Engineering a spatiotemporal macrophage circuit via STING phase separation to override immune suppression in pancreatic cancer

Significance

PDAC remains one of the deadliest malignancies, with over 80% of patients diagnosed at an advanced or metastatic stage, severely limiting treatment options. Through comprehensive analysis of single-cell RNA sequencing datasets, we identified a series of heterogeneous tumor-associated macrophage subpopulations coexpressing mannose receptor c-type 1 (MRC1) which represent the major immunosuppressive population in PDAC, whose abundance progressively increases from normal tissue to primary tumors and metastatic tumors. This work presents a therapeutic strategy that leverages single-cell sequencing–guided immunoengineering and stimulator of interferon genes (STING) phase separation to overcome the immunosuppressive barriers of PDAC.

Abstract

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies, largely due to its highly immunosuppressive tumor microenvironment (TME), which fuels metastasis and resistance to immunotherapy. Through comprehensive analysis of single-cell RNA sequencing datasets, we identified multiple heterogeneous tumor-associated macrophage (TAMs) subpopulations as key regulators of PDAC progression, which coexpress MRC1 and exert their effects by actively suppressing antitumor immune responses. To overcome this barrier, we developed a spatiotemporal macrophage reprogramming platform that leverages STING phase separation to reprogram TAM plasticity and reshape the immune landscape. This system, MRC1-targeting peptide–M@BLZ945 (PMMB), integrates a colony-stimulating factor 1 receptor (CSF-1R) inhibitor and a STING agonist within a macrophage-mimetic nanostructure, enabling sequential, controlled reprogramming of TAMs. By leveraging STING phase separation, PMMB stabilizes TAMs in an antitumor CD80⁺ phenotype while preventing excessive inflammation, achieving durable immune activation. In preclinical models, PMMB not only suppresses both primary and metastatic PDAC but also enhances CD8⁺ T cell infiltration, reinvigorates anti-PD-1 therapy responses, and mitigates immune exhaustion. These findings establish spatiotemporal macrophage circuit engineering via STING phase separation as a cross-scale strategy to override PDAC’s immune barriers and drive next-generation macrophage-targeted immunotherapy. This study paves the way for rationally designed, precision macrophage modulation strategies in solid tumors.

PDAC ranks among the most lethal and treatment-refractory malignancies, with a dismal 5-y survival rate below 10% (1–5). Its clinical course is driven by aggressive tumor biology, a propensity for early dissemination, and a predilection for liver metastasis, which collectively preclude curative interventions for most patients. Consequently, over 80% of cases present at an advanced or metastatic stage, leaving surgical resection as an option for only a small minority of patients (6–9). Current therapeutic strategies for metastatic PDAC provide minimal survival benefit, underscoring the urgent need for innovative approaches that effectively target both primary and metastatic disease (10–14).

A central barrier to effective therapy is the profoundly immunosuppressive tumor microenvironment (TME) that defines PDAC. In contrast to immunogenic tumors, PDAC is largely devoid of cytotoxic T cells and instead dominated by heterogeneous populations of tumor-associated macrophages (TAMs) (15–18). TAMs, enriched within both primary and metastatic lesions, orchestrate tumor progression through multiple complementary mechanisms (19–22). Their heterogeneity enables context-dependent protumorigenic functions: Certain subsets promote desmoplasia and vascular abnormalities, whereas others directly suppress antitumor immunity (23). As such, targeting TAMs represents a compelling therapeutic strategy to reprogram the TME and restore antitumor immunity for both primary and metastatic PDAC.

Despite their protumorigenic functions, TAMs exhibit remarkable plasticity, providing an opportunity for therapeutic reprogramming (24–26). The CSF-1/CSF-1R signaling axis critically regulates immunosuppressive TAMs by driving their recruitment, survival, and polarization toward phenotypes characterized by high expression of PD-L1, IL-10, and transforming growth factor-beta (TGF-β), thus sustaining an immune-excluded TME (27–29). However, CSF-1R inhibitors (e.g., BLZ945) alone are insufficient to achieve durable TAM reprogramming, as their effects are typically transient, and TAMs readily revert to an immunosuppressive state upon treatment withdrawal (30–32). Achieving durable TAM reprogramming therefore requires synergistic approaches. Targeting the cyclic GMP–AMP synthase (cGAS)–STING pathway offers a complementary mechanism, as its activation promotes TAM differentiation toward phenotypes with enhanced antigen-presenting capacity (e.g., upregulation of MHC-II and CD80) and stimulates CXCL10 secretion to recruit cytotoxic T cells (33, 34). Nevertheless, STING agonists alone risk immune overstimulation, leading to inflammation, immune cell death, and genomic instability.

To overcome these limitations, we performed an integrative analysis of scRNA-seq datasets from PDAC patients and identified a set of heterogeneous TAM subpopulations coexpressing MRC1, which represent the main immunosuppressive cells in PDAC and are closely associated with disease progression and poor prognosis (Fig. 1A). Based on this insight, we developed PMMB, an advanced macrophage-reprogramming platform specifically designed to reprogram MRC1⁺ TAMs (Fig. 1B). PMMB is composed of CD80⁺ macrophage membranes—modified with an MRC1-targeting peptide—as the targeted delivery scaffold for MRC1⁺ TAMs, encapsulating the CSF-1R inhibitor (BLZ945) and the STING agonist (MnO₂). The mechanism of action of PMMB involves three key steps: 1) TAM reprogramming via CSF-1R inhibition: BLZ945 shifts MRC1⁺ TAMs into CD80⁺ TAMs, restoring their antitumor function. 2) Sustained immune activation: CD80⁺ TAMs produce high levels of H₂O₂, triggering the release of Mn²⁺, which activates the STING pathway to enhance interferon (IFN) production and drive innate immune activation. 3) Controlled STING activation via phase separation: Mn²⁺ promotes STING phase separation, forming immune signaling compartments that amplify antitumor immunity while preventing excessive inflammation and immune cell death. This spatiotemporal strategy achieves durable TAM reprogramming and reshapes PDAC’s immune landscape, enabling sustained T cell infiltration and synergy with anti-PD-1 therapy. In preclinical models, PMMB suppresses primary and metastatic tumors, restores CD8⁺ T cell function, and enhances checkpoint blockade efficacy. By integrating precision TAM reprogramming with STING phase separation, our approach circumvents PDAC’s immune barriers, offering a powerful strategy to override immune suppression and improve immunotherapy outcomes in PDAC and other immunosuppressive tumors.

