Cancer stem cells (CSCs), also known as tumor initiating (propagating) cells, despite constituting a relatively small population of the tumor cells, play a pivotal role in fueling tumor growth due to their self-renewal and multi-lineage differentiation ability (1–4). This concept, introduced several decades ago, has stimulated extensive studies aimed at decoding the clinical observations through CSCs (5–9). Patients who show initial partial or complete remission through anti-cancer treatments by chemotherapy or radiation may eventually experience tumor relapse or metastasis, conditions that tend to be markedly intractable owing to the increased resistance to therapies (10, 11). This phenomenon is likely attributed to CSCs’ resilience against the therapeutic regimens. Besides, CSCs also contribute to intratumoral heterogeneity (ITH) by differentiating to all kinds of cancer cells with different phenotypic features, some of which can disseminate to other parts of the body (12–14).

CSCs intricately interact with cancer supporting cells especially immune cells in the tumor ecosystem, sculpting a conducive niche and employing mechanisms for immune evasion and immunosuppression, including activating immune escape pathways and suppressing antigen processing and presentation proteins (15). CSCs also overexpress programmed cell death 1 ligand 1 (PD-L1) on the cell surface, an incredibly important immune checkpoint protein that counteract the antitumor immune response (16, 17). Unraveling how CSCs engage with cancer-associated immune cells is gaining increasing attention, as these interactions hold considerable potential as immunotherapeutic targets.

Extracellular vesicles (EVs) are small lipid bilayered membrane vesicles composing of two main subgroups, exosomes and microvesicles (MVs) (18–20). These vesicles range in size from a few tens of nanometers to multiple micrometers (21, 22). Secreted by all cell types, EVs are responsible for the transfer of functional biological cargos including nucleic acids, proteins, and lipids between cells, mediating critical intercellular communication (23, 24). EVs are irreplaceably involved in modulating various innate and adaptive immune processes including antigen presentation, activation of T cells and B cells (25–27). Within the tumor ecosystem, EV-mediated communication is preferentially characterized by EVs produced by CSCs being internalized by other cells, a process integral for the dissemination of CSC-specific traits, which is essential for shaping the tumor immune microenvironments. For example, pancreatic CSC-derived EVs carry agrin protein to increase YAP activation to promote tumor cell proliferation (28). EVs from colorectal CSC could transfer metastatic properties to the non-CSCs via miR-200c (29). CSC-derived EVs are likely to interact with immune cells enriched in the CSC niche, including MHC-II expressing macrophages and programmed cell death 1 (PD1) positive T cells, potentially promoting tumor progression through immunosuppression (30). Conversely, immune cell-derived EVs enhance CSC stemness and propagation (31–33). The reciprocal transfer of EVs between CSCs and immune cells operate in concert to create a tumor-supporting niche.

In this review, we will summarize and discuss the recent findings on the biological functions of CSC-derived EVs, with a focus on their immunological role through engagement with various types of tumor-associated immune cells, and how EV-mediated interactions between CSCs and immune cells contribute to shaping the tumor immune microenvironments.

ITH is a major obstacle in cancer treatment, leading to aggressive tumor growth and treatment failures (34–36). Elucidating the sources of ITH is of great significance to overcome therapy resistance. ITH present in two forms, spatial heterogeneity, which involves the unequal distribution of tumor subpopulations across different regions, and temporal heterogeneity, which refers to genetic diversity within a tumor over time (34, 37). Technological advances such as multiregional sampling, liquid biopsy, single-cell sequencing, and spatial transcriptome provide insights into decoding the intricate compositions of ITH (38–40). Single-cell transcriptomic analysis of various types of tumors have uncovered distinct subpopulations of tumor cells, each characterized by varying levels of differentiation and stemness (41–44). Of note, various driving forces are emerging to be responsible for ITH, encompassing genome instability and clonal evolution (45, 46), metabolic adaptation (47), epithelial-mesenchymal transition (EMT) (48) as well as environmental factors such as hypoxia (49) and inflammation (50). Among these, the existence of CSCs plays a pivotal role in the development and maintenance of ITH (13, 51).

Here comes the concept of CSCs, a subset of tumor cells capable of self-renewal, tumor-initiating, multi-lineage differentiation, therapy-resistance, metastasis, relapse, etc. (7, 52–54) ( Figure 1 ). The CSC theory posits that tumors are hierarchically structured, with CSCs at the top (7, 55–57). Nevertheless, recent findings (2, 13, 58) have revealed that CSCs and non-CSCs are dynamic and can transition between states in response to specific stimuli, complicating tumor eradication. Initially identified in hematological malignancies, CSCs are now recognized in various solid cancers (59–62). Three hypotheses explain CSC origins (63, 64): transformation of non-cancerous stem cells through a series of oncogenic mutations (65), acquisition of pluripotency by progenitor cells (66), and dedifferentiation of differentiated cells (67).

