Surgical tumor–derived nanoplatform targets tumor-associated macrophage for personalized postsurgical cancer immunotherapy

Abstract

Directly activating CD8⁺ T cells within the tumor through antigen-presenting cells (APCs) hold promise for tumor elimination. However, M2-like tumor-associated macrophages (TAMs), the most abundant APCs in tumors, hinder CD8⁺ T cell activation due to inefficient antigen cross-presentation. Here, we demonstrated a personalized nanotherapeutic platform using surgical tumor–derived galactose ligand–modified cancer cell membrane (CM)–coated cysteine protease inhibitor (E64)–loaded mesoporous silica nanoparticles for postsurgical cancer immunotherapy. The platform targeted M2-like TAMs and released E64 within lysosomes, which reshaped antigen cross-presentation and directly activated CD8⁺ T cells, thus suppressing B16-OVA melanoma growth. Furthermore, this platform, in combination with anti–PD-L1 antibodies, enhanced the therapeutic efficacy and substantially inhibited 4T1 tumor growth. CMs obtained from surgically resected tumors were used to construct a personalized nanotherapeutic platform, which, in synergy with immune checkpoint blockade (ICB), effectively inhibited postsurgical tumor recurrence in 4T1 tumor. Our work offered a robust, safe strategy for cancer immunotherapy and prevention of postsurgical tumor recurrence.

E64-loaded biomimetic mesoporous silica reshapes tumor-associated macrophage by inhibiting cysteine protease activity.

INTRODUCTION

Cancer immunotherapy has emerged as an attractive therapeutic modality for addressing cancer in recent years (1–3). Multiple treatment strategies have been extensively studied and clinically evaluated, encompassing immune checkpoint blockade (ICB) therapy, therapeutic vaccines, cytokine therapy, and cell therapy (4–7). Among the various components of the tumor microenvironment (TME), tumor-associated macrophages (TAMs) are one of the most abundant cells and play a crucial role in tumor metastasis, progression, and persistent growth (8–10). In general, TAMs are commonly classified into two main subsets: M1-like macrophages (classically activated) and M2-like macrophage (alternatively activated) (9, 11). M2-like TAMs are the most abundant subtype of TAMs and can effectively shield tumors from the suppressive impact of the adaptive immune response through the expression or secretion of proteases, immunosuppressive factors, and growth factors (12, 13). Hence, M2-like TAMs are increasingly recognized as an important target for immunotherapy.

Currently, several therapeutic strategies have been developed to counteract the tumor-promoting functions of M2-like TAMs. These approaches include the design of inhibitors to block monocyte recruitment, direct elimination of M2-like TAMs, and reeducation of M2-like TAMs to an antitumor M1-like phenotype using agonists (14–19). However, these treatment strategies often overlook the potential of M2-like TAMs as antigen-presenting cells (APCs) (20, 21). Considering the high abundance of M2-like TAMs in tumors compared with other APCs, the strategy that endows M2-like TAMs with antigen cross-presentation ability to activate CD8⁺ T cells in the tumor site would be a more direct and effective immunotherapeutic approach. Regrettably, previous studies indicate that despite their strong phagocytic ability for tumor cells, M2-like TAMs exhibit a very weak antigen cross-presentation capability (22–25). In a recent study, Cui et al. (26) reported that M2-like TAMs exhibited higher lysosomal cysteine protease activity compared with other APCs, and they found that blocking the enzyme activity could effectively enhance antigen cross-presentation and activate CD8⁺ T cells. However, the exogenous lysosome targeted delivery of small-molecule cysteine protease inhibitor E64 is impeded by its low bioavailability, poor drug-like properties, difficulty to cross the cellular plasma membrane, and rapid clearance (27, 28). Therefore, there is a need to design and construct a carrier platform for targeted delivery of cysteine protease inhibitors to M2-like TAMs to boost TAM antigen cross-presentation.

In addition to blocking the lysosomal cysteine protease activity of M2-like TAMs, the availability of a sufficient supply of tumor-associated antigens (TAAs) is also crucial to achieve effective antigen cross-presentation (26, 29–31). Recently, cancer cell membrane–coated nanoparticles (CCNs) represent an emerging class of surface modification strategies that mimic the characteristics of cancer cells, including homologous targeting, immune evasion, prevention of drug leakage, antigen preservation, and long circulation time (32–36). In addition, the cancer cell membranes (CMs) derived from surgically excised cancer cells contain tumor-specific neoantigens that carry patient-specific information and have been used as a valuable antigen source for the development of personalized nanovaccines (37–39). However, overexpression of the “don’t eat me” signal CD47 on CMs may prevent internalization of CCNs by TAMs through interaction with the signal regulatory protein α (SIRPα) receptor on TAMs (40). Although there have been some reports of active targeting of CCNs to TAMs using artificial ligands on CMs, the synthesis and modification of these ligands are complex and inefficient (41–43). Therefore, establishing an efficient and simple artificial ligand modification approach on CMs is crucial for preferential targeting to M2-like TAMs to efficiently deliver cysteine protease inhibitors and TAAs.

In this study, we constructed cysteine protease inhibitor E64-loaded, artificial ligand–modified CM-coated mesoporous silica nanoparticles (MSN-E64@CM, also named as ME@C) for targeted delivery of E64 and TAAs to M2-like TAMs to achieve a personalized tumor-specific immunotherapy. As illustrated in Fig. 1A, CMs were first modified with azide groups through metabolic glycan processing and then decorated with the artificial ligand dibenzocyclooctyne-modified galactose (DBCO-GAL) through gentle and efficient copper-free click chemistry. The extracted CMs were then coated on the surface of E64-loaded MSNs to prepare ME@C nanoparticles (NPs). The NPs could selectively target M2-like TAMs through CD302 receptor–mediated endocytosis. In the lysosome, the NPs released E64 to inhibit the activity of cysteine proteases, thereby enhancing antigen cross-presentation of M2-like TAMs and activating CD8⁺ T cells (Fig. 1B). The CMs provided abundant tumor-specific neoantigens for the process of antigen cross-presentation. In vitro, ME@C effectively restored antigen cross-presentation in M2-like macrophages without altering the macrophage phenotype, thereby facilitating the activation and proliferation of antigen-specific CD8⁺ T cells. In vivo, we elucidated the underlying mechanism of ME@C-mediated antitumor immunotherapy using the B16-F10-OVA tumor model. We further demonstrated that ME@C NPs encapsulated with tumor cell membranes collected from surgically resected autologous tumors could serve as a personalized therapeutic platform, effectively preventing tumor recurrence after the initial resection when combined with anti–programmed cell death ligand 1 (anti–PD-L1) treatment. Our work provides a promising strategy for effective patient-specific postsurgical immunotherapy.

Fig. 1. Fabrication process and treatment mechanism of ME@C NPs.

