Evolution of SPP1+ cavity macrophage-mediated immunotherapy resistance in peritoneal metastasis of colorectal cancer

Xiaomeng Dai; Chunyu Lai; Libing Hong; Yuzhi Jin; Jinlin Cheng; Haimeng Yan; Xuqi Sun; Bin Li; Chuan Liu; Hangyu Zhang; Yanwei Lu; Haibo Zhang; Danyang Wang; Peng Zhao; Yu Shi; Weijia Fang; Shan Xin; Xia Yu; Xuanwen Bao

doi:10.1016/j.xcrm.2026.102930

. 2026 Jul 16;7(8):102930. doi: 10.1016/j.xcrm.2026.102930

Evolution of SPP1⁺ cavity macrophage-mediated immunotherapy resistance in peritoneal metastasis of colorectal cancer

Xiaomeng Dai ^1,^2,^3,^4,^11,^∗, Chunyu Lai ^1,¹¹, Libing Hong ^1,¹¹, Yuzhi Jin ^1,¹¹, Jinlin Cheng ⁵, Haimeng Yan ⁶, Xuqi Sun ¹, Bin Li ¹, Chuan Liu ¹, Hangyu Zhang ¹, Yanwei Lu ⁷, Haibo Zhang ⁷, Danyang Wang ⁸, Peng Zhao ^1,², Yu Shi ⁵, Weijia Fang ^1,^2,^∗∗, Shan Xin ^9,^∗∗∗, Xia Yu ^5,^∗∗∗∗, Xuanwen Bao ^1,^2,^3,^10,^12,^{∗∗∗∗∗}

PMCID: PMC13522804 PMID: 42462726

Summary

The immunometabolic basis of therapy-resistant colorectal cancer (CRC) peritoneal carcinomatosis with malignant ascites remains poorly defined. Here, we profile ascites immune cells from 20 patients across treatment-naive, chemo/targeted therapy-refractory, and immune checkpoint blockade (ICB)/adoptive T cell therapy (ACT)-resistant CRC. Single-cell RNA sequencing identifies SPP1⁺ cavity macrophages as drivers of CD8⁺ T cell dysfunction. Proteomic profiling of 36 patients confirms SPP1 enrichment in ICB/ACT-resistant peritoneal metastases. Mechanistically, SPP1 sustains an M2-like program via PPARγ activation and lipid uptake. SPP1 deficiency reduces PPARγ ligand precursors, triggering NF-κB-driven macrophage reprogramming and reversing CD8⁺ T cell suppression. Supplementation with 15 d-PGJ₂ and fatty acids restores the M2 phenotype. In vivo, macrophage-specific SPP1 knockout enhances cytotoxic T lymphocyte infiltration and ICB efficacy, while SPP1 neutralization overcomes ICB resistance and augments ACT efficacy. Thus, SPP1⁺ cavity macrophages are central immunometabolic regulators, and SPP1 inhibition represents a promising strategy to overcome immunotherapy resistance in this lethal disease.

Keywords: CRC, peritoneal metastasis, Spp1⁺ macrophage, PPARγ, immunometabolism

Graphical abstract

Highlights

•
SPP1⁺ cavity macrophages are enriched in ICB/ACT-resistant CRC peritoneal metastases
•
SPP1 sustains M2-like macrophage polarization via PPARγ activation and lipid uptake
•
SPP1 loss reduces PPARγ ligands, activates NF-κB, and reverses CD8⁺ T cell suppression
•
Macrophage SPP1 knockout/blockade enhances immunotherapy efficacy in vivo

Dai et al. identify SPP1⁺ cavity macrophages enriched in immunotherapy-resistant CRC peritoneal metastasis. These cells suppress CD8⁺ T cells via a PPARγ-lipid axis mediated by lipid-derived PPARγ ligands. SPP1 deficiency reduces these ligands, activates NF-κB, reverses CD8⁺ T cell dysfunction, and restores immunotherapy sensitivity.

Introduction

Peritoneal metastasis represents one of the most lethal manifestations of colorectal cancer (CRC), frequently accompanied by malignant ascites and notorious for its resistance to therapy.¹^,²^,³^,⁴ Malignant ascites not only worsens morbidity but also establishes a distinct tumor microenvironment (TME) shaped by disseminated tumor cells, stromal elements, and infiltrating immune cells.⁵ This fluid-based niche differs fundamentally from primary tumors or solid metastases, yet remains poorly defined, particularly in the setting of therapeutic resistance.⁶^,⁷

Efforts to extend the reach of immune checkpoint blockade (ICB) and adoptive T cell therapies (ACT) to CRC have been met with limited success in patients with peritoneal metastases.⁸^,⁹^,¹⁰^,¹¹ While recent single-cell RNA sequencing (scRNA-seq) studies in ovarian¹² and gastric cancers¹³ have begun to illuminate the ascitic immune landscape—revealing tumor- and therapy-specific remodeling and implicating subsets such as immunosuppressive macrophage populations¹⁴ —whether similar mechanisms drive immune escape in CRC remains unknown.

In CRC, single-cell studies of ascites have so far been confined to treatment-naive patients, where macrophages emerged as the dominant immune population with immunosuppressive features.¹⁵^,¹⁶^,¹⁷ However, the architecture of the ascitic immune ecosystem under therapeutic pressure—particularly in the context of chemo/targeted therapy failure and resistance to ICB or ACT—has yet to be delineated. The cellular and metabolic mechanisms underlying immune escape in this hostile microenvironment remain elusive.

Here, we address this gap by comprehensively characterizing the immune landscape of CRC ascites across treatment-naive, chemo/targeted therapy-refractory, and ICB/ACT-resistant cohorts using scRNA-seq, complemented by proteomic and spatial analyses. We identify SPP1⁺ cavity macrophages (CMs) as an immunosuppressive population enriched under therapeutic selection, positioned in proximity to CD8⁺ cytotoxic T cells, and mechanistically sustained by a PPARγ-dependent metabolic axis. Functional studies demonstrate that disrupting this pathway reprograms macrophages, restores CD8⁺ T cell activity, and sensitizes peritoneal metastases to ICB and ACT. These findings reveal a previously underappreciated mechanism of immune evasion and nominate SPP1⁺ macrophages as a tractable target to overcome immunotherapy resistance in CRC peritoneal metastasis.

Results

Single-cell profiling reveals the cellular architecture of CRC ascites

The detailed workflow of our research is shown in Figure 1A. To dissect the immune landscape of CRC-associated ascites under distinct therapeutic pressures, we performed scRNA-seq on 20 patient samples, encompassing treatment-naive (n = 6), chemo/targeted therapy-refractory (n = 6), and ICB/ACT-resistant (n = 8) cohorts (Figure 1B). Within the ICB/ACT-resistant group, six patients received anti-PD-1 monoclonal antibody therapy, one received chimeric antigen receptor (CAR) T cell therapy, and one received combined CAR T cell and CAR natural killer cell (NK) therapy (Table S1). A total of 205,538 cells were recovered, with comparable cell yields across the groups: 69,652 from treatment-naive, 57,708 from chemo/targeted therapy-refractory, and 78,178 from ICB/ACT-resistant patients.

Single-cell transcriptomic landscape of CRC ascites reveals conserved immune composition across treatment groups

(A) Schematic overview of the study design. Single-cell RNA sequencing (scRNA-seq) was applied to malignant ascites from 20 colorectal cancer (CRC) patients stratified into three cohorts: treatment-naive (n = 6), chemo/target therapy-refractory (n = 6), and immune checkpoint blockade (ICB)/adoptive cell therapies (ACT)-resistant (n = 8).

(B) UMAP embedding of all 205,538 cells color coded by major immune lineages: B/plasma cells, T/NK cells, and myeloid cells. Right panels display absolute cell counts and relative proportions per lineage.

(C) UMAP plots stratified by treatment group, showing a broadly overlapping immune landscape.

(D) Bar plots depicting per-patient relative frequencies of major immune lineages.

(E) Boxplots comparing major immune lineage abundance across groups.

Dimensionality reduction via uniform manifold approximation and projection (UMAP) and unsupervised clustering identified four major immune lineages based on canonical marker expression: B cells (e.g., CD79A and MS4A1), plasma cells (expressing CD38 and JCHAIN), myeloid cells (marked by CD14, CD68, CD300E, and CD163), and T/NK cells (marked by the expression of CD8A, IL7R, and NKG7) (Figure 1B; Figure S1). Notably, UMAP revealed no overt batch effect, and cells from different treatment groups were evenly distributed across transcriptional clusters (Figures 1C and 1D). The relative proportions of these major cell types were similar across all clinical groups (Figure 1E), suggesting that therapeutic resistance is not primarily driven by changes in major immune lineage composition but may instead involve phenotypic remodeling within stable lineages.

Phenotypic and functional characterization of lymphocyte populations in CRC ascites

To further delineate the immune architecture of CRC ascites, we next focused on lymphocyte populations, re-clustering T and NK cells to define phenotypic heterogeneity across clinical groups. A total of eight transcriptionally distinct subpopulations were identified: including NK cells, cytotoxic CD8⁺ T cells, CD8⁺ central memory-like T cells, naive CD4⁺ T cells, Th1-like memory CD4⁺ T cells, regulatory T cells (Tregs), mucosal-associated invariant T cells, and γδ T cells (Figure S2A). The identity of each subset was supported by the expression of established lineage-defining and functional markers projected onto the UMAP embedding and summarized in marker heatmaps (Figures S2B and S2C). For instance, cytotoxic CD8⁺ T cells were characterized by coordinated expression of CD8A, NKG7, GZMK, GZMA, and GZMB, clearly distinguishing them from memory-like and non-cytotoxic T cell populations. CD8⁺ central memory T cells exhibited the expression of CCR7, LEF1, SELL, IFNG-AS1, and TCF7, consistent with an antigen-experienced but non-effector phenotype. Naive CD4⁺ T cells were defined by high expression of CCR7, SELL, LEF1, SESN3, and BACH2, together with low expression of cytotoxic effector genes. Quantitative profiling across clinical groups revealed 34,102 T/NK cells in the treatment-naive group, 24,210 in the chemo/targeted therapy-refractory cohort, and 30,659 in the ICB/ACT-resistant group. Across all cohorts, cytotoxic CD8⁺ T cells and NK cells collectively accounted for nearly 50% of the T/NK population, with Th1-like memory CD4⁺ T cells ranking among the top three most abundant subsets (Figure S2D), indicating a cytotoxic-skewed lymphoid landscape within CRC ascites. Notably, Tregs were significantly increased in the ICB/ACT-resistant group (p = 0.031) (Figure S2E), suggesting a therapy-associated transition toward an immunosuppressive immune milieu. The per-patient cell type proportions are detailed in Figures S3A and S3B. The signaling pathway enrichments associated with each T/NK subset are shown in Figure S2F.

We also analyzed B cell lineages in CRC ascites and identified four subpopulations: memory B cells, mature B cells, plasmablasts, and plasma cells (Figures S2G, S2H, S3C, and S3D). Frequency analysis revealed an increase in plasma cells in the ICB/ACT-resistant group (Figure S2I), suggesting skewing toward terminal differentiation in resistant states. Pathway-level analyses of B lineage cells further supported distinct immunoregulatory states, with functional enrichment profiles indicating differential engagement of humoral immune mechanisms across subsets (Figure S2J).

In summary, our single-cell dissection of lymphocyte populations in CRC ascites reveals an immune microenvironment dominated by cytotoxic and memory-like T and NK cells, with a notable absence of classical exhaustion markers in most T cells. The contrasting dynamics of NK cells and Tregs across clinical groups reflects shifts in immune pressure associated with therapy.

Enrichment and distinct molecular programming of SPP1⁺ cavity macrophages in treatment-resistant CRC ascites

Given that myeloid constituted the dominant immune populations in the ascites of CRC, we next focused on delineating the transcriptional and phenotypic heterogeneity of this compartment. Re-clustering of the 88,088 myeloid cells identified across all samples revealed eight transcriptionally distinct subpopulations, comprising three dendritic cell (DC) subsets (cDC1, cDC2, and LAMP3⁺ DCs), together with neutrophils, monocytes, monocyte-like CMs (mono-CMs), SPP1⁺ CMs, and Ki67⁺ CMs (Figure 2A).

Myeloid cell heterogeneity and enrichment of immunosuppressive SPP1⁺ macrophages in treatment-resistant CRC ascites

(A) UMAP visualization of re-clustered myeloid-lineage cells from CRC ascites (n = 88,088), identifying canonical dendritic cell subsets (cDC1, cDC2, and LAMP3⁺ DCs), monocytes, monocyte-like cavity macrophages (mono-CMs), SPP1⁺ cavity macrophages (SPP1⁺ CM), Ki67⁺ CM, and neutrophils.

