Dexosomes as a cell-free vaccine for cancer immunotherapy

Cross-presentation of TAAs through DCs is mandatory for effective stimulation of T cells and initiation of antitumor cytotoxic effects [23]. cDC1s are dedicated to incorporate TAAs into MHC I proteins and present the TAA-MHC I complexes to CD8⁺ CTLs [24, 25]. Certain proteins associated with membrane trafficking are necessary for this process, including Sec22b (a member of the soluble N-ethyl maleimide (NEM)-sensitive factor attachment protein receptor (SNARE) proteins) and WDFY4. These proteins are not only necessary for controlling tumor growth but also they are required for the effectiveness of immunotherapies based on anti-PD1 (programmed cell death protein 1) agents [26, 27]. In addition to the proteins involved in TAA cross-presentation, there are other cDC1-associated proteins required for promotion of antitumor immune recalls [28]. For efficient stimulation of CD8⁺ T cells, TAAs should be transferred to lymph nodes draining the tumor via migrating CD103⁺ cDC1s in a CCR7-restricted mode [29]. XCR1 expressed by cDC1s contributes to the development of antitumor immunity by orchestrating localization of DCs in response to XCL1 (the XCR1 ligand) expressed by CTLs and NKs [30, 31]. cDC1s also promote local antitumor immune responses via producing CXCL9 and CXCL10 chemokines that stimulate CXCR3⁺ effector T cells and NKs [32, 33]. Moreover, these chemokines coordinate the localization of memory CD8⁺ T cells in cDC1-abundant regions to improve local restimulation of T cells [34, 35]. cDC1s also produce and release a large amount of IL12 which stimulates CTL and NK cytotoxic activity and promotes production of IFNγ [36‐39]. NKs were shown to have the ability of employing circulatory cDC1s to nearby tissues and tumors [40]. Flt3L molecules produced from intratumor cDC1s preserve the viability and activity of cDC1s inside the tumor micro environment and trigger local differentiation of DCs from their precursor cells [41]. Along with the induction of CTL expansion, cDC1s are also capable of stimulating the production of CD4⁺ Th1 cells by cross-presenting TAA-MHC II complexes [42]. The antitumor activity of cDC1s is supposed to be further assisted by pDCs [43]. As main producers of type I IFN, pDCs trigger antigen cross-presentation and CD8⁺ CTL antitumor immune response [44, 45]. Taken together, cDC1s represent an effective system for antitumor CTL stimulation by interacting with components of both the innate and adaptive immune systems.

Due to the absence of selective membrane markers required for precise detection of cDC2s in pathological contexts and the unavailability of sufficient preclinical investigations, the function of cDC2s in cancer immunology remains to be fully elucidated. cDC2s contribute to cross-presentation of TAA-MHC II complexes to CD4⁺ T cells [46‐50]. As a result, the activated CD4⁺ T cells promote antitumor immune responses by secreting IFNγ that triggers macrophages and NKs, blocks angiogenesis, regulates the tumor stroma formation, and leads to direct tumor cell lysis [51]. However, compared to cDC1s, cDC2s are less potent in cross-processing TAAs, migrating to tumor-draining lymph nodes, secreting IL12, and activating CD8⁺ CTLs [29, 32, 36, 52].

The communication and interaction between DC subsets and T cells plays a critical role at different stages of antitumor immunity. Maximal stimulation of CTLs is dependent on the activation of both cDC1s and cDC2s [42, 51, 53]. It was demonstrated that cDC2s lose their ability to induce CD4⁺ T cell differentiation during tumor growth. However, when T regulatory cells (Tregs) were depleted, cDC2 migration and activation capacity of CD4⁺ T cells for producing IFNγ was enhanced [54]. Additionally, cDC2s were reported to activate CD4⁺ T cells toward IL17-producing T cells [46].

