Bacteria-mimicking cancer cells reprogram macrophages via multiple pattern recognition receptor pathways for cancer immunotherapy

Read post on nature.com

IntroductionThe advent of chimeric antigen receptor (CAR)-T cell therapeutics has revolutionized the treatment of hematological cancers.1 Clinical success in leukemia and lymphoma has demonstrated the potential of adoptive cell therapies to selectively eliminate malignant cells and induce durable therapeutic responses. However, effective cell therapies for solid cancers, which account for approximately 90% of all cancers,2 have yet to be successfully developed. Unlike hematological malignancies, solid tumors possess multiple barriers that limit the efficacy of immune cell-based therapies, including poor immune cell infiltration, dense stromal structures, heterogeneous antigen expression, and the highly immunosuppressive tumor microenvironments (TMEs).3 Among the potential candidates for overcoming these limitations, macrophages, a type of innate immune cell, are being actively investigated for their ability to infiltrate solid tumors and exert antitumor effects through phagocytosis and inflammatory responses.4,5 In addition to directly engulfing cancer cells, activated macrophages can modulate the TME through cytokine secretion and recruitment of other immune cells, highlighting their therapeutic potential for solid tumor immunotherapy. Nevertheless, macrophages within the TME are often immunosuppressed and adopt protumoral phenotypes,4 which suppress inflammatory activity and reduce phagocytic function, thereby hindering their therapeutic application.Pathogen-associated molecular patterns (PAMPs) are recognized by pattern-recognition receptors (PRRs) on innate immune cells and initiate robust immunological responses.6,7,8 Although researchers have attempted to reprogram protumoral macrophages into inflammatory phenotypes by leveraging the PAMP-PRR pathways,9,10,11,12,13 sustaining these induced inflammatory states remains difficult due to persistent and dynamic immunosuppressive cues within the TME, which continuously reshape macrophage signaling and can rapidly attenuate activation, metabolic fitness, and functional polarization over time.14 Meanwhile, phagocytosis not only inhibits tumor growth directly through clearance of malignant cells but also promotes enhanced tumor antigen processing and presentation, thereby bridging innate and adaptive immunity to support more durable and systemic antitumor responses.4,5,6,7,8 These premises led to the development of phagocytosis-enhancing strategies via genetic and/or adoptive cell transfer methods, including ex vivo generation and reinfusion of SIRPα-blocked macrophages15 or inflammatory CAR-macrophages.16,17,18,19 However, these approaches often require extensive genetic engineering, multi-step ex vivo manipulation, and complex quality-control processes, and are therefore associated with high production costs, limited scalability, and potential regulatory burdens that hinder broad clinical translation. Moreover, securing a sufficient number of engineered cells for therapeutic application remains particularly challenging for macrophages, as they exhibit minimal proliferative capacity ex vivo,20 making it difficult to generate clinically relevant cell numbers without functional decline. Therefore, the establishment of a macrophage-based solid tumor immunotherapy strategy that can both maintain the inflammatory phenotypes of activated macrophages and promote phagocytosis by naturally tumor-resident macrophages in a simpler, robust, and scalable manner may provide a potential solution to these challenges.Challenges of maintaining the inflammatory phenotypes of macrophages within the immunosuppressive TME and securing high numbers of macrophages can be addressed by delivering and maintaining macrophage stimulants within the TME for prolonged periods, as well as by the utilization of high numbers of tumor-resident immunosuppressed macrophages. In this study, we developed a bacteria-associated molecular pattern-induced recognition enhancement (BAMPIRE) strategy to address these challenges. In the BAMPIRE strategy, bacterial PAMPs are attached to the surface of cancer cells. This approach maintains PAMPs longer within the TME, efficiently reprogramming macrophages into antitumoral phenotypes, and simultaneously inducing the phagocytosis of PAMP-attached pathogen-mimicking cancer cells by macrophages. To the best of our knowledge, this type of strategy is the first demonstration of generating pathogen-like cells via the attachment of PAMPs onto the surface of cancer cells, followed by macrophage-mediated antitumor effects.For the BAMPIRE strategy, positively charged PAMP particles were prepared using bacterial membrane extracts and synthetic lipids. Positively charged particles readily associate with negatively charged cancer cell membranes through electrostatic interactions. Coincubation of the resulting bacteria-mimicking cancer cells with macrophages notably increased the phagocytosis of macrophages to approximately 70%, compared to 20% in the control, along with enhanced inflammatory cytokine production, including interleukin-6 (IL-6) and tumor necrosis factor alpha (TNF-α), through various PRR stimulation. The BAMPIRE strategy demonstrated macrophage-dependent inhibition of tumor growth in a murine cancer model. Moreover, when combined with the conventional chemotherapeutic agent doxorubicin (DOX), the median days of survival for both BAMPIRE and BAMPIRE plus DOX combination groups were compared to the DOX-only group. Tumor regression was observed with complete remissions in some cases. Further immunological analyses supported the involvement of adaptive immunity. Similar trends of therapeutic effects were observed in three different murine cancer models. Overall, we present a strategy for delivering PAMPs to the surface of cancer cells to generate pathogen-mimicking cells as a therapeutic approach for macrophage-driven treatment of solid tumors.ResultsMacrophages are stimulated by bacterial membrane extracts via various signaling pathwaysMacrophages possess several pattern-recognition receptors (PRRs), including Toll-like receptors (TLRs) that sense microbial antigens.21 The phagocytic function of macrophages is significantly suppressed in the TME owing to immunosuppressive cytokines and the expression of CD47 in cancer cells.4To restore this function, we developed the BAMPIRE strategy, which involves decorating the surface of cancer cells with phagocytosis-inducing TLR-stimulating microbial antigens. We hypothesized that this strategy would induce macrophages to recognize cancer cells as bacterial pathogens, thereby triggering inflammatory responses and phagocytosis.Among the surface-localized TLRs, TLRs 2, 4, and 5, which are potent inducers of inflammation and phagocytosis,22,23 were determined to be targets. Staphylococcus aureus (SA) (for TLR224) and Escherichia coli (EC) (for TLRs 425 and 526,27) were selected for bacterial membrane extraction to obtain TLRs 2-, 4-, and 5-agonists. This would allow for multiple PRRs to be engaged upon interaction with particles containing PAMPs and induce inflammatory responses (Fig. 1a).Fig. 1Full size imageBacterial membrane extracts stimulate various PRRs and activate macrophages. a Schematic illustration of activated macrophages stimulated via various PRR pathways upon recognizing BME particles. b Macrophages were treated with different concentrations of BME (2, 10, 20 μg/mL) or LPS (positive control), and the nitrite concentration was quantified using the Griess assay. Data are presented as the mean ± standard error (n = 4 biological replicates). c–e Involvement of various signaling pathways. Macrophages were pretreated with the denoted small molecule blockers with target pathways (c) and treated with BME (20 μg/mL). Proinflammatory cytokines (TNF-α (d) and IL-6 (e)) in the supernatant were quantified using ELISA. Neg negative control, Pos positive control. Data are presented as the mean ± standard error (n = 3 biological replicates). Untreated (PBS) or protumoral (induced with IL-4 treatment) macrophages were treated with 20 μg/mL BME for 6 h or 24 h, and the transcriptional levels of f–i M1 activation markers (Il6, Il1b, Tnfa, and Mhcii) and j, k M2 markers were analyzed using qRT‒PCR. Data are presented as the mean ± standard error (n = 4 biological replicates). P-values were determined using one-way ANOVA with Tukey’s post hoc analysisWe treated macrophages with SA, EC, and mixed (1:1 ratio) bacterial membrane extracts (BMEs) and evaluated both transcriptional and translational inflammatory responses. While the mixed formulation did not significantly upregulate inflammatory responses compared to the single formulations at the transcriptional level (Supplementary Fig. 1a–d), the secretion of representative inflammatory cytokines IL-6 and TNF-α increased at the translational level (Supplementary Fig. 1e, f). Compositional analysis of the mixed BME formulation showed that approximately 36 μg of total protein and 40 nmol of phospholipids were contained in 100 μg BME (Supplementary Fig. 2a–c). Endotoxin and exotoxin (TSST-1) levels were approximately 0.17 endotoxin units (EU) and below a few ng per 100 μg BME (Supplementary Fig. 2d, e), respectively, which were similar to or lower than the concentrations safely used in in vivo studies.28Compared to the well-established macrophage stimulant lipopolysaccharide (LPS), BME did not affect the viability of macrophages (Supplementary Fig. 3a, b) but significantly enhanced the concentration of nitrite, an inflammatory activation molecule (Fig. 1b). Nitrite production was dose-dependent on the BME (Supplementary Fig. 4).We investigated the mechanisms underlying the stronger macrophage-stimulatory effects of the BME. Specific PRRs or signaling pathways of macrophages were first blocked using small molecule inhibitors (Fig. 1c). Following BME treatment, TNF-α secretion was found to be primarily dependent on TLRs 2 and 5 (Fig. 1d), while IL-6 secretion was found to be dependent on multiple pathways, including the nod-like receptor (NLR), nucleotide-binding oligomerization domain (NLRP), and