(B) Dot plot showing the expression of representative lineage-defining and annotation markers across myeloid subpopulations.

(C) Boxplots comparing the relative frequencies of myeloid subpopulations across clinical cohorts.

(D) Gene Ontology (GO) enrichment analysis of monocyte/macrophage subsets, illustrating distinct functional programs associated with each population.

(E) Radar plot summarizing enrichment of single-cell myeloid gene sets, including interferon-primed tumor-associated macrophage (IFN-TAM), inflammatory cytokine-enriched TAM (Inflam-TAM), lipid-associated TAM (LA-TAM), immune regulatory TAM (Reg-TAM), resident-tissue macrophage-like TAM (RTM-TAM), proliferating TAM (Prolif-TAM), and classical tumor-infiltrating monocyte (TIM) programs across monocyte/macrophage subsets.

(F) Heatmap showing expression of representative RTM-TAM and LA-TAM gene signatures across monocyte/macrophage populations.

(G) Dot plot showing pathway-level enrichment analysis among monocytes and macrophages.

(H) Monocle2-based pseudotime analysis of monocyte/macrophage populations, together with density plots showing the distribution of monocyte and macrophage subsets along the inferred trajectory.

Cell yields across clinical groups were comparable, with 25,534 myeloid cells from treatment-naive patients, 29,048 from the chemo/targeted therapy-refractory cohort, and 33,506 from the ICB/ACT-resistant group (Figure 2A; Figures S4A and S4B). Marker gene expression profiles for each subpopulation are shown in Figure 2B. For instance, CLEC9A and CD1C marked cDC1 and cDC2 subsets, respectively. LAMP3⁺ DCs, a population implicated in immune regulation, were defined by elevated LAMP3 expression. Monocytes were characterized by high expression of FCN1 and VCAN. A population of mono-CMs was defined by concurrent expression of VCAN and FCN1, together with CM-associated genes including MARCO, VSIG4, CTSL, and CD68. SPP1⁺ CMs were distinguished by high expression of SPP1 and VISIG, together with immunoregulatory markers such as LYVE1 and CD163, indicating a polarized macrophage state. In addition, a distinct population expressing FCGR3B, CSF3R, S100A8, and S100A9, together with low transcript complexity, was identified and annotated as neutrophils (Figure 2B; Figure S4C).⁷^,¹⁶^,¹⁸^,¹⁹

Quantitative comparisons of subpopulation frequencies revealed marked shifts across clinical groups (Figure 2C; Table S2). While the treatment-naive cohort exhibited a higher proportion of cDC1 cells, the ICB/ACT-resistant cohort displayed a pronounced enrichment of SPP1⁺ CMs, suggesting resistance-associated remodeling of the myeloid landscape in advanced disease.

To explore the functional properties of these monocyte/macrophage subsets, we performed Gene Ontology enrichment analysis (Figure 2D). Monocytes were enriched for pathways related to innate immune defense and leukocyte aggregation, whereas SPP1⁺ CMs exhibited enrichment of complement activation pathways, including genes of the C1Q family, consistent with an immunoregulatory phenotype.

To further characterize transcriptional programs across monocyte/macrophage subsets, we applied curated single-cell gene sets capturing tumor-associated macrophage (TAM) diversity, including inflammatory cytokine-enriched TAMs (Inflam-TAMs), interferon-primed TAMs (IFN-TAMs), immune regulatory TAMs (Reg-TAMs), proliferating TAMs (Prolif-TAMs), resident-tissue macrophage-like TAMs (RTM-TAMs), lipid-associated TAMs (LA-TAMs), and classical tumor-infiltrating monocytes (TIMs) (Table S3).²⁰ Radar plot visualization revealed distinct program-level enrichment patterns across subsets (Figure 2E), which were further supported by heatmap analysis of representative RTM/LA-TAM gene signatures (Figure 2F). Notably, SPP1⁺ CMs showed strong enrichment of RTM-TAM and LA-TAM programs. Pathway-level characterization revealed that SPP1⁺ CMs were specifically enriched for lysosomal activity and complement/coagulation cascades (Figure 2G), consistent with their proposed role in TME remodeling and immune evasion. To investigate lineage relationships, we performed pseudotime analysis. SPP1⁺ CMs were preferentially distributed toward the late/distal end of the inferred pseudotime trajectory. This observation was supported by an increased proportion of SPP1⁺ CMs in late pseudotime bins, which indicated reduced differentiation potential in these population relative to other myeloid subsets (Figure 2H).

In summary, our single-cell atlas of CRC ascites identifies a transcriptionally distinct macrophage population—SPP1⁺ CMs—markedly enriched in ICB/ACT-resistant CRC ascites. These cells exhibit a resident, immunosuppressive phenotype associated with complement activation, potentially contributing to therapeutic failure.

Integrated proteomic analysis validates SPP1⁺ CM accumulation and immunosuppressive TME in ICB/ACT-resistant cohort

To further substantiate the immune microenvironmental features identified in ascitic fluid, we conducted complementary proteomic profiling of formalin-fixed paraffin-embedded (FFPE) peritoneal metastases obtained from three clinically defined patient cohorts: treatment-naive (n = 14), chemo/targeted therapy-refractory (n = 12), and ICB/ACT-resistant individuals (n = 10) (Figure 3A; Table S4). Given that peritoneal metastatic nodules typically arise prior to the development of malignant ascites in CRC, they serve as an independent and pathophysiologically relevant tissue source to validate findings derived from ascites-based single-cell analyses.

Integrated proteomic and spatial analysis revealing immunosuppressive SPP1⁺ macrophage enrichment in ICB/ACT-resistant peritoneal metastases

(A) Schematic overview of the proteomic strategy. Formalin-fixed paraffin-embedded (FFPE) peritoneal metastatic nodules from three cohorts (treatment-naive, chemo/target therapy-refractory, ICB/ACT-resistant) underwent liquid chromatography-tandem mass spectrometry profiling with downstream differential expression, KEGG analysis, and ssGSEA.

(B) Dual volcano plots displaying differentially expressed proteins (DEPs) from pairwise comparisons: treatment-naive vs. chemo/target therapy-refractory (bottom) and treatment-naive vs.

ICB/ACT-resistant (top).

(C) KEGG enrichment of upregulated DEPs in FFPE-derived metastases, highlighting immune regulatory pathways.

(D–F) Boxplots comparing enrichment scores for (D) lipid-associated tumor-associated macrophages (LA-TAMs), (E) resident-tissue macrophage-like TAMs (RTM-TAMs), and (F) SPP1⁺ cavity macrophages (SPP1⁺ CMs), all peaking in ICB/ACT resistance.

(G) Hallmark ssGSEA showing pathway distinctions.

Differential expression analysis, visualized via volcano plots, revealed the pronounced proteomic shifts when comparing treatment-naive tumors with those from chemo/targeted therapy-refractory and ICB/ACT-resistant groups (Figure 3B). Notably, SPP1⁺ CM-associated programs—including SPP1 and APOE—were significantly upregulated in the ICB/ACT-resistant cohort, indicating the role of SPP1⁺ CM in establishing an immunosuppressive TME.

Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of cohort-specific upregulated proteins further delineated functionally distinct TME landscapes (Figure 3C). Both chemo/targeted therapy-refractory and ICB/ACT-resistant groups exhibited enrichment of complement and coagulation cascades as well as PPAR signaling pathways, whereas the NF-κB signaling pathway was predominantly enriched in the treatment-naive group. Quantitative single-sample gene set enrichment analysis (ssGSEA) signature scores of LA-TAMs, RTM-like TAMs, and SPP1⁺ CMs demonstrated the highest levels in ICB/ACT-resistant ascites (Figures 3D–3F), consistent with findings from our scRNA-seq dataset. Additionally, this cohort showed marked enrichment of complement activation components, supporting an association between SPP1⁺ CM-linked programs and immune modulation in immunotherapy-resistant disease (Figure 3G).

Together, these proteomic analyses identify SPP1⁺ CM-associated immune programs that are preferentially enriched in ICB/ACT-resistant peritoneal metastases. Across two independent but clinically related human tissue contexts—ascites-derived single-cell transcriptomes and proteomic profiling of peritoneal metastatic lesions—SPP1⁺ CM-linked signatures consistently associate with complement activation and immunoregulatory features. These findings extend the single-cell observations from ascites to tumor-associated tissues and support a reproducible association between SPP1⁺ CM-associated programs and therapy-resistant disease states in CRC peritoneal metastasis.

SPP1-shaping immunosuppressive macrophage-T cell interactions driving therapy resistance

To investigate the role of SPP1 in regulating macrophage function, we performed genetic deletion of SPP1 in the mouse macrophage ANA-1 cells using CRISPR-Cas9 technology (Figure 4A). SPP1 knockout (KO) led to a substantial reduction in the expression of M2 macrophage-associated markers, including CD206, Arg1, and PD-L1, compared with control cells (Figure 4B; Figure S5A). Transcriptome sequencing analysis revealed downregulation of M2-associated genes (MS4A4 A, SLC4A7, MRC1, and CTSC) and upregulation of M1-associated genes (IRF1, CD86, IL-6, and PDGFA) in the SPP1 KO group (Figure 4C; Figure S5B). Furthermore, ssGSEA and GSEA demonstrated enhanced M1 polarization and increased enrichment of antigen processing and presentation pathways in the SPP1 KO cells, while control cells showed enrichment for M2 polarization and complement activity-related pathways (Figure 4D; Figure S5C).

Consistently, our scRNA-seq analysis revealed prominent ligand-receptor interactions between SPP1⁺ CMs and cytotoxic T lymphocytes (CTLs), primarily driven by the SPP1-CD44 axis (Figure 4E; Figure S5D and S5E). Multiplex immunofluorescence staining of FFPE sections from peritoneal metastatic samples (Figure 4F) revealed significantly greater frequency of SPP1⁺ CM-CTL interactions (defined as <50 μm apart) in the ICB/ACT-resistant cohort compared with the other groups. To quantitatively evaluate SPP1⁺ CM-CTL spatial relationships, we measured the distances between SPP1⁺ CMs and CTLs across six distance intervals: <10 μm, 10–20 μm, 20–30 μm, 30–40 μm, 40–50 μm, and >50 μm. The ICB/ACT-resistant cohort exhibited a marked enrichment of short-range interactions (<30 μm), suggesting close physical proximity between SPP1⁺ CMs and effector T cells (Figure 4G). Moreover, the ICB/ACT-resistant group exhibited the highest abundance of SPP1⁺ CMs (Figure 4H) and the lowest infiltration of CTLs (Figure 4I), indicating a spatially organized immunosuppressive niche.

To validate these findings, we co-cultured SPP1 KO IL-4/13-polarized ANA-1 cells with splenic CD8⁺ T cells in the presence of anti-CD3/CD28 stimulation. We found that the absence of SPP1 remarkably attenuated the immunosuppressive capacity of IL-4/13-polarized macrophages, as evidenced by the enhanced CD8⁺ T cell proliferation and increased GZMB secretion (Figures 4J–4M).

Given that SPP1 is a secreted protein, we first examined its secretion by macrophages. We confirmed that macrophages stimulated with IL-4/IL-13 secreted high levels of SPP1, whereas SPP1-KO macrophages produced negligible amounts (Figure S6A). Additionally, recombinant SPP1 (rhSPP1) protein stimulation partially restored the downregulated expression of Arg1 and CD206 in SPP1 KO IL-4/13-polarized macrophages (Figures S6B and S6C). Notably, rhSPP1 also upregulated CD206 expression in unpolarized ANA-1 cells (Figure S6D). Quantitative real-time PCR analyses demonstrated that rhSPP1 significantly upregulated the mRNA expression of M2-associated markers, including Arg1, Chil3, and VEGF, while concomitantly downregulating the expression of M1-associated markers such as IL-1, IL-6, and Nos2 in ANA-1 cells (Figure S6E).

To further dissect the underlying mechanisms, we investigated whether SPP1 suppresses CD8⁺ T cell function indirectly through macrophage reprogramming or directly via extracellular SPP1. Pretreating SPP1-KO macrophages with rhSPP1 for 24 h, followed by thorough washing prior to co-culture with CD8⁺ T cells, partially restored GZMB expression in CD8⁺ T cells, indicating that SPP1 can modulate T cell function indirectly through sustained reprogramming of macrophages (Figures S6F and S6G). Separately, direct treatment of isolated CD8⁺ T cells with rhSPP1 significantly reduced GZMB expression, an effect reversed by CD44 blocking antibody, confirming that SPP1 can directly inhibit CD8⁺ T cell function via engagement of CD44 on CD8⁺ T cells (Figure S6H). Furthermore, in co-cultures of wild-type macrophages and CD8⁺ T cells, addition of an SPP1 neutralizing antibody significantly enhanced GZMB expression in CD8⁺ T cells, further supporting that macrophage-derived SPP1 actively suppresses T cell function (Figure S6I). Collectively, these results demonstrate that SPP1 exerts its immunosuppressive effects through a dual mechanism: indirectly by reprogramming macrophages toward an M2-like phenotype and directly by engaging CD44 on CD8⁺ T cells to inhibit their effector function.