(A) Fabrication process of ME@C NPs. Initially, tumor cells obtained by surgery were subjected to metabolic glycan labeling using Ac₄ManNAz to introduce azide groups onto the tumor cell membrane (CM). Subsequently, artificial ligand DBCO-GAL was attached through click chemistry. The artificial ligand–modified CMs were then extracted and coated onto the surface of E64-loaded MSN NPs through physical extrusion, ultimately forming ME@C NPs. (B) After arriving the TME, ME@C NPs preferentially targeted M2-like TAMs through CD302 receptor–mediated endocytosis and were translocated to lysosomes. Within the acidic lysosomal environment, ME@C released E64, which inhibited the cysteine protease activity within the lysosomes. This process resulted in a reshaped antigen cross-presentation capability of M2-like TAMs and simultaneously initiated a direct CD8⁺ T cell response to attack tumor cells.

RESULTS

ME@C is an ideal carrier for antigens and E64

The preparation of ME@C involves three steps: First, artificial ligand–modified CMs were extracted from tumor cells. This approach is well established in metabolic engineering techniques, allowing for the introduction of artificial chemical receptors onto the cell surface as reported in prior studies (44, 45). In our specific method, we used tetraacetylated N-azidoacetylmannosamine (Ac₄ManNAz), a metabolic precursor, to incorporate azide groups on the cell membrane (Fig. 2A). To confirm the successful modification of azide groups and assess the efficiency of the bioorthogonal copper-free click reaction, we coincubated B16-F10-OVA cells with Ac₄ManNAz, followed by the addition of DBCO-Cy5.5, which was then imaged using confocal laser scanning microscopy. The resulting images, as anticipated, showed strong fluorescence on the surface of all B16-F10-OVA cells (Fig. 2B). Next, we introduced a previously unidentified artificial chemical ligand, DBCO-GAL. The synthesis of this ligand is detailed in fig. S1A, and the final product was thoroughly characterized using nuclear magnetic resonance (fig. S1B) and mass spectrometry (fig. S1C) to validate its structure. This purified galactose ligand was then used in bioorthogonal copper-free click chemistry, with azide groups present on the surface of B16-F10-OVA cells. Following the modification of the tumor cell surface, the next step involved the successful fabrication of MSNs, which are US Food and Drug Administration–approved nanocarriers known for their safety. This was achieved through established protocols based on the Stöber method (46). Subsequently, we loaded E64, a cysteine protease inhibitor, into MSNs to obtain MSN-E64 NPs (ME). In Fig. 2C, the synthesized MSN-E64 NPs and the artificial ligand–modified CM mixture were subjected to repeated extrusion through 200-nm pores. This process resulted in the uniform coating of artificial ligand–modified CMs on the MSN-E64 NPs.

Fig. 2. Preparation and characterization of ME@C.

(A) Schematic representation of azide labeling on CMs via metabolic glycoengineering and subsequent bioorthogonal labeling of the GAL ligand. (B) Confocal fluorescence images of B16-F10-OVA cells incubated with Ac₄ManNAz for 24 hours, followed by DBCO-Fluor Cy5.5 treatment. Scale bar, 50 μm. (C) Schematic diagram of preparation process of ME@C. (D) TEM images of MSNs, ME, and ME@C. Scale bar, 200 nm (inset scale bar, 50 nm). Size distribution (E) and ζ potentials (F) of MSNs, ME, and ME@C; means ± SEM. n = 3. (G) SDS-PAGE protein identification photograph of B16-F10-OVA CMs, MSNs, and MSN@CM. (H) Western blot analysis of CD47 and OVA for B16-F10-OVA CMs, MSNs, and MSN@CMs. (I) Particle size of MSNs, ME, and ME@C in Dulbecco’s modified Eagle’s medium + 10% FBS at various time intervals. Means ± SEM. n = 3. (J) Loading capacities of E64 in MSNs at different E64/MSN preparative molar ratios. (K) Time-dependent E64 release of ME@C or ME under different pH. Means ± SEM. n = 3. (L) Cell viability of 3T3, B16-F10-OVA, and RAW264.7 after incubation with different concentrations of ME@C. Means ± SEM. n = 3.

Transmission electron microscope (TEM) images revealed that the diameters of MSNs, ME, and ME@C were about 83 ± 2 nm, 86 ± 1 nm, and 98 ± 4 nm, respectively. The presence of a lipid shell, approximately 6 nm in thickness, confirmed the successful formation of membrane coating (Fig. 2D). Dynamic light scattering analysis further characterized the particles, revealing average hydrodynamic diameters of around 90 ± 2 nm, 95 ± 3 nm, and 115 ± 5 nm for MSNs, ME, and ME@C, respectively (Fig. 2E). The ζ potentials, as presented in Fig. 2F, were measured to be 14.9, 7.8, and −14.6 for MSNs, ME, and ME@C, respectively. These data suggested that the coating of CMs on ME involved a combination of physical extrusion and electrostatic attraction. Moreover, SDS–polyacrylamide gel electrophoresis (SDS-PAGE) bands confirmed intact CM proteins on ME@C (Fig. 2G). Western blot analysis further revealed the retention of CD47 protein and OVA protein in ME@C (Fig. 2H), suggesting that ME@C had the capability of functioning as an antigen nanocarrier and evading phagocytosis in vivo. In addition, we tested the stability of these particles over time. MSNs, ME, and ME@C showed no statistically notable changes in size when stored in the medium for an extended period. This result indicated the potential of ME@C as a stable delivery system under physiological conditions (Fig. 2I).

Next, we quantified the amount of E64 loaded in MSNs using high-performance liquid chromatography (HPLC) at various feeding ratios. It was observed that the loading capacity of E64 increased with the feeding ratio of E64/MSNs, reaching a maximum of 21.5% at a feeding ratio of 2:1 (w/w) (Fig. 2J). We also evaluated the release profile of ME@C under different pH conditions. As depicted in Fig. 2K, ME@C exhibited notable release of E64 at pH 5.0, which corresponds to the lysosomal pH, with approximately 80% release within 24 hours. In contrast, minimal release of E64 was observed after pretreatment at pH 6.8 (representing the TME) and pH 7.4 (representing blood pH). Notably, ME@C displayed reduced leakage compared to ME under pH 6.8 and pH 7.4 conditions, a feature attributed to the leak-proof capability of the CMs. The cytotoxicity of ME@C was also evaluated. As shown in Fig. 2L, ME@C exhibited excellent biocompatibility toward 3T3 normal cells, B16-F10-OVA cancer cells, and RAW264.7 macrophages cells. Together, these results presented that ME@C served as an ideal carrier for TAAs and E64, with the potential to maintain stability during in vivo circulation.

