduce the shedding of TDEs.

2.2.3 Nucleic Acids

Importantly, exosomes contain nucleic acids such as microRNA [69], mRNA, DNA, and long non-coding RNAs (lncRNAs). Mitochodrial DNA [161], single-stranded DNA, and chromosomal segments (transposons) [162] have been identified in tumor microvesicles. In 2014, double-stranded DNA was found in exosomes from the serum of pancreatic cancer patients that reflected the mutations in the cells of origin [163]. Since the genetic material can be processed and expressed within the recipient cells, the presence of mutated DNA in TDEs is an important consideration. This cargo is not researched as extensively as RNA content and should not be overlooked.

mRNAs are selectively enriched in exosomes [69]. Microvesicles from embryonic stem cells were able to reprogram hematopoietic progenitors [164] and endothelial microvesicles activated endothelial cells [165] by ‘horizontal transfer’ of mRNA. The sorting of mRNAs normally occurs through coding in the 3'-untranslated region (UTR), as previously described [166]. Trafficking of specific mRNA into exosomes may also be due to coding and potentially subsequent folding within the 3ʹUTR fragments [167]. In an interesting 2013 study, Batagov et al. [168] reported that mRNAs in exosomes exhibit enrichment of 3ʹUTR fragments. These are potential insights into the specific sorting mechanism for mRNA secretion to exosomes, although considerably more information is still sorely needed. The 3ʹUTRs have many regulatory sequences for RNA-binding proteins and microRNAs [167, 169] and the major RNA content of exosomes is microRNAs. MicroRNAs in TDEs are representative of the altered microRNA profile of the tumor. After transport to local and distant sites, the result is influence of gene expression in the target cells of the surrounding stroma and the distant pre-metastatic niche [170]. MicroRNAs packaged in exosomes are very stable in serum and plasma, making the use of exosomes for cancer microRNA profiles very appealing. The idea that mature microRNA, which is normally quickly degraded, is present outside of the cell in the plasma is fascinating. Mitchell et al. [116] showed that these are surprisingly stable and claims that they can be at room temperature for 24 h and can tolerate up to eight freeze/thaw cycles.

Interestingly, processing of pre-microRNAs to mature-microRNAs can occur within exosomes, which allows for direct effects on the recipient cells without further processing [171]. In 2007, Kawahara et al. [172] reported the importance of adenosine-to-inosine (A-I) editing of microRNA. Since, A-I microRNA editing has been related to several human cancers [173, 174]. In fact, microRNA editing has been related to human cancer prognosis with an exhaustive characterization of microRNA sequences for 20 cancer types [175]. These data were further explored to show that A-I microRNA editing is repressed in cancer tissue as compared to normal tissue [176]. Importantly, a recent study has reported, for the first time, that microRNA editing occurs within exosomes which may be predictive of cancer pathogenesis [177]. MicroRNA has become a highly investigated component of cancer at all stages and microRNA in exosomes is becoming more illustrative by the day. For example, in prostate cancer, compared to patients with non-recurrent disease, both miR-141 and miR-375 are increased in the plasma exosomes with recurrent disease after radical prostatectomy [178]. Mitchell et al. [116] also showed increased miR-141 indicative of prostate cancer metastasis with 60% sensitivity and 100% specificity. Table 2 presents some examples of published exosomal microRNA research related to cancer.

Table 2 Examples of published small RNAs found in tumor-derived exosomes

The second most abundant RNA in exosomes is lncRNA [220]. This RNA is responsible for regulating gene expression by binding to DNA, but does not code for proteins, or at least does not demonstrate more than 100 amino acid reading frames. lncRNA has been shown to play a role in tumorigenesis. With consideration of the DNA content in exosomes mentioned previously, this component is an interesting part of the signaling carried by TDEs. It is likely the lncRNA studies will reveal more in-depth information about the transfer of mutated DNA to healthy cells, as well as the protection of that DNA during transport. Some studies of lncRNA have been included in Table 2.

3 Exosomes and Cancer Progression

Exosomes contribute to all phases of tumor progression and metastasis (Fig. 3).