Fig. 1.

Schematic illustration of the spatiotemporal immunotherapy strategy. (A) Using scRNA-seq and spatial transcriptomics of primary and metastatic PDAC samples, we identified multiple subpopulations of TAMs as key regulators of disease progression. These subpopulations coexpress MRC1 and suppress antitumor immune responses. (B) PMMB was developed as an advanced macrophage-reprogramming platform leveraging spatiotemporal immune engineering and STING-mediated phase separation. By incorporating a CD80⁺ macrophage cell membrane coating and MRC1-targeting peptides, PMMB achieves precise delivery to MRC1⁺ TAMs. BLZ945 in PMMB modulates CSF-1R signaling in MRC1⁺ TAMs, potentially driving their phenotypic shift toward CD80⁺ TAMs. Concurrently, H₂O₂ triggers MnO₂ degradation, liberating Mn²⁺, which activates the STING pathway and induces IFN production. Mn²⁺ induces STING phase separation, mitigating excessive inflammation and immune toxicity. This spatiotemporal immunotherapy strategy synergistically combines CSF-1R inhibition and STING activation, leading to sustained TAM reprogramming and robust antitumor immune responses.

Results

Integrated Multiomics Profiling Reveals Specialized Macrophage Subsets as Architects of Pancreatic Cancer Progression.

We harmonized single-cell RNA sequencing (RNA-seq) data from two public GEO repositories (seven normal tissues, nine primary tumors, and four liver metastases), implementing rigorous batch correction to eliminate technical artifacts (Fig. 2A). Unsupervised clustering identified 31 transcriptionally distinct subpopulations across five major lineages (Fig. 2B and SI Appendix, Fig. S1). Cell types were annotated using established marker genes. Uniform Manifold Approximation and Projection (UMAP) visualization revealed epithelial, T and B lymphocyte, myeloid, and stromal cells within the pancreatic tumor microenvironment.

Fig. 2.

Workflow of sample collection, scRNA-seq, and spatial transcriptomics analysis of TAMs in PDAC (A) Schematic workflow of sample collection and data processing. (B) UMAP of the integrated scRNA-seq datasets comprising 31 clusters across five annotated cell types, colored by cluster identity. (C) UMAP of TAM populations from patients with PDAC. (D) Dot plot showing scaled expression of representative marker genes across TAM clusters. (E) Tissue prevalence of each TAM cluster estimated by Ro/e score. (F) Gene Ontology biological process (GO BP) enrichment was assessed by GSEA, with genes ordered according to log₂FC values comparing each TAM subset to the remaining TAMs. Normalized enrichment score (NES) denotes the normalized enrichment score. (G, Left) Spearman correlation between the proportions (%) of PDAC subgroups in 68 tumor regions and the proportions (%) of SPP1⁺ TAMs, MKI67⁺ TAMs, FOLR2⁺ TAMs, and MT⁺ TAMs. (Right) Spatial transcriptomics (Visium) analysis showing normalized mean expression of marker genes for SPP1⁺ TAMs, MKI67⁺ TAMs, FOLR2⁺ TAMs, MT⁺ TAMs, Ductal cell type 1, and Ductal cell type 3 in primary PDAC and liver metastasis regions. (H) Kaplan–Meier survival curves of PDAC patients (TCGA cohort) stratified by expression of SPP1⁺ TAMs, MKI67⁺ TAMs, FOLR2⁺ TAMs, and MT⁺ TAMs marker genes, normalized to CD68. Hazard ratios (HR) and P values were determined by Cox regression. (I–K) IHC imaging and quantification of Ki67 and MRC1 expression in samples from Stage I, II, and III PDAC patients. (Scale bar, 40 μm.) Statistical significance determined by one-way ANOVA with Fisher’s least significant difference (LSD) test.

Within this landscape, TAMs emerged as dominant regulators of pancreatic oncogenesis. High-resolution subclustering delineated six functionally distinct TAM subsets: SPP1⁺ TAMs, MT⁺ TAMs (mitochondrial-high), MKI67⁺ TAMs (proliferative), HSP⁺ TAMs (stress-responsive), FOLR2⁺ TAMs, and IL1B⁺ TAMs (proinflammatory) (Fig. 2C). Marker gene validation confirmed unique expression programs defining each subset (Fig. 2D). Spatiotemporal dynamics of TAMs specialization revealed metastatic niche adaptation: SPP1⁺, MT⁺, and MKI67⁺ subsets exhibited significant enrichment in liver metastases, while FOLR2⁺ TAMs dominated primary tumors (Fig. 2E). Furthermore, functional annotation revealed specialized pathogenic programs across TAM subsets (Fig. 2F). MKI67⁺ TAMs exhibited proliferative dominance through DNA replication, nuclear division, and cell cycle checkpoint activation, concurrently expressing immunosuppressive effectors that facilitate tumor immune evasion. FOLR2⁺ TAMs demonstrated paradoxical functions—displaying negative regulation of cell migration and clathrin-dependent endocytosis alongside tissue-remodeling markers—while actively modulating immune responses through adaptive immunity and leukocyte activation regulation, indicative of their context-dependent roles. SPP1⁺ TAMs drove desmoplastic progression via extracellular matrix-degrading enzymes, coupled with metabolic reprogramming toward glucose catabolism and oxidative phosphorylation to fuel invasion. MT⁺ TAMs manifested extreme metabolic specialization, amplifying oxidative phosphorylation and cytochrome complex assembly while activating protein targeting and ribosome biogenesis machinery to survive nutrient stress.