ME@C selectively targets M2-like BMDMs

In tumor tissues, most of TAMs are in the M2-like phenotype. To evaluate the targeting of ME@C toward M2-like TAMs, we isolated bone marrow–derived macrophages (BMDMs) from C57BL/6J mice. These BMDMs were then polarized into M1-like phenotype using lipopolysaccharide (LPS) and interferon-γ (IFN-γ), or M2-like phenotype using interleukin-4 (IL-4), according to established protocols (47). The expression levels of classical phenotypic markers in the different types of BMDMs were analyzed using quantitative real-time polymerase chain reaction (qRT-PCR). The results, as depicted in Fig. 3A, showed higher expression of TNF-α (tumor necrosis factor–α) in LPS/IFN-γ–activated M1-like BMDMs and Arg-1 (arginase-1) in IL-4–activated M2-like BMDMs, confirming successful activation of the BMDMs. To evaluate the selectivity of ME@C for M2-like BMDMs, we subsequently replaced E64 with fluorescein isothiocyanate (FITC) (MF@C) and conducted a cellular phagocytosis experiment. Confocal microscopy analysis, as presented in fig. S2, indicated that after co-cultivation for 12 hours, M2-like BMDMs exhibited a higher capacity for NP internalization compared to M0-like and M1-like BMDMs. Flow cytometry analysis revealed that the phagocytic ratio of M2-like macrophages was 2.5 times higher than that of M1-like BMDMs and 3.6 times higher than that of M0-like BMDMs (Fig. 3B). Both M1-like and M2-like phenotypes were found to coexist in the TME. To further confirm the specific targeted ability of ME@C toward M2-like TAMs, we prestained the M2-like BMDMs with 4′,6-diamidino-2-phenylindole (DAPI) dye and coincubated them with nonstained M1-like BMDMs in the presence of MF@C. As expected, the results further demonstrated the excellent selectivity of ME@C toward M2-like BMDMs (Fig. 3C).

Fig. 3. ME@C selectively targets the M2-like BMDMs and enhances antigen cross-presentation.

(A) Relative TNF-α and Arg-1 mRNA levels in BMDMs under various polarization conditions. (B) Flow cytometry analysis (left) and quantification (right) of BMDMs with different phenotypes after incubation with MF@C (50 μg/ml). (C) Flow cytometry analysis (left) and quantification (right) of M1/M2-like BMDMs after coculture with MF@C. (D) Flow analysis of MGL expression on BMDMs with different phenotypes. (E) Confocal images of M2-like BMDMs treated with MF@C for 12 hours. Scale bar, 10 μm. (F) Pearson and overlap coefficient analysis of lysosomes with MF@C. (G) Diagram of the OT-I CD8⁺ T cell activation assay. (H) Antigen cross-presentation in M2-like BMDMs coincubated with various formulations for 24 hours. Relative TNF-α (I) and Arg-1 (J) mRNA levels in M2-like BMDMs treated with PBS, OVA (25 μg/ml), E64 (10 μg/ml), MSN (50 μg/ml), MEO (50 μg/ml), MEO@C (50 μg/ml), and LPS (10 ng/ml) for 24 hours. OT-I CD8⁺ T cell priming assay showing CD8⁺ T cell activation (K) and proliferation (L) after coculture with M2-like BMDMs for 72 hours. (M) Diagram of the B3Z CD8⁺ T cell activation assay. (N) Relative percentage of activated B3Z CD8⁺ T cells. All data were expressed as means ± SEM. n = 3. Statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test. ns, not significant, P > 0.05; *P < 0.05; **P < 0.01; and ****P < 0.0001.

Previous extensive research has established that M2-like TAMs residing within the TME exhibited the highest glycolytic activity. This heightened metabolic activity resulted in elevated expression of macrophage galactose-type lectin (CD302, MGL), a phagocytic receptor (48). Flow analysis data in Fig. 3D confirmed that MGL expression in M2-like BMDMs was 4.23-fold higher than that in M0- and M1-like BMDMs. Consequently, it was plausible that the selective phagocytosis of ME@C was primarily attributed to the interaction between the galactose ligand and the MGL receptor. To further investigate the significance of MGL receptors and galactose ligands in the targeting of ME@C on M2-like BMDMs, we preincubated M2-like BMDMs with free galactose and then assessed the uptake of ME@C. As illustrated in fig. S3, the blockage of MGL receptors by free galactose noticeably reduced the phagocytic efficiency of M2-like BMDMs. This result provided further evidence of the pivotal role of the MGL receptor in the preferential targeting of ME@C to M2-like BMDMs. To further investigate the internalization process of ME@C, we used MF@C to visualize the intracellular localization of the delivered drugs. A notable portion of the FITC, delivered by MSNs, exhibited specific localization within lysosomes within a 12-hour timeframe. This observation in Fig. 3 (E and F) and fig. S4 underscores the efficient lysosomal targeting of the cysteine protease inhibitor delivered by our nanoplatform. These results demonstrate that ME@C facilitates sustained delivery of substantial amounts of E64 into the lysosomes of M2-like TAMs, thereby establishing a foundation for modulating the antigen cross-presentation of these TAMs.

ME@C restores antigen cross-presentation in M2-like BMDMs

During the process of antigen cross-presentation, TAAs are internalized by APCs and subsequently presented on major histocompatibility complex class I (MHC-I) molecules, leading to the activation of CD8⁺ T cells (Fig. 3G) (37). To assess the cross-presentation efficiency in M2-like TAMs and the activation of CD8⁺ T cells, we loaded the model antigen ovalbumin (OVA) into ME and encapsulated it with engineered CMs, generating MEO@C. Flow cytometry quantification of MHC-I–associated SIINFEKL peptide and MHC-I molecules on M2-like BMDMs revealed the elevated expression exclusively in the E64-loaded groups (MEO and MEO@C) (Fig. 3H and fig. S5). In contrast, there were no substantial differences in the non-E64–loaded groups (OVA, MO, and MO@C). Moreover, M2-like BMDMs treated with MEO@C displayed a 2.1-fold increase in SIINFEKL peptide expression compared with the MEO-treated group. This enhancement can be attributed to several factors: (i) CMs enhanced NP uptake by M2-like BMDMs; (ii) E64 inhibited the activity of cysteine proteases in the lysosomes of M2-like BMDMs, preventing complete degradation of OVA without altering autophagy (fig. S6) and facilitating its reprocessing and presentation; and (iii) the B16-F10-OVA CMs provided additional OVA antigens. Treatment with MEO@C did not induce an increase in TNF-α levels (Fig. 3I) in M2-like BMDMs while maintaining the high Arg-1 expression levels (Fig. 3J). Furthermore, there was no substantial up-regulation of typical protein expression associated with M1-like macrophage markers, including CD80 and CD86, as shown in fig. S7. These results demonstrated that the specific modulation of lysosomal cysteine protease activity by MEO@C did not substantially alter the phenotype of M2-like BMDMs. Subsequently, we further investigated whether MEO@C would enhance MHC-II–associated antigen presentation. The expression level of MHC-II molecules in M2-like BMDMs after treatment with MEO@C showed no substantial differences compared with the control groups (fig. S8), revealing that MEO@C did not affect MHC-II–associated antigen presentation. Collectively, these results highlighted the remarkable ability of ME@C to substantially enhance the antigen cross-presentation functionality of M2-like TAMs without eliciting unnecessary inflammatory responses. This unique feature sets the ME@C nanoplatform apart from conventional macrophage repolarization approaches and positions it as a promising candidate for clinical applications in postsurgical therapy.