Fig. 3
figure 3

The contributions of exosomes to cancer progression. Exosomes contribute to all phases of tumor progression and metastasis. EMT: tumor cells secrete exosomes to surrounding stromal cells and nearby vasculature, signaling EMT in stromal cells and increasing the porosity of the vessels. Migration: exosomes from different tumors are attracted to specific distal tissue types. Preparation: at the distal sites, exosomes alter the local signaling to suppress the immune response and send inflammatory signals, increasing vascular leakiness for incoming cells (arriving tumor cells). Metastasis: as tumor cells arrive at the prepared sites, their exosomes continue to suppress the immune response and alter the new environment. EMT epithelial to mesenchymal transition, TDEs tumor-derived exosomes

3.1 Epithelial to Mesenchymal Transition

It is believed that the tumor-stroma interaction initiates the EMT, which frequently begins the metastatic process. The tumor cells lose their polarity and junctions, reduce proliferation, and increase migration and invasion [221]. This is reversed when the tumor cells reach their destination niche. TDEs can force other cancer cells to follow this process as well. TDEs, but not exosomes in general, are able to induce EMT and enhance migratory activity. Exosome-depleted cancer cells fail to gain a stroma-mediated growth advantage in vivo [98]. Exosomes from nasopharyngeal carcinoma induce EMT by high hypoxia inducible factor 1a (HIF1a) and high latent membrane protein 1 (LMP1; which reduces degradation of HIF1a) [222]. Exosomes from invasive bladder cancer caused urothelial cells to exhibit EMT features, but exosomes from embryonic kidney cells did not cause these changes [223, 224]. This has also been seen in glioblastoma, lung carcinoma, and gastric cancer cells [55, 225, 226]. In lung carcinoma and melanoma, exosomal miR-23a leads to EMT promotion [227, 228]. Exosomes from melanoma, gastric, and colorectal cancer contain miR-191 and let7a, which modulate EMT [229,230,231]. These examples leave little doubt that stromal EMT, facilitating migration of tumor cells, is controlled in part by TDEs.

3.2 Migration and Metastasis

When tumor cells lose their adhesion to the stroma they may migrate into the bloodstream (migration/metastasis). TDEs have been shown to play a role in this process as well. TDEs promote migration and invasion in oral and esophageal squamous carcinoma [188]. In breast cancer, miR-105 in TDEs influences invasion into blood vessels. In brain cancer, blocking TDEs from astrocytes inhibits metastasis [23]. Zhou et al. [170] showed that miR-105 in patient serum is prognostic for metastasis and miR-105 in the exosomes correlates with miR-105 levels in the primary tumor [170]. Further, miR-21 in serum exosomes correlates with migration, invasion, recurrence, and metastasis of esophageal cancer [232].

The tumor cells must recognize the prepared sites for invasion and, upon arrival to these distant sites, exit the circulation and invade the tissue. Prior to invasion and engraftment, the host organ is prepared by TDEs. Hence, the metastatic site is also a consequence of intricate tumor–stroma interactions [233]. Paget’s [235] 1989 ‘soil and seed’ hypothesis expanded upon Fidler [234], suggesting that circulating tumor cells form metastatic nucleation sites only if the seed (tumor cell) and host (organ) are compatible. A pre-metastatic niche formation is required for tumor cells to engraft a distant organ. It was shown in vivo and in vitro with malignant melanoma that this pre-metastatic niche is partially signaled through TDEs. In fact, TDEs are sufficient to direct tumor cells to a specific organ [155]. In malignant melanoma, melanoma exosomes home to sentinel lymph nodes in vivo and recruit melanoma cells [236]. An elegant study of an organotropic metastasis process in pancreatic cancer by Costa-Silva et al. [9] presented that the TDEs cause induction of inflammatory and fibrotic signaling, which activates migration inhibitory factor, which supports TGF-1 production, which enhances production of fibronectin, which attracts bone marrow dendritic cells to the liver. In support of this process, high levels of TGF-1 have been associated with poor prognosis in pancreatic ductal adenocarcinoma [237] and higher levels of migration inhibitory factor (supports TGF-1 function) were found in the serum of patients with pancreatic ductal adenocarcinoma with liver metastasis and progressive disease than in patients with full remission [9]. Additionally, TGF-1 inhibitors are having positive results in clinical trials [238]. In 2016, Lugini et al. [239] showed that exosomes from tumor cells induce transformation of mesenchymal stem cells, supporting the growing evidence that exosomes are important for metastasis.

