Macrophages exist in almost all tissues of our body and are differentiated from peripheral blood monocytes upon injured tissue recruitment (54). They can not only phagocytose intrusive pathogens but also induce inflammatory response by releasing chemokines (1). Furthermore, they are antigen-presenting cells (APCs) to process and present peptides of phagocytic pathogens on major histocompatibility complex class II (MHC-II) receptor, activating adaptive immune system (54). The macrophages can be reprogramed into two different polarization states to exert their Janus functions, that is classical activated M1φ phenotype (pro-inflammatory) and alternative activated M2φ phenotype (anti-inflammatory) upon the stimulus of cytokines (55). However, the unbalanced M2φ/M1φ ratio or delayed transition from M1φ to M2φ phenotype would lead to continual inflammation (8).

Li et al. observed that EVs derived from human UC-MSCs (hUC-MSC-EVs) reduced the number of macrophages (CD68) to suppress burn-induced inflammation (23). They claimed that miR-181c enriched in hUC-MSC-EVs decreased the expression of toll-like receptor 4 (TLR4), a receptor of lipopolysaccharide (LPS), and subsequently reduced the activation of nuclear factor kappa B (NF-κB)/p65. In another study, hUC-MSC-EVs were also reported to modulate PI3K-AKT signaling pathway to shift macrophage polarization from M1φ to M2φ phenotype by their abundant has-miR-122-5p, -148a-3p, -486-5p, -let-7a-5p, as well as 100-5p (24). Surprisingly, EVs derived from BM-MSCs (BM-MSC-EVs) were reported to achieve “One Stone Two Birds”. They decreased the infiltration number of M1φ phenotype and promoted M2φ phenotype polarization in the plaque of atherosclerosis through miR-let7/IGF2BP1/PTEN and miR-let7/HMGA2/NF-kB signaling pathways, simultaneously (26).

Microglia are a special population of macrophages in the central nervous system. Microglia can not only phagocytose apoptotic neurons but also control synaptic pruning in the brain (56). Microglia can be activated in a phagocytic state under the stimulus of nervous system injuries, releasing lots of cellular inflammatory transmitters (57). Like macrophages, activated microglia can be polarized into pro-inflammatory M1 and anti-inflammatory M2 subtypes, exerting their damage or protection function on neural network (58). Therefore, the proliferation, migration, and activation of microglia are crucial for nerve-related inflammatory responses.

Similarly, BM-MSC-EVs showcased analogical “One Stone Two Birds” modulation on microglia compared with macrophages. BM-MSC-EVs promoted microglial M2 polarization through activating NF-κB pathway and decreased the number of CD68+ microglia (29). In another study, human AD (hAD) derived MSC-EVs were reported to mitigate neuroinflammation mainly by suppressing M1 microglial activation through NF-κB and MAPK pathways and secondarily promoting M2 microglial polarization (30). Moreover, the EVs derived from human retinal progenitor cells were reported to stabilize microglia to avoid its excessive activation, migration, and proliferation, thereby decreasing the secretion of inflammatory cytokines like ionized calcium binding adapter molecule 1 (Iba1) and increasing the anti-inflammatory gene expressions of IL-4, IL-10, and TGF-β, thus mitigating neuroinflammation (59).

In short, the modulation of macrophages or microglia by MSC-EVs is mainly based on two aspects: decreasing their infiltration number and shifting their polarization from M1φ to M2φ subtype, but their potential mechanism still needs to be further explored.

