Mutant p53 sensitizes KRAS-driven lung adenocarcinoma to immunotherapy by repressing SPP1 Authors: Xin Zhang, Qian Hao, Wenyue Yang, Yunbo Yu, Qingya Yan, Zhenhua Wu, Zhihuang Hu, Xiang Zhou, Jialei Wang Published: 2026 ABSTRACT The efficacy of immune checkpoint blockade (ICB) in lung adenocarcinoma (LUAD) is limited by primary or acquired resistance, especially in genetically defined subsets. We investigated the impact of KRAS/TP53 co-mutation on ICB response. Our analysis reveals that KRAS/TP53 co-mutation is associated with favorable ICB outcomes, a phenotype orchestrated by secreted phosphoprotein 1 (SPP1). Mechanistically, mutant p53 and KRAS exert opposing effects on SPP1. Mutant p53 synergizes with transcription factor FOXA1 to repress SPP1 transcription, whereas KRAS activates the mitogen-activated protein kinase (MAPK)–extracellular signal–regulated kinase (ERK)–signal transducer and activator of transcription 1 (STAT1) axis to up-regulate SPP1. Consequently, reduced SPP1 level in KRAS/TP53 co-mutant LUAD impairs myeloid-derived suppressor cell (MDSC) recruitment and increases CD8+ T cell infiltration. Notably, SPP1 blockade synergizes with anti–programmed death-1 (PD-1) therapy to suppress tumor growth in KRAS-mutant xenograft models. Collectively, we delineate a mutant p53–FOXA1–SPP1 regulatory axis that modulates ICB susceptibility and unveil a synergistic combination therapy strategy. INTRODUCTION Lung adenocarcinoma (LUAD), the most prevalent subtype of lung cancer, remains a leading cause of cancer-related mortality worldwide (1). The advent of immune checkpoint blockade (ICB), particularly those targeting programmed death-1 (PD-1) and programmed death-ligand 1 (PD-L1), has improved survival for patients with advanced LUAD (2–4). However, therapeutic efficacy is substantially limited by resistance. Approximately 30 to 40% of patients exhibit primary resistance, whereas many initial responders later develop acquired resistance with disease progression within 12 to 18 months (5, 6). Consequently, only about 20% of patients with advanced non–small cell lung cancer (NSCLC) achieve long-term benefits from ICB (7). These challenges underscore the urgent need to elucidate the mechanisms underlying immunotherapy resistance and to develop rational combination strategies for specific LUAD subtypes. KRAS mutation is one of the most prevalent driver alterations in NSCLC, occurring in ∼20 to 30% of cases (8). Although recent breakthroughs in KRAS-targeted therapy have provided more options for pretreated patients, first-line treatment for advanced KRAS-mutant LUAD still relies on ICB-based combination therapy, following the paradigm for driver mutation-negative NSCLC (9, 10). Optimizing ICB efficacy in this population is therefore critically important. Clinical responses to ICB in KRAS-mutant LUAD are highly heterogeneous, a variability strongly influenced by co-occurring genetic alterations (11, 12). Among these, concurrent TP53 mutation is the most common (13). As a classic tumor suppressor gene, wild-type p53 maintains genomic stability by regulating cell cycle arrest, DNA repair, senescence, and apoptosis (14). However, TP53 mutations are common in lung cancer. Missense mutations within the DNA binding domain (DBD) abrogate DNA binding capacity and disrupt the transcriptional regulatory functions of wild-type p53. In addition, mutant p53 often acquires dominant-negative effects or gain-of-function (GOF) properties (15). Recent studies suggested that mutant p53 could modulate the tumor microenvironment (TME). In NSCLC, mutant p53 has been implicated in both suppressing innate immune signals and shaping adaptive immunity, correlating with features such as elevated PD-L1 expression, increased tumor mutation burden (TMB), and enhanced CD8+ T cell infiltration (16, 17). Despite these insights, previous studies have yielded conflicting conclusions regarding the combined impact of KRAS/TP53 co-mutations