Certain surface markers such as CD133, CD44, epithelial cell adhesion molecule (EPCAM), and intracellular markers such as aldehyde dehydrogenase (ALDH) have been identified for CSCs (68–72). ALDH is an enzyme mediating aldehyde detoxification, which is assessed by the ALDEFLUOR assay, is instrumental in drug resistance (73–75). However, these markers are not solely specific to CSCs, and some CSCs don’t exhibit these markers at all (9, 76, 77). Moreover, CSCs typically constitute less than 1% of the tumor mass, and such scarcity further complicates their isolation and identification (77).

Besides markers, the characterization of CSCs requires surrogate functional assays (78–80). Current well-known surrogate methods include in vitro tumorsphere formation and in vivo limiting-dilution tumorigenicity assays (81–83). In vitro tumorsphere formation assay evaluates the ability of cells to grow and form spheres in a three-dimensional, anchorage-independent culture environment (84, 85). In vivo limiting-dilution tumorigenicity assay, which tests the tumor-initiating capacity of cells by transplanting them into immunocompromised mice and observing tumor formation, is considered the gold standard for CSC research (86, 87).

CSCs contribute to therapy resistance through several mechanisms (88–90), including high levels of multi-drug efflux ATP-binding cassette transporters, slow-cycling state, enhanced DNA repair capacity, apoptosis evasion, immune-privileged property, etc. Furthermore, EMT activation is tightly associated with the formation of CSCs (91). Consequently, CSCs consist of heterogenous subtypes occupying different locations and exhibiting varied EMT characteristics within the primary tumor (92, 93), further complicating their therapy resistance.

Additionally, it is important to emphasize that CSCs have been found to actively reshape the tumor microenvironment into immunosuppressive state, facilitating their own growth and proliferation while evading the therapeutic elimination (15, 94, 95). This interaction is significantly mediated by EVs, powerful cell-cell communicators that plays a crucial role in modulating the immune microenvironment (96). Through EVs, CSCs transferring a diverse array of biological cargos to surrounding or distant immune cells, eliciting immunosuppressive responses that protects CSCs and fosters tumor progression (28).

EVs are lipid-bilayer membrane structures secreted by virtually all living cells, encompassing two main subtypes, exosomes and MVs, whose sizes ranging from about 50 nm to 5 µm (97, 98) ( Figure 2 ). EVs contain cellular bioactive components like proteins, lipids, metabolites, and nucleic acids, reflecting their cell of origin and functioning as mediators of intercellular communication (99, 100). EVs perform multifaceted functions including waste disposal, signal cargo delivery to alter recipient cell physiology, and mediating interactions between cells and extracellular matrix (101–105). Furthermore, they can also facilitate long-distance communication via blood or lymph (106, 107). Separating different EV subtypes is challenging due to overlapping properties, with methods like differential centrifugation, size-exclusion chromatography, and immunoprecipitation used in combination to improve specificity (108).

Exosomes stand out as the most well-studied subtype of EVs with a size ranging between 40–160nm, primarily due to their unique biogenesis, small size, and specific molecular content (109). Exosomes play a pivotal role in cellular communication and cancer research. Originating from endosomal compartments within cells, they carry an array of biomolecules which reflect their cellular origin and can influence recipient cell behavior. This makes them key players in tumor progression and ideal for targeted drug delivery and biomarker discovery in cancer diagnostics and therapeutics (110–112). Their small size, specific content, and stability in bodily fluids enhance their potential, making them promising candidates in the medical and scientific exploration of cancer.

Contrasting with exosomes, which originate from endosomal pathways, MVs are formed by budding directly off the plasma membrane and are generally larger, with sizes spanning from 200–5000 nm (97, 113). MVs refer to a diverse group of membrane-derived vesicles, including microparticles, oncosomes, or ectosomes, and is less characterized EV subtype whose cargo trafficking mechanisms are still under investigation (114). Besides functional cargos as RNAs, proteins, lipids, and metabolites, MVs could also mediate the transfer of mitochondria between cells (115), which boosts ATP production in the recipient cells. Studies have shown that cancer cell-derived MVs participate in tumor progression (116, 117), while MVs from radiation treated cancer cells exert antitumor effect through immunogenic death pathway (118). These findings suggest a promising functional role of MVs in cancer therapeutics.