To improve the immunomodulatory functions of MSC-EVs on macrophages, some preconditioning methods, like the stimulus of hypoxia (27, 60), pharmacological agents (25, 61, 62) or pro-inflammatory cytokines (28, 63, 64), have been developed. For example, Sicco et al. pre-exposed hAD-MSCs to hypoxic condition (1% O₂) and collected the secreted EVs (27). The pretreatment of 1% O₂ elevated the contents of miR-223 and miR-146b in hAD-MSC-EVs. The hypoxia-changed hAD-MSC-EVs promoted the macrophage polarization from M1φ to M2φ phenotype, and then downregulated the production of interleukin (IL)-6 and Nos2, followed by upregulating the expression of Arg1 and Ym1 in vivo and in vitro. As for pharmacological agents, Ti et al. claimed that the pretreating of hUC-MSCs with LPS (100 ng/mL) for 48 h enhanced the expression of let-7b inside hUC-MSC-EVs. Let-7b is reported to regulate TLR4/NF-κB/STAT3/Akt pathway which is the potential controller for the macrophage plasticity (25). The precondition of parent cells with pro-inflammatory cytokines was also observed to increase the shifting effect of MSC-EVs on macrophages by improving the contents of the immunoregulatory microRNAs (miRNAs) (63–65) and proteins (28, 66). For example, IL-1β-primed (10 ng/mL, 12 h) hUC-MSCs were observed to secret miR-146a-enriched EVs, which promoted M2φ polarization, efficiently (63). In an interesting study, hUC-MSCs pretreated with tumor necrosis factor alpha (TNFα, 1 ng/mL, 3 d) generated the EVs containing upregulated miRNA-299-3p which accounted for the inhibiting effect of hUC-MSC-EVs on the activation of NLRP3 in macrophages, partially (65). Additionally, preconditioning hAD-MSCs with the mixture of interferon gamma (IFNγ)/TNFα (40 ng/mL, 48 h) upregulated the contents of miR-34a-5p, miR-21 as well as miR-146a-5p inside hAD-MSC-EVs. The obtained hAD-MSC-EVs switched macrophages from M1φ to M2φ phenotype (64). Similarly, in the treatment of experimental murine colitis, preconditioning cAD-MSCs with the mixture of IFNγ/TNFα (20 ng/mL, 24 h) upregulated the expression of immunosuppressive proteins such as hepatocyte growth factor (HGF) and transforming growth factor beta (TGF-β), thus dominating the polarization of macrophages (28).

Collectively, the existing pretreatment methods can efficiently enrich the immunomodulatory components in MSC-EVs. We believe that the exploration to obtain MSC-EVs with more specific components and greater immunomodulation potential via milder preconditioning may be the research direction in the future.

As the main type of phagocytic granulocytes, short-lived neutrophils are recruited to injury sites firstly (1). They phagocytize and destroy pathogens accurately and rapidly by releasing lytic enzymes and producing reactive oxygen species (ROS), further undergoing neutrophil extracellular traps (NETs) (67). Furthermore, the phagocytosis of apoptotic neutrophils can promote anti-inflammatory responses of M2φ subtype and tissue repairing (9).

However, insufficient function and short lifespan of neutrophils often occurred on those patients with severe congenital neutropenia (SCN) (68) or chronic granulomatous disease (CGD) (69), thus leading to serious pathogen infection. Scientists observed the effects of MSC-EVs on the lifespan and biological behaviors of neutrophils from healthy donors, SCN or CGD patients in vitro systematically (10, 31, 32, 70). In their studies, the apoptosis and biological behaviors like respiratory burst and phagocytosis of neutrophils were characterized by annexin V-propidium iodide, nitro blue tetrazolium assay, and Giemsa staining. MSCs were isolated from Wharton’s jelly (WJ, a mucosal connective tissue of UC) or from hAD. hAD-MSC-EVs could enhance the phagocytosis and ROS production in neutrophils from both CGD patients and healthy volunteers, and decrease the lifespan of neutrophils from CGD patients (70). However, the influence of hAD-MSC-EVs on neutrophils from SCN patients seemed different. In their observation, hAD-MSC-EVs recovered and prolonged the respiratory burst and lifespan of neutrophils from SCN patients or healthy volunteers significantly, while showed limited influence on the phagocytosis percentage of neutrophils from both SCN patients and healthy volunteers (32). For EVs derived from WJ-MSCs (WJ-MSC-EVs), the lifespan and phagocytosis of neutrophils from healthy volunteers were significantly augmented in comparison with their influence on respiratory burst (31). In-depth studies from aspects of genomics and proteomics should be conducted to explain why those two MSC-EVs showed different regulation behaviors on neutrophils from SCN or CGD patients, or from healthy donors. Very recently, Loh et al. reported that EVs from human embryonic stem cell (hESC, cell line of E1-MYC 16.3)-derived MSCs (hESC-MSC-EVs) inhibited terminal complement activation complex C5b-9-mediated neutrophil activation, thus suppressing the release of NETs and IL-17 via a CD59-dependent mechanism (33). This study revealed bright application prospect of hESC-MSC-EVs on treating immune dysregulation in COVID-19 patients.