stimulator of interferon genes (STING) pathways, in addition to TLRs 2 and 5 (Fig. 1e), where the activation of the STING pathway has been demonstrated to be crucial in stimulating tumor-associated macrophages for antitumor functions and subsequent T cell priming.29While LPS is recognized by the TLR4 complex,30 BME recognition involves several signaling pathways. The involvement of multiple pathways may induce stronger activation of macrophages. These observations are consistent with those of previous studies, suggesting that multiple signaling pathways involving TLRs induce stronger inflammatory responses in a combinatorial manner.31,32,33BME treatment reprograms macrophages from protumoral to antitumoral phenotypesSince many tumor-resident macrophages display a protumoral phenotype,4 we assessed the ability of the BME to directly restore the antitumoral phenotype from the protumoral phenotype. We induced protumoral macrophages with IL-4 treatment34 and compared the response with that of untouched (naïve state, phosphate-buffered saline (PBS)) macrophages at 6 and 24 h after treatment. The transcription levels of inflammatory and activation markers (Il6, Il1b, Tnfa, and Mhcii) (Fig. 1f–i) and anti-inflammatory markers (Cd206 and Arg1) (Fig. 1j, k) were measured. Compared to naïve macrophages (PBS), protumoral macrophages (IL-4) expressed similar or lower levels of activation markers. When treated with BME, the expression of these markers was significantly higher in both groups, and the expression levels were not significantly affected by IL-4 treatment (Fig. 1f–i). The translational expression levels of IL-6 and TNF-α correlated with the above observations (Supplementary Fig. 5a–d). The expression of the protumoral macrophage markers Cd206 and Arg1 increased upon treatment with IL-4 but decreased upon treatment with BME (Fig. 1j, k). This suggests that the BME formulation can generate potent antitumor macrophages, irrespective of the current phenotype.BME particles associate with cancer cell membranesHaving confirmed the strong macrophage-stimulating effects of BME, we engineered bacteria-like cancer cells by decorating the surface of cancer cells with BME (Fig. 2a). To electrostatically enable the association of BME particles with cancer cell membranes, we added a positively charged synthetic lipid, 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP), which has been utilized for electrostatic interaction-mediated membrane fusion,35 to the BME formulation. The resulting BME particles had a hydrodynamic diameter of approximately 100 nm and a net positive zeta potential of approximately +25 mV (Supplementary Fig. 6a, b). Both scanning and transmission electron microscopy (SEM and TEM) images confirmed similar particle diameters with spherical morphologies (Supplementary Fig. 6c, d). We also investigated whether electrostatic assembly of BME and DOTAP induces any alterations in the formulation’s conformation or accessibility of PAMPs upon interaction with serum components. We first incubated BMEplus particles with serum and measured the hydrodynamic diameter, which had negligible effects (Supplementary Fig. 6a). Then, we treated macrophages with serum-incubated BMEplus. Quantification of secreted TNF-α and IL-6 concentrations showed that accessibility to PAMPs did not change after incubation with serum (Supplementary Fig. 6e, f). The positively charged BME formulation is referred to as BMEplus, and the charge-control formulation (neutral lipid instead of positively charged DOTAP) is referred to as BMEctrl. Neither the BME nor the positive charge of the BMEplus formulation induced direct cytotoxicity to cancer cells (Supplementary Fig. 7).Fig. 2Full size imageCharge-adjusted BME particles rapidly associate with cell membranes and are maintained. a Schematic illustration of positively charged BME (BMEplus) particles associated with cell membranes. b Confocal microscopy imaging of MC38 cells treated with fluorescent-tagged BMEplus. BMEplus: green, cell membrane: red, nucleus: blue. A similar pattern was observed in at least five biologically independent replicates. Scale bar: 20 μm. c Time-resolved association behavior of BMEplus with MC38 cells. Flow cytometry histogram plot. d Time-resolved association behavior of BME with MC38 cells measured using flow cytometry. Data are presented as the mean ± standard error (n = 3 biological replicates). e, f BMEplus-treated MC38 cells were washed, incubated for 2 days, and subsequently analyzed for BME association. e BME-positive cells (%) and f the amount of remaining BME over time. Data are presented as the mean ± standard error (n = 3–4 biological replicates). g, h Association mechanism of BMEplus particles. MC38 cells were pretreated with small molecule endocytosis blockers (cytochalasin D: phagocytosis and macropinocytosis inhibitor, dynasore: dynamin-mediated endocytosis inhibitor, nystatin: lipid raft-mediated endocytosis inhibitor) and treated with fluorescent-tagged (green) BMEplus. g Flow cytometry quantification (%) and h fluorescence microscopy observation of cells. Data are presented as the mean ± standard error (n = 4 biological replicates). P-values were determined via unpaired two-tailed t-test (d) or one-way ANOVA with Tukey’s post hoc analysis (e–g)To examine the morphological characteristics of BMEplus particle deposition, we incorporated a green fluorescent N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD) conjugate lipid. The cancer cell membrane and nucleus were stained red (CellMaskTM plasma membrane stain) and blue (Hoechst 33342 nucleus stain), respectively. Under confocal microscopy, most of the BMEplus signals colocalized with the plasma membrane signals (Fig. 2b), indicating that the PAMPs contained in the BME formulation were effectively delivered and present on the surface of the cancer cells. The microscopy images of the membrane association highly aligned with those of previous literature reporting membrane fusion strategies.35,36As measured using flow cytometry, the particles associated with cancer cell membranes rapidly in the BMEplus group, reaching approximately 100% within 30 min (Fig. 2c, d). In contrast, the association event was significantly slower in the BMEctrl group, demonstrating the effect of positive charge on cell association.We then investigated the duration of association between the particles and the cancer cells (Fig. 2e, f). The percentage of BME-positive cells at 24 h post-treatment with BMEplus was almost 100%, which then decreased to ~60% at 48 h (Fig. 2e). In contrast to the 100% BME-positive cells, the amount of residual BME in each cell decreased at 24 h post-treatment as the cancer cells proliferated (Fig. 2f). A similar association was observed in other cancer cell lines (Supplementary Fig. 8). Taken together, these results indicate that the BMEplus particles remain associated with cancer cells for tens of hours, making this a broadly applicable strategy, regardless of cell type.To determine whether this association was primarily surface association, as intended, we conducted an endocytosis inhibition assay (Fig. 2g, h). The major particle uptake pathways were blocked with small molecule endocytosis inhibitors (cytochalasin D - micropinocytosis/phagocytosis,37 dynasore - clathrin-mediated endocytosis,38 and nystatin - lipid raft/cholesterol-mediated endocytosis39) prior to BMEplus treatment. Subsequently, we observed 100% association across all the endocytosis inhibitor groups, thereby confirming that the association of BMEplus with cancer cell membranes is primarily due to electrostatic interactions between the negatively charged cell membranes and positively charged BMEplus particles, rather than endocytosis of the particles, indicating the decoration of cancer cell membranes with PAMPs.One concern might be the association of BMEplus with noncancerous cells, such as stromal cells or immune cells, including macrophages themselves, as membrane decoration is attributed to electrostatic interactions. However, stromal cells such as cancer-associated fibroblasts (CAFs) in the TME are reprogrammed into a tumor-promoting status.40 There is also literature that the removal of CAFs provides enhanced therapeutic efficacy,41,42 relieving concerns about stromal cell surface decoration. Then, to test how macrophages react to BMEplus treatment, we treated macrophages with NBD-tagged BMEplus particles and observed their morphologies as well as lysosome colocalization of the BME particles (Supplementary Fig. 9). Unlike when cancer cells were decorated where phagocytic morphologies were observed (Fig. 3g), the treatment to macrophages did not lead to these behaviors. Moreover, we observed that the particle signals were almost exactly colocalized with the lysosome signal, indicating that the BMEplus treatment to macrophages did not lead to phagocytosis of themselves, but the BMEplus particles were uptaken and processed.Fig. 3Full size imageSurface-decorated cancer cells induce the secretion of inflammatory cytokines and are phagocytosed by macrophages. a Schematic illustration of BMEplus-treated cancer cells coincubated with macrophages. MC38 cells treated with macrophages at 0, 24, and 48 h post-treatment with BMEplus and the secretion of proinflammatory cytokines in supernatants b IL-6 and c TNF-α were analyzed using ELISA. Data are presented as the mean ± standard error (n = 3 biological replicates). d–f BME-decorated cancer cells promote phagocytosis of macrophages. BME-decorated MC38 cells were coincubated with macrophages for 2 h. d Representative flow cytometry plots, e quantification of phagocytic macrophages, and f % of dead cancer cells among all cancer cells. Data are presented as the mean ± standard error (n = 3–4 biological replicates). g Representative confocal microscopy images (scale bar: 20 μm) and h Side scatter-area (SSC-A) values of macrophages after different treatments. MC38 cells were stained red, and macrophages were stained green. Data are presented as the mean ± standard error (n = 3–4 biological replicates). P-values were determined using one-way ANOVA with Tukey’s post hoc analysisBMEplus-decorated cancer cells induce inflammatory responses and enhance phagocytosis