SPP1 sustaining macrophage immunosuppression via PPARγ lipid ligand biogenesis

To investigate the molecular mechanisms underlying SPP1-mediated macrophage remodeling, we performed an in-depth analysis of RNA sequencing datasets comparing SPP1 KO and control ANA-1 cells. KEGG pathway enrichment analysis revealed significant downregulation of the PPARγ signaling pathway and upregulation of the NF-κB pathway in SPP1 KO cells compared with control ANA-1 cells (Figure 5A). Western blotting consistently showed a significant reduction in PPARγ protein expression in SPP1 KO cells, accompanied by elevated levels of NF-κB p65 under IL-4/IL-13 polarization. Importantly, IκBα phosphorylation occurred earlier in SPP1-deficient cells, indicating an accelerated activation of the NF-κB signaling cascade (Figures 5B–5D).

To further clarify the role of PPARγ, we employed pharmacological inhibition and activation approaches. Specifically, treatment with T0070907, a PPARγ inhibitor, led to decreased expression of M2 macrophage markers CD206 and Arg1 in IL-4/IL-13-polarized macrophages. Conversely, pioglitazone, a PPARγ agonist, restored the expression of CD206 and Arg1 in SPP1 KO cells (Figure 5E). Additionally, NF-κB inhibitor QNZ rescued CD206 and Arg1 expression in SPP1 KO macrophages (Figure 5F), indicating that SPP1 deficiency activates NF-κB signaling, which antagonizes PPARγ-mediated M2 polarization. Combining pioglitazone and QNZ treatments showed no further rescue compared with single-treatment groups (Figure S6J), suggesting a linear regulatory relationship between PPARγ and NF-κB in maintaining the M2 phenotype. Notably, supplementation of SPP1 KO macrophages with pioglitazone rescued the alteration of PPARγ and NF-κB p65 expression caused by SPP1 deficiency (Figure 5G). These findings collectively suggest the role of SPP1 in PPARγ-dependent modulation of NF-κB signaling.

PPARγ is a critical regulator of macrophage metabolism and the immune response.²¹^,²² Based on our RNA sequencing datasets, KEGG analysis revealed that genes downregulated in SPP1 KO cells were significantly enriched in the biosynthesis of unsaturated fatty acids and fatty acid metabolism (Figure 5A). ssGSEA demonstrated that SPP1 KO cells exhibited decreased activity in pyruvate metabolism, unsaturated fatty acids biosynthesis, sphingolipid metabolism, and steroid biosynthesis compared with control macrophages (Figure 5H). Consistently, CD36 and FABP4 protein levels, key lipid-related markers, were significantly downregulated in SPP1 KO cells (Figures 5I and 5J). Metabolomic analysis further confirmed that SPP1 KO cells had significantly decreased monounsaturated fatty acid content relative to control macrophages (Figure 5K; Figures S7A and S7B).

Supplementing SPP1 KO macrophages with oleic acid (OA)¹² or OA + docosahexaenoic acid (DHA)⁵ did not rescue the downregulation of CD206 and Arg1 (Figures S7C–S7F). Adding pyruvate, dimethyl sulfate (DMS), dimethyl-α-ketoglutarate (DM-α-KG), or DM-α-KG + DMS partially rescued Arg1 downregulation, but not CD206 downregulation, due to SPP1 deficiency (Figures S8A–S8G). This indicates that the metabolic substrate deficit is not the key factor influenced by SPP1 deficiency. We further analyzed lipidomic data and demonstrated that SPP1 deficiency reduced the levels of the PPARγ ligand precursors, namely, the phosphatidylcholine (PC)-derived lipid species PC(PGJ₂/16:1(9Z)) and PC(P-16:0/PGD₂) (Figure 5L), which are metabolically converted to the endognous PPARγ ligand 15-deoxy-Δ^12,14-prostaglandin J₂ (15d-PGJ₂). Supplementation of SPP1-deficient macrophages with 15d-PGJ₂ successfully rescued the downregulated expression of CD206 and Arg1 (Figures ;S8H–S8J). Strikingly, in the presence of 15d-PGJ₂, supplementation with either OA or DHA further enhanced the rescue effect, whereas combined supplementation with OA and DHA completely restored CD206 and Arg1 expression levels (Figure 5M). This indicates that SPP1 deficiency disrupts PPARγ activation by depleting essential lipid-derived ligands and inhibiting metabolic energy pathways.

Targeting SPP1⁺ macrophages enhances the efficacy of immunotherapy

To further validate the role of SPP1⁺ macrophages in determining the efficacy of ICB in vivo, we generated conditional KO transgenic mouse models with selective deletion of SPP1 in macrophages. This was achieved by crossing Spp1-floxed (Spp1^f/f) mice with Lyz2-Cre mice, resulting in a strain with selective ablation of Spp1 in the macrophage compartment (Spp1^f/f; Lyz2-Cre) (Figure 6A). Using both Spp1^f/f; Lyz2-Cre and Spp1^f/f mice, we established a syngeneic CRC peritoneal metastasis model by intraperitoneal inoculation of MC38-luc cells (Figure 6B), a microsatellite instability-high cell line known to be responsive to ICB, thereby enabling assessment of whether macrophage-specific SPP1 KO could further enhance ant-PD-1 (αPD-1) efficacy.²³ Notably, through continuous in vivo bioluminescence imaging and survival monitoring, we observed that compared with Spp1^f/f control mice, Spp1^f/f; Lyz2-Cre mice exhibited significantly reduced growth of peritoneal metastatic tumors and improved survival outcomes. This effect was further enhanced by αPD-1 therapy, with the combination treatment yielding the most pronounced therapeutic benefit across all experimental groups (Figures 6C–6E).

Then, we conducted multicolor immunohistochemistry (mIHC) on peritoneal tumor nodules retrieved from these mice. The analysis demonstrated that macrophages within peritoneal tumor nodules failed to express SPP1 in Spp1^f/f; Lyz2-Cre mice, unlike their SPP1^f/f control counterparts (Figure S9). Further, mIHC revealed that αPD-1-treated SPP1^f/f; Lyz2-Cre mice exhibited elevated GZMB expression in intratumoral CD8⁺ T cells compared with other experimental groups (Figures 6F and 6G), indicative of enhanced cytotoxic T cell activity. To further evaluate the impact of SPP1⁺ macrophages on ACT therapy, we established an MC38-OVA peritoneal metastasis model. On day 16 following intraperitoneal inoculation with MC38-OVA cells, mice were randomly assigned to one of four treatment groups: isotype control antibodies, CD8⁺ T cells derived from OT1 mice, SPP1-neutralizing antibody (αSPP1) therapy, or combination treatment with αSPP1 and ACT via intraperitoneal injection (Figure 6H). Notably, the combination of αSPP1 and ACT resulted in significantly prolonged mouse survival compared with all other treatment regimens (Figure 6I). Together, these findings align with previous reports demonstrating that macrophage-specific SPP1 targeting can enhance immunotherapy efficacy²⁴ and further support the potential of SPP1⁺ macrophage-directed strategies as a promising approach to improve cancer immunotherapy outcomes.

SPP1 blockade reversing ICB resistance in vivo

To determine whether targeting SPP1 in the TME could overcome resistance to ICB with αPD-1 in vivo, we developed an ICB-resistant CRC peritoneal metastasis model by intraperitoneally inoculating mice with CT26-luc cells, a microsatellite-stable cell line known to be intrinsically resistant to ICB.²⁵ Nine days post-injection, mice underwent bioluminescence imaging and were stratified into four groups based on tumor burden. These groups received isotype control antibodies, αSPP1, αPD-1, or combination therapy, with the experimental timeline outlined in Figure S10A. Consistent with prior studies, the CT26 CRC model exhibited inherent resistance to αPD-1 monotherapy.²⁴ Continuous bioluminescence monitoring and survival assessments revealed that both αSPP1 monotherapy and αSPP1 + αPD-1 combination therapy significantly suppressed tumor growth and prolonged survival compared with the control group. The combination therapy achieved the most pronounced tumor suppression and longest survival duration (Figures 7A–7C). Notably, no significant differences in mouse body weight were observed across experimental groups.

To uncover the mechanisms underlying these effects, we conducted flow cytometry analyses on ascites-derived leukocytes from all groups. The gating strategy for TAMs is detailed in Figure S10B. In line with our in vitro findings, αSPP1 therapy and αSPP1 + αPD-1 combination therapy significantly increased the infiltration of M1-like TAMs (dead dye⁻CD11b⁺F4/80⁺CD86⁺CD206⁻) in malignant ascites (MAs), while reducing M2-like TAMs (dead dye⁻CD11b⁺F4/80⁺CD86⁻CD206⁺). Conversely, αPD-1 monotherapy showed no significant impact on TAM polarization compared with controls. Furthermore, combination therapy with αSPP1 + αPD-1 led to a marked reduction in total TAM infiltration and PD-L1⁺ TAMs in MAs compared with the control group (Figures 7D–7G).

To assess the impact of αSPP1 + αPD-1 dual treatment on T cell activation, we analyzed the infiltration and functional characteristics of T helper 1 (Th1) cells and CTLs. The flow cytometry gating strategies for these immune subsets are illustrated in Figure S10C. Notably, combination therapy significantly increased the infiltration of CD4⁺ T cells and CD8⁺ T cells in MAs compared with the control group, although Th1 (dead dye⁻CD45⁺CD3⁺CD4⁺IFN-γ⁺) infiltration remained unchanged across groups (Figures 7H–7J). Importantly, both αSPP1 monotherapy and αSPP1 + αPD-1 combination therapy significantly enhanced the infiltration of CTLs (dead dye⁻CD45⁺CD3⁺CD8⁺GZMB⁺) in MAs compared with the control group (Figure 7K).

Given the substantial CTL infiltration observed in MAs following αSPP1 + αPD-1 treatment, we hypothesized that the therapeutic effects of this combination might depend on CTL activity. To test this, we administered CD8-neutralizing antibodies (αCD8) to deplete CTLs and assessed the impact on tumor growth and survival. As expected, CTL depletion significantly reversed the tumor-suppressive effects and survival benefits conferred by αSPP1 + αPD-1 combination therapy, as evidenced by bioluminescence imaging and survival monitoring (Figures 7L–7N).

Collectively, these findings demonstrate that targeting SPP1, either as monotherapy or in combination with αPD-1, effectively reduces TAM infiltration and their immunosuppressive activity in MAs while promoting their polarization toward a pro-inflammatory M1-like phenotype. This therapeutic strategy enhances CTL-mediated anti-tumor immunity, ultimately inhibiting CRC peritoneal metastasis progression and prolonging survival. These results strongly suggest that combining SPP1 blockade with ICB represents a promising strategy to overcome tumor resistance to immunotherapy.

Discussion

Peritoneal metastasis in CRC remains one of the most lethal and therapy-refractory disease states, where resistance to chemotherapy, targeted therapy, and even modern immunotherapies such as ICB or ACT is nearly universal. Previous single-cell studies in gastric and ovarian cancer ascites have described dynamic shifts in myeloid and lymphoid compartments during disease progression and therapy response.¹³^,¹⁹ In CRC, however, studies have largely been limited to treatment-naive settings.¹⁵ Our comprehensive profiling of malignant ascites across treatment-naive, chemo-refractory, and ICB/ACT-resistant patients provides an unprecedented window into how the ascitic TME evolves under therapeutic pressure. These data reveal not only a distinct immune ecosystem but also a tractable vulnerability that can be leveraged for therapeutic intervention. Within this evolving myeloid landscape, FCN1⁺ monocytes in ascites share features with peripheral blood monocytes; however, their transcriptional enrichment for inflammatory recruitment supports their role as active participants in the peritoneal TME rather than mere blood contamination.⁷ SPP1⁺ macrophages have been implicated across multiple cancers.²⁴^,²⁶^,²⁷^,²⁸^,²⁹^,³⁰^,³¹ Extending this myeloid-centric perspective, we further demonstrate that SPP1⁺ CMs progressively expand across distinct therapeutic states of CRC, from treatment-naive to chemo/targeted therapy-refractory and ICB/ACT-resistant settings, where they emerge as a dominant immunosuppressive population.