To further investigate the spatial architecture and potential interdependence between these TAM subsets and other cell types, we performed integrative spatial transcriptomic profiling (Visium) on samples from five primary PDAC and 10 metastatic lesions. Results showed that FOLR2⁺, SPP1⁺, and MKI67⁺ TAMs were preferentially colocalized with neoplastic Ductal cell 1 populations, whereas MT⁺ TAMs were enriched in regions containing Ductal cell 3 populations (Fig. 2G). These findings highlight a microenvironmental codependency between specific TAM subsets and ductal cell states, suggesting that the physical proximity of these cell types may facilitate reciprocal signaling that sustains tumor progression and metastatic potential. Consistently, CellChat analysis positioned these TAM subsets as central signaling hubs orchestrating tumor-stroma crosstalk (SI Appendix, Fig. S2). Ligand–receptor interrogation identified potentiated oncogenic pathways: a) FOLR2⁺ TAMs → Ductal cell 1: HBEGF-EGFR activation driving epithelial proliferation; b) SPP1⁺ TAMs →Ductal cell 1: TGFβ1-transforming growth factor beta receptor (TGFβR2)-mediated epithelial–mesenchymal transition; c) MKI67⁺ TAMs →Ductal cell 1: SLC16A3-MCT1 metabolic coupling supporting anabolic growth; d) MT⁺ TAMs →Ductal cell 3: RARRES2-CCRL2 chemotaxis promoting metastatic dissemination. Subsequently, we evaluated the clinical relevance of the four TAMs transcriptional programs in The Cancer Genome Atlas (TCGA) PDAC cohort and identified all corresponding gene signatures as independent prognostic indicators of poor survival, thereby substantiating their collective tumor-promoting role in disease progression (Fig. 2H).

Building upon our scRNA-seq findings, which demonstrated the tumor-promoting roles of MKI67⁺ TAMs, FOLR2⁺ TAMs, SPP1⁺ TAMs, and MT⁺ TAMs, we observed that MRC1 (CD206) was consistently and abundantly expressed across all four transcriptionally defined subsets (Fig. 2D). The uniformly high MRC1 expression in these phenotypically and functionally distinct TAM subsets highlights it as a unifying phenotypic hallmark of tumor-promoting TAMs and a potential therapeutic entry point for strategies aiming to broadly target immunosuppressive TAM populations in PDAC. Furthermore, we validated this signature in a clinical cohort of 30 PDAC patients encompassing a broad spectrum of disease stages. Immunohistochemical quantification demonstrated coordinated patterns of tumor cell proliferation (Ki67⁺) and MRC1⁺ macrophage infiltration (Fig. 2 I–K). The above findings establish MRC1 as a viable therapeutic target for macrophage modulation in PDAC.

Construction and Characterization of PMMB.

After validating the pivotal role of MRC1⁺ TAMs in PDAC progression and metastasis, we developed a spatiotemporal immunization strategy aimed at reprogramming MRC1⁺ TAMs for the treatment of metastatic PDAC (Fig. 3A). This strategy was realized through a nanodelivery system named PMMB. PMMB is composed of CD80⁺ macrophage membranes—modified with an MRC1-targeting peptide—as the targeting delivery scaffold, encapsulating the CSF-1R inhibitor BLZ945 and the STING agonist MnO₂ nanoparticles (MRC1-targeting peptide–M@BLZ945, PMMB). The preparation of PMMB was carried out as follows: 1) Bone marrow–derived macrophages (BMDMs) were isolated from mouse tibiae and polarized into CD80⁺ macrophages by stimulation with lipopolysaccharide (LPS) and interferon-gamma (IFN-γ). Cell membranes were extracted from CD80⁺ macrophages using ultracentrifugation. 2) MnO₂ were synthesized via a redox reaction method. 3) MnO₂ and BLZ945 were coencapsulated within CD80⁺ macrophage membranes via extrusion, and 1,2-distearoyl-sn-glycero-3-phosphoethanolamine (DSPE)-polyethylene glycol (PEG)–MRC1-targeting peptides were inserted into the membrane structure for specific recognition of MRC1⁺ TAMs. This design establishes PMMB as a targeted delivery platform for spatiotemporal reprogramming of MRC1⁺ TAMs.

Fig. 3.

Characterization of PMMB and in vitro assessment of the ability to induce durable reprogramming in MRC1⁺ macrophages. (A) Schematic representation of the procedure for the preparation of PMMB. (B) Transmission electron microscopy (TEM) images and size distribution histogram of PMMB. (Scale bar, 200 μm.) (C) XPS spectrum of PMMB. (D) The expression of protein (β-actin and INOS) on PMMB. (E) Detection of Mn²⁺ release from PMMB by inductively coupled plasma-mass spectrometry (ICP-MS) in the presence of H₂O₂. (F) Fluorescent imaging of MRC1⁺ macrophages incubated with Free-Rhb, Rhb@PMB, and Rhb@PMMB for 2 h. (Scale bar, 20 μm.) (G) Schematic representation of the assay for evaluating the reprogramming capacity of different formulations on MRC1⁺ macrophages. (H) Representative flow cytometry plots of CD80⁺ F4/80⁺ and MRC1⁺ F4/80⁺ macrophages across different treatment groups. (I) Quantification of the CD80⁺ F4/80⁺ macrophage population (n = 5). (J) Quantification of the MRC1⁺ F4/80⁺ macrophage population (n = 5). (K) Immunofluorescence analysis of STING phase separation in MRC1⁺ macrophages following PMMB treatment. (Scale bar, 5 μm.) (L) Schematic representation of the assay for evaluating the sustained reprogramming capacity of different formulations on MRC1⁺ macrophages. (M) Quantification of the population of CD80⁺ F4/80⁺ macrophages (n = 5). (N) Quantification of the population of MRC1⁺ F4/80⁺ macrophages (n = 5). (O and P) Representative flow cytometry plots with quantification of T cell apoptosis under different treatment conditions (n = 5). (Q and R) Representative flow cytometry plots with quantification of T cell proliferation under different treatment conditions (n = 5). (S and T) Enzyme-linked immunosorbent assay (ELISA) quantification of granzyme B and perforin released from T cells under different treatment conditions. (U) Heatmaps of TAMs-related genes and STING pathway-related genes in PMMB and PBS groups based on bulk RNA-seq. (V) GSEA enrichment of immune-related pathways in PMMB compared to phosphate buffered saline (PBS), DEGs used for GSEA were defined by |log₂fold change| > 1 and FDR q-value < 0.05. (W) GO enrichment factor analysis was conducted for DEGs between PMMB- and MMB-treated groups using the same DEG thresholds (|log₂fold change| > 1, FDR q-value < 0.05). Data are shown as mean ± SD; n represents the number of biologically independent samples. Statistical significance was calculated by one-way ANOVA with Fisher’s LSD test.