Mast cells are the first responders of long-lived innate immune cells, that release heparin as well as histamine rapidly in response to an external infection (1). Excessive accumulation and activation of mast cells by immunoglobulin E (IgE) lead to allergy, interstitial cystitis, and other inflammatory diseases (71). Mast cells can be stabilized by MSC-EVs through different molecular mechanisms. Cho et al. observed that treatment with hAD-MSC-EVs (injected either intravenously or subcutaneously) ameliorated the infiltration of mast cells in atopic dermatitis in vivo by reducing the level of serum IgE (34). Liu et al. reported that hBM-MSC-EVs suppressed the activation of mast cells (cell line of LAD2) through upregulating the production of prostaglandin E2 (PGE2) and E-prostanoid 4 (EP4) receptors (35). EVs derived from hPT-MSCs (hPT-MSC-EVs) could attenuate TLR7-mediated activation of mast cells. In this study, imiquimod (IMQ, the agonist for TLR7) was employed to activate human mast cell line (HMC-1) (13). The introduction of hPT-MSC-EVs inhibited IMQ-induced HMC-1 activation and the expression of inflammatory cytokines in HMC-1 cells via transferring miRNAs like has-miR-214-3p and has-miR-424-5p. Very recently, a study launched by Lin et al. reported that hUC-MSC-EVs suppressed the activation of IgE-stimulated mast cells (cell line of KU812) and downregulated the expression level of NF-κB, thus inhibiting the degranulation of mast cells and release of IL-1β, TNF-α, and IL-6 (36). Furthermore, hUC-MSC-EVs attenuated IgE-induced STAT5 phosphorylation inside KU812 cells in a dose-dependent manner. Collectively, MSC-EVs can stabilize mast cells to relieve allergies by reducing their infiltration and degranulation through different mechanisms.

NK cells are lymphocytes that can detect and kill neighboring infected cells that don’t express a certain number of MHC molecules on cell surface through the specialized receptors on NK cells (NKG2D, KIR, etc.) without prior sensitization (1). MSC-EVs can exert their modulative function on NK cells by regulating their behaviors such as proliferation, activation, and releasing cytotoxic substances. hUC-MSC-EVs were found to relieve the renal ischemic reperfusion injury (IRI) by decreasing the number of NK cells at injury site. hUC-MSC-EVs downregulated the expression of C-X3-C motif chemokine ligand-1 (CX3CL1) and TLR-2, thus inhibiting the infiltration of CD3-CD161+NK cells (37). In another study, hBM-MSC-EVs were also reported to inhibit the secretion of IFN-γ and TNF-α by activating NK cells, showing potential in treating therapy-refractory graft-versus-host diseases (72). EVs derived from human FL (hFL)-MSCs (hFL-MSC-EVs) showed efficient inhibition on the proliferation, activation, and cytotoxicity of NK cells by transferring thrombospondin 1 (TSP1, a regulatory molecule for TGF-β) to downregulate TGF-β/Smad2/3 signaling pathway in NK cells (14). In short, these reports showcased the therapeutic roles of MSC-EVs in inhibiting the lethality of NK cells.