Studies have shown that the functions of cDC1s and cDC2s may overlap to some extent, for example they both produce IL12 and depend on Flt3L for their development [55, 56]. Moreover, the number of blood-borne cDC1s and cDC2s is generally reduced in cancer patients [57]. However, cDC2s are known for not having a unique pattern of gene expression. Rather, they demonstrate a common gene expression pattern with monocytes with only a number of genes preferentially expressed, e.g. CCL22 which encodes for a chemokine that activates CCR4⁺ T cells [58]. It was shown that tumor-associated cDC2s possess langerin-encoding CD207 gene as a marker both in human and mouse lung tumors [59].

pDCs participate in exerting protective antitumor immune responses by producing IFNα that inhibits tumor growth, angiogenesis, and metastasis [60]. Both ex vivo and in vivo models [61, 62] demonstrated the direct cytotoxic function of pDCs via producing and secreting Granzyme B and TRAIL (TNF-related apoptosis-inducing ligand) molecules [63, 64]. pDCs are also capable of exerting indirect antitumor immunity by the OX40L-mediated production of IFNγ and the CCR5-mediated recruitment of NKs [65]. A unique subset of pDCs were identified in head and neck squamous carcinoma that overexpress OX40 and was reported to demonstrate synergizing effects with cDCs in inducing effective TAA-specific CD8⁺ T cell responses [66].

moDCs.

Due to their overlapping functions of moDCs with other myeloid cells, their role in exerting antitumor immunity in human is not clear yet. However, they probably play significant roles in stimulating the propagation of naïve CD8⁺ T cells [67]. Preclinical investigations suggested central roles for moDCs in regulating antitumor immunity during chemotherapy, cell vaccination, and T cell adoptive therapy [68‐70].

Exosomes produced and released by DCs are termed as ‘dexosomes’. As is the case with other exosomes, dexosomes have a characteristic molecular profile of their own. Human dexosomes contain cargos that together operate as a whole antigen-presenting entity. These include a variety of all the known antigen-presenting molecules, such as MHC I/II proteins and costimulatory factors like CD86 [90], which are employed for cross-presentation of peptide antigens to CD8⁺ and CD4⁺ T cells and subsequent triggering of their proliferation. Dexosomes also harbor CD1a, b, c, and d proteins that are involved in cross-presentation of lipid antigens [90]. Dexosomal ICAM1 (intercellular adhesion molecule 1, CD54) was shown to play pivotal role in regulating DC-T cell communication [91]. As a ligand of Mac1 integrins (CD11b/CD18) [92] and the lymphocyte function-associated antigen 1 (LFA1, CD11a/CD18) [93], ICAM1 can either facilitate dexosomal capture by target DCs or promote the interaction of T cells with dexosome-receiving DCs that hold dexosomes on their external surface. Expressing an abundance of microdomain-organizing tetraspanin proteins including CD9, CD37, CD53, CD63, CD81, and CD82, which regulate dexosome-target DC interactions, are considered the hallmark of dexosomes [94]. The presence of CD55 and CD59 molecules on dexosomal surface prevents complement-mediated degradation of dexosomes throughout their extracellular journey [95]. Tsg101 and Alix proteins determine the sorting of ubiquitinated cargos into ILVs during dexosome generation process [96]. MFGE8, that binds to phosphatidylserine on the external surface of dexosomes, promotes dexosomal uptake through interacting with integrins αvβ3 and αvβ5 on APCs [97]. However, successful capture of MFGE8-deficient dexosomes by bone marrow-derived DCs (BMDCs), which produce little or no αvβ3 or αvβ5 integrins in vitro, proved the presence of MFGE8-independent machineries involved in dexosomal uptake [98]. While HSPs, FasL, and CD11b and c are common between human and mouse dexosomes, MFGE8 was only detected in murine monocyte-derived DC (MCDC) dexosomes [98]. HSC73, a member of the HSP70 family, is also abundantly present within the dexosomal cytosol [97]. In cooperation with the members of the HSP90 family, HSC73 probably regulates the immunogenicity of dexosomes by triggering different cells of the immune system and playing pivotal roles in MHC loading and as antigen chaperones [99]. In addition to proteins, dexosomes also harbor various RNA species with the aim of intercellular communication and to induce certain posttranslational modifications in the recipient DCs. For example, it was suggested that dexosomal transfer of miRNAs could suppress the targeted mRNAs in DCs [100], indicating that certain RNA profiles of dexosomes, or particularly those of parent DCs, can impact dexosome immunogenicity.