These macrophages exhibit M2-like polarization, complement activation, and notable enrichment of resident-like and lipid-associated macrophage programs. Moreover, they suppress CD8⁺ T cell function through a dual mechanism: indirectly by reprogramming macrophages toward an immunosuppressive phenotype and directly through engagement of CD44 on CD8⁺ T cells. Importantly, despite the abundance of T and NK cells in CRC ascites, their cytotoxicity is functionally restrained by the expanding SPP1⁺ macrophage compartment, underscoring the centrality of this myeloid barrier in therapy resistance.

Our multi-level validation—scRNA-seq, proteomics, spatial analysis, and functional assays—affirms that SPP1⁺ CMs are not merely correlative but act as active regulators of immune evasion. Spatial mapping in peritoneal metastases demonstrated their juxtaposition to CD8⁺GZMB⁺ T cells, suggesting localized suppression. In vitro co-cultures confirmed impaired CD8⁺ T cell proliferation and cytotoxicity in the presence of SPP1⁺ macrophages. These convergent findings elevate SPP1⁺ macrophages from descriptive biomarkers to functional effectors of immunotherapy resistance in CRC peritoneal metastasis.

Mechanistically, our work establishes a previously unrecognized metabolic axis whereby SPP1 sustains PPARγ-driven macrophage polarization and restrains NF-κB activation. SPP1 deficiency disrupted lipid uptake pathways (CD36/FABP4) and reduced endogenous PPARγ ligands, particularly 15d-PGJ₂, thereby destabilizing the M2-like state. Supplementation with 15d-PGJ₂ alone partially restored M2 programming, which was further enhanced by OA or DHA and fully restored by combining 15d-PGJ₂ with both OA and DHA. These findings position SPP1 as a metabolic gatekeeper linking lipid-derived ligands to macrophage polarization and highlight a therapeutically actionable axis.

From a translational perspective, these results provide strong preclinical rationale for therapeutically targeting SPP1⁺ CMs. Macrophage-specific SPP1 KO or pharmacologic blockade significantly potentiated anti-tumor responses to αPD-1 and ACT in transgenic and ICB-resistant CRC peritoneal metastasis models, while simultaneously reprogramming TAMs toward pro-inflammatory states and unleashing cytotoxic T cell activity. Importantly, the efficacy of this approach was abrogated upon CD8⁺ T cell depletion, confirming that the therapeutic benefit operates through restoration of antitumor immunity rather than nonspecific effects. These findings suggest that SPP1 blockade could serve as a rational adjunct to immunotherapy in CRC peritoneal metastasis, a setting where patients currently have no effective options. Beyond its role as a therapeutic target, SPP1 may also serve as a clinically informative biomarker. A recent study by Liu et al. demonstrated that plasma SPP1 levels are elevated in metastatic CRC and are associated with poor response to immunotherapy,³¹ supporting its prognostic and predictive potential.

In summary, our study uncovers SPP1⁺ CMs as central drivers of therapy resistance in CRC peritoneal metastasis through an immunometabolic program linking PPARγ and NF-κB signaling. By demonstrating that pharmacologic or antibody-mediated SPP1 blockade restores sensitivity to ICB and ACT, we highlight a tangible therapeutic opportunity. Targeting SPP1⁺ macrophages could, therefore, extend the benefits of immunotherapy to a patient population that currently faces dismal outcomes, bridging mechanistic discovery with translational potential in one of the most challenging subtypes of CRC.

Limitations of the study

Several limitations should be acknowledged. The ICB/ACT-resistant cohort was limited in size, which restricted statistical power for subgroup analyses and correlation with clinical parameters. The absence of a therapy-responsive comparator group precluded definitive assignment of resistance-specific features. Ascites and peritoneal nodules represent distinct biological compartments; although we used them complementarily, direct extrapolation between compartments should be avoided. Future prospective studies with larger, well-annotated cohorts and samples collected across different treatment phases, including multicenter cohort validation and further clinical trials targeting SPP1, will be required to validate our findings and further delineate the clinical relevance of the SPP1⁺ macrophage axis. This study did not characterize the pharmacokinetics and tissue distribution of the anti-SPP1 antibody used in preclinical experiments. Future clinical translation studies will need to further optimize antibody dosing regimens and tumor targeting properties.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Xuanwen Bao (xuanwen.bao@zju.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

The proteomics data have been deposited in the Integrated Proteome Resources (https://www.iprox.cn) (Database: IPX0012394000). The scRNA-seq data have been deposited in the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA011790), and are publicly accessible at https://ngdc.cncb.ac.cn/gsa-human. This study does not report custom code. Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (82473399 to X.D., 82473471 to X.B., and 82373428 to W.F.), the Zhejiang Provincial Natural Science Foundation of China (LY23H160013 to X.B.), the Major Scientific Project of Zhejiang Province (2023C03061 to W.F.), the “Pioneer” and “Leading Goose” R&D Program of Zhejiang Province (2023C03061 to W.F. and 2025C02073 to P.Z.), the Open Research Fund of Hubei Key Laboratory of Precision Radiation Oncology (jzfs015 to X.D.), and Beijing Science And Technology Innovation Medical Development Foundation (KC2023-JX-0288-FM50 to X.D.). We thank the Core Laboratory, First Affiliated Hospital, Zhejiang University School of Medicine, for technical support.

Author contributions

X.D., C. Lai, L.H., and Y.J. contributed equally to this work. Conceptualization, X.D., X.Y., W.F., S.X., and X.B.; methodology, X.D., C. Lai, J.C., Y.J., and Haibo Zhang; investigation, X.D., C. Lai, L.H., H.Y., X.S., and B.L.; formal analysis, X.D., C. Lai, L.H., C. Liu, and Hangyu Zhang; data curation, Y.L., D.W., and P.Z.; validation, J.C., Y.S., and Haibo Zhang; resources, W.F., Y.S., and X.B.; writing – original draft, X.D., C. Lai, and L.H.; writing – review & editing, X.D., X.Y., W.F., S.X., and X.B.; visualization, X.D. and Y.J.; supervision, W.F., S.X., X.Y., and X.B.; project administration, W.F. and X.B.; funding acquisition, X.D., W.F., P.Z., and X.B.

Declaration of interests

The authors declare no potential conflicts of interest.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies

Rabbit monoclonal Anti-NF-kB p65	ABclonal	Cat#A19653;RRID:AB_2862717
Rabbit monoclonal Anti-IκBα(44D4)	CST	Cat#4812T;RRID:AB_10694416
Mouse monoclonal Anti-Phospho-IκBα	CST	Cat#9246T;RRID:AB_2267145
Rabbit monoclonal Anti-CD36	CST	Cat#28109T;RRID:AB_3675401
Rabbit monoclonal Anti-FABP4	CST	Cat#50699T;RRID:AB_3752313
Rabbit monoclonal Anti-SPP1	CST	Cat#88742S;RRID:AB_3107209
Rabbit monoclonal Anti-PPARγ	CST	Cat#2435T;RRID:AB_2166051
Rabbit monoclonal Anti-CD8	CST	Cat#98941T;RRID:AB_2756376
Rabbit monoclonal Anti-CD4	CST	Cat#48274S;RRID:AB_3076699
Mouse monoclonal Anti-β-Actin	Proteintech	Cat#66009-1-Ig;RRID:AB_2782959
Rabbit polyclonal Anti-PCNA	Proteintech	Cat#10205-2-AP;RRID:AB_2160330
Mouse monoclonal Fc-blocking Anti-CD16/32	BioLegend	Cat#156604;RRID:AB_2783138
Mouse monoclonal anti-CD206-APC	BioLegend	Cat#141708;RRID:AB_10900231
Mouse monoclonal anti-Arginase 1-PE	BioLegend	Cat#165804;RRID:AB_3068116
Mouse monoclonal anti-CD86-PE	BioLegend	Cat#159204;RRID:AB_2832568
Mouse monoclonal anti-CD45-BV785	BioLegend	Cat#103149;RRID:AB_2564590
Mouse recombinant anti-CD45-FITC	BioLegend	Cat#157608;RRID:AB_2832554
Mouse recombinant anti-Granzyme B-BV421	BioLegend	Cat#396414;RRID:AB_2810603
Mouse monoclonal anti-IFN-γ-PE-Cy7	BioLegend	Cat#505826;RRID:AB_2295770
Human monoclonal anti-CD4-PerCP/Cyanine5.5	BD	Cat# 550954;RRID:AB_393977
Mouse monoclonal anti-CD3-AF700	BioLegend	Cat#100216;RRID:AB_493697
Mouse monoclonal anti-PD-L1-BV711	BioLegend	Cat#124319;RRID:AB_2563619
Mouse monoclonal anti-CD44	BioLegend	Cat#103046;RRID:AB_2561491
InVivoMAb anti-SPP1	BioXCell	Cat#BE0382;RRID:AB_2927519
InVivoMAb anti-mouse PD-1	BioXCell	Cat#BE0146;RRID:AB_10949053
InVivoMAb isotype control	BioXCell	Cat#BE0089;RRID:AB_1107769
Rabbit monoclonal Anti-Granzyme B	abcam	Cat#ab255598;RRID:AB_2860567
Rabbit multiclonal Anti-SPP1	abcam	Cat#ab283656;RRID:AB_2894861
Rabbit monoclonal Anti-FOXP3	abcam	Cat#ab215206;RRID:AB_2860568
Rabbit monoclonal Anti-CD68	abcam	Cat#ab213363; RRID:AB_2801637

Biological samples

Conditional SPP1^ff; Lyz2-Cre mice	Shanghai Model Organisms Center	N/A
C57BL/6 mice	Zhejiang Center of Laboratory Animals	N/A
OT-1 transgenic mice	Zhejiang Center of Laboratory Animals	N/A

Chemicals, peptides, and recombinant proteins

Recombinant Mouse IL-2	BioLegend	Cat#575404
Recombinant Mouse IL-4	BioLegend	Cat#574304
Recombinant Mouse IL-13	BioLegend	Cat#575904
Recombinant Mouse Osteopontin	BioLegend	Cat#763606
DHA	Beyotime	Cat#ST1267
OA	Aladdin	Cat#O108484
BSA	Aladdin	Cat#B741824

Critical commercial assays

MojoSort™ Mouse CD8 T cell Isolation Kit	BioLegend	Cat#480008
Mouse 15 days-PGJ₂ elisa Kit	Crgent Biotech	Cat#ERJ0708
Zombie NIR™ Fixable Viability Kit-APC-Cy7	BioLegend	Cat#423105
Zombie Aqua™ Fixable Viability Kit-BV510	BioLegend	Cat#423101
Dynabeads® Mouse T-Activator CD3/CD28	Life Technologies	Cat#11452D

Deposited data

Single-cell sequencing data	This paper	HRA011790
Proteomics data	This paper	IPX0012394000

Experimental models: Cell lines

Mc38 cell line	Cell Resource Center, Institute of Basic Medicine of the Chinese Academy of Medical Sciences.	Cat#1101MOU-PUMC000523
CT26 cell line	Cell Resource Center, Institute of Basic Medicine of the Chinese Academy of Medical Sciences.	Cat# 1101MOU-PUMC000275
ANA-1 cell line	Cell Resource Center, Institute of Basic Medicine of the Chinese Academy of Medical Sciences.	Cat#3101MOUGNM 2

Oligonucleotides

See Table S5 for primer

Recombinant DNA

psPAX2	Miaoling Biotech	Cat#P0261
pMD2.G	Miaoling Biotech	Cat#P0262
LentiCas9-Blast	Miaoling Biotech	Cat#P41051
pLV3-U6-SPP1(mouse)-sgRNA-Puro	Miaoling Biotech	Cat#G50693
pLV3-CMV-OVAL (chicken)-3×FLAG-Puro	Miaoling Biotech	Cat#P46845
pLV3-CMV-MCS-Fluc-Puro	Miaoling Biotech	Cat#P21253

Software and algorithms

R software (version 4.4.1)	R Core	https://www.r-project.org/
ImageJ software (version 4.0)	NIH	https://imagej.nih.gov/ij/
GraphPad Prism (v8.0)	Dotmatics	https://www.graphpad.com/
Python (v3.12.7)	Python	https://www.python.org/downloads/release/python-3127/
ConsensusClusterPlus (version 3.16)	Matt Wilkerson, Peter Waltman	https://git.bioconductor.org/packages/ConsensusClusterPlus
Image-Pro Plus software (version 6.0)	Media Cybernetics	https://www.mediacy.com/78-products/image-pro-plus
Flowjo	Flowjo	https://www.flowjo.com/

Other

Gene sets	The Broad Institute	https://www.gsea-msigdb.org/gsea/msigdb/

Open in a new tab

Experimental model and study participant details

Cell lines and cell culture

The mouse macrophage cell lines ANA-1, CRC cell lines MC38 and CT26 were obtained from Cell Resource Center, Institute of Basic Medicine of the Chinese Academy of Medical Sciences and used within 20 passages. The cell lines were authenticated by the supplier and tested negative for mycoplasma contamination. Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) or Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco), supplemented with 10% fetal calf serum (FCS) (Gibco) and 1% penicillin-streptomycin, under standard conditions (37°C and 5% CO₂).