on ICB efficacy in LUAD (13, 18, 19), and the underlying molecular mechanisms remain poorly delineated, hindering patient stratification and the development of tailored combination therapies. Secreted phosphoprotein 1 (SPP1) is a key extracellular matrix glycoprotein that orchestrates immunosuppressive TME remodeling (20). Several studies have demonstrated that SPP1 promotes tumor stemness via the CD44-ERK axis and drives macrophage polarization toward an immunosuppressive M2-like phenotype (21). Preclinical studies further show that targeting SPP1 enhances anti–PD-1 efficacy by increasing cytotoxic T cell infiltration and reducing cancer-associated fibroblasts (CAFs). In addition, SPP1 facilitates immune evasion by recruiting myeloid-derived suppressor cells (MDSCs) (22, 23). Despite its pivotal role in these processes, the relationship between SPP1 and gene mutational status remains largely unexplored. In this study, we first validated the association between KRAS/TP53 co-mutations and improved ICB outcomes in patients with LUAD. We then identified SPP1, a key immunoregulator in the TME, as a critical downstream effector comodulated by mutant p53 and mutant KRAS. These two mutations exert opposing effects on SPP1 expression. Mechanistically, mutant p53 cooperates with the transcription factor (TF) FOXA1 to repress SPP1 transcription, whereas oncogenic KRAS up-regulates SPP1 by activating the mitogen-activated protein kinase (MAPK)–extracellular signal–regulated kinase (ERK)–signal transducer and activator of transcription 1 (STAT1) axis. Functionally, in KRAS-mutant LUAD, mutant p53 remodels the TME by reducing SPP1 secretion, which impairs monocytic-derived myeloid-derived suppressor cell (M-MDSC) recruitment and enhances CD8+ T cell infiltration. Our results demonstrate that SPP1 blockade synergizes with anti–PD-1 therapy to suppress tumor growth in a p53-null, KRAS-mutant xenograft model. Collectively, this study provides a mechanistic insight and an actionable therapeutic strategy for personalized immunotherapy in KRAS-driven LUAD. RESULTS Co-mutation of KRAS and TP53 correlates with favorable prognosis in patients with LUAD receiving immunotherapy To investigate the impact of KRAS mutation alone and KRAS/TP53 co-mutations on the efficacy of immunotherapy in patients with LUAD, we retrospectively enrolled 120 patients with advanced LUAD harboring KRAS mutation or KRAS/TP53 co-mutations who received immunotherapy at the FUSCC. Patients in the co-mutation group were younger, whereas no significant differences were observed in gender, smoking history, clinical stage, or PD-1 inhibitor regimen between the two groups (Fig. 1A). Survival analyses indicated that patients with KRAS/TP53 co-mutations had significantly longer progression-free survival (PFS) and overall survival (OS) compared with those with KRAS mutation alone (Fig. 1B and fig. S1A). This finding was validated using data from the MSKCC cohort (Fig. 1C and fig. S1B). In addition, univariate and multivariate Cox regression analyses confirmed that KRAS/TP53 co-mutation serves as an independent predictor of prognosis (table S1). Notably, we did not observe this survival benefit in a comparable pancreatic cancer immunotherapy cohort (24) (fig. S1, C to E). Collectively, these results demonstrate that KRAS/TP53 co-mutation is associated with favorable prognosis in patients with LUAD treated with immunotherapy. KRAS/TP53 co-mutation predicts favorable prognosis in immunotherapy treated LUAD. (A) Clinical characteristics associated with KRAS and TP53 mutation status in the FUSCC cohort (n = 120). Statistical significance was determined using the two-sided chi-square test. IQR, interquartile range. (B) Survival analysis of PFS between patients with KRAS mutation and KRAS/TP53 co-mutations in the FUSCC cohort (n = 120). Statistical significance was determined using the log-rank test. HR, hazard ratio. (C) Survival analysis of PFS between patients