DCs are recognized as the most efficient and professional APCs. They ingest antigens by internalizing invaders and process antigens, followed by presenting antigens to T cells (1). During this process, the antigen-processing immature DCs (iDCs) are transformed to mature DCs (mDCs, antigen-presenting cells) and migrate to secondary lymphoid organs, activating adaptive immune system. Therefore, the immunomodulation on DCs can be achieved by regulating their maturation and migration behaviors. HBM-MSC-EVs induced the hypoactive phenotype of DCs with repressed allorecognition and downregulated expression of costimulatory molecules and MHC-II, subsequently inhibiting the development of Th1 and Th17 cells in the in-vivo mouse models of type 1 diabetes and experimental autoimmune uveoretinitis (73). In another study, hBM-MSC-EVs were observed to induce the generation of iDCs, which were characterized by the reduced expression of IL-6 and IL-10 (74). However, the regulated expression of costimulatory markers and MHC-II seemed different in the researches of mAD-MSC-EVs. Cho et al. observed that mAD-MSC-EVs induced the maturation of DCs, characterized by the increased expression of co-stimulatory molecules (38). While, Shahir et al. claimed that the treatment of immature or LPS-induced mature DCs with mAD-MSC-EVs led to the tolerogenic DC population with downregulated expression of costimulatory markers (39). As for underlying molecular mechanisms, several key miRNAs or proteins in MSC-EVs may be involved in the regulation of DC behavior. MBM-MSC-EVs were found to impair the maturation of iDCs and the ability of mDCs to produce IL-12 by transferring miR-146a (the potential miRNA controlling the survival and maturation of human DCs) to iDCs (40). MiR-21-5p was another miRNA that regulated the maturation and function of DCs (41). HBM-MSC-EVs enriched with miR-21-5p degraded the C-C chemokine receptor type 7 gene (CCR7 gene, modulating the homing of DCs to the lymph nodes) and hampered the migration toward the CCR7-ligand CCL21. Furthermore, the treatment with hBM-MSC-EVs restricted the antigen uptake capacity of iDCs and downregulated the secretion of IL-6 and IL-12p70 as well as upregulated the secretion of TGF-β. In another study, the upregulated miR-540-3p and immunoregulatory four-and-a-half LIM domain protein 1 (FHL-1) in EVs derived from indoleamine 2,3-dioxygenase (IDO)-pretreated BM-MSCs were reported to regulate Janus kinase 3 (JAK3, an immune activator) protein negatively and inhibit activation of AKT, respectively, inhibiting the maturation and functions of DCs (42).

T cells showcase multi-biofunctions including directly killing target cells, regulating antibody production of B cells and secreting lymphokines after the adaptive immunity system is activated (1). Generally, T cells are classified into two subpopulations based on the CD4 and CD8 receptors expressed on their surfaces, called helper and killer T cells respectively. Furthermore, in the presence of IL-12 and IFN-γ, CD4+ T cells are activated into Th1 subtype to induce inflammation and kill pathogens. In the presence of IL-4, CD4+ T cells are activated into Th2 subtype to support the antibody production of B cells (1).

According to literatures, MSC-EVs exerted their immunoregulation functions on T cells with the assistance of APCs. Zhang et al. found that EVs derived from hESC (cell line of huES9.E1)-MSCs (hESC-MSC-EVs) mediated the polarization of Tregs from CD4+ T cells with the assistance of monocytes (75). In their observation, the differentiation of CD4+CD25+FoxP3+ Tregs was observed distinctly by co-incubating hESC-MSC-EVs with CD4+ T cells and human macrophages (THP-1) for 24 h. Furthermore, they found that hESC-MSC-EVs increased the generation of CD4+CD25+Foxp3+ Tregs from CD4+ T cells (activated by APC-enriched spleen cells) through a dose–dependent manner, indicating that MSC-EVs enhanced the production of Tregs through an APC-mediated pathway (76).

A lot of active proteins including tumor necrosis factor-α-stimulated gene/protein 6 (TSG-6) (46), TGF-β (44, 45), adenosine (45, 48), CD73 (48, 49), programmed death ligand-1 (PD-L1) (5), granulocyte-macrophage colony stimulating factor (GM-CSF) (47), and β-catenin (50) loaded in or expressed on the surface of MSC-EVs were involved in the modulation on T cells. As for proteins, TSG-6 in cAD-MSC-EVs was the key protein to increase Tregs by upregulating forkhead box P3 (FOXP3) protein (stabilizes precursor cells of Tregs) (46). TGF-β displayed on the surface of BM-MSC-EVs decreased the number of Th17 cells and increased FoxP3+ Tregs in GAD65-stimulated peripheral blood mononuclear cells (PBMCs) (45). Additionally, TGF-β surface‐bounded or encapsulated in hEND-MSC-EVs exhibited significant inhibition on the activation of CD4+ T cells (44). Mokarizadeh et al. claimed that PD-L1 and TGF-β in mBM-MSC-EVs were responsible for suppressing the activation and proliferation of CD4+ T cells and promoting the generation of Tregs (5). Adenosine and adenosinergic signaling are efficient immunosuppressor and pathway employed by immunosuppressive Tregs by neutralizing pro-inflammatory adenosine 5′-triphosphate (ATP) in the extracellular environment, especially in injured tissues (77). CD73 expressed on the surface of hUC-MSC-EVs (48) or hBM-MSC-EVs (49) catalyzed the production of adenosine from adenosine 5′-monophosphate (AMP), suppressing the proliferation of T cells or promoting the apoptosis of CD39+ Th1 cells. Additionally, Crain et al. claimed that cWJ-MSC-EVs inhibited the mitogen-induced proliferation of CD4+ T cells in a dose-dependent manner with the synergistic effect between TGF-β and adenosine (43). GM-CSF (47) and β-catenin (50) inside MSC-EVs play an important role in immunoregulating T cells associated with tumor therapies. A prophylactic anticancer vaccine composed of GM-CSF-overexpressing mESC-MSC-EVs were reported to enhance the migration of CD8+ T effector cells to rise the expression of pro-inflammatory TNF-α and IFN-γ and restrict the migration of immunosuppressive Tregs (47). Besides, β-catenin inside mAD-MSC-EVs showed significant promotion effect on the migration and circulation of natural killer T cells (50). Furthermore, the stimulus of pro-inflammatory TNF-α and IFN-γ on cAD-MSCs increased the expression of immunosuppressive proteins such as TSG-6, PGE2, and TGF-β inside the derived EVs (28).