Animal model

All animal husbandry and experimental procedures, including animal housing and diet, were performed under the guidelines, and were approved by the Institutional Animal Care and Use Committee (IACUC) and ZJLA (ethical number: ZJCLA-IACUC-20010711).

Model 1: In vivo syngeneic tumor therapy model

Conditional SPP1^ff; Lyz2-Cre mice on a C57BL/6 J genetic background were generated by Shanghai Model Organisms Center, Inc., subsequently bred and maintained at ZJCLA. Cohorts of age-matched SPP1^ff; Lyz2-Cre mice and control SPP1^ff littermates received intraperitoneal injections of 5 × ×10⁶ MC38-luc tumor cells. Ten days postinoculation, tumor-bearing mice were randomized into designated treatment groups and subjected to intraperitoneal administration of therapeutic agents every 72 h, comprising either αPD-1 monoclonal antibody (100 μg), αSPP1 blocking antibody (100 μg), isotype control antibody (100 μg), or specified combinations thereof. Tumor burden was longitudinally monitored via sequential bioluminescence imaging, and survival endpoints were recorded for Kaplan-Meier analysis. Upon completion of therapeutic interventions, peritoneal metastatic lesions were harvested for multiplex immunohistochemical profiling.

Model 2: In vivo adoptive T cell therapy model

C57BL/6 mice (6–7 weeks old) were obtained from ZJCLA and housed in pathogenfree facilities at ZJCLA. Mice received intraperitoneal injections of 10 7 MC38-OVA tumor cells. CD8⁺ T cells were isolated from spleens of OT-1 transgenic mice [C57BL/6-Tg (TcraTcrb) 1100Mjb/J] using the MojoSort Mouse CD8⁺ T cell Isolation Kit, followed by activation with Dynabeads Mouse T-Activator CD3/CD28 and recombinant IL-2 (20 ng/mL, 575404, BioLegend). Sixteen days post-tumor inoculation, tumor-bearing mice were randomized into treatment cohorts and administered activated OT-1 CD8⁺ T cells (10⁶ cells per mouse), αSPP1 blocking antibody (100 μg q3d × 3), isotype control antibody (100 μg q3d × 3), or combination therapies via intraperitoneal injection. Survival was tracked for Kaplan-Meier analysis.

Model 3: In vivo model of resistance to ICB therapy

BALB/c mice (6–7 weeks old) received intraperitoneal injections of 5 × 10⁵ CT26-luc tumor cells. For therapeutic assessment, nine days post-inoculation, tumor-bearing mice were randomized into cohorts and administered αPD-1 monoclonal antibody (100 μg q3d × 3), αSPP1 blocking antibody (100 μg q3d × 3), isotype control antibody (100 μg q3d × 3), or combinations thereof intraperitoneally. Longitudinal tumor burden was quantified by sequential bioluminescence imaging, survival endpoints were recorded for Kaplan-Meier analysis, and ascites fluid was collected for flow cytometric analysis upon treatment completion. For the CD8⁺ T cell deletion assay, using identical inoculation, tumor-bearing mice were randomized at eight days and intraperitoneally administered αPD-1 (100 μg q3d × 3), αSPP1 (100 μg q3d × 3), αCD8 blocking antibody (100 μg q3d × 3), isotype control (100 μg q3d × 3), or combinations thereof, with longitudinal tumor burden and survival similarly monitored.

Human subjects

Study cohorts

This study utilized two in-house CRC peritoneal metastasis cohorts. The ascites scRNA-seq cohort included 20 participants (11 male, 9 female). The peritoneal metastasis cohort used for proteomic analysis comprised 36 participants (53% male, 47% female). Detailed demographic and clinical information, including age, gender, disease stage, treatment history, and group assignment, is provided in Tables S1 and S4. Peritoneal metastasis tissues were retrieved from the pathology department archives.

Ethical approval

All ascites and peritoneal metastasis tissue collections and experiments were approved by the Ethics Committee of the First Affiliated Hospital, Zhejiang University School of Medicine (IIT2025B0248) and the Research Ethics Board of Zhejiang Provincial People’s Hospital (KY2025187). The study was conducted in accordance with the International Ethical Guidelines for Biomedical Research Involving Human Subjects (CIOMS). The authors are accountable for all aspects of the work, ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Inclusion criteria

Patients were eligible if they met the following criteria: (1) written informed consent obtained; (2) age ≥18 years at study entry; (3) histopathologically confirmed primary colorectal cancer with peritoneal metastases and malignant ascites confirmed by imaging (e.g., CT) or other examinations; (4) for the treatment-naïve group, no anti-cancer therapy (including chemotherapy, immunotherapy, or targeted therapy) within at least one year prior to sample collection; (5) for the chemo/targeted therapy-refractory group, samples collected at least one month but less than six months after the most recent chemotherapy or targeted therapy, with disease progression confirmed by RECIST 1.1 criteria; (6) for the ICB/ACT-resistant group, samples collected at least one month but less than six months after the most recent immunotherapy (e.g., anti-PD-1 antibody, CAR-T, or CAR-NK), with disease progression confirmed by RECIST 1.1 criteria; (7) no contraindications to paracentesis (e.g., absence of coagulopathy or other conditions precluding safe needle insertion).

Exclusion criteria

Patients were excluded if they met any of the following criteria: (1) lack of written informed consent; (2) age <18 years at study entry; (3) receipt of systemic or intraperitoneal anti-cancer therapy (including chemotherapy, immunotherapy, or targeted therapy) within one month prior to sample collection; (4) presence of contraindications to paracentesis, such as coagulopathy, active abdominal infection, or inability to safely perform the procedure; (5) pregnancy or lactation; (6) concurrent severe illness, such as active infection; (7) presence of other primary malignancies; (8) presence of hemorrhagic ascites or chylous ascites.

Method details

Ascites processing and single-Cell RNA sequencing analysis

Ascites samples were processed through sequential filtration and centrifugation steps to obtain high-quality single-cell suspensions. Initial filtration through a 40 μm cell strainer (Falcon) removed debris, followed by centrifugation at 300 × g for 5 min at 4°C to pellet cells. After two washes in culture medium, cells were resuspended in RPMI 1640 medium supplemented with 0.04% BSA to minimize aggregation. CD45⁺ immune cell enrichment was performed using species-specific MicroBeads (Miltenyi Biotec) according to the manufacturer’s protocol, with cell concentration and viability assessed by automated counting or trypan blue exclusion.

For single-cell transcriptome analysis, cell suspensions were adjusted to 700–1200 cells/μL and processed using the MobiCube High-throughput Single Cell 3′ Transcriptome Set V2.1. Libraries were prepared following the manufacturer’s instructions and sequenced on an Illumina NovaSeq 6000 platform (PE150 configuration). Raw sequencing reads were aligned to the GRCh38 human reference genome using MobiVision software (v3.2), with UMI counting and barcode assignment performed during data processing.

The resulting UMI count matrix underwent rigorous quality control in Seurat (v4.0.0), where cells were filtered based on multiple parameters including gene content (<200 genes), UMI counts (<1000), gene-to-UMI ratio (log10GenesPerUMI <0.7), and mitochondrial/hemoglobin gene expression (>10% and >5%, respectively). Potential doublets were identified and removed using DoubletFinder (v2.0.3) prior to data normalization with the LogNormalize method.

Following quality control, the top 2000 highly variable genes were selected using Seurat’s FindVariableGenes function with FastExpMean and FastLogVMR parameters. Dimensionality reduction was achieved through principal component analysis, with batch effects corrected using Harmony (v1.0). Graph-based clustering identified distinct cell populations that were visualized in two dimensions using uniform manifold approximation and projection (UMAP). Cluster-specific marker genes were identified through differential expression testing (presto method), with statistical significance determined using Bonferroni-adjusted p-values. Differential expression analysis between conditions was performed using the same statistical framework, maintaining consistency across all comparative analyses. Cell differentiation trajectories were reconstructed using Monocle2 to model the dynamic transitions between transcriptional states. The top differentially expressed genes identified during Seurat clustering were selected for dimensionality reduction with DDRTree algorithm. Pseudotime values were calculated by ordering cells along the minimum spanning tree rooted in the monocyte population.

Proteomics processing

The formalin-fixed paraffin-embedded (FFPE) samples mounted on staining racks were first equilibrated in an oven at 37°C for 30 min to improve subsequent solvent penetration. The samples were then subjected to a sequential dewaxing procedure involving two 10-min incubations in heptane at room temperature, with complete solution replacement between incubations to ensure thorough paraffin removal. Following dewaxing, a graded ethanol series (100%, 90%, and 75% ethanol, 5 min each) was used to gradually rehydrate the tissue sections while maintaining structural integrity. The rehydrated samples were carefully transferred to PCT tubes using a blade, and 12.5 μL of 100 mM Tris-HCl buffer (pH 10) was added to each tube. Protein extraction was performed under denaturing conditions at 95°C with constant agitation (600 rpm) for 30 min to ensure complete solubilization.

The extracted proteins underwent reduction and alkylation in PCT tubes containing 6 M urea, 2 M thiourea, and 100 mM TEAB, supplemented with 200 mM TCEP and 800 mM IAA. This process was carried out through 90 pressure cycles (45,000 psi for 30 s followed by ambient pressure for 10 s) at 30°C. Following reduction/alkylation, the urea concentration was reduced below 1.5 M by adding 85 μL of 100 mM TEAB. Enzymatic digestion was then performed by adding trypsin (0.5 μg/μL) and Lys-C (0.25 μg/μL) under 20,000 psi pressure through 120 cycles (50 s high pressure/10 s ambient pressure) at 30°C. The digestion was terminated by acidification with 10% trifluoroacetic acid, and the resulting peptides were transferred to clean microcentrifuge tubes.

Prior to desalting, the sample pH was confirmed to be between 2 and 3 using pH indicator strips. Peptide cleanup was performed using SOLAμ SPE cartridges (Thermo Fisher Scientific) following the manufacturer’s protocol. The purified peptides were then labeled using the TMTpro 16plex Isobaric Label Reagent Set according to the manufacturer’s instructions, with careful monitoring of labeling efficiency.

Peptide fractionation was performed using a Waters XBridge Peptide BEH C18 column (5 μm, 300 Å Å, 4.6 × 250 mm) on a Dionex UltiMate 3000 system. The mobile phases consisted of 10 mM ammonium hydroxide (pH 10) as buffer A and 98% acetonitrile/10 mM ammonium hydroxide (pH 10) as buffer B. Peptides were separated using a 60-minute gradient from 5% to 35% buffer B at a flow rate of 0.5 mL/min, with fractions collected every minute. The collected fractions were pooled into 30 final fractions, dried by SpeedVac centrifugation, and reconstituted in 2% acetonitrile/0.1% formic acid for LC-MS/MS analysis.

Mass spectrometric analysis was performed on an Orbitrap Exploris 480 mass spectrometer equipped with a FAIMS Pro interface and coupled to a Dionex UltiMate 3000 RSLCnano system. Peptides were loaded onto a trap column (3 μm, 100 Å Å, 20 × 75 μm) at 6 μL/min for 4 minutes and then separated on an analytical column (1.9 μm, 120 Å Å, 150 × 75 μm) using a 30-minute gradient from 7% to 30% buffer B (98% acetonitrile/0.1% formic acid) at 300 nL/min. MS1 spectra were acquired at 60,000 resolution (m/z 375–1800) with an AGC target of 300% and maximum injection time of 50 ms. Data-dependent MS/MS scans were performed at 30,000 resolution with an isolation window of 0.7 m/z, AGC target of 200%, and maximum injection time of 86 ms. All solvents used were mass spectrometry grade to minimize background interference.

Macrophage-CD8⁺ T cell co-culture

CD8⁺ T cells were isolated from the spleens of C57BL/6 mice using the MojoSort Mouse CD8⁺ T cell Isolation Kit (480008, BioLegend). ANA-1 control and ANA-1 SPP1-KO macrophages were stimulated with IL-4 (40 ng/mL, 574304, BioLegend) and IL-13 (20 ng/mL, 575904, BioLegend) for 24 h. CD8⁺ T cells were labeled with CFSE and activated using Dynabeads Mouse T-Activator CD3/CD28 (11452D, Life Technologies). The activated CD8⁺ T cells were then co-cultured with IL-4/IL-13-pretreated ANA-1 control or ANA-1 SPP1-KO macrophages for 3 days. Proliferation and GZMB expression of CD8⁺ T cells were subsequently assessed by flow cytometry.