with KRAS mutation and KRAS/TP53 co-mutations in the MSKCC cohort (n = 211). Statistical significance was determined using the log-rank test. Mutant p53 down-regulates SPP1 expression To elaborate the molecular mechanisms by which mutant p53 modulates downstream gene expression and signaling pathways in the context of KRAS mutations, we conducted transcriptome sequencing in H1299 cells harboring KRAS Gly12Cys (G12C) (K), as well as KRAS G12C co-mutated with either p53-Arg175His (R175H) (KPR175H) or p53-Arg248Gln (R248Q) (KPR248Q) (fig. S2A). Differentially expressed genes (DEGs) were identified from the K versus KPR175H and K versus KPR248Q comparisons using thresholds of fold change (FC) > 2 and adjusted P values < 0.05. The intersection of these two DEG sets identified 23 genes commonly regulated by p53-R175H and p53-R248Q (Fig. 2A), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the DEGs revealed that transcriptional misregulation in cancer was significantly enriched in both groups (fig. S2, B to E). To validate mutant p53–mediated regulation of these candidate genes, we conducted reverse transcription quantitative polymerase chain reaction (RT-qPCR) and confirmed consistent down-regulation of SPP1 (fig. S2, F to H). Furthermore, transient knockdown of mutant p53 in PC9 and H1975 led to up-regulation of both SPP1 mRNA and protein levels (Fig. 2, B to E). Conversely, ectopic expression of mutant p53 in H1299K (KRASG12C-mutant H1299 cells), A549p53−/−, and KP (KRASGly12Asp (G12D)/TP53−/− mouse adenocarcinoma cell line) cells resulted in down-regulation of SPP1 mRNA and protein expression (Fig. 2, F to K). To systematically assess the regulatory effects of different p53 mutants on SPP1 expression, we overexpressed p53-Ser241Phe (S241F), p53-Tyr220Cys (Y220C), p53-Arg249Ser (R249S), p53-Arg273His (R273H), and p53-Arg248Trp (R248W) in H1299K cells. These mutant p53 variants suppressed SPP1 mRNA and protein expression (Fig. 2, L and M). However, overexpression of wild-type p53 in H1299K cells, pharmacological activation via Nutlin-3, and small interfering RNA (siRNA)–mediated knockdown of endogenous wild-type p53 in A549 and H460 cells consistently failed to alter SPP1 protein levels (Fig. 2, N to P, and fig. S2, K to N). Together, these findings demonstrate that mutant p53 represses SPP1 expression in a manner independent of its specific mutation site. Inhibition of SPP1 expression by mutant p53. (A) Schematic diagram depicting the identification of SPP1. (B to E) RT-qPCR and WB assays validating the expression levels of SPP1 in PC-9 [(B) and(D)] cells and H1975 [(C) and (E)] cells transfected with siNC and sip53. (F to K) RT-qPCR and WB assays validating the expression levels of SPP1 in H1299K [(F) and (I)] and A549p53−/−[(G) and (J)] cells transfected with Flag-Ctrl, Flag-p53-R175H, and Flag-p53-R248Q; KP cells transfected with Flag-Ctrl, Flag-p53-R172H, and Flag-p53-R270H [(H) and (K)]. Data are means ± SD (n = 3). **P < 0.01; ***P < 0.001; ****P < 0.0001, two-tailed unpaired t test. (L and M) RT-qPCR (L) and WB (M) assays validating the expression levels of SPP1 in H1299K cells transfected with Flag-Ctrl, Flag-p53-S241F, Flag-p53-Y220C, Flag-p53-R249S, Flag-p53-R273H, and Flag-p53-R248W. Data are means ± SD (n = 3). **P < 0.01, one-way ANOVA. (N) WB assay validating the expression level of SPP1 in H1299K cells transfected with Flag-Ctrl and Flag-wtp53. (O and P) WB assay validating the expression level of SPP1 in A549 (O) and H460 (P) cells treated with dimethyl sulfoxide (DMSO) or Nutlin-3 (20 μM). Mutant p53 cooperates with FOXA1 to repress SPP1 transcription Mutant p53 modulates gene transcription via interactions with TFs or cofactors (15). To explore how mutant p53 regulates SPP1, we used JASPAR and PROMO databases to predict candidate TFs regulating SPP1. Intersecting the results yielded nine candidate TFs: CEBPB, CEBPA, MYB, FOXA1, NFIA, STAT1, GATA2, SPI1, and HNF1A (Fig. 3A). We then