EVs are one kind of nanometer-sized spherical hollow structures that can carry bioactive molecules and deliver them to recipient cells (8). Classic EVs can be broadly classified into exosomes, microvesicles, and apoptotic bodies. Exosomes, with a diameter of 50–100 nm, can be released by resting or stimulated cells and can transfer mRNA, miRNA, and oncogenic receptors, exhibiting antigen presentation, immune activation, and immune suppression activities (9) (10). Microvesicles, measuring 100–1000 nm in diameter, are produced by platelets (11), red blood cells (12), or epithelial cells (13) through outward budding of cell membranes and have procoagulant functions. Apoptotic bodies, with a diameter of 1–5 micrometers, are generated during cell apoptosis and can transfer DNA and oncogenes, presenting T-cell epitopes to immune cells when ingested by phagocytes, thereby exerting immune suppressive effects (8).

Although most cells possess the ability to produce extracellular vesicles (EVs), not all EVs derived from cells can be used as carriers for drug delivery. The standards for drug delivery included the production yield, surface protein properties, size, and the composition in the vesicles. Currently, several cell types have been explored as potential donor sources for EVs used in drug delivery, such as dendritic cells (DC) (14), macrophage (15), tumor cells (9), red blood cells et al. (16) (12).

Tumor cell-derived EVs (TEVs), especially autologous TEVs, carry a similar repertoire of tumor antigens, co-stimulatory molecules, and DNA fragments as their parental cells (17, 18). This property can elicit robust T-cell-dependent anti-tumor immune responses and has demonstrated therapeutic effects in melanoma mouse models (19), hepatocellular carcinoma (20), and colon cancer (21). Compared to EVs produced by non-cancerous cells, TEVs can achieve tumor cell-specific targeting through intrinsic homotypic adhesion mediated by membrane surface antigens (22). In tumor therapy, TEVs play a significant role, such as enabling deep tumor penetration for drug delivery (23), exhibiting high specific homing capabilities, and activating the signal transducer and activator of transcription 3 pathway (24). Additionally, the in situ generation of micron-sized tumor cell-derived vesicles serves as an autologous tumor vaccine to enhance systemic immune responses (25), and functional DNA-modified cancer cell membrane vesicles are used as targeted vaccines for tumor immunotherapy (26). However, the role of TEVs in promoting cancer progression by enhancing cell proliferation and evading apoptosis, inducing angiogenesis, metabolic reprogramming, enhancing invasion and metastasis, and evading immune surveillance has been well-documented (27). Therefore, unlike exosomes from other sources, TEVs can be a double-edged sword when used as therapeutic agents in cancer treatment. A thorough elucidation of their formation, secretion, and network functions is urgently needed to realize this attractive and promising cancer treatment strategy, requiring more extensive in vivo studies with larger sample sizes to investigate the effectiveness and safety of TEVs as future drug delivery systems (DDS) (28).

Dendritic cells (DCs) are fundamental immune cells essential for antigen presentation and T cell activation. DC-derived vesicles (DEVs) maintain the basic immune stimulating ability of DCs (e.g., antigen presentation to T cells) (29). DEVs production processes are amenable to strict regulation and monitoring (e.g., easy determination of their composition and MHC-I and MHC-II contents) and pose lower risks associated with feasible cell or viral therapies (e.g., in vivo replication risks) (30, 31). In recent decades, DEV-based therapies have been widely applied in immunotherapy and drug delivery. For instance, in breast cancer treatment, DEVs can enhance cancer cell sensitivity to immune checkpoint inhibitors and prevent recurrence of resected tumors (15). DEVs can overcome biological barriers like the blood-brain barrier (BBB), making them attractive for future drug delivery (32).

Macrophage-derived EVs express functional immune regulatory proteins including MHC class I and II (33), preferentially inducing Th1-type (cell-mediated) immune responses that direct T cells to attack abnormal cells (e.g. cancer cells) or cells infected by intracellular parasites (34, 35). Macrophage-derived EVs also have extensive applications in tumor treatment. For example, macrophage-derived exosomes are thought to transfer miR-365, a key regulator of gemcitabine resistance in pancreatic cancer (36). M1-like macrophage-derived EVs (M1 EVs) are used to treat glioblastoma multiforme (37). In photodynamic therapy (PDT), fusion of M1 EVs with thylakoid membranes of natural plants imparts active tumor targeting ability to M1 EVs. Therefore, macrophage-derived EVs offer promising new strategies for tumor treatment.

Besides these cells, there are other candidates for drug delivery vesicles, such as those derived from red blood cells (RBCs) (38, 39), natural killer (NK) cells (40) and T cells (41). The CD47 on RBC-derived EVs interacts with its receptor, signal regulatory protein alpha (SIRPa) on macrophages, protecting the RBC-derived EVs from clearance by initiating a “don’t eat me” signal (42). NK cell-derived EVs contain tumor necrosis factor-α and granzyme B, exhibiting cytotoxic effects on glioma cells (40) and melanoma- cells (43) with no significant side effects in vitro and in vivo. Furthermore, studies have found that activated CD8⁺ T cells from healthy mice release cytotoxic EVs, leading to a significant reduction in tumor invasion and metastasis (44). EVs derived from CD4⁺ T cells enhance the anti-tumor response of CD8⁺ T cells by augmenting their proliferation and activity without affecting regulatory T cells.