SPP1 rescue experiment in macrophage-CD8⁺ T cell co-culture

CD8⁺ T cells were isolated from the spleens of C57BL/6 mice using the MojoSort Mouse CD8⁺ T cell Isolation Kit. ANA-1 control and ANA-1 SPP1-KO macrophages were stimulated with IL-4 (40 ng/mL) and IL-13 (20 ng/mL) for 24 h. For the rescue group, ANA-1 SPP1-KO macrophages were additionally treated with recombinant SPP1 (1 μg/mL, 763606, BioLegend) during the stimulation period. CD8⁺ T cells were activated using Dynabeads Mouse T-Activator CD3/CD28. Prior to co-culture, all macrophage groups were washed thoroughly with PBS to eliminate residual factors. The activated CD8⁺ T cells were then co-cultured with the pretreated macrophages for 3 days, and GZMB expression in CD8⁺ T cells was assessed by flow cytometry.

Effect of SPP1 neutralization in macrophage-CD8⁺ T cell co-culture

CD8⁺ T cells were isolated from the spleens of C57BL/6 mice using the MojoSort Mouse CD8⁺ T cell Isolation Kit. ANA-1 macrophages were stimulated with IL-4 (40 ng/mL) and IL-13 (20 ng/mL) for 24 h. CD8⁺ T cells were activated using Dynabeads Mouse T-Activator CD3/CD28. The activated CD8⁺ T cells were then co-cultured with IL-4/IL-13-pretreated ANA-1 macrophages for 3 days in the absence or presence of an anti-SPP1 neutralizing antibody (100 ng/mL, BE0382, BioXCell). GZMB expression in CD8⁺ T cells was subsequently assessed by flow cytometry.

Direct effect of SPP1 on CD8⁺ T cells via CD44 engagement

CD8⁺ T cells were isolated from the spleens of C57BL/6 mice using the MojoSort Mouse CD8⁺ T cell Isolation Kit. The cells were activated using Dynabeads Mouse T-Activator CD3/CD28 and then treated with control medium, recombinant SPP1 (1 μg/mL, 763606, BioLegend), or recombinant SPP1 combined with an anti-CD44 blocking antibody (100 ng/mL, 103046, BioLegend). After culture, GZMB expression in CD8⁺ T cells was assessed by flow cytometry.

Bulk RNA-seq data processing

Total RNA was extracted using TRIzol reagent (Invitrogen) following the manufacturer’s protocol. RNA concentration and purity were assessed with the NanoDrop ND-1000 spectrophotometer (Thermo Fisher), and RNA integrity was confirmed using the Bioanalyzer 2100 system (Agilent Technologies), retaining only samples with RNA integrity number (RIN) > 7.0. Additional quality confirmation was performed via denaturing agarose gel electrophoresis. Poly(A)+ mRNA was isolated from 1 μg of total RNA using Dynabeads Oligo(dT)₂₅ (Thermo Fisher) with two rounds of purification. Purified mRNA was fragmented at 94°C for 5–7 min using the Magnesium RNA Fragmentation Module (NEB), followed by firststrand cDNA synthesis with SuperScript II Reverse Transcriptase (Invitrogen). Secondstrand synthesis incorporated dUTP using E. coli DNA Polymerase I and RNase H (NEB), generating U-labeled double-stranded cDNA. Blunt-ended fragments were Atailed and ligated to indexed adapters with T-overhangs. After size selection using AMPure XP beads (Beckman Coulter), heat-labile UDG enzyme (NEB) was applied to degrade the U-labeled second strand. The ligated DNA was PCR-amplified under the following cycling conditions: 95°C for 3 min; 8 cycles of 98°C for 15 s, 60°C for 15 s, and 72°C for 30 s; followed by a final extension at 72°C for 5 min. The resulting libraries had an average insert size of ∼300 ± 50 bp. Sequencing was performed using paired-end 150-bp reads (PE150) on the Illumina NovaSeq 6000 platform (LC-Bio, Hangzhou, China) according to the manufacturer’s instructions. Raw reads were first processed using fastp to remove adapter contamination, lowquality reads, and undetermined bases. Clean reads were then aligned to the GRCh38 human reference genome using HISAT2, and aligned reads were assembled into transcripts with StringTie using default parameters. Transcriptomes from all samples were merged using gffcompare to generate a unified transcript annotation. Transcript expression was quantified as Fragments Per Kilobase of transcript per Million mapped reads (FPKM). Differential expression analysis was conducted using the deseq2 R package, applying a parametric F-test based on nested linear models.

Metabolite extraction, LC-MS acquisition, and data analysis

Cell samples were thawed on ice, and metabolites were extracted using a cold-phase lipid extraction buffer (isopropanol:acetonitrile:water, 2:1:1, v/v/v). Briefly, 100 mg of tissue was homogenized in 1 mL of extraction buffer, vortexed for 1 min, incubated at room temperature for 10 min, and stored overnight at −20°C. After centrifugation at 4,000 × g for 20 min at 4°C, supernatants were transferred to 96-well plates and stored at −80°C until analysis. A pooled quality control (QC) sample was generated by mixing 10 μL aliquots from each extract to monitor instrument stability and reproducibility. Chromatographic separation was performed on an ACQUITY UPLCsystem (Waters) equipped with a Kinetex C18 column (100 mm × 2.1 mm, 100 Å Å; Phenomenex). The column temperature was maintained at 55 °C with a flow rate of 0.3 mL/min. Mobile phases were composed of solvent A (acetonitrile: water = 6:4, v/v, with 0.1% formic acid) and solvent B (isopropanol: acetonitrile = 9:1, v/v, with 0.1% formic acid). Gradient elution proceeded as follows: 0–0.4 min, 30% B; 0.4–1 min, 30–45% B; 1–3 min, 45–60% B; 3.5–5 min, 60–75% B; 5–7 min, 75–90% B; 7–8.5 min, 90–100% B; 8.5–8.6 min, 100% B; 8.6–8.61 min, 100–30% B; 8.61–10 min, re-equilibration at 30% B.

Eluted metabolites were detected using a Q Exactive high-resolution mass spectrometer (Thermo Scientific) operated in both positive and negative ion modes. Full MS scans (m/z 70–1,050) were acquired at a resolution of 70,000 with an AGC target of 3 × 10⁶ and a maximum injection time of 100 ms. Data-dependent MS/MS was performed in Top-3 mode at 17,500 resolution with an AGC target of 1 × 10⁵ and 80 ms maximum injection time. QC samples were injected every 10 runs to monitor signal stability and batch effects.

Raw data files were converted to mzXML format and processed using the XCMS and metaX R packages. Preprocessing included peak detection, retention time correction, and grouping of isotopes and adducts. Features were annotated by matching accurate mass (±10 ppm) against KEGG and HMDB databases, and confirmed with isotopic pattern validation. Additional confirmation of metabolite identity was performed using an in-house MS/MS spectral library.

Downstream statistical analysis was performed in R v4.4.1. Metabolites with fold change >1.2 and p < 0.05 (two-sided t test) were considered significantly altered.

Gene set enrichment and functional annotation

Pathway analysis was conducted using the Molecular Signatures Database hallmark gene sets. GSEA was performed with 1,000 permutations using weighted enrichment statistics and signal-to-noise ratio for gene ranking, with significance thresholds set at FDR <0.25 and nominal p-value <0.05. For single-sample analysis, ssGSEA scores were calculated using gsva package. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were carried out to identify significantly enriched biological terms, with Benjamini-Hochberg adjusted pvalues <0.05 considered significant.

Multiplex immunofluorescence staining and spatial analysis

Paraffin-embedded tissue sections (4–5 μm thickness) were processed through a comprehensive multiplex immunofluorescence staining protocol. Following standard deparaffinization, antigen retrieval was performed using microwave treatment (100% power for 5 min followed by 60% power for 15 min) in citrate-based buffer. Non-specific binding sites were blocked with protein blocking solution for 30 min at room temperature prior to primary antibody incubation (1–2 h at room temperature).

Immunofluorescence staining employed iterative cycles of: (i) primary antibody incubation; (ii) secondary antibody detection (30 min, RT); and (iii) tyramide signal amplification (TSA) with fluorophore-conjugated tyramides (TissueGnostics), each followed by rigorous PBS washes. Nuclear counterstaining used DAPI (10 min). Targeted markers included: anti-Granzyme B (ab255598), anti-SPP1 (ab283656), antiFOXP3 (ab215206), anti-CD8 (98941 T, CST), anti-CD4 (48274 S, CST), and anti-CD68 (ab213363) (all antibodies from Abcam unless noted). Spatial relationships between SPP1⁺ macrophages and CD8⁺ T cells were quantified using HALO™ for colocalization analysis and cell-to-cell distance measurements.

Lentiviral transduction for stable cell line

Lentiviral vectors for luciferase and ovalbumin overexpression and SPP1 konckout were obtained from Miaoling Biotech, with an empty vector serving as the control. The vectors were co-transfected with packaging plasmids psPAX2 and pMD2.G into HEK293T cells using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. Viral supernatants were harvested 48 h post-transfection and used to transduce MC38, CT26 and ANA-1 cells in six-well plates in the presence of Polybrene (10 μg/mL, P8230, Solarbio). Transduced cells were selected using puromycin.

Quantitative real-time PCR

Total RNA was isolated from cells using RNA Rapid Extraction Kit (Biosharp, #RN001), with concentration determined via NanoDrop ND-1000 spectrophotometry (Thermo Fisher Scientific). Purified RNA was reverse-transcribed into cDNA using Evo M-MLV RT Premix for qPCR (AG, #AG11706). Quantitative real-time PCR (qRT-PCR) analyses were performed on a Bio-Rad CFX system with SYBR Green Premix Pro Taq HS qPCR Kit (AG, #AG11706), employing the comparative Cq (threshold cycle; 2−ΔΔCt) method for relative quantification. Expression values were normalized to ACTIN as the endogenous control, with primer sequences detailed in Table S5.

Western blot analysis

Nuclear proteins were isolated using a Membrane and Nuclear Protein Extraction Kit (C500009-0050; Sangon Biotech). Cells were lysed in RIPA buffer, and proteins separated by SDS-PAGE were transferred onto polyvinylidene fluoride (PVDF) membranes. After blocking with 5% skim milk/TBST for 1 h at room temperature, membranes were incubated with primary antibodies overnight at 4°C. Following TBST washes, membranes were probed with diluted secondary antibodies and imaged using NcmECL Ultra substrate. Primary antibodies included: anti-NF-κB p65 (A19653; ABclonal), anti-IκBα (4812 T; CST), anti-Phospho-IκBα (Ser32/36) (9246 T; CST), anti-CD36 (28109 T; CST), and anti-FABP4 (50699 T; CST), anti-SPP1 (88742 S; CST), anti-PPARγ (2435 T; CST), anti-β-ACTIN (66009-1-Ig; Proteintech), anti-PCNA (10205-2-AP; Proteintech).

Preparation of BSA-conjugated free fatty acids

Free fatty acids (oleic acid (OA) and docosahexaenoic acid (DHA)) and 15-deoxyΔ12,14-prostaglandinJ₂ (15 days-PGJ₂) were dissolved in 150 mM NaCl (final concentration 25 mM, pH 7.4). Specifically, 50 mg aliquots were dissolved in 7 mL NaCl solution containing 48 μL NaOH, followed by vortexing and heating at 65°C (37°C for 15days-PGJ₂) until complete dissolution. These 25 mM solutions were mixed with ice-cold 24% bovine serum albumin (BSA) at a 54:46 volume ratio (10% BSA for 15 days-PGJ₂), yielding ∼12.5 mM conjugates (pH 7.4). Conjugates were stored at −20°C and vortex-mixed for 10 min at room temperature before use.

Flow cytometric analysis

Single-cell suspensions were prepared using standardized protocols. Cells were simultaneously stained with Fc-blocking anti-CD16/32 antibody (BioLegend #156604), viability dyes [Zombie NIR APC-Cy7 (BioLegend #423105), and Zombie NIR BV510 (BioLegend #423101)], and the following fluorescently conjugated antibodies: APCCD206(BioLegend #141708), PE-arginase 1 (BioLegend #165804), PE-CD86 (BioLegend #159204), BUV395-CD11b (BD Biosciences #743983), BV786-CD45 (BioLegend #103149), FITC-CD45 (BioLegend #157608), BV421-GZMB (BioLegend #396414), PE-Cy7-IFN-γ (BioLegend #505826), PerCP/Cyanine5.5-CD4 (BD #550954), AF700-CD3 (BioLegend #100216), BV711-PD-L1 (BioLegend #124319). Data acquisition was performed on a 5-laser BD Fortessa flow cytometer (BD Biosciences).