3.3 Engraftment

When exosomes reach their pre-metastatic niche, they either become dormant tumor cells or proliferate. The TDEs recruit bone marrow dendritic cells by upregulating inflammation, which enhances vascular leakiness at the metastatic sites [240]. Hood et al. [236] showed that exosomes from highly metastatic melanoma cell lines localize to the lungs and increase the permeability of lung endothelium (likely by enhancing the expression of HSP90 and HSP70), which promotes invasion of tumor cells. These exosomes were also seen to localize to bone marrow, kidney, liver, and spleen, establishing metastatic niches, but do not remain in the blood. Bone marrow that has been exposed to metastatic melanoma exosomes had three times more metastatic spread in a mouse model. In pancreatic cancer, the exosomes from the pancreatic tumor interact with the liver Kupffer cells to promote engraftment of tumor cells in the liver [9]. Engraftment begins with reversion of EMT to an epithelial, proliferative, less migratory phenotype with establishment of stable cell junctions. In the recipient organ, TDEs help maintain the tumor cells by promoting immunosuppression. They inhibit effector cells and stimulate regulatory T cells [241, 242], reduce NK cell cytotoxicity [243, 244], and suppress T cell activation by activating myeloid-derived suppressor cells [245, 246].

4 Acquired Resistance Through Exosomes

Recognizing the response of tumors to stress and their subsequent exosome release is not only important for directing discovery, but also for understanding therapeutic responses. Hepatocellular carcinoma cells respond to anticancer drugs by releasing HSP-bearing exosomes, which stimulate the NK cell response [247]. Exosomes may also be involved in removing chemotherapeutic agents from the tumor cells. Exosomes from drug-resistant cells have been shown to contain multidrug resistance-related proteins [248] which help sequester the drugs into vesicles. P-glycoprotein and multidrug resistance protein-1 can be transferred from drug-resistant to drug-sensitive cells through exosomes [249, 250], resulting in an increase in the substrates efflux. For example, after treatment with cisplatin, ovarian carcinoma cells show an increase in the secretion of exosomes carrying cisplatin [251]. Irradiation can induce a senescent phenotype associated with increased production and release of exosome-like vesicles into the microenvironment that can potentially influence tumor progression [252]. For example, proton irradiation has the potential to increase the release of exosomes with survivin, an anti-apoptotic protein involved in cellular proliferation, survival, and tumor cell invasion [253]. The transfer of multidrug resistance-related mRNAs and microRNAs from drug-resistant to drug-sensitive cells may also play a role in the spread of resistance [254]. Interestingly, shear stress has been shown to induce the transfer of athero-protective microRNA from endothelial cells to smooth muscle cells, where they reduce atherosclerotic lesion formation [255].

5 Exosomes for Therapeutics

A significant response against cancer cells is antibody-dependent cytotoxicity, a process that is the basis for therapeutic antibodies [256]. Unfortunately, TDEs have been shown to neutralize antibodies against tumors by forming a complex at the surface [257], potentially reducing the effects of anticancer drugs [258]. For example, in breast cancer it was shown that TDEs express HER2 (human epidermal growth factor receptor 2) and epithelial cell adhesion molecule (EpCAM) antigens. These bind antibodies and render them ineffective [258]. B cell lymphoma secretes exosomes with CD20. The rituximab CD20 antibody therapy for B cell lymphoma is neutralized by these exosome–antigens. In support of this finding, depleting exosomes from the plasma of B cell lymphoma patients increased the efficacy of rituximab [257].