of macrophagesFollowing cell surface decoration, we examined the inflammatory and phagocytic response of macrophages toward surface-decorated bacteria-mimicking cancer cells (Fig. 3a).To examine the inflammatory response, we treated macrophages with PBS, cancer cells alone, BMEctrl-treated cancer cells, and BMEplus-treated cancer cells at 0, 24, and 48 h post-treatment with BMEctrl or BMEplus. The levels of secreted IL-6 and TNF-α were then analyzed (Fig. 3b, c). Both representative inflammatory cytokines displayed similar results, where cancer cells only did not induce any elevated secretory responses compared to the PBS group. The BMEctrl group induced slightly increased secretion of both cytokines only at 0 h. In contrast, the concentrations of both cytokines were significantly higher at 0, 24, and 48 h. A similar trend of cytokine induction was observed in another cancer cell line (Supplementary Fig. 10). These observations suggest that the surface-decorated BMEplus formulation is not only sustained but that the macrophage-stimulating function of the surface-decorated PAMPs is also well preserved.The phagocytosis flow cytometry analysis was also consistent with the above findings (gating in Supplementary Fig. 11a). At a macrophage-to-cancer-cell ratio of 1:2 (E:T ratio of 1:2), the percentage of phagocytic macrophages in the cancer-cell-alone and BMEctrl groups was approximately 20%, which increased to approximately 70% in the BMEplus group (Fig. 3d, e). Corresponding to % phagocytosis, BMEplus group macrophages showed the highest cytotoxicity against cancer cells (Fig. 3f). Over other E:T ratios of 1:1 and 2:1, BMEplus also demonstrated the highest phagocytic macrophage fraction (Supplementary Fig. 11b, c).To determine whether enhanced phagocytosis leads to enhanced antigen presentation, which is crucial for engaging adaptive immunity,4,5,6 we treated model antigen ovalbumin (OVA)-expressing MC38-OVA cells with PBS, BMEctrl, or BMEplus and treated them to macrophages. The macrophage population was quantified for H-2kb bound to the SIINFEKL peptide (major histocompatibility complex class 1 (MHCI)-OVA peptide complex) (Supplementary Fig. 11d, e). The fractions of OVA peptide-presenting macrophages did not differ between the PBS and BMEctrl groups, but the percentages were ~30% after 2 h and ~60% after 6 h in the BMEplus group. This indicates that, along with splenocyte activation by BME treatment, including T cell activation marker CD69 (Supplementary Fig. 12a–c), the BAMPIRE strategy involves adaptive immunity.Confocal microscopy images demonstrate almost no observable interactions between macrophages (green) and cancer cells (red) in the BMEctrl group. However, in the BMEplus group, most macrophages phagocytosed the cancer cells (Fig. 3g) in a fragmented manner, which correlated with an increased side scatter (SSC) value in flow cytometry (Fig. 3h).Locally injected BME decorates cell membranes only at the target regionWith enhanced inflammation and phagocytosis following the decoration of the cancer cell membrane with BMEplus in vitro, we assessed the in vivo applicability of the BAMPIRE strategy.Similar to Fig. 2b, BME particles were labeled green with NBD-lipids. BMEctrl or BMEplus was then locally injected into the tumor tissue; 24 h post-injection, the tumor was isolated, and tissue sections were stained with membrane and nuclear stains (red and blue, respectively) (Fig. 4a). BME fluorescence was not observed in the BMEctrl group, whereas a strong green fluorescence signal was observed in the BMEplus group. The BME signal overlapped significantly with membrane staining, as indicated by the intensity line profile along the arrow in the white rectangular area (Fig. 4b). Similarly, we intratumorally injected fluorescently labeled BMEctrl or BMEplus particles for biodistribution studies; 24 h post-injection, five major organs (lymph nodes, liver, spleen, kidney, and lungs) were homogenized, and the particle signal was measured (Fig. 4c). Signals were observed only at the tumor site, suggesting that the acute and strong inflammatory response demonstrated in vitro is a local response in vivo, avoiding the potential adverse effects of systemic inflammation. Histological analysis of these organs further supported the absence of systemic inflammation (Fig. 4d and Supplementary Fig. 13). Quantitative polymerase chain reaction (qPCR) of the tumor tissue homogenate (Fig. 4e) confirmed local responses represented by the upregulation of inflammatory cytokine genes, including il1b, il6, and il12a (Fig. 4f).Fig. 4Full size imageLocal injection of BMEplus, association behavior, and local responses. a Locally injected BMEplus formulation (green) displays colocalization with the cell membrane (red) under a fluorescence microscope, while BMEctrl does not show any signal (24 h post-local injection). Scale bar: 50 μm. b Line profile of BMEplus and cell membrane fluorescence signal colocalization in the white box in (a). The cell membrane signal (red line) and BMEplus signal (green line) show similar intensity profiles. c Particle localization in major organs 24 h post-local injection into tumor. Data are presented as the mean ± standard error (n = 3–4 biological replicates). d Histology analysis (H&E) of major organs 24 h post-local injection into tumor. Scale bar: 10 μm. e Tumor tissue cytokine analysis of 84 innate immune- and inflammation-related genes (qRT‒PCR). (n = 3 mice per group). f Volcano plot demonstrating significantly upregulated and downregulated genes in (e). P-values were determined using two-sided t-testLocally administered BMEplus induces macrophage-mediated antitumor effectsWe evaluated the in vivo therapeutic response of BME formulations using a mouse model. Mice were subcutaneously inoculated with MC38 murine cancer cells in the right flank, and either BMEctrl or BMEplus was locally injected. Injections were administered every other day from day 9, for a total of five injections (Fig. 5a). The every other day injection regimen was determined based on in vitro results where the fraction of BME-positive cancer cells decreased and had very little remaining BME on day 2 (Fig. 2e, f). The five injections were decided as the first humane endpoint was reached in some of the mice around day 19. Tumor volume monitoring revealed that both the BMEctrl and BMEplus formulations inhibited tumor growth; however, the effect was more pronounced with the BMEplus formulation (Fig. 5b). We analyzed the tumor tissues 24 h after the first injection of each modality. Hematoxylin & eosin (H&E) staining showed that tumor tissues from the PBS and BMEctrl groups had a similar morphology with intact nuclei, whereas the BMEplus group had fragmented and/or abnormal nuclear morphology (Fig. 5c). The results of the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay corresponded with those of the H&E staining assay, wherein a more fragmented DNA structure (green) was observed (Fig. 5d, e), indicating a more apoptotic environment.43Fig. 5Full size imageTherapeutic efficacy of the BME formulation following local injection and histological and transcriptome analysis. a Comparison of treatment regimens for BMEctrl and BMEplus and macrophage depletion mechanism study. The BME formulation was administered every 2 days for a total of five injections. Clodronate liposomes were given every 3 days for a total of three injections. b Tumor volume monitoring of BMEctrl and BMEplus formulations for evaluating therapeutic efficacy. Data are presented as the mean ± standard error (n = 5 mice for PBS, n = 8 mice for BMEctrl and BMEplus). c Histological analysis (H&E staining) of the tumor tissue 24 h after injection for each modality. d Histological analysis (TUNEL assay) of the tumor tissue 24 h after injection for each modality. Blue fluorescence: nucleus, green fluorescence: fragmented DNA. Scale bar: 50 μm. e Quantification of the TUNEL-positive area in (d). Two random points of view were quantified from a single tissue slice. Data are presented as the mean ± standard error (n = 2 mice for PBS and BMEctrl, n = 3 mice for BMEplus). f Tumor volume monitoring during treatment with the BMEplus formulation with and without macrophage depletion using clodronate liposomes for mechanistic study. Data are presented as the mean ± standard error (n = 3 mice for PBS and depletion + PBS, n = 4 for BMEplus and depletion + BMEplus). g Gene ontology analysis from sequencing data of isolated macrophages from the tumor tissue 24 h post-injection with BMEplus. h Heatmap gene expression analysis from sequencing. n = 3 mice per group. i Volcano plot analysis of upregulated and downregulated genes from sequencing results. n = 3 mice per group. P-values were determined using one-way ANOVA with Tukey’s post hoc analysis for (e) and two-way repeated-measure ANOVA with Bonferroni’s post hoc analysis for (b, f)In addition to the systemic toxicity tests (shown in Fig. 4c, d), we assessed the levels of serum inflammatory markers in this therapy regimen after five injections (Supplementary Fig. 14a). The results showed that the markers indicative of systemic damage (alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), lactate dehydrogenase (LDH), blood urea nitrogen (BUN), creatinine, and uric acid (UA)) did not change compared to those in the PBS control group (Supplementary Fig. 14b–h).To confirm whether the therapeutic effect originated from macrophages, we performed a macrophage depletion assay using commercial clodronate liposomes.44 Local macrophage depletion was performed three times as indicated (Fig. 5a). Macrophage-depleted tumors that received BMEplus treatments did not show any therapeutic effects and indeed grew at a similar rate to PBS-treated control tumors (Fig. 5f). This indicates that the therapeutic effects were mainly attributed to macrophages.Macrophages sorted from tumor tissue were subjected to transcriptome analysis. Gene ontology analysis revealed the upregulation of most inflammatory pathways, including innate immune response and chemokine/cytokine activity (Fig. 5g). A heatmap representation of gene expression showed the upregulation of immune response-related genes (Fig. 5h). The volcano plot represented 