In comparison with the cellular membrane, dexosomal membrane shows an increased transverse diffusion of phospholipids (flip-flop movements) which results in a loss of lipid asymmetry in the membrane structure. The elevated transbilayer movements of phospholipids along with the rigidity of dexosomal membrane at neutral pH control their fusion with other membranes which in result guarantees the stability of dexosomes in circulation [101]. The phospholipid composition of dexosomes is also distinct from their donor DCs. While the amount of sphingomyelin is twice as high in dexosomes, phosphatidylcholine is much lower and cholesterol is absent from dexosomal membrane. A remarkable enrichment of disaturated molecular species such as phosphatidylethanolamines was also distinguished in dexosomal membranes in addition to a 50% decrease in molar ratio of diglicerides:phospholipids. Further investigations demonstrated the abundance of phospholipase D2 in dexosomes, and that phospholipid D probably mediates the putative signaling properties of dexosomes either by cooperating with a second messenger like phosphatidic acid or by association of dexosomes with the target DCs through the fusogenic quality of phosphatidic acid [101, 102].

Several mechanisms have been suggested on the topic of how dexosomes contribute to cross-presentation of TAAs by means of MHC proteins and trigger T cell immunity in lymph nodes. One theory suggests direct triggering of T cells by dexosomes in vitro (Fig. 2a). However, the direct dexosome-T cell route was reported unsuccessful in inducing naïve T cells and is less likely to arise extensively in vivo [120]. Most probably, dexosomes are not able to interact with T cells until they are captured by other DCs which extract and process the antigenic material from the TAA-MHC complexes and use them for priming specific T cells [121]. Furthermore, it is more probable that the direct mechanism is only efficient in restimulating memory T cells, formerly-activated T cells, or T cell clones, lines and hybrids [122]. Therefore, dexosomes possess less T cell stimulatory capacity than their parent DCs [117], although their stimulation potency can be promoted when they are immobilized or when their concentration is enhanced [117, 122].

Rather than a direct route for dexosome-T cell activation, accumulating evidence revealed that dexosomes or any other APC-originated exosomes initiate T cell immunity via indirect presentation of antigens to the adjacent APCs [7]. According to this hypothesis, the biological cycle of dexosome release continues as follows (Fig. 2b): immature naïve DCs regularly circulate within the body in order to identify exogenous molecules and antigens originating either from infectious sources or tumors. Once encountered, DCs present at the locality of infection or tumors capture the antigens, e.g. TAAs, and process and convert them to smaller peptide fragments which are then incorporated into MHC I and II molecules. Upon maturation, the antigen-MHC complexes are located on the external surface of DCs and they start to travel to local lymph nodes where they prime antigen-specific T cell responses. Throughout this migration, the antigen-MHC molecules and other immunostimulatory factors are sorted into dexosomes and released to the extracellular space. The secreted dexosomes then proceed to lymph nodes where they transfer their components to the resident naïve DCs. ICAMs and integrins contribute to uptake of dexosomes by bystander DCs. After binding to DCs, some of dexosomes (and not all of them) are internalized and the remaining vesicles probably retain on the external surface of DCs. The fraction of internalized dexosomes depends on the maturation status of the target DCs. Mature DCs maintain dexosomes mostly on their external surface whereas immature DCs tend to internalize them. It is assumed that the surface-retained dexosomes still have the capacity to stimulate T cells [120]. If internalized, the peptide-MHC complexes are processed via the endosomal route, leading to the transfer of dexosome-borne antigenic peptides to MHCs [120, 122, 123]. Afterwards, these antigen-MHC complexes are carried back to the DC membrane surface where they are presented to T cells [90]. Stimulation of naïve T cells was demonstrated to take place only in the presence of APCs [120]. The activation status of parent DCs affects the efficiency of indirect T cell stimulation mechanism to a great extent. For instance, LPS- or IFNγ-matured DCs release dexosomes that abundantly express ICAM1 which improves dexosomal capture, MHC and CD86 molecules which facilitate T cell priming [111].

A second indirect route proposed for dexosome-mediated T cell priming is a process called ‘dexosome-to-DC cross-dressing’ in which dexosomes convey their peptide-MHC molecules to DCs by directly fusing with their plasma membrane (Fig. 2c) [124]. This allows T cells to immediately recognize MHC-incorporated peptides and take advantage of the costimulatory factors and additional molecules on DC surface without the necessity for internalizing or processing of the antigens by DCs. To support this pattern, Thery et al. demonstrated that dexosomes could prime T cells only in the presence of mature CD8α⁻ DCs, even if the mature DCs were MHC II-deficient [125]. This proves the presence of dexosome-to-DC cross-dressing of antigen-MHC molecules which depends greatly on CD80 and CD86 costimulatory molecules for stimulation of T cells [125]. However, dexosome MHC I cross-dressing of adjacent DCs did not trigger the stimulation of ovalbumin-specific CD8⁺ T cells. Instead, dexosomes were internalized (as mentioned before) and the antigen-MHC I complexes were subsequently presented on the DC surface [126].