The TME is a complex and dynamic battlefield consisting of cancer cells, CSCs and cancer supporting cells, serving as a critical zone for the interplay between immune cells and CSCs ( Figure 3A ). This environment is a hub where diverse immune cells from both innate and adaptive immune systems converge (151, 152). Dendritic cells (DCs) are key in antigen presentation and the initiation of immune responses, while macrophages, also involved in antigen presentation, display ambiguous effects on tumors by either fostering or inhibiting their growth. MDSCs predominantly suppress immune activity, facilitating tumor immune evasion. Natural killer (NK) cells, adept at autonomously destroying cancer cells, and neutrophils, whose impact on cancer can vary from hindering to promoting tumor progression. In terms of the T cell population, subsets including cytotoxic T lymphocytes (CTLs) are directly responsible for recognizing and eradicating cancer cells, highlighting their critical role in antitumor immunity. T helper 17 (Th17) cells, a subset of CD4+ helper T cells, known for their secretion of interleukin-17 (IL-17), exhibit a dual role in cancer by either promoting inflammation that can support tumor growth or recruiting effector T cells, NK cells and DCs into TME that enhance antitumor responses.

Mounting evidence have shown that EVs released by cancer cells, especially CSCs, are capable in modulating both innate and adaptive immune responses, thereby facilitating to establish pro-tumorigenic and pro-metastatic immune niches through their interactions with various immune cell types within TME (153–156). Understanding the diverse functions and interactions of these immune cells with CSCs through EVs provides insight into the complex nature of cancer and opens avenues for innovative therapeutic strategies ( Figure 3B , Table 3 ).

EVs are favored for therapeutic delivery due to their superior biocompatibility and ability to penetrate biological barriers (191, 192). These years, EVs have been developed as targeted delivery systems to disrupt CSC functions by transporting RNA-based therapeutics into CSCs. EVs can be engineered to carry siRNAs that target key signaling pathways such as Wnt/β-catenin in liver cancer, leading to the suppression of CSC proliferation (193). In non-small cell lung cancer, EVs delivering APE1 shRNA have demonstrated potential in overcoming drug resistance (127). Furthermore, EVs designed to silence genes can reduce resistance to sorafenib treatments in liver cancer, promising to enhance patients’ clinical outcomes (194).

Notably, researchers have been making persistent efforts to develop engineered EVs as immunotherapeutic strategies to combat immunosuppressive state fueling by cancer cells and immune cells (159, 195–198). Xu et al. has developed a bispecific EVs (BsEVs) engineered from DCs that target tumor antigen CD19 on tumor cells and block the PD-1 checkpoint, thus bolstering cancer immunotherapy (197). These BsEVs show remarkable tumor-homing capabilities and can substantially remodel the tumor’s immune landscape, demonstrating their potential in personalized and versatile cancer treatments. Innovatively, chimeric exosomes, produced by M1 macrophage-tumor hybrid cells, naturally targeting to lymph nodes and tumors, enhancing T cell response, and overcoming immunosuppression. This dual-action immunostimulatory exosome strategy can alleviate tumors and enhance survival in animal studies and shows potential in personalized immunotherapy, especially when used with PD-1 inhibitors (199). Moreover, a dual-functional exosome delivery system that employs bone marrow mesenchymal stem cell-derived exosomes loading with galectin-9 siRNA and the immunogenic cell death trigger oxaliplatin, promises to enhance immunotherapy. This system aims to achieve win-win idea of both counteracting the immunosuppressive actions of M2 macrophages and simultaneously improving tumor targeting (200).

However, the immunotherapeutic approaches using EVs to target CSCs is still in its early stage and requires further in-depth and comprehensive research. Naseri et al. established a DC-based therapeutic strategy using colon CSC-derived exosomes as antigen sources, aiming to increase proliferation and activation of T cells specifically for killing CSCs (159). Besides leverage EVs for support, disrupting the interactions of CSC-derived EVs with macrophages emerges as a potential therapeutic choice. Colon CSC-derived exosomes, containing molecules like IL-6, p-STAT3, TGF-β1, and beta-catenin, are known to promote the generation of cancer-associated fibroblasts and M2 macrophages. Ovatodiolide, a bioactive compound, has been found to reduce these harmful components in exosomes, consequently weakening chemotherapy resistance (201). This suggest that ovatodiolide could serve as an effective agent against colon cancer through disrupting the exosomal supply CSCs provide to immunosuppressive cells.