Next, we further investigated whether M2-BMDMs, whose antigen-presenting function was restored by MEO@C treatment, could effectively activate T cells in vitro. As shown in Fig. 3K, in contrast to the other groups, M2-like BMDMs treated with MEO@C stimulated increased secretion of IFN-γ by OT-I CD8⁺ T cells. Consistently, to assess the potential of ME@C-treated M2-like BMDMs to induce T cell proliferation, we cocultured OT-I CD8⁺ T cells prestained with carboxyfluorescein diacetate succinimidyl ester (CFSE) with M2-like BMDMs treated with different groups. As depicted in Fig. 3L, M2-like BMDMs treated with MEO@C resulted in over 75% proliferation of OT-I CD8⁺ T cells. In addition, MEO@C could not directly activate OT-I CD8⁺ T cells due to the lack of costimulatory molecules (fig. S9). These results provide strong evidence that M2-like BMDMs treated with MEO@C are highly effective in inducing subsequent OVA-specific T cell responses. To further validate the CD8⁺ T cell activation of M2-like BMDMs following functional restoration, we conducted experiments using the B3Z CD8⁺ T cell activation system (Fig. 3M). The B3Z CD8⁺ T cells activate IL-2 promoter elements and downstream lacZ gene expression upon T cell receptor (TCR) recognition of SIINFEKL-H-2Kb complexes. The product of the lacZ reporter gene, β-galactosidase, converts the chlorophenol red-β-d-galactopyranoside (CPRG) to chlorophenol red (CPR). As shown in Fig. 3N, M2-like BMDMs treated with MEO@C exhibited robust activation of B3Z CD8⁺ T cells, surpassing other control groups. This result is consistent with the results observed in the OT-I CD8⁺ T cell system. In summary, the ME@C nanoplatform can effectively modulate the lysosomal activity of M2-like macrophages through E64 while delivering TAAs. This dual action can increase MHC-I–associated antigen peptides in M2-like BMDMs, subsequently triggering the proliferation and activation of specific CD8⁺ T cells.

ME@C preferentially targets M2-like TAMs in vivo

The CMs have intercellular homologous binding capability, and therefore, ME@C is expected to be enriched in tumor site. To confirm this, we intravenously injected Cy5.5-loaded MSNs (MCy5.5) coated with artificial ligand–modified 4T1 CMs (MCy5.5@C) into the 4T1 tumor model. The NPs without coating (MCy5.5) were used as controls. Compared to MCy5.5, MCy5.5@C demonstrates superior long-term accumulation capability. The highest tumor accumulation is observed at 24 hours after injection, and detectable fluorescence signals persist up to 48 hours (fig. S10, A and B). This prolonged retention can be attributed to the enhanced permeability and retention (EPR) effect, as well as the homologous binding capability (49). Furthermore, tumor tissues were harvested after injection. As depicted in fig. S10 (C and D), MCy5.5@C exhibited significantly higher enrichment content and longer retention time at the tumor site compared to MCy5.5, further indicating that the CM encapsulation enhances NP accumulation in the tumor region.

TME is composed of various cell types, including tumor cells, TAMs, T cells, and others, forming a complex cellular composition (50). We explored whether MF@C could selectively target TAMs. As illustrated in Fig. 4A and fig. S11, although CD45⁻ cells exhibited some phagocytic ability, MF@C was predominantly internalized by TAMs. Notably, the accumulation of ME@C in M2-like TAMs (CD206⁺) was approximately four times higher than that in M1-like TAMs (CD206⁻). This preferential accumulation can be attributed to the overexpression of MGL receptors in M2-like TAMs (Fig. 4, B and C). This result is consistent with the findings in vitro. Together, these results indicated that the artificial ligand CMs played a pivotal role in the accumulation of ME@C within tumor regions and its preferential engulfment by M2-like TAMs.

Fig. 4. ME@C preferentially targets M2-like TAMs and elicits immune responses in tumors.

(A) Flow cytometry analysis of MF@C uptake by various cell types in tumor 24 hours after intravenous injection. Means ± SEM. n = 6. (B and C) Flow cytometric analysis of MF@C uptake by M1-like or M2-like TAMs 24 hours after injection. Representative flow images of CD206 gating logic (B) and quantification (C) are shown. Means ± SEM. n = 6. (D) Scheme and timeline of the experimental design to evaluate the immune responses elicited by OVA, MO, MO@C, MEO, and MEO@C. (E and F) Proportions of CD11b⁺ F4/80⁺ SIINFEKL⁺ OVA (SIINFEKL)–presenting TAMs in the tumors. Representative flow cytometry data (E) and quantification (F) are shown. Means ± SEM. n = 6. (G to J) Abundance of CD45⁺ cells (G), TAMs (H), CD8⁺ T cells (I), and CD8⁺ T effector cells (J) in tumors. Means ± SEM. n = 6. (K and L) Frequency of SIINFEKL-MHC-I tetramer⁺ CD8⁺ T cells in tumor by flow cytometry analysis of tetramer⁺ CD8⁺ T cells. Representative quantification (K) and flow dot plot (L) are shown. Means ± SEM. n = 6. Abundance of CD4⁺ T cells (M) and DCs (N) in tumors. Means ± SEM. n = 6. Statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test. ns, not significant, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001.

ME@C restores antigen cross-presentation in TAMs in vivo

We next established a B16-F10-OVA tumor model in C57BL/6 mice and administered NPs twice within a week and collected tumors to analyze the immunization efficacy of ME@C (Fig. 4D). First, we assessed the cross-presentation of the OVA antigens in TAMs. The treatment of MEO@C produced significantly higher levels of antigen cross-presentation in TAMs compared with the control group, free OVA, MO, MO@C, and MEO groups (Fig. 4, E and F, and fig. S12). The MEO group did not show an improvement in antigen cross-presentation, similar to that observed in the cell experiments. This result can be attributed to the absence of CMs coated on MEO, which led to the lack of targeting specificity toward TAMs and the potential leakage of E64 and OVA in vivo. Furthermore, treatment with MEO@C significantly increased the infiltration of immune cells, such as CD8⁺ T cells and CD8⁺ effector T cells within the tumor (Fig. 4, G to J, and fig. S13). This was accompanied by a noticeable up-regulation of the activation marker CD69 and proliferation marker Ki67 in CD8⁺ T cells (fig. S14). Notably, the frequency of SIINFEKL–MHC-I tetramer⁺ CD8⁺ T cells was approximately twofold higher in the MEO@C group compared with other groups (Fig. 4, K and L, and fig. S15). Furthermore, no substantial changes were observed in the frequency of CD4⁺ T cells and dendritic cells (DCs) compared with the control groups (Fig. 4, M and N). This observation may be attributed to the fact that MEO@C preferentially targeted TAMs and was primarily directed toward MHC-I–restricted presentation of TAMs (26). Overall, these results indicated that ME@C had the capacity to modulate TAM cross-presentation capabilities and stimulate a robust, antigen-specific T cell immune response, presenting a unique approach to potentiate the antitumor effectiveness.