A third mechanism through which dexosomes can induce T cells occurs via cancer cells (Fig. 2d). Based on recent observations, dexosome-treated human breast adenocarcinoma cells (in comparison with untreated cells) were able to restimulate formerly activated T cells, resulting in extensive proliferation of IFNγ-secreting T cells [127]. Reception of dexosomes by cancer cells indicates the feasibility of converting cancer cells to more powerful immunogenic targets which presents novel avenues for development of therapeutics that enhance tumor immunotargeting.

In a cancer setting, a significant role of DCs and dexosomes is to identify and destroy tumor cells. Indeed, DCs represent the initial bridge between the host immune system and the ongoing oncogenesis process. This is the first stage of cancer-immunity cycle that attempts to eradicate tumor cells by activating T cells [128]. During oncogenesis, TAAs are produced and secreted and then captured and processed by proximal DCs for cross-presentation to T cells, which results in anti-TAA T cell priming. However, T cell propagation is only stimulated when given additional prerequisites are met within the local tumor microenvironment [128]. These include local immunogenic signals, i.e. proinflammatory cytokines and pathogen- or damage-associated molecular patterns (PAMPs or DAMPs), which force DCs to present the received TAAs to cognate T cells through MHC I/II and costimulatory proteins [128]. Dexosome-based antitumor signaling route is able of modulating tumor cells beyond the level of conventional ligand-receptor signaling routes, leading to elaborate modifications which control tumor progression and antitumor immune responses. Munich et al. revealed that dexosomes can stimulate caspase activity and result in tumor cell apoptosis via expressing the ligands of TNF superfamily such as TNF, FasL and TRAIL on their external surface [129]. Consistent with these findings, a study demonstrated that hyperthermic CO₂-treated dexosomes were able to suppress proliferation and trigger apoptosis to a certain extent in gastric cancer cells [130]. Moreover, dexosomes were shown to induce propagation of splenic cells and promote cytotoxic ability against L1210 tumor cells [131].

Accumulating evidence have demonstrated that dexosomes mediate interactions that result in stimulation of cells of the innate immune system. Of note, in addition to MHCs, dexosomes carry proteins that are able to trigger or inhibit immune recalls in an antigen-independent manner. Dexosomes can stimulate NKs by providing ligands that bind to NK-activating receptors. This process is either mediated by dexosomal HLA-B-associated transcript 3 (BAT3; BAG6 (BCL2-associated athanogene 6); a ligand for natural cytotoxicity triggering receptor 3 (NCR3)) [132] or by UL16-binding molecules, i.e. MHC I polypeptide-related sequence A and B (MICA and MICB; ligands of NKG2D (natural killer group 2 member D)), on dexosomes [133]. In a murine model of advanced melanoma, dexosomes were shown to induce IL15Rα and NKG2D-dependent proliferation of NKs and promote IFNγ release which result in antimetastatic effects of NKs within the local tumor environment [134]. In human melanoma, dexosomes expressing NKG2D ligands on their surface could directly interact with NKG2D and NKs, supporting the hypothesis that dexosomes are capable of stimulating antimetastatic immune responses through a non-MHC-dependent manner. Similar to DCs, dexosomes can affect NKs to produce IFNγ by the interaction of dexosomal TNF with TNF receptor on NKs [129]. Dexosomes also express TLR4 (Toll-like receptor 4) and TLR1/2 ligands on their surface which cause enhanced expression of TNF and subsequent activation of NKs [135]. These findings describe novel characteristics of dexosomes in regulation and elicitation of NK-related immune responses, and propose new approaches for evaluating dexosome-mediated antitumor efficacy.