Future directions for targeting CSCs and the tumor immune microenvironment are pointed to be focused on integrating EVs with cutting-edge cancer therapeutic strategies, such as differentiation therapy and synthetic lethality, aiming to provide more effective and precise cancer treatments. Differentiation therapy is an innovative approach that exploits the plasticity of CSCs by inducing them to differentiate into less malignant, more differentiated cells, making them more susceptible to cytotoxic drugs (202, 203). Acting as biocompatible natural carriers, EVs are promising to be engineered to deliver differentiation-inducing and immune activating molecules to target CSCs and immune microenvironment, thereby synergistically eliminating refractory CSC pool. Synthetic lethality (SL) is defined as the simultaneous inactivation of two genes lead to cell death, whereas the loss of either gene alone is not lethal (204, 205). A prime example is the use of PARP inhibitors in combination with BRCA1/2 mutations, which results in the targeted death of cancer cells (206). Multifunctional engineered EVs present a promising delivery mechanism for SL-based therapy, as co-delivering multiple therapeutic agents that target separate pathways essential CSC survival and immune responding within the same EVs can enhance the efficacy of SL approaches. To conclude, EV-based therapies offer a versatile and targeted approach to treating cancer by addressing both CSCs and the tumor immune microenvironment.

CSCs have been recognized as crucial targets in oncology due to their irreplaceable roles in tumor growth, metastasis, and the potential for developing more effective cancer therapies by disrupting CSC-specific pathways. The interaction between CSCs and immune cells significantly influences cancer progression, offering therapeutic avenues to modify the immune environment and exploit immune cells for CSC eradication. EVs, as masters of intercellular communication, not only shed light on the complex dynamics of cell interactions but also offer platforms for bioengineering as cutting-edge immunotherapeutic tools, harnessing their natural communication ability to modulate immune responses and precisely combat against CSCs. Specifically, compared with EVs derived from non-malignant stem cell such as mesenchymal stem cells which are known for regenerative and anti-inflammatory properties (207, 208), CSC-derived EVs possess superior cancer therapeutic capacities due to their inherent tumor-targeting specificity, tumor immune modulation activity, and higher uptake efficiency by cancer cells (209). These characteristics makes CSC-derived EVs exceptionally advantageous as potential mediators for cancer therapeutic payload delivery.

In this review, we have provided current research on EV-mediated communications of CSCs with individual subtype of immune cells in a wide spectrum of cancers. In terms of clinical translation, it is noteworthy that preclinical investigations into ovatodiolide have highlighted its potential as a disruptor of the deleterious feedback loop between CSCs and immunosuppressive macrophages, heralding an augmentation in the therapeutic efficacy for patients with colon cancer undergoing chemotherapy.

Despite the advancements discussed in our review, it is critical to emphasize that our understanding of the interactions between CSCs, EVs, and immune cells is still in its infancy. Especially understudied are the roles of NK cells, Th17 cells, and B cells in this tripartite communication, which is surprising given their crucial roles in the immune response to cancer. Take NK cells for example, as cytotoxic lymphocytes in the innate immune system, they possess the ability to eliminate cells infected by viruses or cancer cells (210). NK cells are powerful in cancer immunotherapy because of being able to swiftly attack cancer cells, thereby boosting both the immediate and long-term immune defense against tumors (211, 212). Previous studies (213–216) have found that while CSCs with reduced MHC-I expression and certain CSC markers can activate NK cells’ cytotoxic functions, leading to their effective elimination in various cancers, CSCs also employ numerous mechanisms to suppress NK cell-mediated immune responses, such as downregulating activation ligands or entering a dormant state to evade detection. This intricate dynamic between NK cells and CSCs suggests a potential role for EVs in mediating their interactions. Recognizing this, we advocate for a concerted effort to deepen the investigation into CSC-EV-immune cell interactions. Furthermore, the field of CSC research necessitates to advance precise isolation techniques that consider the heterogeneity of CSCs, including the development of comprehensive identification strategies including refined cell surface markers. Timely initiation of preclinical and clinical trials, grounded in laboratory findings, is imperative to substantiate the therapeutic efficacy and expedite the translation of research into improving patient survival and quality of life.

YJ: Writing – original draft, Writing – review & editing. CZ: Writing – review & editing. WY: Conceptualization, Supervision, Writing – review & editing. XL: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – review & editing.