Abundant miRNAs inside MSC-EVs were also involved in the regulation process. MiR-125a-3p, which suppresses the proliferation of several cells (78), is the most highly upregulated miRNA inside hBM-MSC-EVs and is recognized to account for the suppression effect of hBM-MSC-EVs on the functional differentiation of T cells (51). MiR-540-3p is another involved miRNA responsible for the regulatory functions of MSC-EVs. He et al. engineered mBM-MSCs by gene transfection to get IDO1-overexpressing mBM-MSC-EVs with upregulated miR-540-3p and FHL-1 protein (42). MiR-540-3p could regulate JAK3, an immune activator, negatively. While, FHL-1 protein was found to suppress IGF/PI3K signaling and activate endoplasmic reticulum (ER) signaling. They cooperated with each other to mediate immunotolerance associated with APCs and T cells after organ transplantation (42).

The roles of T cells and the cooperation between T cells and other immune cells are elaborate and complex. The regulations of the proliferation, migration, activation, apoptosis, and homeostasis of T cells by MSC-EVs should be carefully considered.

Upon activation and proliferation, B cells can produce large quantities of highly-specific antibodies and secrete them into blood or tissue fluid. Analogously, B cells also include two subtypes, B1 cells (primary producer of natural antibodies, like immunoglobulin M (IgM)) and B2 cells. Furthermore, B2 cells have two subsets including marginal zone B (MZB) cells and follicular B (FOB) cells, which participate in innate and adaptive immune responses, respectively (1).

Controversial results exist about how MSC-EVs regulate B cells. Budoni et al. explored the role of BM-MSC secreted membrane vesicles (MVs, BM-MSC-MVs) in the inhibition of B cells (52). They observed that BM-MSC-MVs specifically inhibited proliferation and differentiation of B cells and suppressed IgM secretion in a dose-dependent manner, compared with T lymphocytes or NK cells. In another study, BM-MSC-EVs were found to decrease the proliferation of B cells in vitro, which might be attributed to the upregulated expression level of marginal zone B1 (MZB1) and B cell receptor (BCR)-mediated Ca mobilization in certain subsets of B-lymphocytes (53, 79). Additionally, BM-MSC-EVs promoted migration and chemoresistance of chronic lymphocytic leukemia (CLL) B cells, decreasing their apoptosis in a contact-independent manner by inducing BCR-like activation (80). However, a study launched in 2019 claimed that AD-MSC-EVs isolated by size-exclusion chromatography showed minimal effects on activated B cells, compared with the effects of AD-MSCs, AD-MSC-conditioned medium and AD-MSC derived soluble protein-enriched fractions. AD-MSC-EVs just induced the production of similar CD24^hiCD38^hi B cells, but not real ones because these cells could not produce IL-10 (81). These controversial results may be attributed to different isolation techniques and origins for MSC-EVs in literatures. A standard isolation technology and culturing approach should be established to explore the impacts of MSC-EVs on B cells.