Several studies revealed that dexosomes have the potential of activating CD8⁺ T cell clones in vitro either alone [136] or when incubated with DCs that produce allogeneic MHC I proteins [137]. These findings indicate functionality of dexosomal antigen-MHC I assemblies. The first evidence supporting dexosome-mediated triggering of CD8⁺ T cells was reported in 1998 by Zitvogel and colleagues [107]. They showed that dexosomes isolated from BMDCs containing TAA-MHC I complexes acted efficiently in both suppression and eradication of an established malignancy in immune-competent but not immune-deficient mice [107]. The efficiency of this process was improved when dexosomes were administered concomitantly with mature DCs or chemical adjuvants that encouraged DC maturation [118]. Zitvogel et al. also demonstrated that dexosomes were more effective than their parent DCs in terms of tumor suppression, and that autologous (not allogeneic) dexosomes could induce TAA-targeted CTL response ex vivo, emphasizing the role of dexosomal MHC I throughout this process [107]. Dexosomes of mature DCs (in comparison with the dexosomes of immature DCs) were more effective in triggering CD8⁺ T cell immunity, indicating the importance of costimulatory factors present on dexosomes of mature DC origins [138]. Another study demonstrated that human dexosomes loaded directly with MART1 peptide (melan-A antigen; a TAA) harbored intact functional peptide-MHC I assemblies to target DCs ex vivo [118]. Here, it was further shown that dexosome-pulsed DCs were more effective in activation of CD8⁺ T cells than peptide-pulsed DCs. Induction of CD8⁺ T cell recalls was also confirmed when DCs lacking the TAP molecules (transporter associated with antigen processing) were used as the recipient [139]. In endocytic pathway, internalized antigens are carried from endosomes into the cytosol for proteasomal degradation [140]. Antigen-derived peptides are then carried by the TAP molecules into the endoplasmic reticulum or back into the antigen-containing endosomes, where they can be incorporated onto MHC I molecules [141]. However, Lawand et al. recently reported that some antigens may enter the endocytic pathway in a TAP-independent manner, indicating the possibility of other transporters dedicated to antigen translocation into endosomes for cross-presentation to CTLs [139].

Dexosome-mediated transference of peptide-MHC I assemblies to DCs was also observed in vivo. Autologous dexosomes loaded with antigenic peptides were able to transfer them to allogeneic DCs and initiate peptide-specific stimulation of CD8⁺ T cells in mouse. Intriguingly, intravenous administration of autologous dexosomes alone did not trigger any CD8⁺ T cell response [118]. Likewise, no noticeable level of antigen-specific CD8⁺ T cells was observed when MHC I-restricted peptide of ovalbumin (OVA, SIIN-FEKL) was loaded onto dexosomes [142]. Conversely, whole OVA protein-loaded dexosomes (with indirect method) were capable of initiating protein-targeted CD8⁺ T cell response. This effect depended mostly on CD4⁺ T cells and partly on B cells, particularly marginal zone B cells [142]. Further investigation by Hao et al. revealed that CD8⁺ T cell propagation induced by dexosomes loaded with protein relied on CD4⁺ T cells [143] and CD11c⁺ DCs [144].

A number of strategies was adopted in order to promote dexosomal-mediated antigen-specific CD8⁺ T cell responses. Viaud et al. introduced IFNγ as a key cytokine that stimulates dexosomal expression of CD40, CD80, CD86, and CD54 molecules which result in induction of direct and powerful antigen-dependent CD8⁺ T cell responses by dexosomes derived from IFNγ-matured MCDCs [111]. Another strategy is to inject dexosomes comprising a danger signal like a TLR ligand, such as polyinosinic:polycytidylic acid (poly (I:C)) or CpG-ODN, that boost DC maturation [118]. Here, a humanized MHC I-deficient murine model was used for administration of two therapeutic dosages of dexosomes pulsed with human peptides and coinjected with CpG-ODN, and it was found that tumor development was significantly decreased compared to mice that received tumor peptide-CpG-ODN [118]. αGC-loaded dexosomes were also reported to promote CD8⁺ T cell responses against a concurrently loaded antigen [145]. When DCs were exposed to the lysates of B16F10 melanoma cells with the aim of loading TAAs onto dexosomes, the prepared vaccine led to stimulation of melanoma-specific CD8⁺ T cells and recruitment of CTLs, NKs, and NKTs in the subcutaneously grafted melanoma tumors in mice. Consequently, tumor development was remarkably decreased and survival prolonged. DCs loaded with gastric TAAs demonstrated the ability to trigger the proliferation of CTLs. Additionally, by binding to TLR ligands, dexosomes can activate adjacent DCs to express transmembrane TNF and produce pro-inflammatory cytokines [146].