1097 significantly upregulated and 327 significantly downregulated genes (Fig. 5i). The heatmap and volcano plot of the transcripts are shown in Supplementary Fig. 15. Collectively, these results indicate that the responses of macrophages to PAMP-decorated cancer cells resembled their responses to bacteria.BME elicits synergistic antitumor effects in combination with conventional chemotherapeutic agentWe investigated whether the BMEplus formulation could synergize with one of the most widely available chemotherapeutic agents, DOX. BMEplus was administered locally every 2 days for a total of five times, and DOX liposomes were intravenously injected every 3 days for a total of three injections (Fig. 6a). The doses and dosing schedule were determined based on previous studies,45,46,47,48 and the dose was approximately half the maximum tolerated dose to avoid systemic toxicity and weight loss.49 The tumor volumes of DOX and BMEplus increased similarly over time with significant growth inhibition effects, whereas the DOX and BMEplus combination group demonstrated more pronounced therapeutic efficacy (Fig. 6b, c) with prolonged survival (Fig. 6d). The average tumor volumes of DOX and BMEplus were similar, but complete remission and median survival were more prominent in BMEplus (Fig. 6c, d). To evaluate systemic inflammation following this therapy regimen, we measured the concentrations of the cytokines IL-1β, IFN-γ, TNF-α, and IL-6 in blood after all therapeutic agent injections, which are associated with cytokine storm.50 The results show that all four groups (PBS, DOX, BMEplus, and Combi) did not show any symptoms of elevated systemic cytokine levels (mostly below the detection limit, a few pg/mL) (Supplementary Fig. 16a–d), confirming the therapeutic strategy’s safety profile.Fig. 6Full size imageTherapeutic efficacy of the BMEplus formulation combined with the conventional chemotherapeutic drug DOX. a Treatment regimen for BMEplus and DOX combined therapy in the MC38 tumor model. BMEplus was administered every 2 days for a total of five injections. DOX liposomes were administered every 3 days for a total of three injections. b Tumor volume monitoring during combination treatment in the MC38 tumor model. Data are presented as the mean ± standard error (n = 7 mice for PBS, n = 9 mice for DOX, n = 12 mice for BMEplus, and n = 12 mice for combi). c Individual tumor growth curves for (b), CR complete remission. d Long-term survival rate for (b). e Mice with complete remission from the BMEplus (n = 2 CR) and Combi (n = 3 CR) groups were rechallenged with MC38 cells. f Treatment regimen for combination therapy of BMEplus and DOX in a two-tumor MC38 model. The same regimen as that in (a) was applied to the primary tumor alone. The secondary tumor remained untreated. g Tumor volume monitoring of primary (treated) tumors. Data are presented as the mean ± standard error (n = 5 mice for PBS, n = 7 mice for DOX, n = 9 mice for BMEplus, and n = 11 mice for combi). h Tumor volume monitoring of secondary (untreated) tumors. Data are presented as the mean ± standard error (same number of mice as g). i Tumor volume monitoring during combination treatment in the CT26 tumor model. The same treatment regimen as that in (a) was applied from day 6. Data are presented as the mean ± standard error (n = 5 mice for PBS, n = 4 mice for DOX, n = 7 mice for BMEplus, and n = 4 mice for combi). j Long-term survival rate for (i). P-values were determined using two-way repeated-measure ANOVA with Bonferroni’s post hoc analysisAdditionally, we performed flow-based immune cell analysis. The results show that innate myeloid immune cells, including macrophages, DCs, and monocytes, were recruited to the tumor in the BMEplus group, indicating active innate immune responses toward pathogen-mimicking cancer cells due to BMEplus treatment. The relative abundance of CD86+ activated innate immune cell populations corresponded to this finding (Supplementary Fig. 17a–f). In terms of T cell immunity, there was an increased % of T cells in both the BMEplus and Combi groups, but the increase was much more prominent in the Combi group (Supplementary Fig. 17g, h). The relative abundances of memory phenotype T cells were also greater in the BMEplus and Combi groups (Supplementary Fig. 17i–l). The total splenic T cell population as well as the memory T cell population were mostly significantly increased in the Combi group (Supplementary Fig. 18a–g). These results collectively indicate that while the therapeutic efficacy of BMEplus can be attributed to macrophages and myeloid lineage cells, that of Combi strongly involves adaptive immunity as determined by enhanced T cell populations and activation.51When the therapeutic efficacy was compared with two types of established antibody-based immune checkpoint inhibitor (ICI) therapies using anti-CD47 (αCD47) and αPD-L1 antibodies, the Combi group showed the highest efficacy, while the BMEplus single treatment group still surpassed the ICI therapy groups (Supplementary Fig. 19a–c). Mice with complete remission from the BMEplus and Combi groups were rechallenged with MC38 cells (Fig. 6e). One in two mice from the BMEplus group and all three mice from the Combi group were protected against rechallenge, indicating the formation of immune memory.Given that macrophages are antigen-presenting cells that can coordinate innate and adaptive immunity,7 and considering that we found significant upregulation of Mhcii (Fig. 1h, i) and enhanced antigen presentation in the MHCI complex upon BME treatment (Supplementary Fig. 11d, e), we evaluated the abscopal effect to assess whether these observations could lead to systemic immunity in vivo. In this regimen, only the primary tumor was treated with BMEplus (Fig. 6f). Tumor volumes were affected in both primary (Fig. 6g) and secondary (Fig. 6h) tumors, indicating the development of systemic immunity following local administration of the BMEplus formulation.Finally, we tested whether our BAMPIRE strategy in combination with DOX could be used in other cancer models. We generated a CT26 colon cancer model as well as an orthotopic EMT6 triple-negative breast cancer model and administered the same treatment regimen (five BMEplus and three DOX injections). Similar to the results obtained in the MC38 cancer model (Fig. 6a), BMEplus alone had therapeutic effects, and synergism was observed in combination with DOX (Fig. 6i, j, Supplementary Fig. 20a, b). This suggests that our strategy may be applicable to other cancer types, as it does not require specific antigens from certain cancers or target certain cell death pathways.DiscussionAlthough macrophage-based platforms hold great promise for solid tumor immunotherapy, their clinical translation is significantly hindered by the highly immunosuppressive TME and the difficulty of securing sufficient numbers of functionally active macrophages ex vivo. In this study, we demonstrate macrophage-mediated solid cancer immunotherapy using the BAMPIRE strategy. Bacteria-derived PAMPs were attached to the surface of cancer cells to render them pathogen-like and were then phagocytosed by macrophages via PAMP-PRR recognition. Macrophages were highly phagocytic and inflammatory toward these surface-decorated cancer cells, enhancing phagocytosis by ~3.5-fold and releasing significantly elevated levels of inflammatory cytokines such as IL-6, IL-12, and TNF-α. Following the local injection of the BMEplus formulation into the tumor, the particles attached to the cell membranes and remained at the tumor site. Locally injected BMEplus particles showed therapeutic potential with tumor growth-inhibitory effects. Tumor remission and abscopal effects were observed when coadministered with the chemotherapeutic agent DOX.Cancer regression through combination therapy is desirable, as it can prevent the development of drug resistance and has the potential to synergize through compensatory antitumor mechanisms.51,52,53 We observed significantly enhanced antitumor efficacy in all three different animal models when BMEplus and DOX were used together compared to monotherapies of BMEplus or DOX. A combination of PRR agonists and immunogenic cell death (ICD) inducers, such as PAMPs and DOX, as in our study, can be an especially potent and compensatory combination. PRR agonists directly reprogram innate immune cells, including macrophages, into antitumor and inflammatory phenotypes,4,5,6,7 yet there may be insufficient killing due to the relatively low abundance of cancer antigens.51 ICD inducers, on the other hand, do not stimulate immune cells as strongly as PRR agonists. However, ICD of cancer cells releases cancer antigens, including damage-associated molecular patterns (DAMPs), often leading to adaptive antitumor responses.54 Thus, the use of PRR agonists and ICD inducers in combination allows both the activation of immune cells and the abundance of cancer antigens, resulting in durable and synergistic antitumor responses through two distinct mechanisms.There have been reports utilizing more than one PRR agonist, such as dual or synthetic triagonists,32,55,56 or bacterial outer membrane vesicles that contain a number of PRR agonists.57 These approaches significantly increased the activation of immune cells, including macrophages, compared to a single type of PRR agonist treatment. Our current work was in line with this, where there was greater macrophage activation when macrophages were treated with combined EC and SA extracts compared to EC or SA single treatment. While most strategies utilizing PRR agonists mainly focus on immune activation itself, our approach additionally attached these PRR agonists on the surface of cancer cells. This additional feature granted PRR agonists to act as direct macrophage phagocytosis inducers, leading to not only activating macrophages but also exerting direct phagocytic cytotoxicity as well as antigen presentation, achieving durable antitumor responses. For dosing and injections, our study employed repeated dosing and intratumoral injections, which might not be the most convenient and accessible way to deliver therapeutic cargos. However, both repeated dosing and intratumoral injections are common