Among various sources of EVs, RBCs exhibit distinct advantages in safety and scalable production due to their relatively low content of cellular components and ease of procurement (45). TEVs, while amenable to scalable production through in vitro expansion of tumor cells, possess surface markers enriched with tumor-specific antigens that may enhance tumor-targeting efficacy (23). However, their content (proteins, nucleic acids, etc.) may carry oncogenic risks, necessitating further improvements in biosafety (46). Immune cell-derived EVs (e.g., DCs, macrophages, or NK cells) retain functional biological properties inherent to their parent cells, enabling tailored therapeutic applications. Nevertheless, the scalability of immune cell-derived EVs remains constrained due to stringent ex vivo expansion requirements and high costs associated with isolating immune cells from biological systems (47). These factors collectively highlight the need to balance source-specific advantages with technical and safety considerations for clinical translation.

Peptide vaccines can be divided into two categories based on their activation functions, one group stimulates the innate immune system by interacting with tumor-associated macrophages (TAM), dendritic cells (DC), neutrophils, and natural killer (NK) cells, while the other group can activate the adaptive immune system by interacting with T cells and B cells (48–50).

For tumor-associated macrophages (TAM) in the innate immune system, TAMs exhibit two phenotypic activation states: the antitumor M1 and the protumor M2 (51–55). Currently, the main strategies to block M2-TAM activity involve inhibiting the recruitment of macrophages to tumors and converting M2-TAMs to M1-TAM. For example, researchers developed a biohybrid material with the ability to immunologically regulate TAM cell populations, using vascular endothelial growth factor (VEGF) mRNA interference-M2 targeting peptide. This material primarily blocks M2-TAM activity and cancer cell growth by inhibiting VEGF-related signaling pathways and triggering host immune responses that lead to sustained tumor regression, and it can also generate long-lasting antitumor immune memory (56).

As for DCs, Wang et al. selected the TRP2 peptide and the dodecamer CPP (AAVLLPVLLAAP) to prolong the presentation of MHC class I-restricted self-peptides on dendritic cells (DCs), thereby enhancing antitumor immune responses. CPP1 can effectively deliver the TRP2 peptide into mature DCs and retain the full capacity of DCs to present MHC-peptide complexes to antigen-specific T cells over an extended period. They demonstrated that immunizing mice with DCs loaded with TRP2-CPP1 conjugate led to complete protection against B16 tumor suppression, and lung metastasis inhibition (57).

The human body has three principal subtypes of mature T cells: cytotoxic T lymphocytes (CTLs), helper T cells, and regulatory T cells (Tregs).Immune checkpoint blockade is one of the major immunotherapies (58–64), which precisely targets tumor cells by blocking immune checkpoints such as CTLA-4 and PD-1. Researchers have discovered peptides that inhibit the PD-1/PD-L1 interaction and reactivate T cell function against tumor cells, including peptide-57, CLP001/CLP002, and PD-L1 Pep-1/PD-L1 Pep-2. These peptides not only reawaken T cells via their PD-L1 inhibiting activity but also utilize PD-L1 as a tumor target to deliver chemotherapeutic agents specifically to tumors exhibiting elevated PD-L1 expression.

Anti-cancer peptide(ACP) are bioactive peptides that inhibit cancer cell growth via a spectrum of mechanisms. Lytic peptides are toxic molecules that kill cancer cells by disrupting cell membrane. Recent studies revealed that the cell fragments by lytic peptides can act as tumor antigens to trigger the immune response. For example, lytic peptide EP-100 offers a unique therapeutic option for patients demonstrating insufficient responses to immunotherapy for ovarian cancer (65). EP-100 is a synthetic fusion peptide composed of an LHRH ligand and a lytic peptide (CLIP-71) that specifically binds the LHRH receptor (LHRH-R) (66). As an immune enhancer, it induces PD-L1 synthesis in neoplastic cells, therefore altering the tumor microenvironment. This leads to an augmentation of immune cells that facilitate tumor lysis (CD8⁺ T cells, NK cells, dendritic cells, and macrophages) while diminishing immunosuppressive cells (Tregs, B cells, and mMDSCs). Targeted ACPs can inhibit immune-related signal pathways to modulate immune response. For example, A new peptide-based PROTAC has been developed to combat the prevalent resistance to PD-1/PD-L1 inhibitors in clinical contexts by degrading PD-1 or PD-L1, thus inducing cancer cell apoptosis (67).