Quantification and statistical analysis

Analyses were conducted using R (v4.4.1) and Python (v3.12.7) for sequencing data, and GraphPad Prism (v8.0) for experimental data. Two-group comparisons used unpaired two-tailed Student’s t-tests. Single-variable multiple comparisons employed one-way ANOVA with Dunnett’s test. Two-variable comparisons used two-way ANOVA with Tukey’s test. Survival curves were analyzed by log rank test. Data represent mean ± SD or mean ± SEM. Significance: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (ns, not significant).

Published: July 16, 2026

Footnotes

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.xcrm.2026.102930.

Contributor Information

Xiaomeng Dai, Email: dxm1106@zju.edu.cn.

Weijia Fang, Email: weijiafang@zju.edu.cn.

Shan Xin, Email: shan.xin@yale.edu.

Xia Yu, Email: zjuyuxia@zju.edu.cn.

Xuanwen Bao, Email: xuanwen.bao@zju.edu.cn.

Supplemental information

Document S1. Figures S1–S10 and Tables S1–S5

mmc1.pdf^{(5.2MB, pdf)}

Document S2. Article plus supplemental information

mmc2.pdf^{(39.6MB, pdf)}

References

1.Fakih M.G. Metastatic colorectal cancer: current state and future directions. J. Clin. Oncol. 2015;33:1809–1824. doi: 10.1200/JCO.2014.59.7633. [DOI] [PubMed] [Google Scholar]
2.Ceelen W.P., Bracke M.E. Peritoneal minimal residual disease in colorectal cancer: mechanisms, prevention, and treatment. Lancet Oncol. 2009;10:72–79. doi: 10.1016/S1470-2045(08)70335-8. [DOI] [PubMed] [Google Scholar]
3.Berlin C., Mauerer B., Cauchy P., Luenstedt J., Sankowski R., Marx L., Feuerstein R., Schaefer L., Greten F.R., Pesic M., et al. Single-cell deconvolution reveals high lineage- and location-dependent heterogeneity in mesenchymal multivisceral stage 4 colorectal cancer. J. Clin. Investig. 2023;134 doi: 10.1172/JCI169576. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Li Q., Xiao Y., Han L., Luo W., Dai W., Fang H., Wang R., Xu Y., Cai S., Goel A., et al. Microbiome dysbiosis, neutrophil recruitment and mesenchymal transition of mesothelial cells promotes peritoneal metastasis of colorectal cancer. Nat. Cancer. 2025;6:493–510. doi: 10.1038/s43018-025-00910-9. [DOI] [PubMed] [Google Scholar]
5.Bhat A.A., Nisar S., Singh M., Ashraf B., Masoodi T., Prasad C.P., Sharma A., Maacha S., Karedath T., Hashem S., et al. Cytokine- and chemokine-induced inflammatory colorectal tumor microenvironment: Emerging avenue for targeted therapy. Cancer Commun. 2022;42:689–715. doi: 10.1002/cac2.12295. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Ceelen W., Ramsay R.G., Narasimhan V., Heriot A.G., De Wever O. Targeting the Tumor Microenvironment in Colorectal Peritoneal Metastases. Trends Cancer. 2020;6:236–246. doi: 10.1016/j.trecan.2019.12.008. [DOI] [PubMed] [Google Scholar]
7.Zhang Y., Qian L., Chen K., Gu S., Meng Z., Wang J., Li Y., Wang P. Oncolytic adenovirus in treating malignant ascites: A phase II trial and longitudinal single-cell study. Mol. Ther. 2024;32:2000–2020. doi: 10.1016/j.ymthe.2024.04.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Li B., Zhu X., Ge J., Zhang H., Gao Y., Bao X., Dai X., Du Z., Tu X., Tong Z. Intraperitoneal infusion of NKG2D CAR-NK cells induces endogenous CD8+ T cell activation in patients with advanced colorectal cancer. Mol. Ther. 2025;33:4509–4528. doi: 10.1016/j.ymthe.2025.05.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Zhang H., Yang Z., Zhu X., Li J., Gao Y., Zhang Y., Tong Z., Fu Q., Bao X., Li B. American Society of Clinical Oncology; 2024. Phase I Trial of Hypoxia-Responsive CEA CAR-T Cell Therapy in Patients with Heavily Pretreated Solid Tumor via Intraperitoneal or Intravenous Transfusion. [Google Scholar]
10.Zhang H., Hong L., Dong Z., Xin S., Lin B., Cheng J., Tian W., Li B., Wang J., Liu X., et al. Spatially resolved C1QC+ macrophage-CD4+ T cell niche in colorectal cancer microenvironment: implications for immunotherapy response. Cell Discov. 2025;11:60. doi: 10.1038/s41421-025-00811-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Bao X., Wang D., Dai X., Liu C., Zhang H., Jin Y., Tong Z., Li B., Tong C., Xin S., et al. An immunometabolism subtyping system identifies S100A9+ macrophage as an immune therapeutic target in colorectal cancer based on multiomics analysis. Cell Rep. Med. 2023;4 doi: 10.1016/j.xcrm.2023.100987. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Loret N., Vandamme N., De Coninck J., Taminau J., De Clercq K., Blancke G., Jonckheere S., Goossens S., Lemeire K., De Prijck S., et al. Distinct transcriptional programs in ascitic and solid cancer cells induce different responses to chemotherapy in high-grade serous ovarian cancer. Mol. Cancer Res. 2022;20:1532–1547. doi: 10.1158/1541-7786.MCR-21-0565. [DOI] [PubMed] [Google Scholar]
13.Huang X.-Z., Pang M.-J., Li J.-Y., Chen H.-Y., Sun J.-X., Song Y.-X., Ni H.-J., Ye S.-Y., Bai S., Li T.-H., et al. Single-cell sequencing of ascites fluid illustrates heterogeneity and therapy-induced evolution during gastric cancer peritoneal metastasis. Nat. Commun. 2023;14:822. doi: 10.1038/s41467-023-36310-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Xu J., Fang Y., Chen K., Li S., Tang S., Ren Y., Cen Y., Fei W., Zhang B., Shen Y., Lu W. Single-cell RNA sequencing reveals the tissue architecture in human high-grade serous ovarian cancer. Clin. Cancer Res. 2022;28:3590–3602. doi: 10.1158/1078-0432.CCR-22-0296. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Zhou H., Yin J., Wang A., Yin X., Jin T., Xu K., Zhu L., Wang J., Wang W., Zhang W., et al. Single-cell landscape of malignant ascites from patients with metastatic colorectal cancer. Cancer Commun. 2024;44:713–717. doi: 10.1002/cac2.12541. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Saris J., Li Yim A.Y.F., Bootsma S., Lenos K.J., Franco Fernandez R., Khan H.N., Verhoeff J., Poel D., Mrzlikar N.M., Xiong L., et al. Peritoneal resident macrophages constitute an immunosuppressive environment in peritoneal metastasized colorectal cancer. Nat. Commun. 2025;16:3669. doi: 10.1038/s41467-025-58999-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Zhang L., Li Z., Skrzypczynska K.M., Fang Q., Zhang W., O'Brien S.A., He Y., Wang L., Zhang Q., Kim A., et al. Single-Cell Analyses Inform Mechanisms of Myeloid-Targeted Therapies in Colon Cancer. Cell. 2020;181:442–459.e29. doi: 10.1016/j.cell.2020.03.048. [DOI] [PubMed] [Google Scholar]
18.Han J., Gallerand A., Erlich E.C., Helmink B.A., Mair I., Li X., Eckhouse S.R., Dimou F.M., Shakhsheer B.A., Phelps H.M., et al. Human serous cavity macrophages and dendritic cells possess counterparts in the mouse with a distinct distribution between species. Nat. Immunol. 2024;25:155–165. doi: 10.1038/s41590-023-01688-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Zheng X., Wang X., Cheng X., Liu Z., Yin Y., Li X., Huang Z., Wang Z., Guo W., Ginhoux F., et al. Single-cell analyses implicate ascites in remodeling the ecosystems of primary and metastatic tumors in ovarian cancer. Nat. Cancer. 2023;4:1138–1156. doi: 10.1038/s43018-023-00599-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Ma R.Y., Black A., Qian B.Z. Macrophage diversity in cancer revisited in the era of single-cell omics. Trends Immunol. 2022;43:546–563. doi: 10.1016/j.it.2022.04.008. [DOI] [PubMed] [Google Scholar]
21.Ishtiaq S.M., Arshad M.I., Khan J.A. PPARγ signaling in hepatocarcinogenesis: Mechanistic insights for cellular reprogramming and therapeutic implications. Pharmacol. Ther. 2022;240 doi: 10.1016/j.pharmthera.2022.108298. [DOI] [PubMed] [Google Scholar]
22.Tate T., Xiang T., Wobker S.E., Zhou M., Chen X., Kim H., Batourina E., Lin C.S., Kim W.Y., Lu C., et al. Pparg signaling controls bladder cancer subtype and immune exclusion. Nat. Commun. 2021;12:6160. doi: 10.1038/s41467-021-26421-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Xie M., Yuan K., Zhang Y., Zhang Y., Zhang R., Gao J., Wei W., Jiang L., Li T., Ding Y., et al. Tumor-resident probiotic Clostridium butyricum improves aPD-1 efficacy in colorectal cancer models by inhibiting IL-6-mediated immunosuppression. Cancer Cell. 2025;43:1885–1901.e10. doi: 10.1016/j.ccell.2025.07.012. [DOI] [PubMed] [Google Scholar]
24.Liu Y., Xun Z., Ma K., Liang S., Li X., Zhou S., Sun L., Liu Y., Du Y., Guo X., et al. Identification of a tumour immune barrier in the HCC microenvironment that determines the efficacy of immunotherapy. J. Hepatol. 2023;78:770–782. doi: 10.1016/j.jhep.2023.01.011. [DOI] [PubMed] [Google Scholar]
25.Fong W., Li Q., Ji F., Liang W., Lau H.C.H., Kang X., Liu W., To K.K.W., Zuo Z., Li X., et al. Lactobacillus gallinarum-derived metabolites boost anti-PD1 efficacy in colorectal cancer by inhibiting regulatory T cells through modulating IDO1/Kyn/AHR axis. Gut. 2023;72:2272–2285. doi: 10.1136/gutjnl-2023-329543. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Sathe A., Mason K., Grimes S.M., Zhou Z., Lau B.T., Bai X., Su A., Tan X., Lee H., Suarez C.J., et al. Colorectal Cancer Metastases in the Liver Establish Immunosuppressive Spatial Networking between Tumor-Associated SPP1+ Macrophages and Fibroblasts. Clin. Cancer Res. 2023;29:244–260. doi: 10.1158/1078-0432.CCR-22-2041. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Bill R., Wirapati P., Messemaker M., Roh W., Zitti B., Duval F., Kiss M., Park J.C., Saal T.M., Hoelzl J., et al. CXCL9:SPP1 macrophage polarity identifies a network of cellular programs that control human cancers. Science (New York, N.Y.) 2023;381:515–524. doi: 10.1126/science.ade2292. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Zhu Y., Yang J., Xu D., Gao X.M., Zhang Z., Hsu J.L., Li C.W., Lim S.O., Sheng Y.Y., Zhang Y., et al. Disruption of tumour-associated macrophage trafficking by the osteopontin-induced colony-stimulating factor-1 signalling sensitises hepatocellular carcinoma to anti-PD-L1 blockade. Gut. 2019;68:1653–1666. doi: 10.1136/gutjnl-2019-318419. [DOI] [PubMed] [Google Scholar]
29.Nardo A.D., Grün N.G., Zeyda M., Dumanic M., Oberhuber G., Rivelles E., Helbich T.H., Markgraf D.F., Roden M., Claudel T., et al. Impact of osteopontin on the development of non-alcoholic liver disease and related hepatocellular carcinoma, Liver international. Liver Int. 2020;40:1620–1633. doi: 10.1111/liv.14464. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Lyu A., Fan Z., Clark M., Lea A., Luong D., Setayesh A., Starzinski A., Wolters R., Arias-Badia M., Allaire K., et al. Evolution of myeloid-mediated immunotherapy resistance in prostate cancer. Nature. 2025;637:1207–1217. doi: 10.1038/s41586-024-08290-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Liu S., Zhang Z., Wang Z., Liu C., Liang G., Xu T., Li Z., Duan X., Xu G., Feng X., et al. SPP1 Drives Colorectal Cancer Liver Metastasis and Immunotherapy Resistance by Stimulating CXCL12 Production in Cancer-Associated Fibroblasts. Cancer Res. 2025;86:58–79. doi: 10.1158/0008-5472.CAN-24-4916. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Figures S1–S10 and Tables S1–S5

mmc1.pdf^{(5.2MB, pdf)}

Document S2. Article plus supplemental information

mmc2.pdf^{(39.6MB, pdf)}

Data Availability Statement

[bib1] 1.Fakih M.G. Metastatic colorectal cancer: current state and future directions. J. Clin. Oncol. 2015;33:1809–1824. doi: 10.1200/JCO.2014.59.7633. [DOI] [PubMed] [Google Scholar]