and standard in preclinical as well as clinical studies. For example, FDA-approved oncolytic virus T-VEC therapy requires repeated intratumoral injections.58 There are also many approved and trial-stage therapies requiring repeated dosing and local injections.59,60,61 Real-time ultrasound image-guided monitoring of local injections also greatly improves the clinical relevance of local injections of deep tissues and is used in the clinic.62 Therefore, repeated dosing and local injections can efficiently be managed to achieve clinically sound treatment regimens.Recently, there have been a large number of studies utilizing bacterial systems – not only bacteria-derived materials such as PAMPs but also intact dead/attenuated as well as live bacteria. Moreover, many clinical trials employing TLR agonists and bacteria themselves are ongoing.14,63,64,65 In our study, damage to major organs, systemic inflammation, or body weight loss was not observed following local injections of the BMEplus formulation. As evaluated in our and others’ studies, the use of bacterial materials can be a promising approach to developing cancer treatment strategies without significant side effects. Our BME formulation from independently prepared batches had a narrow distribution of protein and lipid compositions as well as bacterial toxins, emphasizing concerns about batch consistency, immunogenicity, and safety. For clinical aspects, establishing compositional standardization metrics such as specific protein concentration and/or endotoxin concentration would further enhance the translational relevance of the formulation. Determining the optimal PRR agonist formulation would also provide valuable insight into designing potent cancer therapeutics. The variables can include the number and type of PRRs stimulated, as exemplified in dual or triagonist systems,32,56 and/or the number of different agonists stimulating the same type of PRR. Identifying this optimal formulation would allow for the selection of only required PRR agonists and thus the development of synthetic systems with fully known compositions for enhanced clinical relevance. Moreover, the utilization of the PAMP-PRR mechanism can be broadly applied to various types of cancers, as this strategy does not target specific antigens such as CD19 or HER2 in CAR therapies. Our study demonstrated this broad applicability by attaching BMEplus to various cancer cell lines as well as by performing an in vivo study with multiple cancer models.Overall, our work provides insights into the development of macrophage-mediated cancer therapy approaches, where reprogramming or restoring antitumoral phenotypes of macrophages from protumoral phenotypes is a desirable outcome in view of potential phagocytosis and inflammatory functions,66 and targeting the abundance of cancer cells as a therapeutic strategy holds promise for cancer immunotherapy.67,68,69 The synergistic efficacy observed in combination with DOX emphasizes the value of integrating innate immune stimulation with therapies that increase tumor antigen availability through immunogenic cell death. Importantly, because this approach does not depend on targeting specific antigens, it may offer broad applicability across diverse solid tumor types. Collectively, these findings suggest that engineering tumor cells to become more readily recognized by innate immune cells represents a promising direction for cancer immunotherapy, and further optimization of PRR agonist composition, formulation standardization, and delivery strategies may enable the development of durable and clinically effective macrophage-mediated therapeutic platforms for solid cancers.Materials and methodsAnimalsAll animal experiment protocols were approved by the Institutional Animal Care and Use Committee (IACUC) (INHA 230827-886) of Inha University. All animals were purchased from Orient Bio Inc. (Charles River, Korea). Animals were housed at the university-operated housing facility and monitored daily by a professional. Animals were additionally monitored by the researchers at least once every two days and euthanized upon reaching predefined humane endpoints approved by the IACUC, including significant body weight loss or excessive tumor burden. The tumor volumes were measured using calipers. Tumor volumes were calculated as {(length) × (width)2 × 0.5}, where length and width denote the longest and shortest diameters of the tumor, respectively. Bone marrow cells collected from the tibia and femur were flushed using syringes and strained with a 70 μm cell strainer. Red blood cells were lysed with RBC lysis buffer (#420301, BioLegend). The resulting cells were handled as described in the “Cells and cell culture” section.Cells and cell cultureB16F10 and CT26 cells were purchased from the Korean Cell Line Bank (KCLB, Seoul, Korea). MC38 and MC38-OVA cells were maintained in Roswell Park Memorial Institute medium (RPMI 1640, #22400105) supplemented with 10% fetal bovine serum (FBS, #10082147) and 1% P/S (v/v) (#15140122), while other cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (#11995073) supplemented with 10% FBS and 1% P/S (v/v). All media and supplements were purchased from GIBCO (Carlsbad, CA, USA). Cells were incubated at 37 °C in a 5% CO2 atmosphere. Bone marrow cells were differentiated into macrophages in macrophage medium (RPMI 1640) supplemented with 10% FBS, 1% P/S, and 20 ng/mL macrophage colony-stimulating factor (M-CSF) (#576404, BioLegend, San Diego, CA, USA). Macrophage media were exchanged on day 3. Only the adherent cell fraction was used on day 6.Flow cytometry and antibodiesSingle-cell suspensions of cell samples were prepared in cell staining buffer (#420201, BioLegend). Viability staining was performed using Zombie NIR fixable dye (#423105, BioLegend) for 15 min at room temperature. The cell suspension was treated with an anti-mouse CD16/32 antibody (1:50 dilution) (clone 93, BioLegend) for 10 min at room temperature prior to antibody staining. PerCP anti-mouse F4/80 antibody (clone BM8), fluorescein isothiocyanate (FITC) anti-mouse I-A/I-E (clone M5/114.15.2), Pacific Blue anti-mouse CD69 (clone H1.2 F3), and PerCP anti-mouse CD86 (clone GL-1) antibodies were all purchased from BioLegend, San Diego, CA, USA. Flow cytometry and cell sorting were performed using a FACSMelody (BD Biosciences). For immunological analysis, tumors and spleens were isolated on day 22. Tissues were made into single-cell suspensions using a gentleMAC dissociator (Miltenyi Biotec). Samples were stained with a viability dye, blocked with anti-mouse CD16/32 antibody, and stained with fluorophore-tagged antibodies (Supplementary Table 1). Stained samples were analyzed with an Attune Cytpix flow cytometer (Thermo Fisher).Bacterial membrane extraction70E. coli (ATCC #25922) and S. aureus (ATCC #29213) were grown on Luria–Bertani (LB) medium for 12 h at 37 °C, 200 rpm. Bacterial cells were purchased from ATCC, USA. The bacterial cell concentration was adjusted to an optical density (OD)600 of 0.7 and harvested by centrifugation (8000 × g, 5 min). The supernatant was discarded, and the cell pellet was resuspended with 200 μg/mL lysozyme (#L4919, Sigma‒Aldrich, St. Louis, MO, USA) (in 200 mM Tris-HCl, pH 8.0) and kept at room temperature for 10 min. The same volume of 10 μg/mL DNase I (#11284932001, Roche, Mannheim, Germany) (in 50 mM Tris-HCl, 2% Triton, pH 8.0) was added and incubated at 4 °C for 30 min. The samples were centrifuged at 1633 × g for 5 min. The supernatant was collected and centrifuged at 22,967 × g for 10 min. The pellet was resuspended in excess PBS and centrifuged at 22,967 × g for 10 min. The supernatant was discarded, and the pellets were collected in the required volume of PBS. The centrifugation temperature was maintained at 4 °C. The absorbance was measured using a V-730 spectrophotometer (Jasco, Japan).BME charge adjustments and membrane associationDOTAP was first deposited as a thin film in a glass vial via argon purging. The extracted BME was mixed with DOTAP at the indicated ratios, and the mixture was vigorously vortexed. The mixture was sonicated (30 s, 2/2 s pulse, 25% amplitude) on an ice block using a probe-tip sonicator (VCX500; Sonics & Materials, Newtown, CT, USA). A BME:DOTAP ratio of 1:1.67 (wt.) was used for all subsequent experiments. For association quantification experiments, 18:1 NBD-PE was added at a 1% molar ratio to DOTAP. The same amount of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) was used for charge-control particles. The resulting fluorescent-tagged particles were added to cell-seeded wells (2 × 105 cells overnight in 6-well plates) at 20 μg/mL BME concentration. After incubation for 4 h, the medium containing free (unassociated) particles was removed, the cells were washed three times with PBS, and fresh medium was added. Cells were harvested for flow cytometry quantification at the denoted time points (0, 24, and 48 h). For the assessment of BMEplus treatment on macrophages, 2 × 105 macrophages were seeded in 6-well plates and treated the same way as above. After incubation and washing, macrophages were stained with LysoTracker (#L7528, Invitrogen). All lipids (DOTAP, DOPC, DPPC, 18:1 NBD-PE) were purchased from Avanti Polar Lipids (Alabaster, AL, USA).Primers (5′ ->3′)Il6 forward/reverse: TGCTGGTGACAACCACGGCC/GCCACTCCTTCTGTGACTCCAGCArg1 forward/reverse: ACACTCCCCTGACAACCAGC/AGGGTCTACGTCTCGCAAGCIl10 forward/reverse: ATCATGCCTGGCTCAGCACT/AAGGCAGTCCGCAGCTCTAGIl1b forward/reverse: CTGCAGTGGTTCGAGGCCCTA/GGACAGCCCAGGTCAAAGGTTnfa forward/reverse: GGGACTAGCCAGGAGGGAGA/AGTTCCACGTCGCGGATCATMhcii forward/reverse: GAAAGGACCTCGAGGCCCTC/CTCTGAGTCGGAGTGGCAGGIl12p40 forward/reverse: TGCAGGGTCCGATCCTAGGA/ATTAGCTCCCTGGCTCTGCG18S rRNA forward/reverse: ATGGTAGTCGCCGTGCCTAC/CGGGAGTGGGTAATTTGCGCPrimers were purchased from Integrated DNA Technology, Singapore.Griess assayThe Griess assay was performed according to the manufacturer’s instructions (Griess Reagent Kit, #G7921; Invitrogen, Eugene, OR, USA). The components N-(1-naphthyl)ethylenediamine and sulfanilic acid were mixed in a 1:1 (vol.) ratio to obtain Griess Reagent. In each well of a 96-well plate, 130 μL of deionized water, 150 μL of the sample, and 