MEO@C inhibits B16-F10-OVA tumor growth

We further evaluated the therapeutic efficacy of the MEO@C-based nanotherapeutic platform in a B16-F10-OVA melanoma model. On the 4th and 11th day after the tumor volume reached approximately 80 mm³, the MEO@C or control formulations was intravenously injected into the corresponding experimental groups of mice (Fig. 5A). The administration of OVA alone, MO, MO@C, or MEO did not provide any notable benefit for tumor growth suppression compared with the phosphate-buffered saline (PBS) control. In contrast, the MEO@C group exhibited a potent inhibitory effect on tumor growth, with an average tumor volume of only 390.68 mm³ on the 14th day (Fig. 5, B to H). Furthermore, in the untreated group, all animals died within 25 days, whereas the survival rate of mice treated with the MEO@C nanoplatform was remarkably improved (Fig. 5I). These results collectively underscore the potential of MEO@C as a nanotherapeutic platform for cancer treatment.

Fig. 5. MEO@C inhibits tumor growth and prolongs survival in B16-F10-OVA tumor mice.

(A) Schemes of tumor challenge experiment design. (B) Average tumor growth curves for B16-F10-OVA tumors on mice. Means ± SEM. n = 6 from independent animals. Tumor growth kinetics for individual mice in control (C), OVA (D), MO (E), MO@C (F), MEO (G), and MEO@C (H). (I) Kaplan-Meier survival curves for B16-F10-OVA tumor–bearing mice. Means ± SEM. n = 6 from independent animals. Statistical significance was analyzed by the log-rank test. ***P < 0.001.

ME@C and ICB combination therapy enhances tumor suppression

Although MEO@C resulted in the inhibition of B16-F10-OVA tumor growth, the therapeutic efficacy was moderate, and all mice died within 35 days. Considering that ME@C induced TAMs to up-regulate MHC-I expression and fostered the activation and proliferation of CD8⁺ T cells, we sought to explore whether augmenting CD8⁺ T cell viability through the combination of ICB therapy could further enhance the antitumor immune response (Fig. 6A). Anti–PD-L1 antibody specifically binds to PD-L1, thereby alleviating immune suppression within the TME and facilitating the elimination of tumor cells by CD8⁺ effector T cells (4). Here, in our experiments, each BALB/c mouse was subcutaneously inoculated with 4T1 tumors. After the tumor volume reached approximately 80 mm³, the different formulations were intravenously injected into the mice. Compared with untreated mice, the mice treated with ME@C or anti–PD-L1 alone showed the moderate tumor growth suppression. In contrast, the combination treatment using ME@C and anti–PD-L1 (ME@C-A) substantially improved the therapeutic efficacy and remarkably inhibited the tumor growth (Fig. 6, B to F). Moreover, all mice in the ME@C or anti–PD-L1 monotherapy groups died within 50 days. On the contrary, the combination treatment significantly prolonged the survival time, with a remarkable 67% of mice surviving up to 60 days (Fig. 6G). To further investigate the antitumor effects achieved by combination therapy, immunofluorescence analysis was conducted using tumor tissues. Mice treated with either ME@C or combination therapy exhibited a marked increase in intratumoral CD8⁺ T cell infiltration, as shown in Fig. 6H. However, despite reshaping M2-like TAMs and promoting CD8⁺ T cell infiltration, ME@C monotherapy displayed underwhelming antitumor efficacy. This limitation might stem from persistent PD-L1expression on tumor cells, which triggered premature exhaustion of CD8⁺ T cells. The synergistic ICB combination therapy elegantly addressed this obstacle. By effectively abrogating the PD-L1/PD-1 axis, the combination therapy prevented premature exhaustion of CD8⁺ T cells, thereby unleashing their full cytotoxic potential and maximizing therapeutic efficacy. In sum, these findings demonstrated the considerable efficacy of ME@C in synergy with ICB therapy, which effectively triggered a robust CD8⁺ T cell–mediated antitumor immune response in a complementary manner.

Fig. 6. ME@C and ICB combination therapy enhances 4T1 tumor suppression.

(A) Schemes of 4T1 tumor challenge experiment design. (B) Average tumor growth curves for 4T1 tumors on mice. Means ± SEM. n = 6 from independent animals. Tumor growth kinetics for individual mice in control (C), ME@C (D), ME@C-A (E), and anti–PD-L1 (F). (G) Kaplan-Meier survival curves for 4T1 tumors on mice. Means ± SEM. n = 6 from independent animals. (H) Immunofluorescence images of CD8⁺ T cell infiltration in 4T1 tumor from mice treated with different formulations. Scale bars, 50 μm. Statistical significance was calculated by the log-rank test. ***P < 0.001.

ME@C nanotherapeutic platform has excellent biosafety

There were no notable changes observed in the body weight of mice over time, indicating low biotoxicity of the ME@C and ME@C-A treatment (fig. S16). It is crucial to have a comprehensive understanding of the biosafety of nanomaterials for their clinical applications. To further investigate the biosafety of ME@C, we evaluated its potential toxicity in healthy mice. The liver and kidney function parameters as well as the blood cell parameters showed no noticeable changes after treatment with nanomaterials compared with the PBS control group, as depicted in figs. S17 and S18. These results demonstrated the excellent biosafety of the ME@C treatment. Notably, histological analysis of organ sections (heart, liver, lung, spleen, and kidney) stained with hematoxylin and eosin showed no substantial changes in each treatment group (fig. S19), suggesting that ME@C did not induce inflammation in vivo. Furthermore, systemic cytokine levels (including TNF-α, IL-1β, IL-12, and IFN-γ) in mice treated with ME@C exhibited no substantial difference with those in PBS-treated mice (fig. S20), indicating that ME@C therapy did not induce systemic inflammation. These above results demonstrate that the ME@C nanotherapeutic platform not only exhibits low toxicity but also does not induce unnecessary inflammation in vivo, providing a solid foundation for its potential clinical applications.