As reported in Sect. 3.1 and  3.2, inhibiting tumor exosome release may be a therapeutic approach in cancer treatment. The synthesis of ceramide (a component of the exosomal membrane) or the secretion of microRNAs could be chemically blocked. GW4869, a chemical blocker, impairs release of both exosomes and pro-inflammatory cytokines through blockage of neutral sphingomyelinase 2, which regulated both ceramide and microRNA secretion. GW4869 has been shown to inhibit tumor exosome release in vitro [259]. Unfortunately, concerns remain about possible systemic off-target effects of inhibiting exosome release or potential undesirable effects of altering microRNA secretion. It would be a lofty aim, and potentially beneficial, to find/develop exosome blockades that are cancer specific or only elicit local inhibition. With evidence emerging regarding the risk linked to pathway inhibitor metagenomics, and assays designed to detect associated exosome profiles, the future of synthetic nucleotides, encapsulation, and vesicle-based gain-of-function strategies would seem both opportune and obvious from the perspective of minimalist bacterial genomes and cells that have been built since 2010 by Craig Venter and his team at Synthetic Genomics (https://www.syntheticgenomics.com/).

APC-derived exosomes carry MHC-I molecules loaded with antigen. These are presented to dendritic cells, mediating T cell activation and T cell-dependent immune responses against the tumor cells. These properties with exosomes as antitumor vaccines support the added benefit of being cell-free [44]. The difficulty with this method is that APCs are very difficult to culture. The specific tumor antigen must be MHC-I haplotype matched and the specific tumor antigens need to be transferred to the APCs [260]. Hence, the exosomes would need to be prepared and collected from APCs for each individual patient based on their MHC-I haplotype and their specific cancer. TDEs can carry shared tumor antigens, not just antigens specific to one tumor [46], which may provide cross-protection against various cancer types. Importantly, these do not require MHC-I haplotyping. For example, human Mucin 1 (hMUC1) has been shown to indicate aggression and poor outcome. Exosomes carrying hMUC1 could be used as a vaccine against MUC1-expressing tumors independent of MHC-I [261]. Another example is the heat treatment of tumor cells to induce HSP70 release in the tumor exosomes. HSP70 stimulates monocytes and dendritic cells, amplifying the immune response further [260]. The HSP70-loaded exosomes could also directly activate NK cells to initiate the apoptosis of tumor cells [10], which has been shown to decrease the size of primary tumors and reduce metastasis in a mouse model [262]. Andre et al. [6] showed that exosomes from melanoma patient ascites activated dendritic cells to produce an antitumor T cell response [6]. Phase I clinical trials are showing promising results for stage III/IV malignant melanoma [263] and non-small lung carcinoma [264].

Competitive implant (metastatic trap [M-Trap]) is an interesting approach where exosomes are embedded in a three-dimensional scaffold and implanted into the peritoneum to establish an artificial pre-metastatic niche. This implant competes with the natural peritoneum for metastatic cells. This process has demonstrated significant efficacy in increasing survival of ovarian cancer patients [265]. An additional mode of treatment currently being investigated is transport of chemo-, microRNA, and biological therapeutics to the tumor cells using natural or synthetic nanoparticles. An example of natural nanoparticle transportation is the packing of paclitaxel into exosomes, which was shown to be 50 times more cytotoxic to drug-resistant lung cancer than conventional paclitaxel [266]. Synthetic nanoparticles could transport biologicals or tumor-suppressive microRNA to tumor cells via liposome preps or mimetics [267, 268] to reduce the toxicity of the chemotherapeutic agents [269]. For example, in a phase III metastatic breast cancer study, administration of liposome-encapsulated doxorubicin HCl with cyclophosphamide allowed a higher median cumulative dose (> 1260 vs. 480 mg/mL) to be safely administered [270, 271]. This methodology allows the use of higher concentrations for release at the tumor sites and has also been investigated in metastatic breast cancer and non-Hodgkin lymphoma. Cells can be transfected with therapeutic microRNAs to change the exosome cargo profile. This re-program could induce apoptosis or increase p53 signaling or direct macrophages [272], to name a few. The exosomes collected from these transfected cells may be used induce desired effects on recipient tissues. Alternatively, the use of anti-acidic approaches for targeting as well as treatment may be combined with known properties of exosome release and delivery [129, 130]. For example, the delivery of proton pump inhibitors (which correct the acidic pH) to reduce exosome release from tumor tissue [130] has been shown to re-sensitize melanoma to cisplatin treatment both in vitro and in vivo [273]. Another approach might utilize pH to control the release of therapeutic substances, such as CO2 [129] or acridine orange [274], to create a hostile tumor environment or targetable environment (e.g., photosensitization) [275].