20 μL of Griess Reagent were added and incubated for 30 min at room temperature. For blank measurement, Griess Reagent was added to 150 μL of media mixed with 130 μL of deionized water immediately before the measurement. The absorbance was measured at 548 nm using a microplate reader (Varioskan Lux; Thermo Fisher Scientific, Bothell, WA, USA). For sample preparation, macrophages were seeded in a 24-well plate at 105 cells in 1 mL of medium and incubated overnight. Then, 100 μL of PBS containing BME or LPS (#L2880, Sigma‒Aldrich) was added to the wells. The supernatant was collected, and the assay was performed as described above.Macrophage transcriptional and translational response to BMEOne milliliter of 105/mL bone marrow-derived macrophages (BMDMs) was seeded per well in a 24-well plate. To induce the protumor phenotype, IL-4 (I1020, Sigma‒Aldrich) was added at a final concentration of 20 ng/mL. After 48 h of incubation, BME was added at a final concentration of 20 μg/mL. After 6 h or 24 h of incubation, the supernatant was analyzed using enzyme-linked immunosorbent assay (ELISA) (IL6 (#88-7064-88), and TNF-α (#88-7324-88), Invitrogen), and the RNA was extracted from cells and further processed for quantitative real-time polymerase chain reaction (qRT‒PCR).qRT‒PCRTotal RNA was extracted using commercial kits (#9767, Takara, Shiga, Japan; or #K-3140, Bioneer, Daejeon, Korea) according to the manufacturer’s instructions. cDNA was synthesized (#RR047A, Takara, Japan, or #K-2249, Bioneer, Daejeon, Korea) using 500 ng of extracted total RNA, followed by RT‒PCR reaction (#K-6252, Bioneer). Forward and reverse primers (10 μM, 0.8 μL) were mixed with the synthesized cDNA (1 μL of total cDNA reaction mixture). Reactions were performed using the CFX Opus 96 RT‒PCR system (Bio-Rad) for 40 cycles, and mRNA expression was calculated through the ΔΔCt model.71Endocytosis inhibition assayA total of 2 × 105 MC38 cells were seeded in each well of a 6-well plate. After overnight incubation, the cells were pretreated with cytochalasin D (#C8273, Sigma‒Aldrich), dynasore (#D7693, Sigma‒Aldrich), or nystatin (#N1400000, Sigma‒Aldrich) for 30 min. Fluorescently tagged (with NBD-PE, 1 mol%) BMEplus was added at a 20 μg/mL concentration. After incubation for 30 min, the cells were analyzed using flow cytometry for NBD-positive cells or imaged under a fluorescence microscope (Evos M7000, Invitrogen) after staining with Hoechst 33342 (H3570, Thermo Fisher Scientific).Signaling pathway blockadeMacrophages were plated at a density of 105 cells/well in a 24-well plate and incubated overnight. The cells were treated with PRR antagonists such as TH1020 (#HY-116961), ODN 24991 (#HY-150746), TLR4-IN-C34 (#HY-107575), C29 (#HY-100461), CY-09 (#HY-103666), NOD-IN-1 (#HY-100691), and H-151 (#HY-112693) at final concentrations of 2, 1, 10, 70, 5, 15, and 0.5 μM and incubated for 30 min. BME was added to the wells at a final concentration of 20 µg/mL. Supernatants were collected at 6 h after treatment and analyzed using ELISA. All PRR antagonists were purchased from MedChem Express (Monmouth Junction, NJ, USA).Phagocytosis analysisFor the phagocytosis assay, macrophages were seeded at a density of 2 × 105 cells/well in a 6-well plate in 1 mL of RPMI 1640 medium (10% FBS, 1% P/S, and 10 ng/mL M-CSF). After overnight incubation, the cells were stained with 1 μM CellTracker Green CMFDA (#C7025, Invitrogen) for 15 min in an incubator. Stained macrophages were washed, and 1 mL of RPMI 1640 medium (supplemented with 10% FBS, 1% P/S, and 10 ng/mL M-CSF) was added. MC38 cells were collected and stained similarly with 5 μM CellTracker Red CMTPX (#C34552, Invitrogen). After washing, 2.5 × 106 MC38 cells (in PBS) were transferred to a tube, and PBS, BMEctrl, or BMEplus was added at a final concentration of 50 µg/mL in 1 mL. After 15 min of incubation at room temperature, the cells were washed and added to macrophage-seeded wells at the indicated densities. After 2 h of incubation, adherent cells were scraped off, stained with a fixable viability dye, and fixed with 4% paraformaldehyde PFA for 20 min at room temperature. The fixed samples were analyzed using a cytometer to determine the percentage of phagocytic macrophages as well as the percentage of dead cancer cells among all cancer cells to evaluate direct cytotoxicity. The same procedure was repeated for confocal imaging, except that the cells were plated on a confocal imaging dish. Confocal images were acquired using an LSM 980 microscope.Macrophage ovalbumin antigen peptide presentationMacrophages were seeded in a 6-well plate at 2 × 105 cells in 1 mL of RPMI 1640 medium (10% FBS, 1% P/S, and 10 ng/mL M-CSF) and incubated overnight. Then, 4 × 105 MC38-OVA, MC38-OVA-BMEctrl, or MC38-OVA-BMEplus cells (prepared as described for the phagocytosis assay) were added, and the plates were incubated for 2 h or 6 h. Cells were scraped off and stained with APC anti-mouse H-2Kb bound to the SIINFEKL antibody (BioLegend #141606, clone 25-D1.16, 50:1 dilution).Cytokine secretion analysisMC38 cells were prepared as described in the “BME charge adjustments and membrane association” section, except that no NBD-tagged lipid was used. Cells harvested at different time points after membrane association (0, 24, and 48 h) were added to macrophage-seeded wells (2 × 105 cells). After 6 h of coincubation, the supernatant was collected and analyzed for IL-6, TNF-α, and IL-12 (#88-7121-88, Invitrogen) using ELISA, following the manufacturer’s instructions.Doxorubicin liposomesDoxorubicin liposomes were prepared using the thin-film method. Initially, DPPC and cholesterol (#C8667; Sigma‒Aldrich) were dissolved in chloroform at a molar ratio of 2:1. The solvent was evaporated under argon gas, and the remaining solvent was completely removed under vacuum. The resulting film was hydrated with an (NH4)2SO4 solution adjusted to pH 5.5, followed by 11 cycles of extrusion through a 100 nm polycarbonate membrane. The buffer was subsequently replaced with PBS using a PD-10 desalting column (Cytiva). Doxorubicin hydrochloride (EDQM) was loaded into the liposomes by mixing them with the prepared solution at a drug-to-lipid ratio of 1:10 and incubating them for 1 h at 47 °C. To remove unencapsulated doxorubicin, the solution was passed through a PD-10 desalting column, and the encapsulated doxorubicin content was quantified using UV‒vis spectrometry at 480 nm.Tumor models and treatmentsFor the MC38 tumor models, 6–8-week-old male C57BL/6 mice were inoculated with 5 × 105 MC38 cells in the right flank on D + 0. For the macrophage depletion study, clodronate liposomes (#F70101C-N; FormuMax Scientific, Sunnyvale, CA, USA) were intratumorally injected on D + 8, D + 11, and D + 14. For therapy, treatments began from D + 9. For the two-tumor models, 3 × 105 MC38 cells were inoculated in the left flank. For the CT26 tumor models, 6–8-week-old female BALB/c mice were inoculated with 1 × 106 CT26 cells in the right flank on D + 0. Treatments began from D + 6. For EMT6 orthotopic tumors, 6- to 8-week-old female BALB/c mice were inoculated with 5 × 105 EMT6 cells in the 2nd mammary fat pad. Treatments began from D + 6. PBS, BMEctrl, or BMEplus was intratumorally injected every 2 days for a total of 5 times, 20 μg per injection. For the combination therapy, doxorubicin liposomes were injected into the tail vein at 4 mg/kg every 3 days for a total of 3 times. For anti-PD-L1 therapy, 200 μg of atezolizumab was intraperitoneally injected on days 9, 12, 15, and 18 (total of 4 times). For anti-CD47 therapy, 50 μg of anti-CD47 antibody (#127518, clone miap301, BioLegend) was intratumorally injected on days 9, 11, 13, and 17 (total of 4 times). Injection routes, number of injections, and doses per injection for both ICIs were determined based on previous studies.44,72,73In vivo particle distributionAfter inoculation with 5 × 105 MC38 cells, 50 μL of PBS, BMEctrl, or BMEplus was intratumorally injected on D + 9. BMEctrl and BMEplus were prepared as described above, except for the addition of 1 mol% (to DOTAP) 18:1 NBD-PE for fluorescence labeling. After 24 h, the mice were euthanized, and the tumor, inguinal and axillary lymph nodes (total 4), spleen, liver, kidney, and lungs were isolated. Each organ was transferred to absolute ethanol and homogenized by mechanical grinding, followed by tip sonication. The homogenized tissues were centrifuged at 10,000 × g for 3 min. The supernatants were transferred to black well plates, and NBD fluorescence was read at 488 nm excitation and 533 nm emission on a microplate reader.Serum biochemical analysisBlood samples were collected from mice, transferred to serum separator tubes (SST) (BD Microtainer), and allowed to clot at room temperature for 30 min. After clotting, the tubes were centrifuged to separate the serum from cellular components. The resulting serum supernatant was collected and sent to Dooyeol Biotech, Korea, for further sample preparation and biochemical analysis.Tissue imagingFor confocal imaging, tumors were treated as previously described. Isolated tumor tissues were embedded in optimal cutting temperature (OCT) compound and immediately frozen at −80 °C. The cryopreserved tissues were sectioned into 10 μm slices and transferred to a slide glass. The tissues were washed with PBS, stained with Hoechst 33342 for 5 min, and then stained with CellMask Deep Red Plasma Membrane Stain (#C10046, Invitrogen) for 20 min. After washing, the stained tissues were embedded in mounting medium and imaged using an LSM 980 confocal microscope (Carl Zeiss). For histological analyses, the tumors were treated similarly, except that the injected BME particles did not contain 18:1 NBD-PE lipids. For hematoxylin & eosin (H&E) staining, the isolated organs were fixed in 4% PFA and sent to Celltis Bio (Cheongju, Korea) for further processing. For the TUNEL assay, the samples were stained and prepared by Histoire (Seoul, Korea), and the prepared TUNEL slides were imaged under an LSM 980 confocal microscope.Transcriptome analysisFollowing inoculation with 5 × 105 MC38 cells, 50 µL of PBS or BMEplus was intratumorally injected on D + 9. After 24 h, the tumor tissue was isolated and digested into