ME@C and ICB combination therapy prevents postsurgical tumor recurrence

Surgical intervention is widely regarded as the primary treatment option for early-stage solid tumors in clinical practice (51). However, one major challenge is the presence of microscopic residual tumors following tumor resection, which can contribute to tumor recurrence and adversely affect patient outcomes. To address this issue, conventional postsurgical cancer treatments, such as chemotherapy and radiotherapy, have been used to suppress disease recurrence (52). Unfortunately, these therapies often induce additional inflammation at the surgical site, causing further harm to patients. In recent years, the collection of patient-specific information through CMs from resected tumors has gained considerable attention as a promising and highly personalized approach for developing innovative treatment strategies (37, 49). Given the low-inflammatory properties of the ME@C treatment, we implemented the collection of CMs from surgically resected tumors for ex vivo labeling of artificial receptors. Subsequently, ME@C was prepared and combined with ICB therapy to address residual tumors after surgery. The surgical procedure and treatment regimen are illustrated in Fig. 7A. The individual treatments of ME@C and anti–PD-L1 were ineffective in inhibiting tumor recurrence after tumor resection (Fig. 7, B to F). In stark contrast, the combination therapy of ME@C and anti–PD-L1 demonstrated excellent inhibitory effects, resulting in a low tumor recurrence rate of 20% (Fig. 7E). Furthermore, the survival rate of mice treated with ME@C-A was significantly higher than that obtained in other treatment groups (Fig. 7G). The consistent body weights of mice during the treatment underline the favorable biosafety of ME@C and ME@C-A treatments in postsurgical mice (Fig. 7H). Together, these data suggested that the combination of ME@C and ICB therapy offered a highly personalized and inflammation-free therapeutic strategy, with the potential to be an effective patient-specific postsurgical immunotherapy approach.

Fig. 7. ME@C and ICB combination therapy prevents postsurgical tumor recurrence in 4T1 tumors.

(A) Schematic illustration of experimental design for postsurgical recurrence challenge in 4T1 tumors. (B) Average tumor growth curves for 4T1 tumors on mice. Means ± SEM. n = 6 from independent animals. Tumor growth kinetics for individual mice in control (C), ME@C (D), ME@C-A (E), and anti–PD-L1 (F). (G) Kaplan-Meier survival curves for 4T1 tumors after surgery. Means ± SEM. n = 6 from independent animals. (H) Postsurgical treatment relative body weight of mice bearing 4T1 tumors. Statistical significance was calculated by the log-rank test. ***P < 0.001.

DISCUSSION

In summary, this study introduced a straightforward and promising nanotherapeutic platform, offering a feasible personalized immunotherapy approach that effectively targeted and restored antigen presentation capacity in M2-like TAMs. ME@C treatment led to the reshaping of M2-like TAM antigen cross-presentation ability and the direct activation of CD8⁺ T cell. Loaded with the model antigen OVA and E64, this nanotherapeutic platform demonstrated potent inhibition of B16-OVA tumor growth. Moreover, in the 4T1 tumor model, the combination of ME@C with ICB treatment substantially enhanced antitumor immune efficacy. In particular, CMs obtained from surgically resected tumors were used to construct a personalized nanotherapeutic platform, which, in synergy with ICB therapy, effectively suppresses tumor recurrence after 4T1 tumor removal, leading to prolonged survival in mice. The safety and stability of this nanotherapeutic platform offered valuable insights for the design of future nanoplatforms. We anticipate that this personalized nanotherapeutic platform holds great promise for clinical translation in the postsurgical setting, representing an exciting candidate therapeutic approach.

MATERIALS AND METHODS

Reagents

Sodium hydroxide (NaOH), (3-aminopropyl)triethoxysilane (APTES), cetyltrimethylammonium bromide (CTAB), tetraethyl orthosilicate (TEOS), methanol (MeOH), acetonitrile, dimethyl sulfoxide (DMSO), hydrochloric acid, triethylamine, Ac₄ManNAz, DBCO-Cy5.5, phenylmethylsulfonyl fluoride (PMSF), FITC, sulfo-cyanine5.5 (Cy5.5), galactose, magnesium chloride (MgCl₂), trifluoroacetic acid, glycine, and EDTA were obtained from Aladdin Reagent Company (Shanghai, China). NP-40 was from MedChemExpress Company (Shanghai, China). CPRG, d-galactosamine hydrochloride, DBCO–N-hydroxysuccinimidyl ester (DBCO-NHS), triethylamine, methanol, acetonitrile, and DMSO-d₆ were obtained from Merck KGaA Company (Darmstadt, Germany). OVA was offered by Thermo Fisher Scientific Inc. (Waltham, USA).

Synthesis of DBCO-GAL

d-Galactosamine hydrochloride (0.1 mM) and triethylamine (0.2 mM) were dissolved in methanol, followed by the addition of DBCO-NHS (0.12 mM). The reaction mixture was stirred at room temperature for 72 hours. The solvent was removed, and the crude product was purified via silica gel chromatography (Dichloromethane:MeOH = 10:1), giving the desired DBCO-GAL as a light yellow solid in 80% yield. ¹H nuclear magnetic resonance (400 MHz, DMSO-d₆) δ 7.73 to 7.55 (m, 2H), 7.53 to 7.42 (m, 3H), 7.42 to 7.23 (m, 3H), 6.41 to 6.23 (m, 1H), 5.03 (d, J = 13.9 Hz, 1H), 4.83 (dt, J = 15.3, 4.0 Hz, 1H), 4.60 (dt, J = 25.3, 5.7 Hz, 1H), 4.48 to 4.36 (m, 1H), 4.31 (t, J = 6.6 Hz, 1H), 3.92 to 3.82 (m, 1H), 3.80 to 3.65 (m, 2H), 3.65 to 3.54 (m, 3H), 3.44 to 3.20 (m, 2H), 2.62 to 2.51 (m, 1H), 2.41 to 2.21 (m, 1H), 2.11 to 1.92 (m, 1H), 1.85 to 1.70 (m, 1H).

Preparation of MSN, ME, MF, MCy5.5, and MEO

To synthesize the amino-functionalized MSNs, we dissolved 1 g of CTAB in 480 ml of deionized water (DIW) and added 3.5 ml of 2.0 M NaOH solution. The temperature was then raised to 70°C. Next, we added 5.0 ml of TEOS (22 mM) and 0.22 ml of APTES (1.10 mM) dropwise, and the mixture was stirred vigorously at 70°C for 2 hours. After filtration, the precipitate was washed with methanol and dried under vacuum at 100°C for 12 hours. To remove excess CTAB, the crude product was stirred in a concentrated hydrochloric acid methanol solution (2 ml) at 80°C for 10 hours. Last, the precipitate was filtered and washed with DIW, and MSNs were obtained.

In short, we weighed 2 mg of E64 and dissolved it in a DMSO solution. After complete dissolution, an equal volume was added to the previously prepared MSN (10 mg/ml) solution, which was stirred overnight at 4°C. Subsequently, the solution was centrifuged to collect ME, followed by dispersal in DIW or PBS for future use. The same procedure was followed to synthesize MF or MCy5.5 by substituting E64 with FITC or Cy5.5, respectively. Furthermore, we added 1 mg of OVA protein to the resulting ME suspension (10 mg/ml), stirring it overnight at 4°C. Next, we centrifuged the solution to collect MEO, dispersing it in PBS for future use.