knocked down each of these candidate TFs and found that FOXA1, NFIA, and STAT1 are key SPP1 regulators (Fig. 3, B to D, and fig. S3, A to F). Given that mutant p53 exerts its functions via interactions with TFs, we further performed coimmunoprecipitation (co-IP) assays and found that only FOXA1 specifically interacted with mutant p53 (Fig. 3, F to H, and fig. S3, G and H). Mapping experiments revealed that FOXA1 specifically interacts with the N-terminal domain (1 to 100 amino acids) of mutant p53 (Fig. 3, I and J). In addition, knockdown of FOXA1 increased SPP1 mRNA and protein levels, and TCGA-LUAD data confirmed a negative correlation between the expression of SPP1 and FOXA1, suggesting that FOXA1 is an inhibitory TF of SPP1 (Fig. 3E and fig. S3, I and J). Notably, knockdown of mutant p53 did not affect FOXA1 expression (fig. S3, K to N). Furthermore, single knockdown of mutant p53 or FOXA1 up-regulated SPP1, whereas co-knockdown did not enhance this effect, indicating that mutant p53 regulates SPP1 expression through its interaction with FOXA1 (Fig. 3, K to L). To further validate this observation, we used the JASPAR database to analyze the genomic sequence upstream of the transcription start site (TSS) of SPP1, and identified six possible FOXA1 response elements (REs) (Fig. 3M). The dual luciferase reporter assay and chromatin immunoprecipitation quantitative polymerase chain reaction (ChIP-qPCR) were conducted to verify that FOXA1 specifically binds to RE 5, which is located at −2118 to −2104 upstream of the SPP1 TSS, to inhibit the SPP1 promoter-driven luciferase activity (Fig. 3, N to P). Furthermore, we tested whether FOXA1 influences the recruitment of mutant p53 to the SPP1 promoter through the ChIP-qPCR assay. Our results showed that FOXA1 depletion reduced the association of mutant p53 in the SPP1 promoter, whereas the overexpression of FOXA1 increased mutant p53 binding to SPP1 promoter (Fig. 3, Q to T). Collectively, our results demonstrate that mutant p53 binds to the SPP1 promoter indirectly by interacting with FOXA1, which is dependent on FOXA1’s DNA binding capacity, to cooperatively repress SPP1 transcription. Mutant p53 cooperates with FOXA1 to repress SPP1 transcription. (A) Venn diagram of SPP1 candidate TFs predicted by JASPAR and PROMO databases. (B to D) RT-qPCR analysis of candidate TFs and SPP1 mRNA levels in PC9 cells transfected with siRNAs targeting FOXA1 (B), STAT1 (C), or NFIA (D). Data are means ± SD (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, two-tailed unpaired t test. (E) Western blot analysis of FOXA1 and SPP1 protein levels in PC9 cells transfected with two independent siRNAs targeting FOXA1 (siFOXA1#1 and siFOXA1#2). (F) Exogenous co-IP assay in HEK293T cells cotransfected with Flag-tagged control plasmid or Flag-tagged mutant p53 (p53-R175H or p53-R248Q) and Myc-tagged FOXA1. Input: total cell lysate; IP: immunoprecipitation with anti-Flag antibody. Western blot (IB) was performed with anti-Myc or anti-Flag antibody. (G and H) Endogenous co-IP assay in PC9 [p53-R248Q mutant; (F)] and H1975 [p53-R273H mutant; (G)] cells. Endogenous FOXA1 was immunoprecipitated (IP: anti-FOXA1), and coprecipitated mutant p53 was detected by IB (anti-p53). IgG: isotype control. (I and J) Mapping experiments in HEK293T cell cotransfected with Flag-tagged mutant p53 domain and Myc-tagged FOXA1. Input: total cell lysate; IP: immunoprecipitation with anti-Flag antibody. Western blot (IB) was performed with anti-Myc or anti-Flag antibody. aa, amino acids. (K and L) Western blot (K) and RT-qPCR (L) analysis of SPP1, p53, and FOXA1 mRNA and protein levels in PC9 cells transfected with siNC, sip53, siFOXA1, or sip53+siFOXA1. Data are means ± SD (n = 3). *P < 0.05; **P < 0.01; n.s., not significant, two-tailed unpaired t test. (M and N) Schematic of the SPP1 promoter region. Predicted FOXA1 REs in the SPP1 promoter (M) and amplicon region [−2390 to −2115 base pairs (bp)] containing