Peptide nanoparticles are widely acknowledged as an effective approach in cancer immunotherapy because of their exceptional stability and significant capability for delivering peptide antigens and immunological adjuvants (68, 69). Peptides and their derivatives can self-assemble into one-dimensional fibers or nanofibers, which can then interlace to form hydrogels or nanoparticles, enabling the targeted release of peptides and adjuvants (69). Collier et al. developed a vaccine utilizing the Q11 self-assembling domain (ac-qqkfqfqqfeqq-am) produced from chicken ovalbumin (OVA323-339) to incorporate MUC1-derived peptides (70). These immunizations stimulate the production of potent antibodies specifically targeting breast cancer cells. To improve the application of this technique in clinical therapy, Huang et al. created a novel synthetic self-adjuvant vaccine using a self-assembling Q11 domain (71). This vaccine can produce fibrous structures under mild conditions and display multivalent B-cell epitopes, thereby markedly enhancing their immunogenicity.

The principal techniques for constructing peptide-based vesicles encompass direct loading via phosphatidylation or click chemistry; surface modification of gene-edited cellular vesicles; and the administration of peptides to immune cells, followed by the preparation of vesicles from these cells to commence the antigen presentation process (72).

Zhu et al. chemically crosslinked a dibenzobicyclooctyne (DBCO) moiety to the surface of dendritic cell-derived extracellular vesicles (EVs) and subsequently reacted it with azide-functionalized MUC1 glycopeptide by click chemistry, thereby covalently affixing MUC1 to the EV surface (73). As for lipid insertion, Ye et al. introduced a noteworthy methodology (74). The 4F-KLA-LDL peptide was synthesized by combining the pro-apoptotic peptide KLA with an LDL-targeting peptide, which selectively binds to the overexpressed LDL receptors on blood-brain barrier (BBB) and glioblastoma (GBM) cell lines. Based on the molecular recognition between phospholipids on EV and ApoA-I mimetic peptides, They developed methotrexate (MTX)-loaded EVs functionalized with 4F-KLA-LDL peptide, which can target low-density lipoprotein (LDL) on GBM cells and enhance the transport of the pro-apoptotic peptide KLA and methotrexate (MTX) to U87 glioma cells. As for genetic manipulation, EVs are typically equipped with these peptides in EV donor cells using transfection or retroviral/lentiviral infection (75). For instance, Ohno et al. reported a technique involving the expression of a fusion protein within HEK-293T cells using a retroviral plasmid (76). This fusion protein consists of the transmembrane domain of the platelet-derived growth factor receptor and a peptide that targets the epidermal growth factor receptor (EGFR), resulting in EGFR-targeted extracellular vesicles (EVs). These electric vehicles are engineered to transport the anti-cancer miRNA let-7 straight to breast tumor cells. As for the chemical engineering approach to covalently conjugate peptides to EVs, Nakase et al. chemically synthesize stearyl-modified octaarginine peptide solid-phase peptide synthesis. The stearyl group functioned as an anchoring unit for membrane insertion. This method facilitated straightforward alteration of the exosome membrane to promote macropinocytosis, markedly increasing cellular absorption of extracellular vesicles (EVs) and enabling efficient intracellular transport of the artificially encapsulated ribosome-inactivating protein saporin through EVs, therefore resulting in tumor cell apoptosis (77). Nevertheless, the severe chemical treatment of EV surfaces, which may result in detrimental functional degradation, has hindered the widespread adoption of these methods (78). As for affinity conjugation, there are various methods that use EV-binding peptides or antibodies to coat EVs. However, these conjugations are transient and unstable. He et al. designed chiral peptide Au (I) infinite covalent polymers (DPAICP) using D-peptides and Au³⁺²⁶. They then ultracentrifuged milk-derived extracellular vesicles (ME) membranes with lactoprotein, embedded the chiral peptide supramolecular assemblies into the ME membrane, and obtained an artificial milk DPAICP@ME with pharmaceutical and absorbable properties. This approach restores the p53 signaling pathway for cancer therapy while further activating T cells and enhancing the efficacy of anti-PD-1 immunotherapy (79). As mentioned earlier, existing surface modification methods for EVs have various drawbacks in terms of safety, stability, and integrity (80). A stable and gentle method for EV coupling has been developed by Pham et al. who devised a novel technique utilizing protein ligases (including sortase A and OaAEP1 ligase) to covalently attach EVs to high-copy-number targeting moieties. The conjugation of EVs with EGFR-targeting peptides or anti-EGFR nanobodies facilitates their accumulation in EGFR-positive cancer cells both in vitro and in vivo. Furthermore, this methodology is applicable for conjugating EVs with peptides and nanobodies targeting other receptors, such as HER2 and SIRPα (81).

Different peptide based vesicles had quite different functions for cancer immunotherapy, which are concluded in Table 1 . The synergistic effect of peptide based vesicles can be categorized into three parts: enhancing immune response by peptide antigen or targeted peptide, augmenting anti-cancer effect of toxic peptides, improving tumor targeted delivery by peptide ligands.