[bib2] 2.Ceelen W.P., Bracke M.E. Peritoneal minimal residual disease in colorectal cancer: mechanisms, prevention, and treatment. Lancet Oncol. 2009;10:72–79. doi: 10.1016/S1470-2045(08)70335-8. [DOI] [PubMed] [Google Scholar]

[bib3] 3.Berlin C., Mauerer B., Cauchy P., Luenstedt J., Sankowski R., Marx L., Feuerstein R., Schaefer L., Greten F.R., Pesic M., et al. Single-cell deconvolution reveals high lineage- and location-dependent heterogeneity in mesenchymal multivisceral stage 4 colorectal cancer. J. Clin. Investig. 2023;134 doi: 10.1172/JCI169576. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] 4.Li Q., Xiao Y., Han L., Luo W., Dai W., Fang H., Wang R., Xu Y., Cai S., Goel A., et al. Microbiome dysbiosis, neutrophil recruitment and mesenchymal transition of mesothelial cells promotes peritoneal metastasis of colorectal cancer. Nat. Cancer. 2025;6:493–510. doi: 10.1038/s43018-025-00910-9. [DOI] [PubMed] [Google Scholar]

[bib5] 5.Bhat A.A., Nisar S., Singh M., Ashraf B., Masoodi T., Prasad C.P., Sharma A., Maacha S., Karedath T., Hashem S., et al. Cytokine- and chemokine-induced inflammatory colorectal tumor microenvironment: Emerging avenue for targeted therapy. Cancer Commun. 2022;42:689–715. doi: 10.1002/cac2.12295. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib6] 6.Ceelen W., Ramsay R.G., Narasimhan V., Heriot A.G., De Wever O. Targeting the Tumor Microenvironment in Colorectal Peritoneal Metastases. Trends Cancer. 2020;6:236–246. doi: 10.1016/j.trecan.2019.12.008. [DOI] [PubMed] [Google Scholar]

[bib7] 7.Zhang Y., Qian L., Chen K., Gu S., Meng Z., Wang J., Li Y., Wang P. Oncolytic adenovirus in treating malignant ascites: A phase II trial and longitudinal single-cell study. Mol. Ther. 2024;32:2000–2020. doi: 10.1016/j.ymthe.2024.04.029. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib8] 8.Li B., Zhu X., Ge J., Zhang H., Gao Y., Bao X., Dai X., Du Z., Tu X., Tong Z. Intraperitoneal infusion of NKG2D CAR-NK cells induces endogenous CD8+ T cell activation in patients with advanced colorectal cancer. Mol. Ther. 2025;33:4509–4528. doi: 10.1016/j.ymthe.2025.05.026. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] 9.Zhang H., Yang Z., Zhu X., Li J., Gao Y., Zhang Y., Tong Z., Fu Q., Bao X., Li B. American Society of Clinical Oncology; 2024. Phase I Trial of Hypoxia-Responsive CEA CAR-T Cell Therapy in Patients with Heavily Pretreated Solid Tumor via Intraperitoneal or Intravenous Transfusion. [Google Scholar]

[bib10] 10.Zhang H., Hong L., Dong Z., Xin S., Lin B., Cheng J., Tian W., Li B., Wang J., Liu X., et al. Spatially resolved C1QC+ macrophage-CD4+ T cell niche in colorectal cancer microenvironment: implications for immunotherapy response. Cell Discov. 2025;11:60. doi: 10.1038/s41421-025-00811-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] 11.Bao X., Wang D., Dai X., Liu C., Zhang H., Jin Y., Tong Z., Li B., Tong C., Xin S., et al. An immunometabolism subtyping system identifies S100A9+ macrophage as an immune therapeutic target in colorectal cancer based on multiomics analysis. Cell Rep. Med. 2023;4 doi: 10.1016/j.xcrm.2023.100987. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] 12.Loret N., Vandamme N., De Coninck J., Taminau J., De Clercq K., Blancke G., Jonckheere S., Goossens S., Lemeire K., De Prijck S., et al. Distinct transcriptional programs in ascitic and solid cancer cells induce different responses to chemotherapy in high-grade serous ovarian cancer. Mol. Cancer Res. 2022;20:1532–1547. doi: 10.1158/1541-7786.MCR-21-0565. [DOI] [PubMed] [Google Scholar]

[bib13] 13.Huang X.-Z., Pang M.-J., Li J.-Y., Chen H.-Y., Sun J.-X., Song Y.-X., Ni H.-J., Ye S.-Y., Bai S., Li T.-H., et al. Single-cell sequencing of ascites fluid illustrates heterogeneity and therapy-induced evolution during gastric cancer peritoneal metastasis. Nat. Commun. 2023;14:822. doi: 10.1038/s41467-023-36310-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] 14.Xu J., Fang Y., Chen K., Li S., Tang S., Ren Y., Cen Y., Fei W., Zhang B., Shen Y., Lu W. Single-cell RNA sequencing reveals the tissue architecture in human high-grade serous ovarian cancer. Clin. Cancer Res. 2022;28:3590–3602. doi: 10.1158/1078-0432.CCR-22-0296. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] 15.Zhou H., Yin J., Wang A., Yin X., Jin T., Xu K., Zhu L., Wang J., Wang W., Zhang W., et al. Single-cell landscape of malignant ascites from patients with metastatic colorectal cancer. Cancer Commun. 2024;44:713–717. doi: 10.1002/cac2.12541. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] 16.Saris J., Li Yim A.Y.F., Bootsma S., Lenos K.J., Franco Fernandez R., Khan H.N., Verhoeff J., Poel D., Mrzlikar N.M., Xiong L., et al. Peritoneal resident macrophages constitute an immunosuppressive environment in peritoneal metastasized colorectal cancer. Nat. Commun. 2025;16:3669. doi: 10.1038/s41467-025-58999-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib17] 17.Zhang L., Li Z., Skrzypczynska K.M., Fang Q., Zhang W., O'Brien S.A., He Y., Wang L., Zhang Q., Kim A., et al. Single-Cell Analyses Inform Mechanisms of Myeloid-Targeted Therapies in Colon Cancer. Cell. 2020;181:442–459.e29. doi: 10.1016/j.cell.2020.03.048. [DOI] [PubMed] [Google Scholar]

[bib18] 18.Han J., Gallerand A., Erlich E.C., Helmink B.A., Mair I., Li X., Eckhouse S.R., Dimou F.M., Shakhsheer B.A., Phelps H.M., et al. Human serous cavity macrophages and dendritic cells possess counterparts in the mouse with a distinct distribution between species. Nat. Immunol. 2024;25:155–165. doi: 10.1038/s41590-023-01688-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] 19.Zheng X., Wang X., Cheng X., Liu Z., Yin Y., Li X., Huang Z., Wang Z., Guo W., Ginhoux F., et al. Single-cell analyses implicate ascites in remodeling the ecosystems of primary and metastatic tumors in ovarian cancer. Nat. Cancer. 2023;4:1138–1156. doi: 10.1038/s43018-023-00599-8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] 20.Ma R.Y., Black A., Qian B.Z. Macrophage diversity in cancer revisited in the era of single-cell omics. Trends Immunol. 2022;43:546–563. doi: 10.1016/j.it.2022.04.008. [DOI] [PubMed] [Google Scholar]

[bib21] 21.Ishtiaq S.M., Arshad M.I., Khan J.A. PPARγ signaling in hepatocarcinogenesis: Mechanistic insights for cellular reprogramming and therapeutic implications. Pharmacol. Ther. 2022;240 doi: 10.1016/j.pharmthera.2022.108298. [DOI] [PubMed] [Google Scholar]

[bib22] 22.Tate T., Xiang T., Wobker S.E., Zhou M., Chen X., Kim H., Batourina E., Lin C.S., Kim W.Y., Lu C., et al. Pparg signaling controls bladder cancer subtype and immune exclusion. Nat. Commun. 2021;12:6160. doi: 10.1038/s41467-021-26421-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] 23.Xie M., Yuan K., Zhang Y., Zhang Y., Zhang R., Gao J., Wei W., Jiang L., Li T., Ding Y., et al. Tumor-resident probiotic Clostridium butyricum improves aPD-1 efficacy in colorectal cancer models by inhibiting IL-6-mediated immunosuppression. Cancer Cell. 2025;43:1885–1901.e10. doi: 10.1016/j.ccell.2025.07.012. [DOI] [PubMed] [Google Scholar]

[bib24] 24.Liu Y., Xun Z., Ma K., Liang S., Li X., Zhou S., Sun L., Liu Y., Du Y., Guo X., et al. Identification of a tumour immune barrier in the HCC microenvironment that determines the efficacy of immunotherapy. J. Hepatol. 2023;78:770–782. doi: 10.1016/j.jhep.2023.01.011. [DOI] [PubMed] [Google Scholar]

[bib25] 25.Fong W., Li Q., Ji F., Liang W., Lau H.C.H., Kang X., Liu W., To K.K.W., Zuo Z., Li X., et al. Lactobacillus gallinarum-derived metabolites boost anti-PD1 efficacy in colorectal cancer by inhibiting regulatory T cells through modulating IDO1/Kyn/AHR axis. Gut. 2023;72:2272–2285. doi: 10.1136/gutjnl-2023-329543. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib26] 26.Sathe A., Mason K., Grimes S.M., Zhou Z., Lau B.T., Bai X., Su A., Tan X., Lee H., Suarez C.J., et al. Colorectal Cancer Metastases in the Liver Establish Immunosuppressive Spatial Networking between Tumor-Associated SPP1+ Macrophages and Fibroblasts. Clin. Cancer Res. 2023;29:244–260. doi: 10.1158/1078-0432.CCR-22-2041. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] 27.Bill R., Wirapati P., Messemaker M., Roh W., Zitti B., Duval F., Kiss M., Park J.C., Saal T.M., Hoelzl J., et al. CXCL9:SPP1 macrophage polarity identifies a network of cellular programs that control human cancers. Science (New York, N.Y.) 2023;381:515–524. doi: 10.1126/science.ade2292. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] 28.Zhu Y., Yang J., Xu D., Gao X.M., Zhang Z., Hsu J.L., Li C.W., Lim S.O., Sheng Y.Y., Zhang Y., et al. Disruption of tumour-associated macrophage trafficking by the osteopontin-induced colony-stimulating factor-1 signalling sensitises hepatocellular carcinoma to anti-PD-L1 blockade. Gut. 2019;68:1653–1666. doi: 10.1136/gutjnl-2019-318419. [DOI] [PubMed] [Google Scholar]

[bib29] 29.Nardo A.D., Grün N.G., Zeyda M., Dumanic M., Oberhuber G., Rivelles E., Helbich T.H., Markgraf D.F., Roden M., Claudel T., et al. Impact of osteopontin on the development of non-alcoholic liver disease and related hepatocellular carcinoma, Liver international. Liver Int. 2020;40:1620–1633. doi: 10.1111/liv.14464. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] 30.Lyu A., Fan Z., Clark M., Lea A., Luong D., Setayesh A., Starzinski A., Wolters R., Arias-Badia M., Allaire K., et al. Evolution of myeloid-mediated immunotherapy resistance in prostate cancer. Nature. 2025;637:1207–1217. doi: 10.1038/s41586-024-08290-3. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] 31.Liu S., Zhang Z., Wang Z., Liu C., Liang G., Xu T., Li Z., Duan X., Xu G., Feng X., et al. SPP1 Drives Colorectal Cancer Liver Metastasis and Immunotherapy Resistance by Stimulating CXCL12 Production in Cancer-Associated Fibroblasts. Cancer Res. 2025;86:58–79. doi: 10.1158/0008-5472.CAN-24-4916. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Evolution of SPP1+ cavity macrophage-mediated immunotherapy resistance in peritoneal metastasis of colorectal cancer

Xiaomeng Dai

Chunyu Lai

Libing Hong

Yuzhi Jin

Jinlin Cheng

Haimeng Yan

Xuqi Sun

Bin Li

Chuan Liu

Hangyu Zhang

Yanwei Lu

Haibo Zhang

Danyang Wang

Peng Zhao

Yu Shi

Weijia Fang

Shan Xin

Xia Yu

Xuanwen Bao