Bioorthogonal copper-free click chemistry

To confirm the successful modification of azide groups, B16-F10-OVA cells were seeded in a 12-well plate (Thermo Fisher Scientific) at a density of 4 × 10⁴ cells/ml. After 24 hours, the cells were treated with Ac₄ManNAz (50 μM) for 2 days. Then, the cells were washed with PBS and exposed to DBCO-Cy5.5 (50 μM) for 1 hour. Subsequently, the cells were collected for confocal imaging to visualize the azide groups. Likewise, we used DBCO-GAL, prepared at the same concentration as DBCO-Cy5.5, to replace DBCO-Cy5.5, resulting in B16-F10-OVA cells with a galactose ligand.

Preparation of galactose ligand–modified CMs

To obtain CMs modified with galactose ligands, we followed the previously mentioned modification method in cell culture dish (10 cm). After scraping off the galactose-modified cells using a cell scraper, they were collected with PBS and centrifuged. The collected cell pellet was then resuspended in a low-permeability lysis buffer containing membrane protein extraction reagent (Thermo Fisher Scientific) and PMSF. The suspension was incubated on ice for 10 to 15 min. Next, the cells in the solution were repeatedly broken by freeze-thawing and then centrifuged at 700g for 10 min at 4°C. The supernatant was further centrifuged at 14,000g for 60 min to isolate the cell membrane fragments. The membrane products were freeze-dried overnight, weighed, and stored at −80°C. Before use, the lyophilized membrane material was rehydrated in DIW or PBS.

Preparation of ME@C, MF@C, MCy5.5@C, and MEO@C

To coat modified galactose ligands onto ME, ME (6 mg) and CMs (12 mg) were suspended in 20 ml of DIW. The suspension was then extruded 12 times through a 200-nm porous membrane using a micro-extruder (Avanti) and centrifuged to remove excess CMs. The resulting ME@C was resuspended in PBS and stored at 4°C for later use. Similarly, MF@C, MCy5.5@C, and MEO@C were prepared using the same extrusion method.

Mice

Five- to 6-week-old female C57BL/6 mice and female BALB/c nude mice were obtained from Hunan SJA Laboratory Animal Company in Beijing, China. OT-I transgenic mice (5 to 6 weeks old) were obtained from Jiangsu Aniphe Biolaboratory Animal Company in Jiangsu, China. The use of these mice was approved by the Experimental Animal Ethical Committee of Hunan University [SYXK (xiang) 2022-0007] in Changsha, China and was conducted in accordance with the guidelines outlined in the Guide for the Care and Use of Laboratory Animals. The mice were housed and cared for in a specialized pathogen–free animal facility at Hunan University.

MHC-I–restricted antigen cross-presentation assays

To assess MHC-I–restricted antigen cross-presentation by M2-like BMDMs, cells were seeded in 96-well plates at a density of 200,000 cells per well and incubated overnight to allow cell attachment. They were then cocultured with PBS, OVA (25 μg/ml), MO (50 μg/ml), MO@C (50 μg/ml), MOE (50 μg/ml), or MOE@C (50 μg/ml) for 24 hours. Afterward, the cells were washed three times with PBS and fixed with 4% paraformaldehyde for 15 min. Subsequently, the cells were stained with anti-OVA 257–264 antibody (13-5743-82; eBioscience, 1:1000 dilution) for 30 min to detect MHC-I–bound OVA 257–264 signaling on the surface of M2-like BMDMs using flow cytometry.

OT-I CD8⁺ T cell priming assay

M2-like BMDMs (200,000 cells per well, 96-well plates) were cocultured with PBS, OVA, MO, MO@C, MOE, or MOE@C for 24 hours. Afterward, the cells were washed with PBS. CD8⁺ T lymphocytes from OT-I mice were selected using the magnetic-activated cell sorting system (Invitrogen, 11333D) following the manufacturer’s instructions. The T cells were then stained with the CellTrace CFSE Cell Proliferation Kit (Invitrogen, C34570) as per the supplier’s protocol. The treated M2-like BMDMs were mixed with CFSE-stained splenic CD8⁺ T cells at a ratio of 1:10 and incubated in round-bottomed 96-well plates for 72 hours. Proliferation of OT-I CD8⁺ T cells was detected by flow cytometry. Supernatants were collected and analyzed for the cytokine IFN-γ using an enzyme-linked immunosorbent assay (ELISA) kit according to the manufacturer’s protocol (Invitrogen).

B3Z CD8⁺ T cell activation assay

M2-like BMDMs (200,000 cells per well, 96-well plate) were cocultured with PBS, OVA, MO, MO@C, MOE, or MOE@C for 24 hours. After rinsing with PBS, the treated M2-like BMDMs were mixed with B3Z CD8⁺ T cells at a ratio of 1:10 and incubated in round-bottomed 96-well plates for 72 hours. Subsequently, the cells were resuspended in 100 μl of CPRG working buffer (PBS containing 0.15 mM CPRG, 1 mM MgCl₂, and 0.125% NP-40) and incubated at 37°C for 4 to 6 hours. The color development reaction was stopped by adding 50 μl of CPRG termination buffer (containing 300 nM glycine and 15 mM EDTA in PBS), and the optical density (OD) value of the released CPR was measured at 595 nm using enzyme-labeled instrument (Synergy2, Bio-Tek, USA).

In vivo characterization of M2-like TAM targeting

To assess the targeting of ME@C to the tumor region, we first administered intravenous injections of MCy5.5@C to 4T1 tumor–bearing BALB/c mice. The whole-body Cy5.5 spectral fluorescence intensity of the mice was measured at specific time intervals. In addition, the mice were humanely euthanized at various injection times, and tumor tissues were subjected to fluorescent imaging (Lumina XR, Caliper, USA).

To further confirm the in vivo targeting of M2-like TAMs, we injected MF@C (25 μg) intravenously into mice bearing 4T1 tumors. The tumors were harvested 24 hours after injection, and the uptake of ME@C was assessed using flow cytometry. Fluorescently labeled antibodies are as follows: 7-Aminoactinomycin D (7-ADD) [peridinin chlorophyll protein (PerCP)–Cy5.5, BD OptiBuild, 559925], CD45 (allophycocyanin-Cy7, BD OptiBuild, 557659), CD11b [phycoerythrin (PE), BD OptiBuild, 553311], F4/80 (Brilliant Violet, BD OptiBuild, 565411), and CD206 (Alexa Fluor, BD OptiBuild, 565250).

Analysis of immune system in vivo

Female C57BL/6 mice, aged 6 to 8 weeks and bearing B16-F10-OVA tumors, were randomly divided into six groups (n = 6) and treated with intravenous injections of PBS (100 μl), OVA [100 μl at a dose of 500 μg/kg body weight (b.w.)], MO (100 μl at a dose of 10 mg/kg b.w.), MO@C (100 μl at a dose of 10 mg/kg b.w.), MEO (100 μl at a dose of 10 mg/kg b.w.), or MEO@C (100 μl at a dose of 10 mg/kg b.w.) on days 0 and 4. Tumors were collected on day 7 to assess the immune response.