As shown in Fig. 3B and SI Appendix, Fig. S3 A and B, PMMB displayed a particle size of 146.5 ± 12.3 nm and a zeta potential of approximately −28.4 mV, consistent with parameters for effective tumor-targeted delivery. X-ray photoelectron spectroscopy (XPS) confirmed the structural integrity of MnO₂ nanoparticles within PMMB (Fig. 3C). Furthermore, sodium dodecyl sulfate polyacrylamide gel electrophoresis and western blot analyses demonstrated substantial retention of CD80⁺ TAMs membrane proteins on PMMB (Fig. 3D and SI Appendix, Fig. S3C). Stability studies revealed that PMMB maintained a consistent particle size and polydispersity index over 24 h under pH 7.4/PBS conditions, indicating excellent stability (SI Appendix, Fig. S3D). Mn²⁺ release from PMMB was investigated, with approximately 96% of Mn²⁺ released within 1 h of incubation in H₂O₂ (Fig. 3E). To evaluate the targeting efficiency of PMMB toward MRC1⁺ macrophages, BLZ945 was replaced with Rhodamine B (Rhb), generating Rhb@PMMB (with targeting peptide) and Rhb@MMB (without targeting peptide). Cellular uptake studies (Fig. 3F) revealed significantly enhanced uptake of Rhb@PMMB compared to Rhb@MMB, confirming the critical role of the targeting peptide in directing PMMB to MRC1⁺ macrophage. These findings underscore the efficacy of PMMB’s spatiotemporal immunization strategy in reprogramming MRC1⁺ macrophages, achieving stable immune activation, and transforming the tumor microenvironment, offering a promising therapeutic approach for metastatic PDAC.

In Vitro Evaluation of the Functionalities of PMMB.

In this study, PMMB exploits the tumor-targeting properties of CD80⁺ macrophage membranes and MRC1 targeting peptides to selectively and effectively localize MRC1⁺ TAMs within the tumor microenvironment. BLZ945, incorporated into PMMB, is proposed to modulate CSF-1R signaling in MRC1⁺ TAMs, potentially contributing to their phenotypic shift toward CD80⁺ TAMs. Concurrently, H₂O₂ released during this process triggers the degradation of MnO₂ nanoparticles, liberating Mn²⁺, which activates the STING signaling pathway and induces IFN production. This further reprograms TAMs, stabilizing their phenotypic transformation and amplifying the immune response. Remarkably, Mn²⁺ facilitates the formation of “STING phase separators” via phase separation of the STING protein. These specialized membrane structures sequester and regulate immune molecules, preventing excessive innate immune activation, immune inflammation, and immune cell death. This ensures PMMB’s spatiotemporal immunization strategy achieves robust, widespread, and sustained immune activation, effectively converting the tumor microenvironment into a hot phenotype and suppressing both primary and metastatic PDAC.

To further assess the functional effects of PMMB on MRC1⁺ TAMs in vitro, BMDMs were isolated from mouse bone marrow and induced to differentiate into MRC1⁺ macrophages. Subsequently, the cells were treated with PBS, PMB, PMM, or PMMB. PMM stands for MRC1-targeting peptide–modified CD80 macrophage membrane–coated MnO₂ (MRC1-targeting peptide–M@MnO₂). PMB stands for MRC1-targeting peptide–modified CD80 macrophage membrane–coated BLZ945 (MRC1-targeting peptide–M@BLZ945). Flow cytometry, RT-PCR, and RNA-seq were employed to systematically evaluate their phenotypic transformation and potential molecular mechanisms. (Fig. 3G).

The results analysis showed that PMMB effectively reprogrammed MRC1⁺ macrophages into CD80⁺ macrophages (Fig. 3 H–J). In the PBS group, 74.4% of macrophages retained the MRC1⁺ phenotype, while the PMMB group reduced this proportion to 20.8%. Conversely, the proportion of CD80⁺ macrophages increased from 6.6% in the PBS group to over 50% in the PMMB group, representing a 7.4-fold increase. Compared with the PMB and PMM groups, the proportion of MRC1⁺ macrophages decreased from 43.7% and 41.4 to 20.8% in the PMMB group, whereas the proportion of CD80⁺ macrophages was 1.6-fold higher. These results highlight the superior efficacy of PMMB compared to PMB and PMM, confirming the advantages of combining CSF-1R inhibitors (BLZ945) with STING agonists (Mn²⁺). PMMB’s ability to activate the cGAS-STING signaling pathway was further confirmed by increased expression of pathway-related genes and substantial IFN-1β release in MRC1⁺ macrophages following PMMB treatment (SI Appendix, Fig. S4 A and B). Confocal microscopy revealed that Mn²⁺ triggered STING protein phase separation, forming droplet-like aggregates on the endoplasmic reticulum membrane. This compartmentalization prevented excessive immune activation, ensuring a controlled and effective immune response (Fig. 3K). To assess the durability of MRC1⁺ macrophage reprogramming, we introduced KPC-conditioned medium (KPC-CM) to mimic the tumor microenvironment and evaluate phenotypic stability post–PMMB treatment (Fig. 3 L–N and SI Appendix, Fig. S4 C and D). While KPC-CM disrupted macrophage reprogramming in monotherapy groups (PMB and PMM), the PMMB group maintained a stable CD80⁺ macrophage population (~43.5%), indicating durable phenotypic transformation.

MRC1⁺ macrophages are known to induce T cell apoptosis, suppress proliferation, and impair cytotoxic function, thereby facilitating immune evasion (35, 36). The results demonstrated that the T cell apoptosis rate significantly decreased from 39.3% in the PBS group to 7.7% in the PMMB group, indicating that PMMB markedly alleviated MRC1⁺ macrophage-mediated immunosuppression (Fig. 3 O and P and SI Appendix, Fig. S4E). In addition, MRC1⁺ macrophages strongly suppressed the proliferation of T cells (only 11.0%), whereas PMMB treatment restored T cell proliferation to 47.1%, representing a 3.3-fold increase compared to the PBS group (Fig. 3 Q and R). Furthermore, levels of granzyme B and perforin in the T cell supernatants were significantly elevated in the PMMB group, approximately 3.6- and 6.1-fold higher than those in the PBS group, further suggesting a restoration and enhancement of T cell cytotoxic function (Fig. 3 S and T). In general, these findings indicate that PMMB enhances antitumor immunity by modulating the immunosuppressive phenotype of MRC1⁺ macrophages, thereby improving T cell immune responses and cytotoxicity.