6 Conclusion

A significant response against cancer cells is offered in antibody-dependent cytotoxicity. As a basis for therapeutics, this process embodies the specificity of the tumor that might accentuate the host immune systems to recognition, regulation, and enhancement against the hostile defense of transitional tumor cells. In this review, the exosomal contribution to cell reaction, immune dodging, and metastatic transition is explored with details, generalities, and specificities. Therein lies the challenge: specificity is a multi-gene transcriptional process, and while there are clearly checkpoint inhibition points that might be exploited in treatment, the accountable differences existing within distinct populations underpin the precision medicine that is being sought.

As a case in point comparing populations of European descent with African Americans, 2210 genes express more than two-fold increases or decreases in non-small cell lung cancer (NSCLC) from African Americans compared with matched normal tissue, while for European American samples, 2921 genes were differentially expressed by more than two-fold simply in NSCLC and normal tissue. These numbers indicate both overlap and non-overlapping specificity. Interestingly, 637 genes were differentially expressed only in African Americans and 1844 in European Americans. These numbers are likely different when derived from the cargo of TDEs, but the questions for precision medicine remain, including “should we concentrate on the overlapping signals for pan-tumor indicators?” or “should we concentrate on the patient-specific indicators?” I suggest that both are equally important approaches for different purposes. Pan-tumor indicators for specific cancer groups throughout ethnicities greatly simplifies the diagnostic biomarker process and could lead to more timely indications of developing cancers. The development of patient-specific indicators is crucial to prognostics and treatments for obvious reasons. Further, does the use of absolute genomics and a full read make sense, or do the transitional states of epitope morphogenesis noted herein bear a possible diferent strategy for exosomal seeding and membrane delivery?

It may be time to rethink how targeted therapy is initiated and monitored when managing cancer patients. Driven by multiple genetic changes and blood-based tests, a changing complex genomic landscape suggests that new first-line combination therapies target a fuller array of mutations to eradicate patient’s cancer before drug resistance can arise [276]. Theorizing that drugs targeting an EGFR mutation, as an example, may be able to wipe out the cells carrying that mutation, this approach risks leaving cells behind that might enhance additional mutations; contrasting the prevailing dogma that some cancers (e.g., NSCLC) are ‘driven’ by only one primary genetic mutation [186]. The single-driver view of cancer was buttressed by cell-free genomic studies to analyze patient blood samples for any mutations in 73 genes known to contribute to cancer. Overall, the researchers analyzed liquid biopsy data from 1122 patients whose tumors contained a mutated EGFR gene and 944 patients whose tumors did not have this mutation. Utilizing basic principles of the exosome platform of cell-free exchange and circulation, it was possible to determine that 92.9% of tumors from patients with advanced-stage lung cancer harbored multiple changes in cancer-related genes in addition to the EGFR driver mutation. On average, tumors contained two to three altered genes in addition to EGFR; however, some contained as many as 13 altered genes. What is the vector that might offer a successful incorporation of something akin to a manifold of identities so that each is seeded under the same tumor diagnostic? Perhaps it is first-line combination therapies capable of targeting multiple genetic pathways in patients’ tumors, rather than waiting for resistance to develop before initiating a subsequent drug. Alternatively, the use of more basic and global tumor properties to diagnose, target, and treat malignant tissue may be a preferred approach [277]. Examples include quantification of TDEs in the circulation and pH targeting or adjustment [132]. The International Society for Extracellular Vesicles (ISEV) has published a paper presenting the organizational opinions regarding clinical and regulatory foresight for EV-based therapeutics [278].

Recognition of the value of exosomes in the context of amplified release, early signaling to distal cells, harbingers of metastatic potential, and opportunity for multiple, minimally invasive access to tumor expression is increasing exponentially. Of note, the very simple property of increased secretion of exosomes from malignant tissue may be indicative as a stand-alone property. Furthermore, use of exosomes as a cell-free delivery of therapeutics might offer broader success and less advantage for tumor-evading genetic shifts. In some ways, the only meaningful strategy is to address the evolving genetic complexity with combinations of post-transcriptional and translational interception of undesirable cell–cell signaling.