Several investigations attempted to change the molecular make-up of dexosomes to generate tolerogeneic vesicles with immunosuppressive features. Genetically-modified BMDCs produce IL4, IL10 or FasL molecules that repress the inflammation caused by DTH in a murine model of collagen-induced arthritis [147, 148]. Similarly, when donor dexosomes were injected to a rat model of cardiac transplantation, chronic allograft rejection response was remarkably delayed [149]. Dexosomes produced from TGFβ1- and IL10-matured DCs could also induce immune tolerance in a skin allograft murine model [150]. Moreover, DCs overexpressing indoleamine 2,3-dioxygenase (IDO) molecules produced dexosomes that reduced inflammation in a rheumatoid arthritis model [151]. Lu et al. recently demonstrated that in a murine model of autochthonous hepatocellular carcinoma, mice treated with dexosomes isolated from DCs that expressed α-fetoprotein (AFP) had remarkably more IFNγ-producing CD8⁺ T cells, enhanced levels of IL2 and IFNγ, fewer Tregs and reduced levels of IL10 and TGFβ [152]. Therefore, dexosomes are capable of promoting T cell stimulation along with downregulating immunosuppressive responses, which serve as a promising tool to create efficient antitumor vaccines.

While dexosomes can present antigens and directly trigger cognate CD8⁺ T cell clones, lines [138], or primed CD4⁺ T cells [153], they need to be captured by bystander DCs to activate naïve CD4⁺ T cells [91, 125, 153]. Dexosomes were reported to transfer peptide-MHC II complexes to MHC II-deficient DCs, and allowed them to trigger antigen-specific CD4⁺ T cells [91, 125]. As reported for CD8⁺ T cells, dexosomes of mature DCs were also more effective in CD4⁺ T cell activation in vitro [91]. Peptide- or protein-loaded dexosomes could hinder tumor growth by recruiting CD4⁺ and CD8⁺ T and B cells in vivo [142]. Further in vivo studies demonstrated that dexosomes loaded with antigens and generated using TLR3 agonist, poly (I:C), and OVA initiated the propagation of OVA-specific CD4⁺ and CD8⁺ T cells [154]. These results were remarkably superior compared to using CpG-B and LPS for TLR9 and TLR4, respectively. When antigen-pulsed dexosomes from mature, but not immature, DCs were injected from male into female mouse models, the male skin grafts were rejected because activated CD4⁺ T cells were differentiated into effector CD4⁺ T cells in vivo [91].

As previously mentioned for CD8⁺ T cells, dexosomes are able to initiate propagation of antigen-specific CD4⁺ T cells once they are loaded with whole protein antigens. It was assumed that this effect depended upon a functional compartment belonging to B cells, since the proliferation of CD4⁺ T cells was not detected in B cell receptor signaling deficient btk^−/− mice [155]. These findings imply that dexosome-borne antigens are ingested, processed, and presented by DCs to T and B cells, a hypothesis further approved in an investigation where allogeneic I-Ad⁺ dexosomes could stimulate allo-specific CD4⁺ T cell proliferation in I-Ab⁺ mice [121]. Dexosomes loaded directly or indirectly with peptide-MHC II assemblies could trigger specific CD4⁺ T cell responses in vivo [121, 125]. In vitro, dexosomes could not trigger antigen-specific T cell activation unless mature CD8α⁻ DCs were also present in the culture environment. These mature DCs could be MHC II-deficient, but had to express CD80 and CD86 costimulatory factors. Moreover, in comparison with CD8α⁺ DCs, CD8α⁻ DCs were more powerful in induction of CD4⁺ T cell immunity [125]. In another study, it was reported that CD8α⁺, but not CD8α⁻, DCs purified from the lymph nodes of dexosome-treated mice were capable of stimulating antigen-targeted CD4⁺ T cell proliferation ex vivo [121, 156]. Qazi et al. compared the efficiency of dexosomes directly loaded with OVA peptide with dexosomes derived from OVA-pulsed DCs in initiating specific CD4⁺ T cell responses in vitro and in vivo, and showed that both dexosomes could elicit T cell proliferation in vitro, with peptide-loaded dexosomes being more effective. Conversely, in vivo, only dexosomes produced from OVA-pulsed DCs could induce CD4⁺ T cell proliferation, emphasizing the significance of indirect antigen-loading approaches in clinical applications. Moreover, these dexosomes were able to induce the polarization of T cells to the Th1 type in a B cell-dependent fashion, which highlights the importance of B cells in producing T cell responses through a dexosome-dependent pathway [155].