a single-cell suspension by treatment with collagenase type IV (#C4-28; Sigma‒Aldrich) and DNase I. The cells were stained with a viability dye and anti-mouse F4/80 antibody. Live F4/80-positive macrophages were sorted using the FACSMelody system and immediately frozen. The samples were sent to Bioneer Inc. for further preparation and analysis. Briefly, a library was prepared using the SMARTer Stranded Total RNA-Seq kit v2–Pico input mammalian kit (Takara). Transcriptome sequencing was performed on an Illumina NovaSeq 6000, 150PE platform with 6 GB of data output per sample.Particle characterizationThe hydrodynamic diameter of the particles was measured using dynamic light scattering (DLS). DLS and zeta potential were measured using a Zetasizer Pro (Malvern Panalytical). SEM samples were prepared as previously described.74 Samples were deposited onto a cover glass and dried. The samples were Pt-coated using Q15T-S (Quorum Technologies) at 20 mA current for 120 s. The coated samples were imaged with SU8010 SEM (Hitachi) at an accelerating voltage of 15.0 kV. For TEM, samples were deposited onto a copper TEM grid. The samples were negative-stained with Uranyless (#22409, Electron Microscopy Sciences) and imaged with field emission (FE)-TEM (JEM2100F, Jeol).Software and statistical analysisFlow cytometry data were analyzed using the web-based tools floreada.io and FlowJo (v10.10.0). Volcano plots were created using a web-based tool developed by the Molecular and Genomics Informatics Core Facility (MaGIC; Rutgers NJMS, https://volcano.bioinformagic.tools/).75 All other flow cytometry graphs and schematic figures were created using Biorender.com with statistical analysis, except Fig. 4a, and tumor volume line graphs and corresponding statistical analysis (Origin Pro ver. 2024b (10.15)). ImageJ software was used for line profile extraction, as shown in Fig. 4b.Data availabilityThe sequencing data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE327369 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE327369). All other data necessary to evaluate the conclusions of this study are included in the article or the Supplementary Materials. Requests for further information and resources should be directed to and will be fulfilled by the corresponding authors.ReferencesLarson, R. C. & Maus, M. V. Recent advances and discoveries in the mechanisms and functions of CAR T cells. Nat. Rev. Cancer 21, 145–161 (2021).Article  PubMed  PubMed Central  Google Scholar Siegel, R. L., Giaquinto, A. N. & Jemal, A. Cancer statistics, 2024. CA Cancer J. Clin. 74, 12–49 (2024).PubMed  Google Scholar Torabi, S. et al. Immune cell-based therapies for solid tumors, current challenges and therapeutic advances. Cell Commun. Signal. 24, 73 (2025).Article  PubMed  PubMed Central  Google Scholar Mantovani, A., Allavena, P., Marchesi, F. & Garlanda, C. Macrophages as tools and targets in cancer therapy. Nat. Rev. Drug Discov. 21, 799–820 (2022).Article  PubMed  PubMed Central  Google Scholar Anderson, N. R., Minutolo, N. G., Gill, S. & Klichinsky, M. Macrophage-based approaches for cancer immunotherapy. Cancer Res. 81, 1201–1208 (2021).Article  PubMed  Google Scholar Li, D. & Wu, M. Pattern recognition receptors in health and diseases. Signal Transduct. Target. Ther. 6, 291 (2021).Article  PubMed  PubMed Central  Google Scholar Mogensen, T. H. Pathogen recognition and inflammatory signaling in innate immune defenses. Clin. Microbiol. Rev. 22, 240–273 (2009).Article  PubMed  PubMed Central  Google Scholar Nam, J. et al. Engineered polysaccharides for controlling innate and adaptive immune responses. Nat. Rev. Bioeng. 2, 733–751 (2024).Article  PubMed  PubMed Central  Google Scholar Yang, K. et al. Zinc cyclic di-AMP nanoparticles target and suppress tumours via endothelial STING activation and tumour-associated macrophage reinvigoration. Nat. Nanotechnol. 17, 1322–1331 (2022).Article  PubMed  Google Scholar Lee, D. Y. et al. Synthetic cationic helical polypeptides for the stimulation of antitumour innate immune pathways in antigen-presenting cells. Nat. Biomed. Eng. 8, 593–610 (2024).Article  PubMed  PubMed Central  Google Scholar Liang, J. et al. Outer membrane vesicle-based nanohybrids target tumor-associated macrophages to enhance trained immunity-related vaccine-generated antitumor activity. Adv. Mater. 35, 2306158 (2023).Article  Google Scholar Feng, Q. et al. Engineered bacterial outer membrane vesicles as controllable two-way adaptors to activate macrophage phagocytosis for improved tumor immunotherapy. Adv. Mater. 34, 2206200 (2022).Article  Google Scholar Wang, Y. et al. Engineering endogenous tumor-associated macrophage-targeted biomimetic nano-RBC to reprogram tumor immunosuppressive microenvironment for enhanced chemo-immunotherapy. Adv. Mater. 33, 2103497 (2021).Article  Google Scholar Hu, A. et al. Harnessing innate immune pathways for therapeutic advancement in cancer. Signal Transduct. Target. Ther. 9, 68 (2024).Article  PubMed  PubMed Central  Google Scholar Dooling, L. J. et al. Cooperative phagocytosis of solid tumours by macrophages triggers durable anti-tumour responses. Nat. Biomed. Eng. 7, 1081–1096 (2023).Article  PubMed  PubMed Central  Google Scholar Klichinsky, M. et al. Human chimeric antigen receptor macrophages for cancer immunotherapy. Nat. Biotechnol. 38, 947–953 (2020).Article  PubMed  PubMed Central  Google Scholar Kang, M. et al. Nanocomplex-mediated in vivo programming to chimeric antigen receptor-M1 macrophages for cancer therapy. Adv. Mater. 33, 2103258 (2021).Article  Google Scholar Shah, Z. et al. Human anti-PSCA CAR macrophages possess potent antitumor activity against pancreatic cancer. Cell Stem Cell 31, 803–817 (2024).Article  PubMed  PubMed Central  Google Scholar Lei, A. et al. A second-generation M1-polarized CAR macrophage with antitumor efficacy. Nat. Immunol. 25, 102–116 (2024).Article  PubMed  Google Scholar Ding, T. et al. Comprehensive review of macrophage models: primary cells and immortalized lines across species. Front. Immunol. 16, 1640935 (2025).Article  PubMed  PubMed Central  Google Scholar Kawai, T. & Akira, S. The role of pattern-recognition receptors in innate immunity: update on toll-like receptors. Nat. Immunol. 11, 373–384 (2010).Article  PubMed  Google Scholar Fu, Y. L. & Harrison, R. E. Microbial phagocytic receptors and their potential involvement in cytokine induction in macrophages. Front. Immunol. 12, 662063 (2021).Article  PubMed  PubMed Central  Google Scholar Wang, W. et al. A small secreted protein triggers a TLR2/4-dependent inflammatory response during invasive Candida albicans infection. Nat. Commun. 10, 1015 (2019).Dziarski, R. & Gupta, D. Staphylococcus aureus peptidoglycan is a toll-like receptor 2 activator: a reevaluation. Infect. Immun. 73, 5212–5216 (2005).Article  PubMed  PubMed Central  Google Scholar Raetz, C. R. H. & Whitfield, C. Lipopolysaccharide endotoxins. Annu. Rev. Biochem. 71, 635–700 (2002).Article  PubMed  Google Scholar Tan, A. et al. Evasion of toll-like receptor recognition by Escherichia coli is mediated via population level regulation of flagellin production. Front. Microbiol. 14, 1093922 (2023).Hajam, I. A., Dar, P. A., Shahnawaz, I., Jaume, J. C. & Lee, J. H. Bacterial flagellin-a potent immunomodulatory agent. Exp. Mol. Med. 49, e373 (2017).Article  PubMed  PubMed Central  Google Scholar Chen, M. Y. et al. Endotoxin-free outer membrane vesicles for safe and modular anticancer immunotherapy. ACS Synth. Biol. 14, 148–160 (2025).Article  PubMed  PubMed Central  Google Scholar Li, T. et al. Bioinspired magnetic nanocomplexes amplifying STING activation of tumor-associated macrophages to potentiate cancer immunotherapy. Nano Today 43, 101400 (2022).Article  Google Scholar Park, B. S. & Lee, J. O. Recognition of lipopolysaccharide pattern by TLR4 complexes. Exp. Mol. Med. 45, e66 (2013).Article  PubMed  PubMed Central  Google Scholar Wang, C. et al. Oncolytic mineralized bacteria as potent locally administered immunotherapeutics. Nat. Biomed. Eng. 8, 561–578 (2024).Article  PubMed  Google Scholar Albin, T. J. et al. Linked toll-like receptor triagonists stimulate distinct, combination-dependent innate immune responses. ACS Cent. Sci. 5, 1137–1145 (2019).Article  PubMed  PubMed Central  Google Scholar Rosenberger, K. et al. The impact of single and pairwise Toll-like receptor activation on neuroinflammation and neurodegeneration. J. Neuroinflammation 11, 166 (2014).Article  PubMed  PubMed Central  Google Scholar Orecchioni, M., Ghosheh, Y., Pramod, A. B. & Ley, K. Macrophage polarization: different gene signatures in M1(LPS+) vs. Classically and M2(LPS-) vs. Alternatively activated macrophages. Front. Immunol. 10, 1084 (2019).Article  PubMed  PubMed Central  Google Scholar Kolašinac, R., Kleusch, C., Braun, T., Merkel, R. & Csiszár, A. Deciphering the functional composition of fusogenic liposomes. Int. J. Mol. Sci. 19, 346 (2018).Article  PubMed  PubMed Central  Google Scholar Yang, J. et al. Drug delivery via cell membrane fusion using lipopeptide modified liposomes. ACS Cent. Sci. 2, 621–630 (2016).Article  PubMed  PubMed Central  Google Scholar Rennick, J. J., Johnston, A. P. R. & Parton, R. G. Key principles and methods for studying the endocytosis of biological and nanoparticle therapeutics. Nat. Nanotechnol. 16, 266–276 (2021).Article  PubMed  Google Scholar Macia, E. et al. Dynasore, a cell-permeable inhibitor of dynamin. Dev. Cell 10, 839–850 (2006).Article  PubMed  Google Scholar Ros-Baro, A. et al. Lipid rafts are required for GLUT4 internalization in adipose cells. Proc. Natl. Acad. Sci. USA. 98, 12050–12055 (2001).Article  PubMed  PubMed Central  Google Scholar Masuda, H. Cancer-associated fibroblasts in cancer drug resistance and cancer progression: a review. Cell Death Discov. 11, 341 (2025).Article  PubMed  PubMed Central  Google Scholar Das, S., Valton, J., Duchateau, P. & Poirot, L. Stromal depletion by TALEN-edited universal hypoimmunogenic FAP-CAR T cells enables infiltration and anti-tumor cytotoxicity of tumor antigen-targeted CAR-T immunotherapy. Front. Immunol. 