Bulk RNA-seq was performed on MRC1⁺ macrophages to investigate transcriptomic changes induced by PMMB. This transcriptomic analysis provided deeper insights into PMMB-induced phenotypic transformation. A GO circle plot highlighted differentially expressed genes (DEGs) associated with immune processes between the PMMB and PBS groups (SI Appendix, Fig. S5A). Compared to the PBS group, PMMB treatment enriched pathways related to immune responses and immune system processes (SI Appendix, Fig. S5 B and C). Specifically, DEGs in the PMMB group were significantly enriched in pathways such as the immune response and nuclear factor kappa B (NF-κB) signaling pathways. Notably, genes associated with the MRC1⁺ macrophages were downregulated, while those linked to CD80⁺ macrophage and STING pathway activation were upregulated (Fig. 3 U–W). Gene Set Enrichment Analysis (GSEA) further confirmed heightened activity in immune-related pathways in the PMMB group compared to PBS (Fig. 3V).

Taken together, PMMB effectively reprogrammed MRC1⁺ macrophages into an antitumor phenotype through a spatiotemporal immunization strategy, while simultaneously reducing T cell apoptosis and enhancing antitumor immunity. These results suggest that PMMB may serve as an innovative macrophage-centered therapeutic modality with translational potential for PDAC diagnosis and therapy.

Immunotherapeutic Evaluation of PMMB in KPC Tumor–Bearing Mice.

Next, we examined the biodistribution of PMMB in PDAC mice to evaluate its tumor-targeting capabilities. In vivo imaging at various time points revealed that PMMB exhibited superior tumor accumulation compared to MMB (without MRC1 targeting peptide) attributable to the incorporation of CD80⁺ macrophage membranes, which enhanced tumor-targeting properties (SI Appendix, Fig. S6 A and B). Fluorescence imaging of isolated organ tissues further confirmed PMMB’s tumor-targeting efficiency (SI Appendix, Fig. S6 C and D). Quantification of Mn levels in tumors using ICP-MS and fluorescence spectrophotometry corroborated these findings, with peak accumulation consistently observed at 24 h (SI Appendix, Fig. S6E).

To evaluate the in vivo immunotherapeutic efficacy of PMMB, we administered intravenous injections via the tail vein of PBS, Free BLZ945, PMM, PMB, MMB, or PMMB to KPC tumor–bearing mice (Fig. 4A). PMM stands for MRC1-targeting peptide–modified CD80 macrophage membrane–coated MnO₂ (MRC1-targeting peptide–M@MnO₂). PMB stands for MRC1-targeting peptide–modified CD80 macrophage membrane–coated BLZ945 (MRC1-targeting peptide–M@BLZ945). MMB refers to a CD80 macrophage membrane–coated formulation of both MnO₂ and BLZ945 (M@MnO₂-BLZ945). PMMB demonstrated the most significant tumor growth inhibition, as evidenced by the smallest tumor volumes and weights among all groups, underscoring its superior anticancer efficacy (Fig. 4B and SI Appendix, Fig. S7 A and B). The combination therapy in PMMB, integrating BLZ945 and MnO₂, outperformed single-agent therapies (PMM and PMB), emphasizing the critical role of CSF-1R inhibition and STING pathway activation. Importantly, PMMB treatment significantly prolonged survival in tumor-bearing mice (Fig. 4C).

Fig. 4.

Antitumor efficacy and immunotherapeutic evaluation of PMMB in KPC tumor–bearing mice. (A) A schematic representation of the experimental workflow in the PDAC model. (B) Growth curves of KPC tumors in mice with different treatments (n = 5). (C) Survival percentage of KPC tumor–bearing mice under different treatment conditions (n = 8). (D and E) Representative flow cytometry plots of CD80⁺ F4/80⁺TAMs and MRC1⁺ F4/80⁺ TAMs (gated on CD45⁺CD11b⁺) in tumors of different treatment groups. (F and G) Quantification of the population of CD80⁺ F4/80⁺TAMs and MRC1⁺ F4/80⁺ TAMs in tumors of different treatment groups (n = 5). (H) Quantification of the population of CD8⁺ T cells (gated on CD45⁺CD3⁺) in tumors (n = 5). (I) Heatmaps showing the expression of selected TAM-related genes and STING pathway-associated genes in tumors from PMMB- and PBS-treated mice, as measured by bulk RNA-seq. (J) GO enrichment factor plot of PMMB versus PBS group of DEGs. (K) KEGG pathway enrichment analysis indicating immunomodulatory signaling pathways enriched in PMMB-treated tumors. (L) GSEA enrichment upregulated immune response-related pathways in the PMMB group relative to PBS controls. (M) Schematic of the treatment regimen for subcutaneous PDAC tumor–bearing mice, including anti-PD-1 therapy. (N) Growth curves of KPC tumors in mice subjected to different treatments (n = 5). (O) Survival percentage of KPC tumor–bearing mice under different treatment conditions (n = 8). (P and Q) Quantification of the population of CD80⁺ F4/80⁺TAMs and MRC1⁺ F4/80⁺ TAMs in tumors of different treatment groups (n = 5). (R) Quantification of the population of CD8⁺ T cells in tumors of different treatment groups (n = 5). (S and T) Quantification of the population of PD-1⁺ and Tim-3⁺ T cells (gated on CD45⁺CD3⁺CD8⁺) in tumors (n = 5). (U) Quantification of the population of CD4⁺ T cells (CD45⁺CD3⁺) in tumors (n = 5). Data are shown as mean ± SD; n represents the number of biologically independent samples. Statistical significance was calculated by one-way ANOVA with Fisher’s LSD test.