Exosomes originated from various APC origins contribute to the elicitation of B cell immunity both ex vivo and in vivo. Segura et al. demonstrated that dexosomes can harbor both antigen-MHC assemblies and ICAM1 molecules to less-efficient APCs, such as B cells, leading to T cell stimulation in an indirect way [91]. Naslund et al. also demonstrated B cells were essential for optimal triggering of CD8⁺ T cells via dexosomes [157]. Mycoplasma-infected BMDCs release dexosomes that are able to initiate polyclonal propagation of primary B cells independent of CD40, LPS, or CpG signaling pathways in vitro [158]. In another study, allogeneic BMDC dexosomes were administered systematically (intravenous/intraperitoneal injection) to rats before transplantation with the aim of exploring anti-allograft immune responses. Here, dexosomes could initiate the in vivo production of IgG2a and b antibodies (type I antibodies) specific to dexosomal antigens, and resulted in extended survival of the allograft [159]. Likewise, BMDCs pulsed with diphtheria toxoid or OVA antigen resulted in the production of antigen-specific type I antibodies [155, 160]. Titers of specific antibodies can be amplified by simultaneous pulsing of BMDCs with αGC and an antigenic protein [145].

Taken together, all the above-mentioned reports confirm that dexosomes are perfectly capable of exerting potent immune responses and hence possess great therapeutic values against a variety of immune-related diseases, including malignancies.

Autologous MCDC cultures loaded with HLA-elicited MAGE-A3, -A4, -A10, and MAGE-3DPO4 peptides (melanoma-associated antigens) were employed in the primary phase I studies. In both trials, four doses of dexosome vaccinations were administered at weekly intervals. Thirteen HLA-A2⁺ participants with pretreated final stage (IIIb and IV) non-small cell lung carcinoma (NSCLC) overexpressing MAGE-A3 or -A4 antigens were qualified to receive dexosome immunotherapy in the first line of phase I trials [161], and nine participants completed the therapy. One week following the last vaccination, three of the tested patients, who had not shown MAGE-sensitive immune responses before immunization, exhibited systemic MAGE-specific immune reactivity as confirmed by DTH response. Increased MAGE-specific T cell function was only observed in one of five patients, as determined by enzyme-linked immunospot (ELISPOT) assay. This low-level T cell reactivity was attributed to the possible suppression of Tregs (CD4⁺ CD25⁺ T cells). In two of three participants, Tregs were amplified (compared to the baseline levels) as a percentage of total CD4⁺ T cells following dexosome therapy. An improvement of the NK lytic activity was detected in two of four tested samples. Taken as a whole, the NSCLC phase I trial was well tolerated and showed an acceptable safety profile, with disease stability shown in two participants who had progressive tumors at diagnosis. Additionally, disease stability continued for over twelve months in two of four participants who had initially stable disease [161].

In the second study of phase I trials, fifteen participants with the following criteria were enrolled: stage IIIb/IV, HLA-A I⁺ or HLA-B35⁺ and HLA-DPO4⁺ leukocyte phenotype, MAGE-3-overexpressed malignant melanoma (MM) [162]. The MCDC dexosome administration format included four vaccine doses at weekly intervals. Two dosages of either MHC II proteins (0.13 and 0.40 × 10¹⁴ molecules) or peptides (10 and 100 µg/mL) were administered. Evaluation of the therapy efficiency was conducted two weeks postimmunization. One of the patients showed a partial reactivity to dexosome therapy. In this patient, a depigmentation halo surrounding naevi was observed, and the arterial neovasculature was disappeared and tumor retracted. This participant received four months of continuation therapy with dexosomes which resulted in the disease stabilization and toxicity reduction. Stabilization of the disease for up to twenty four months was too observed in another participant who was given continued dexosome therapy. Taken together, this study resulted in two stabilized diseases along with one minor, one partial, and one mixed reactivity at lymph nodes or skin. A number of these results were achieved in participants with invasive tumors who had formerly gone through other therapies or received alternate anticancer vaccines. As observed in the first phase I trial, neither MAGE-specific CD4⁺ and CD8⁺ T cell activation nor DTH response were identified in the peripheral blood of patients [162]. However, it was demonstrated that immature MCDC dexosomes express NKG2D ligands on their membrane which bind to NKG2D on NKs, resulting in pro-NK effects [134]. After four weeks of vaccine administration, the number of circulatory NKs were remarkably enhanced. After dexosome immunotherapy, the expression of NKG2D and NK cytotoxicity were maintained in 50% of patients who had NK function deficit at the beginning of the study [134]. Moreover, it was found that dexosome therapy could trigger NK proliferation in vivo in an IL15Rα-dependent fashion. These findings were in consistence with the improved control of tumor metastasis in B16F10 melanoma cell-inoculated mice by NK1.1⁺ cells [134]. Human dexosomes derived from immature DCs were reported to present BAG6 molecules on their surface [164], which is a ligand for NKp30 receptors expressed on NKs [165]. Cytokine release from NKs was reported to be directly correlated with the expression levels of BAG6 on dexosomes [164]. Furthermore, dexosomal surface expression of TNF were able to trigger the production of IFNγ by NKs [129].