14, 1172681 (2023).Article  PubMed  PubMed Central  Google Scholar Zhao, L. et al. Antitumor efficacy and potential mechanism of FAP-targeted radioligand therapy combined with immune checkpoint blockade. Signal Transduct. Target. Ther. 9, 142 (2024).Article  PubMed  PubMed Central  Google Scholar Darzynkiewicz, Z., Galkowski, D. & Zhao, H. Analysis of apoptosis by cytometry using TUNEL assay. Methods 44, 250–254 (2008).Article  PubMed  PubMed Central  Google Scholar Osorio, J. C., Smith, P., Knorr, D. A. & Ravetch, J. V. The antitumor activities of anti-CD47 antibodies require Fc-FcγR interactions. Cancer Cell 41, 2051–2065 (2023).Article  PubMed  PubMed Central  Google Scholar Jin, S. M. et al. A nanoadjuvant that dynamically coordinates innate immune stimuli activation enhances cancer immunotherapy and reduces immune cell exhaustion. Nat. Nanotechnol. 18, 390–402 (2023).Article  PubMed  Google Scholar Mastria, E. M. et al. Doxorubicin-conjugated polypeptide nanoparticles inhibit metastasis in two murine models of carcinoma. J. Control. Release 208, 52–58 (2015).Article  PubMed  PubMed Central  Google Scholar Yoo, Y. J. et al. Nanoengineered macrophages armed with TLR7/8 agonist enhance remodeling of immunosuppressive tumor microenvironment. Small 20, 2307694 (2024).Article  Google Scholar Lee, S. N. et al. Focused ultrasound-triggered doxorubicin liposomes reshape tumor microenvironment to boost checkpoint depletion in triple-negative breast cancer model. J. Control. Release 388, 114389 (2025).Article  PubMed  Google Scholar Aston, W. J. et al. A systematic investigation of the maximum tolerated dose of cytotoxic chemotherapy with and without supportive care in mice. BMC Cancer 17, 684 (2017).Article  PubMed  PubMed Central  Google Scholar Karki, R., Netea, M. G., Diorio, C. & Kanneganti, T.-D. Cytokine storm. Nat. Rev. Dis. Prim. 12, 1 (2026).Article  PubMed  Google Scholar Ye, J., Hou, B., Saeed, M., Xu, Z. & Yu, H. Engineering bioinspired nanomedicines to mitigate the resistance to cancer immunotherapy. Acc. Mater. Res. 3, 697–708 (2022).Article  Google Scholar Al-Lazikani, B., Banerji, U. & Workman, P. Combinatorial drug therapy for cancer in the post-genomic era. Nat. Biotechnol. 30, 679–692 (2012).Article  PubMed  Google Scholar Plana, D., Palmer, A. C. & Sorger, P. K. Independent drug action in combination therapy: implications for precision oncology. Cancer Discov. 12, 606–624 (2022).Article  PubMed  PubMed Central  Google Scholar Fucikova, J. et al. Detection of immunogenic cell death and its relevance for cancer therapy. Cell Death Dis. 11, 1013 (2020).Article  PubMed  PubMed Central  Google Scholar Schunke, J. et al. Co-delivery of STING and TLR7/8 agonists in antigen-based nanocapsules to dendritic cells enhances CD8+ T cell-mediated melanoma remission. Nano Today 57, 102365 (2024).Article  Google Scholar Tom, J. K. et al. Modulation of innate immune responses via covalently linked TLR agonists. ACS Cent. Sci. 1, 439–448 (2015).Article  PubMed  PubMed Central  Google Scholar Liu, G. et al. Bacteria-derived nanovesicles enhance tumour vaccination by trained immunity. Nat. Nanotechnol. 19, 387–398 (2024).Article  PubMed  Google Scholar Rehman, H., Silk, A. W., Kane, M. P. & Kaufman, H. L. Into the clinic: talimogene laherparepvec (T-VEC), a first-in-class intratumoral oncolytic viral therapy. J. Immunother. Cancer 4, 53 (2016).Article  PubMed  PubMed Central  Google Scholar Runcie, K. et al. Phase I study of intratumoral injection of talimogene laherparepvec for the treatment of advanced pancreatic cancer. Oncologist 30, oyae200 (2025).Article  PubMed  PubMed Central  Google Scholar Huppert, L. A. et al. Phase Ib study of intratumoral talimogene laherparepvec (T-VEC) in combination with chemotherapy or endocrine therapy in patients with advanced HER2-negative breast cancer. NPJ Breast Cancer 11, 130 (2025).Article  PubMed  PubMed Central  Google Scholar Jiang, M., Hu, Y., Lin, G. & Chen, C. Dosing regimens of immune checkpoint inhibitors: attempts at lower dose, less frequency, shorter course. Front. Oncol. 12, 906251 (2022).Article  PubMed  PubMed Central  Google Scholar Bitar, G. G. et al. Ultrasound-guided intra-tumoral administration of directly-injected therapies: a review of the technical and logistical considerations. Cancer Imaging 24, 145 (2024).Article  PubMed  PubMed Central  Google Scholar Rannikko, J. H. & Hollmén, M. Clinical landscape of macrophage-reprogramming cancer immunotherapies. Br. J. Cancer 131, 627–640 (2024).Article  PubMed  PubMed Central  Google Scholar Gulig, P., Swindle, S., Fields, M. & Eisenman, D. A review of clinical trials involving genetically modified bacteria, bacteriophages and their associated risk assessments. Appl. Biosaf. 29, 186–206 (2024).Article  PubMed  PubMed Central  Google Scholar Zhou, M. et al. Bacteria-based immunotherapy for cancer: a systematic review of preclinical studies. Front. Immunol. 14, 1140463 (2023).Article  PubMed  PubMed Central  Google Scholar DeNardo, D. G. & Ruffell, B. Macrophages as regulators of tumour immunity and immunotherapy. Nat. Rev. Immunol. 19, 369–382 (2019).Article  PubMed  PubMed Central  Google Scholar Ascic, E. et al. In vivo dendritic cell reprogramming for cancer immunotherapy. Science 386, eadn9083 (2024).Article  PubMed  PubMed Central  Google Scholar Zimmermannova, O. et al. Restoring tumor immunogenicity with dendritic cell reprogramming. Sci. Immunol. 8, eadn4817 (2023).Article  Google Scholar Yang, K., Halima, A. & Chan, T. A. Antigen presentation in cancer — mechanisms and clinical implications for immunotherapy. Nat. Rev. Clin. Oncol. 20, 604–623 (2023).Article  PubMed  Google Scholar Park, M. et al. Isolation and characterization of the outer membrane of Escherichia coli with autodisplayed Z-domains. Biochim. Biophys. Acta Biomembr. 1848, 842–847 (2015).Article  Google Scholar Livak, K. J. & Schmittgen, T. D. Analysis of relative gene expression data using real-time quantitative PCR and the 2-ΔΔCT method. Methods 25, 402–408 (2001).Article  PubMed  Google Scholar Li, M. et al. Next generation of anti-PD-L1 Atezolizumab with enhanced anti-tumor efficacy in vivo. Sci. Rep. 11, 5774 (2021).Article  PubMed  PubMed Central  Google Scholar Liu, X. et al. CD47 blockade triggers T cell-mediated destruction of immunogenic tumors. Nat. Med. 21, 1209–1215 (2015).Article  PubMed  PubMed Central  Google Scholar Lee, D. et al. Multicomponent-loaded vesosomal drug carrier for controlled and sustained compound release. Biomacromolecules 24, 3898–3907 (2023).Article  PubMed  Google Scholar Lemenze, A. MaGIC-Analytics/Magic-Volcanoes: V1 Release. (2024). https://doi.org/10.5281/zenodo.10845738Download referencesAcknowledgementsThis work was supported by a National Research Foundation of Korea (NRF) grant funded by the Ministry of Science and ICT (MSIT) (RS-2024-00353900, RS-2025-24534134) and by a grant from the Korea Environmental Industry & Technology Institute (KEITI) funded by the Ministry of Environment (MOE) (No. 2022002980004) of the Republic of Korea. Atezolizumab was kindly provided by Prof. Sang Taek Jung (Seoul National University, Korea).Author informationAuthor notesThese authors contributed equally: Seoyoon Song, Dongjun Yu, Haneul Kang.Authors and AffiliationsDepartment of Biological Sciences and Bioengineering, Inha University, Incheon, 22212, Republic of KoreaSeoyoon Song, Dongjun Yu, Haneul Kang, Deborah Lee, Jiye Lee, Sungjun Kwak, Jinhee Park, Yun Suk Huh, Sun Min Kim & Tae-Joon JeonWallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University School of Medicine, Atlanta, GA, 30332, USASeoyoon Song & Stanislav Y. EmelianovSchool of Pharmacy, Faculty of Medicine and Health, University of Sydney, Pharmacy and Bank Building A15, Sydney, NSW, 2006, AustraliaDeborah Lee, Lifeng Kang & Tae-Joon JeonSydney Nano Institute, University of Sydney, Sydney, NSW, 2006, AustraliaLifeng KangSchool of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, 30332, USAStanislav Y. EmelianovDepartment of Biomedical Sciences, College of Medicine, Inha University, Incheon, 22212, Republic of KoreaJong-Ho ChaBiohybrid Systems Research Center, Inha University, Incheon, 22212, Republic of KoreaJong-Ho Cha, Yun Suk Huh, Sun Min Kim & Tae-Joon JeonDepartment of Biological Engineering, Inha University, Incheon, 22212, Republic of KoreaYun Suk Huh & Tae-Joon JeonDepartment of Mechanical Engineering, Inha University, Incheon, 22212, Republic of KoreaSun Min KimDepartment of Medicinal Biosciences and Bioengineering, Inha University, Incheon, 22212, Republic of KoreaTae-Joon JeonAuthorsSeoyoon SongView author publicationsSearch author on:PubMed Google ScholarDongjun YuView author publicationsSearch author on:PubMed Google ScholarHaneul KangView author publicationsSearch author on:PubMed Google ScholarDeborah LeeView author publicationsSearch author on:PubMed Google ScholarJiye LeeView author publicationsSearch author on:PubMed Google ScholarSungjun KwakView author publicationsSearch author on:PubMed Google ScholarJinhee ParkView author publicationsSearch author on:PubMed Google ScholarLifeng KangView author publicationsSearch author on:PubMed Google ScholarStanislav Y. EmelianovView author publicationsSearch author on:PubMed Google ScholarJong-Ho ChaView author publicationsSearch author on:PubMed Google ScholarYun Suk HuhView author publicationsSearch author on:PubMed Google ScholarSun Min KimView author publicationsSearch author on:PubMed Google ScholarTae-Joon JeonView author publicationsSearch author on:PubMed Google ScholarContributionsS.S., D.Y., H.K., D.L., S.M.K. and T.-J.J. designed the study. S.S., D.Y., H.K., D.L., J.L., S.K. and J.P. performed the experiments. S.S., D.Y., H.K., and D.L. analyzed the data. L.K., S.Y.E., J.-H.C. and Y.S.H. contributed to data interpretation and provided resources. S.M.K. and T.-J.J. supervised the study. S.S. and D.Y. wrote the manuscript. All authors reviewed and approved the final manuscript.Corresponding authorsCorrespondence to Sun Min Kim or Tae-Joon Jeon.Ethics declarationsCompeting interestsThe authors declare no competing interests.Additional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary informationSupplemental Material (download DOCX )Rights and permissionsOpen Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.Reprints and permissionsAbout this article