To understand the mechanisms underlying PMMB’s antitumor efficacy, we analyzed macrophage subtypes in PDAC tumors via immune flow cytometry. Compared with the PBS group, the proportion of MRC1⁺ TAMs in PMMB-treated tumors decreased from 50.3 to 4.9%, while the proportion of CD80⁺ TAMs increased from 6.2 to 32.2%, a 4.2-fold increase. Furthermore, relative to the PMB and PMM groups, the proportion of MRC1⁺ macrophages in PMMB decreased from 13.9% and 45.1 to 4.9%, while the proportion of CD80⁺ macrophages was 1.5 and 4.3 times that of the other two groups, respectively (Fig. 4 D–G). This demonstrates PMMB’s superior ability to induce the phenotypic transformation of MRC1⁺ TAMs into CD80⁺ TAMs compared to monotherapies.

Effector T cells and NK cells are pivotal to antitumor immunity. In the PBS group, PDAC tumors exhibited a low proportion of CD8⁺ T cells (~5.2%), reflecting the low immunogenicity typical of cold tumors. Following PMMB treatment, the proportion of CD8⁺ T cells rose markedly (5.2 to 29.4%), accompanied by a notable expansion of CD4⁺ T cells (3.0 to 21.5%, nearly a 6.2-fold increase). In contrast, the proportion of regulatory T cells decreased from 21.8 to 4.3%, while the proportion of NK cells increased from 5.6 to 27.2% (a 3.9-fold increase) (Fig. 4H and SI Appendix, Fig. S7 C–I). Moreover, T cell apoptosis in PMMB-treated tumors decreased from 51.9% (PBS) to 10.0%, demonstrating PMMB’s ability to mitigate T cell apoptosis and restore T cell functionality (SI Appendix, Fig. S7 J and K). Of note, activation of the cGAS-STING pathway and enhanced secretion of intratumoral immune factors were also observed following PMMB treatment (SI Appendix, Fig. S7L), highlighting the induction of a robust innate immune response. Furthermore, elevated levels of immune cytokines, including IL-6, IFN-1β, IFN-γ, and TNF-α, were detected in PMMB-treated tumors, further corroborating its immunotherapeutic efficacy (SI Appendix, Fig. S7M).

A comprehensive transcriptomic analysis of tumor tissues resected from KPC tumor–bearing mice revealed significant gene expression changes following PMMB treatment. Specifically, genes associated with the STING pathway and antitumor macrophages were significantly up-regulated, while those linked to MRC1⁺ TAMs (protumor macrophages) were down-regulated (Fig. 4I). GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of DEGs identified enrichment in immune-related pathways, including signal transduction pathways such as mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase-Akt (PI3K-Akt) (Fig. 4 J and K and SI Appendix, Fig. S8A). Moreover, GSEA further demonstrated significant activation of adaptive immune response pathways, including T cell receptor signaling and pathways responsive to IFN-γ (Fig. 4L and SI Appendix Fig. S8 B and C).

To more accurately assess the immunotherapeutic efficacy of PMMB, we conducted an immune analysis of the overall tumor microenvironment on Day 8, when there was no significant difference in tumor volume between the PBS group and the PMMB group (SI Appendix, Fig. S9). The results showed that compared with the PBS group, the proportion of MRC1⁺ TAMs in the PMMB group decreased significantly from 41.1 to 12.1%, a decrease of 3.4-fold; while the proportion of CD80⁺ TAMs increased significantly from 12.6 to 47.8%, an increase of 2.8-fold. Additionally, PMMB treatment increased CD8⁺ T cell infiltration by 2.8-fold to 34.9%, while CD4⁺ T cell infiltration increased from 5.1 to 24.3%, representing a 3.8-fold increase compared to the PBS group. These results further validate that PMMB exhibits significant immune microenvironment reprogramming effects at the early stage of treatment, even before tumor volume differences become apparent.

Finally, PMMB’s biosafety was evaluated in healthy mice. No significant differences were observed in the morphology of major organs (heart, spleen, lungs, liver, kidneys) between PMMB-treated and untreated mice, confirming its excellent biosafety profile (SI Appendix, Fig. S10).

Therefore, PMMB achieves effective tumor immunotherapy by combining CSF-1R inhibition and STING pathway activation to reprogram MRC1⁺ TAMs, enhance T cell infiltration, and mitigate T cell apoptosis, thereby overcoming the immunosuppressive tumor microenvironment and suppressing PDAC progression with high safety and efficacy.

Immunotherapeutic Evaluation of PMMB Combined with Anti-PD-1 Therapy in KPC Tumor–Bearing Mice.

After confirming the exceptional antitumor immune efficacy of PMMB in primary PDAC models, we investigated its potential to enhance the efficacy of clinical immunotherapy by combining it with anti-PD-1 therapy (Fig. 4M). The PMMB+Anti-PD-1 combination demonstrated a significantly stronger antitumor effect compared to the anti-PD-1 monotherapy group, markedly delaying tumor growth and prolonging survival in the PDAC-bearing mice (Fig. 4 N and O and SI Appendix, Fig. S11 A and B). This evidence suggests that PMMB can cooperate with PD-1 blockade to achieve enhanced therapeutic benefit.

To elucidate the immunological mechanisms underlying this synergy, we analyzed the infiltration of immune cells within the TME. Compared with the anti-PD-1 monotherapy, the proportion of MRC1⁺ TAMs in the PMMB+Anti-PD-1 group decreased from 48.7 to 7.5%, a 6.5-fold decrease, while the proportion of CD80⁺ TAMs in tumor increased significantly from 10.9 to 50.7%, a 3.7-fold increase. (Fig. 4 P and Q and SI Appendix, Fig. S11 C and D). This indicates that PMMB effectively enhances TAMs reprogramming when combined with PD-1 blockade.

The combined treatment (PMMB+Anti-PD-1) markedly enhanced CD8⁺ T cell infiltration, rising from 7.8 to 28.1%, a 2.6-fold increase compared with anti-PD-1 monotherapy (Fig. 4R and SI Appendix, Fig. S11E). Importantly, the proportion of exhausted CD8⁺ T cells, characterized by PD-1⁺CD8⁺ and Tim-3⁺CD8⁺ phenotypes, was reduced by 46.8% and 80.6%, respectively, in the PMMB+Anti-PD-1 group (Fig. 4 S and T and SI Appendix, Fig. 11 F and G). Additionally, the proportion of effector CD4⁺ T cells increased significantly, from 9.6% in the anti-PD-1 group to 30.5% in the PMMB+Anti-PD-1 group, suggesting enhanced activation of the adaptive immune response (Fig. 4U and SI Appendix, Fig. S11H).