Defective antitumor effects of first-generation dexosomes (IFNγ-free dexosomes) in the induction of NKs and T cells in the phase I trials encouraged researchers to design and develop innovative strategies in order to promote dexosome-dependent antitumor host immune responses. As mentioned before, one strategy is to utilize dexosomes originated from LPS- or IFNγ-matured DCs, which exhibit greater T cell immunity [111]. A clinical-grade process for production of dexosome vaccines was developed when these results were applied to human DC cultures [111]. Here, IFNγ was employed for stimulating human DCs in culture, and subsequently, dexosomal costimulatory factors and ICAMs were upregulated, resulting in second-generation dexosomes (IFNγ dexosome) with increased immunostimulatory capacity [111, 166]. The phase II clinical trial aimed to investigate whether maintenance immunotherapy of advanced NSCLC patients using IFNγ dexosomes could result in progression-free survival (PFS) at four months postchemotherapy [163]. Twenty two HLA-A2⁺ patients who had inoperable (stage IIIb or IV) NSCLC with neutrophils ≥ 1.5 × 10⁹/L and showed immune responses or disease stabilization following four rounds of a first-line platinum-based chemotherapy were qualified for receiving IFNγ dexosome [163]. The utilized TAAs included MAGE-A1, -A3, NY-ESO, MART1 (all MHC I-restricted peptides) and EBV (MHC II-restricted peptides). Patients first received three weeks of metronomic oral low-dose cyclophosphamide (CTX). This regimen was proven to decrease Treg activity and induce IFNγ- and IL17-generating T cell clones according to several preclinical [167‐169] and clinical investigations [168, 170]. Therefore, dexosome-mediated priming of T cells is facilitated, and NK and T cell functions are preserved. Of these participants, seven patients (32%) showed disease stability following nine times of dexosome vaccination, and proceeded to receive the therapy at three-week intervals. Unfortunately, the main clinical outcome of the phase II trial, a PFS of 50%, could not be achieved and no remarkable immune reactivity was reported in the study. Despite loading of multiple epitopes and CTX adjuvant therapy, the use of IFNγ dexosomes was insufficient to manifest TAA-specific T cell reactivity [163]. However, one participant elicited a long-term disease stabilization which allowed for surgical removal of the tumor and the eligibility for local adjuvant radiotherapy. Moreover, second-generation dexosome therapy was reported to promote NKp30-mediated NK function. Although downregulated in stage IV NSCLC, NKp30 activation improved the release of TNFα and IFNγ by blood NKs, which was detected after four cycles of dexosome vaccination. More significantly, this augmented NKp30-mediated NK function was associated with longer PFS [163]. Moreover, the aforementioned membrane-associated NKp30 ligand, BAG6, was detected on the membrane of dexosome vaccine preparations and was reported to be responsible in NKp30-dependent pro-NK activity. To further support this theory, it was shown that the concentration of MHC II molecules in dexosome inocula and NKp30-dependent NK activity correlated with the levels of dexosomal BAG6. This is different from the results obtained in MM phase I trial where NKG2D (and possibly IL15 or IL15Rα) mediated dexosome-related NK responses [134, 162]. Since dexosome preparations employed in the phase I MM study were not obtained from IFNγ-matured DCs (where IFNγ results in BAG6 overexpression [171]), the NKG2D ligand-mediated NK activation was probably manifested more dominantly in the absence of BAG6-NKp30 pathway.