For investigating the antigen cross-presentation ability of M2-like TAMs in vivo, tumors were dissected into small pieces and digested in RPMI 1640 medium containing 2% fetal bovine serum (FBS), 1% penicillin-streptomycin, collagenase I (1.5 mg/ml), collagenase IV (1.5 mg/ml), hyaluronidase (1.5 mg/ml), and deoxyribonuclease (DNase) I (0.2 mg/ml) at 37°C for 30 min to obtain a single-cell suspension. The obtained mixture was filtered through a 200-mesh filter, and the separated cells were stained with fluorescent antibodies, including 7-ADD (PerCP-Cy5.5, BD OptiBuild, 559925), CD45 (allophycocyanin-Cy7, BD OptiBuild, 557659), CD11b (FITC, BD OptiBuild, 561688), F4/80 (BV-421, BD OptiBuild, 565411), and SIINFEKL-H-2Kb (allophycocyanin, eBioscience, 13-5743-82), for 30 min at 4°C, and analyzed by flow cytometry.

To investigate the proportion of various immune cells in the tumor after treatment, the single-cell suspension of tumor tissue was stained with fluorescent antibodies, including 7-ADD (PerCP-Cy5.5, BD OptiBuild, 559925), CD45 (allophycocyanin-Cy7, BD OptiBuild, 557659), CD11b (FITC, BD OptiBuild, 561688), F4/80 (BV-421, BD OptiBuild, 565411), CD3 (PE-CF594, BD OptiBuild, 566699), CD4 (BV-786, BD OptiBuild, 742655), CD8 (BV-480, BD OptiBuild, 746832), MHC-II (BV-605, BD OptiBuild, 743325), CD11c (PE, BD OptiBuild, 561044), CD62L (BV-650, BD OptiBuild, 563808), and CD44 (PE-Cy7, BD OptiBuild, 560569), for 30 min at 4°C, and analyzed by flow cytometry. To analyze the percentage of OVA-specific CD8⁺ T cells, the single-cell suspension of tumor tissue was stained with FVS520 (FITC, BD OptiBuild, 564407), CD45 (allophycocyanin-Cy7, BD OptiBuild, 557659), H-2Kb OVA Tetramer SIINFEKL (PE, ProImmune), anti-CD8 (BV-480, BD OptiBuild, 746832), and anti-CD3 (PE-CF594, BD OptiBuild, 566699) and then analyzed by flow cytometry.

Therapeutic studies with MEO@C

In the therapeutic studies, female C57BL/6 mice (6 weeks old) were subcutaneously injected with B16-F10-OVA cells (2 × 10⁵ cells). Treatment was initiated when the tumors reached 80 mm³ (day 0). The mice were divided into six groups and intravenously injected on days 0, 4, and 8 with PBS (100 μl), OVA (100 μl at a dose of 1 mg/kg b.w.), MO (100 μl at a dose of 10 mg/kg b.w.), MO@C (100 μl at a dose of 10 mg/kg b.w.), MEO (100 μl at a dose of 10 mg/kg b.w.), or MEO@C (100 μl at a dose of 10 mg/kg b.w.) as described in the main text. Mice were euthanized when the tumor volume reached 1500 mm³.

ME@C synergy with anti–PD-L1 therapeutic study

To assess the synergistic therapeutic effect of ME@C and immune checkpoint inhibitors on tumors, 4T1 cells (2 × 10⁵ cells) were subcutaneously implanted into the right dorsal surface of female BALB/c mice (six mice per group). Treatment was initiated once the tumor volume reached approximately 80 mm³ (day 0). The mice were divided into four groups and received intravenous injections on days 0, 4, and 12 with PBS (100 μl), ME@C (100 μl at a dose of 10 mg/kg b.w.), ME@C (100 μl at a dose of 10 mg/kg b.w.) combined with anti–PD-L1 (20 μg per mouse for each injection) (ME@C-A), or anti–PD-L1 alone (20 μg per mouse for each injection), respectively. Anti–PD-L1 was administered intraperitoneally 1 day after the nanoplatform injection. Mice were euthanized when the tumor volume reached 1500 mm³.

Personalized ME@C synergy with anti–PD-L1 therapeutic assay after surgery

First, cell membranes were extracted from surgically excised tumors and functionalized. Specifically, 4T1 cells (2 × 10⁵ cells) were subcutaneously implanted into the right dorsal surface of female BALB/c mice (six mice per group). On day 12, the mice were randomly divided into four groups: (i) control (100 μl of PBS) group, (ii) ME@C (100 μl at a dose of 10 mg/kg b.w.) group, (iii) ME@C (100 μl at a dose of 10 mg/kg b.w.) + anti–PD-L1 (20 μg per mouse for each injection) group, and (iv) anti–PD-L1 (20 μg per mouse for each injection) group. All mice underwent surgical excision of tumors within a controlled laminar flow hood environment, preserving approximately 1% residual tissue to replicate the presence of microscopic tumors commonly encountered on surgical beds. Using PBS (penicillin-streptomycin solution, 2%), tumor samples were collected, and subsequent procedures were conducted within a new laminar flow hood to maintain sterility. The tumors were dissected into fragments and subjected to gentle compression with DNase I and collagenase I (1 mg/ml) over a 2-hour duration. Following treatment with PBS, a single-cell suspension was obtained through a 70-mm cell strainer, followed by the addition of collagenase I (0.5 mg/ml). By observing under an inverted microscope, the digestion was terminated by 10% FBS when the fibrocytes were found to be broken. After washing with PBS buffer solution and replacement of the medium six times to eliminate other cell types, the pure tumor cells can be obtained with better adherence. After 24 hours, the cells were treated with Ac₄ManNAz (50 μM) for 2 days. Then, the cells were washed with PBS and exposed to DBCO-GAL (50 μM) for 1 hour. Subsequently, the tumor cell membranes obtained from the preparation were used to encapsulate the ME, resulting in a personalized nanotherapeutic platform.

Next, once the surgical wound was no longer visible, the surgical tumor–derived CM-coated ME (ME@C) was injected into the tail vein of groups 2 and 3 on day 8 (day 20), day 12 (day 24), and day 20 (day 32) after the surgery. On days 9 (day 21), 13 (day 25), and 21 (day 33) after the surgery, groups 3 and 4 were injected intraperitoneally with anti–PD-L1. Group 1 served as a control and was injected with an equal volume of PBS. Mice were euthanized when the tumor volume reached 2000 mm³.

Statistical analysis

Methods and sample sizes (n) for statistical analyses are specified in the results section or graphic legends for all quantitative data. The results are presented as means ± SEM. Animal treatment studies were conducted with six mice per group, randomly assigned. Statistical differences between two groups were analyzed using one-way analysis of variance (ANOVA) with Tukey’s post hoc test. Survival differences among groups were evaluated using the Kaplan-Meier method, and P values were calculated using the log-rank test. GraphPad Prism 9 software was used for statistical analyses. ns, not significant, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001.

Supplementary Materials

This PDF file includes:

Materials and Methods

Figs. S1 to S20