In conclusion, the combination of PMMB with anti-PD-1 markedly promotes the infiltration and activation of effector T cells and CD80⁺ TAMs, while reducing the abundance of exhausted CD8⁺ T cells and immunosuppressive MRC1⁺ TAMs. These synergistic effects result in a potent antitumor immune response, highlighting the potential of PMMB as an adjuvant to improve the efficacy of PD-1 blockade in PDAC therapy.

Immunotherapeutic Efficacy of PMMB in the PDAC Metastasis Model.

Building on the observed efficacy of PMMB in KPC tumor–bearing mice, we further assessed its therapeutic potential in a liver metastasis model of PDAC (Fig. 5A). In this highly malignant model, control mice (PBS group) exhibited a median survival of less than 30 d (Fig. 5B), underscoring the aggressive nature of PDAC liver metastasis. Remarkably, PMMB treatment significantly extended survival, with hematoxylin and eosin (H&E)-stained liver sections and bioluminescence imaging confirming the smallest number of metastatic lesions in the PMMB group compared to other treatment groups (Fig. 5 C–F).

Fig. 5.

Antimetastatic and immunotherapeutic efficacy of PMMB in the PDAC metastasis model. (A) Schematic diagram of the experimental procedure for the PDAC liver metastatic model. (B) The survival percentage of mice from different groups (n = 8). (C and D) H&E staining and quantification of the liver in different administration groups (n = 5). (Scale bar, 1,000 μm.) (E and F) Bioluminescence imaging and quantification of PDAC liver metastases. (G and H) Quantification of the population of CD80⁺ F4/80⁺ TAMs and MRC1⁺ F4/80⁺ TAMs (gated on CD45⁺CD11b⁺) in tumors of different treatment groups (n = 5). (I) Quantification of the population of CD8⁺ T cells (gated on CD45⁺CD3⁺) in tumors (n = 5). (J) Schematic diagram of the experimental procedure for the PDAC lung metastatic model. (K) The bioluminescence imaging of the PDAC lung metastasis model. (L) Survival percentage of KPC tumor–bearing mice under different treatment conditions (n = 8). (M) H&E staining of lungs in different administration groups (n = 5). (Scale bar, 3,000 μm.) (N and O) Quantification of the population of CD80⁺ F4/80⁺ TAMs and MRC1⁺ F4/80⁺ TAMs (gated on CD45⁺CD11b⁺) in tumors of different treatment groups (n = 5). (P) Representative t-SNE plots showing the distribution of tumor-infiltrating CD4⁺ T cells and CD8⁺ T cells. Data are shown as mean ± SD; n represents the number of biologically independent samples. Statistical significance was calculated by one-way ANOVA with Fisher’s LSD test.

Flow cytometric analysis revealed that PMMB treatment resulted in a 6.0-fold increase in CD80⁺ TAMs abundance (56.5%) compared to the PBS group and a 1.8-fold increase compared to the PMB group (Fig. 5G and SI Appendix, Fig. S12A). Concurrently, the abundance of MRC1⁺ TAMs decreased from 49.8% in the PBS group and 26.0% in the PMB group to just 12.0% in the PMMB group (Fig. 5H and SI Appendix, Fig. S12B), demonstrating the superior efficacy of PMMB in inducing TAM phenotypic transformation within the liver metastasis microenvironment. Moreover, the percentage of CD8⁺ T cells increased significantly, from 6.4% in the PBS group and 24.5% in the PMB group to 40.2% in the PMMB group, indicating that PMMB enhances cytotoxic T cell infiltration and activation (Fig. 5I and SI Appendix, Fig. S12C).

Recognizing that the lung is another common site of PDAC metastasis, we further evaluated PMMB’s efficacy in a PDAC lung metastasis model (Fig. 5J). The bioluminescence imaging results and survival curves revealed that PMMB significantly inhibited lung metastasis and extended survival compared to MMB and PMB groups (Fig. 5 K and L). H&E-stained lung sections and quantitative analyses confirmed the lowest number of metastatic lesions in the PMMB-treated group, providing compelling evidence for its anti-lung metastasis efficacy (Fig. 5M).

In the lung metastasis model, the immunosuppressive microenvironment was marked by a low proportion of CD80⁺ TAMs (6.0%) and a high proportion of MRC1⁺ TAMs (46.8%) in the PBS group (Fig. 5 N and O and SI Appendix, Fig. S13). While PMB treatment decreased MRC1⁺ TAMs to 24.3% and increased CD80⁺ TAMs to 30.5%, PMMB treatment achieved superior results, reducing MRC1⁺ TAMs to 12.4% and increasing CD80⁺ TAMs to 44.8%. These findings highlight PMMB’s exceptional ability to reprogram TAMs, thereby initiating adaptive immune activation. Analysis of T cell populations further revealed that the PMMB group exhibited the highest percentage of CD8⁺ T cells in lung metastases, followed by the MMB and PMB groups, with the PBS group showing the lowest percentage (Fig. 5P). This suggests that PMMB effectively stimulates immune responses and enhances antitumor immunity.

Overall, PMMB demonstrated superior therapeutic efficacy in PDAC metastasis models. This efficacy is attributed to its dual mechanism of TAMs reprogramming and STING-mediated activation of systemic antitumor immunity. These findings position PMMB as a promising candidate nanodrug for the treatment of metastatic PDAC, with significant potential to transform the management of this highly aggressive disease.

MRC1-Targeted Modulation by PMMB Reprograms TAMs to Enhance Cytotoxic Immune Responses in PDAC.

To further elucidate the underlying mechanisms of PMMB-mediated antitumor immunity, we performed scRNA-seq on subcutaneous PDAC tumors from PMMB-treated and PBS-control mice. Focusing on the primary cellular target, TAMs were resolved into four transcriptionally distinct subsets defined by conserved marker genes (Fig. 6 A and B). UMAP visualization further illustrated the distribution of these TAM subsets across treatment groups (SI Appendix, Fig. S14A). Recognizing the importance of translational relevance, we next performed cross-species alignment between murine and human TAM subsets. This analysis revealed striking transcriptional conservation (Pearson r > 0.85, P < 1e−15) in both marker gene expression and functional programs (Fig. 6C), underscoring the validity of our murine model for studying clinically relevant macrophage biology in PDAC. PMMB treatment intervention significantly depleted protumorigenic Spp1⁺ and Mki67⁺ TAMs populations while increasing Folr2⁺ TAMs frequency (Fig. 6D), with concomitant attenuation of MRC1 expression across all subsets. Moreover, compared to the PBS group, TAMs subpopulations in the PMMB group exhibited distinct functional profiles, with enhanced antitumor activity scores and reduced protumor signatures (Fig. 6E), reflecting a global shift toward an antitumor phenotype. Pseudotime trajectory analysis demonstrated global functional reprogramming toward an antitumor phenotype, marked by elevated antitumor scores and suppressed protumor signatures in residual TAMs (Fig. 6F and SI Appendix, Table S1).

Fig. 6.

ScRNA-seq analysis of murine PDAC revealed potential mechanisms underlying the antitumor immune response induced by PMMB. (A) UMAP of scRNA-seq data showing TAMs highlighted in the integrated atlas (Left) and in the reclustered macrophage-only subset (Right) from murine PDAC (n = 3). (B) Dot plot showing scaled expression of representative marker genes for each mouse macrophage cluster. (C) GSEA of human TAM subset marker genes using genes ranked by log₂ fold change (log₂FC) between each mouse TAM subset and all other TAMs. (D) Distribution of TAM subsets (Left) and Mrc1 expression in the PBS versus PMMB groups (Right). (E) ssGSEA of key signaling pathways across mouse TAM subsets. (F) Multimodal analysis of macrophage functional dynamics and therapeutic responses in a murine PDAC model, integrating ssGSEA scores, DEGs, and functional gene signatures as described above. (G) UMAP of scRNA-seq data for lymphocytes from mouse PDAC. (H) Dot plot showing scaled expression of representative marker genes for each mouse T lymphocyte cluster. (I) Composition of T lymphocyte subsets in pancreatic tumors from the PBS versus PMMB mice. (J) Therapy-dependent regulation of NK and CD8⁺ T cell signaling pathways by SPP1⁺ and FOLR2⁺ TAM subsets. Violin plots overlaid with box plots showing functional scores for cytotoxicity and exhaustion in tumor-infiltrating CD8⁺ T cells in the PBS and PMMB groups. Data are shown as mean ± SD; n represents the number of biologically independent samples. Statistical significance between treatment groups was assessed using the Mann–Whitney U test. **P < 0.01 and ***P < 0.001.

Moreover, we further analyzed the immune landscape of T cells and NK cells following PMMB treatment. Compared with the PBS group, PMMB therapy reduced immunosuppressive Tregs while expanding cytotoxic CD8⁺ T cells and NK cells (Fig. 6 G–I and SI Appendix, Fig. S14B), with CD8⁺ T cells exhibiting enhanced cytotoxicity and reduced exhaustion scores. CellChat analysis revealed a rewired intercellular communication network, in which SPP1⁺ TAMs potentiated CD8⁺/NK cell activation through the CXCL9–CXCR3 and IFN-γ–IFNGR axes, whereas attenuation of Folr2⁺ TAM-derived TGF-β–TGFβR signaling relieved the suppression of CD8⁺ T cell cytotoxicity (Fig. 6J). Collectively, these findings indicate that MRC1-targeting PMMB functions as a spatiotemporal macrophage reprogramming platform, effectively reshaping the tumor immune microenvironment. This coordinated immunomodulation supports MRC1 targeting as a promising therapeutic strategy to overcome macrophage-mediated immunosuppression in PDAC.

Discussion

PDAC remains a highly aggressive and therapeutically intractable malignancy, with conventional treatments offering minimal efficacy in controlling disease progression (37–39). The majority of PDAC patients present with locally advanced or metastatic disease at diagnosis (40–42), severely limiting curative options. Through integrative scRNA-seq analysis, we identified progressive enrichment of heterogeneous TAM populations in primary and metastatic PDAC lesions. MRC1 was consistently expressed across these subsets, designating it as a prominent therapeutic target. Here, we introduce a spatiotemporal macrophage reprogramming strategy (PMMB) leveraging STING phase separation to achieve durable TAM modulation. This spatiotemporal immunoengineering strategy effectively remodels the PDAC immune landscape, transforming the TME from immunosuppressive to immune-activated. In preclinical models, PMMB not only suppresses both primary and metastatic PDAC but also synergizes with anti-PD-1 therapy, significantly increasing CD8⁺ T cell infiltration and reversing T cell exhaustion. Unlike conventional TAM-targeting approaches, our strategy achieves long-term immune reprogramming, ensuring a sustained therapeutic effect against metastatic PDAC. Moving forward, refining STING activation thresholds and optimizing clinical translation in patient-derived models will be critical next steps.

Materials and Methods

Experimental Methods and Materials.

The experimental procedures of this study included detailed descriptions of materials, cell lines and mouse strains, reagents, and animal ethics approval, as well as scRNA-seq, spatial transcriptomics, and TCGA data integration. The study also encompassed PMMB construction and characterization, in vitro cellular uptake, functional evaluation of PMMB-treated macrophages (bulk RNA-seq, qPCR, and flow cytometry), T cell function assessment (proliferation, apoptosis, and cytotoxicity), in vivo PMMB biodistribution and biosafety, and therapeutic efficacy in multiple murine pancreatic cancer models, complemented by scRNA-seq analyses to systematically elucidate mechanisms. Full details of all methods are provided in SI Appendix.

Ethics.

Thirty deidentified patient tumor samples from PDAC across various stages were obtained from Zhongda Hospital, Southeast University, and reviewed by both pathologists and surgeons. Tumors were staged by pathologists according to World Health Organization criteria. The ethical approval number was 2021ZDSYLL029-Y01.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. The bulk RNA-seq and scRNA-seq data generated in this study are deposited in Dryad: https://doi.org/10.5061/dryad.kkwh70shs (43). This study also used previously published data from GEO: GSE197177 (44), GSE212966 (45), and GSE217847 (46).