FOXA1-RE 5 used for ChIP assays (N). (O) Dual-luciferase reporter assay in PC9 cells transfected with pcDNA3.1-FOXA1 and pGL3-SPP1-luc. Relative luciferase activity (normalized to Renilla) was measured. Data are means ± SD (n = 3). **P < 0.01; n.s., not significant, two-tailed unpaired t test. (P) Dual-luciferase reporter assay in PC9 cells cotransfected with pGL-SPP1-luc2 promoter luciferase and FOXA1-RE 3/4/5/6 deletion constructs and pcDNA3.1-FOXA1. Relative luciferase activity (normalized to Renilla) was measured. Data are means ± SD (n = 3). *P < 0.05; **P < 0.01; n.s., not significant, two-tailed unpaired t test. (Q and R) ChIP-qPCR analysis of FOXA1 (Q) and p53 [(R); detected by Do-1 antibody] recruitment on the SPP1 promoter region (FOXA1-RE 5). Data are means ± SD (n = 3). ***P < 0.001, two-tailed unpaired t test. (S) Western blot analysis of FOXA1 protein levels in PC9 cells transfected with shFOXA1 or shNC. (T) ChIP-qPCR analysis of p53 recruitment on the SPP1 promoter region (FOXA1-RE 5) in PC9 cells transfected with shNC, shFOXA1, or shFOXA1+FOXA1. *P < 0.05; **P < 0.01; n.s., not significant, one-way ANOVA. Mutant p53 impairs monocyte migration by reducing SPP1 secretion Recent findings indicated that SPP1 promotes the formation of an immunosuppressive TME (25–27). To characterize the association between SPP1 expression and immune cell subsets in NSCLC, we first analyzed TCGA-LUAD transcriptomic data and found that SPP1 is highly expressed in LUAD tumor tissues and is associated with poor prognosis (Fig. 4A and fig. S4A). Next, we conducted xCell algorithm and found that SPP1 expression was negatively correlated with CD8+ T cell infiltration and positively correlated with monocyte lineages (Fig. 4, B and C). As a secreted protein, SPP1 exerts chemokine-like functions that influence monocyte migration (23, 28). To test whether mutant p53 remodels the KRAS-mutated NSCLC TME by suppressing SPP1 secretion, we collected conditioned medium (CM) from the human H1299 cell lines K, KPR175H, and KPR248Q cells, as well as from mouse cell lines KP−/−, KPR172HArg172His(R172H), and KPR270HArg270His(R270H), and determined the levels of SPP1 by enzyme-linked immunosorbent assay (ELISA). Our results showed that SPP1 concentrations were reduced in CM from KRAS/TP53 co-mutant cells compared to KRAS-mutant cells (Fig. 4, D and E, and fig. S4, B and C). We further performed the Transwell migration assay using THP-1 and RAW264.7 monocytes cocultured with tumor cells and found that KRAS/TP53 co-mutant cells induced less monocyte migration than KRAS-mutant cells; this impairment was rescued by supplementation with either human or mouse recombinant SPP1 (Fig. 4, F to H). However, coculture assays using THP-1 and RAW264.7 with KRAS-mutant and KRAS/TP53 co-mutant tumor cells revealed no statistically significant differences in IRF8 or SPP1 expression levels in the monocytes (fig. S4, D and E). We further examined whether tumors harboring KRAS/TP53 co-mutations modulate the TME differently than those with KRAS mutation alone in vivo. C57BL/6 mice were subcutaneously inoculated with KP cell lines overexpressing p53-R172H or p53-R270H (Fig. 4I). Our finding demonstrated that tumor volume did not significantly differ among groups (Fig. 4J and fig. S4F), but flow cytometry analysis revealed increased CD8+ T cell infiltration and decreased M-MDSC infiltration in KRAS/TP53 co-mutant tumors (Fig. 4, K to L, and fig. S4, G and H). Overall, these results indicate that mutant p53 reshapes the KRAS-mutant NSCLC immune microenvironment by inhibiting SPP1 secretion, thereby reducing M-MDSC chemotaxis and promoting CD8+ T cell infiltration. Mutant p53 impairs monocyte migration by reducing SPP1 secretion. (A) SPP1 mRNA expression level in tumor tissue versus adjacent normal tissue from the TCGA LUAD cohort. ***P < 0.001, two-tailed unpaired t test. (B and C) Infiltration abundance of monocyte (B) and CD8+ T cells (C) in TCGA-LUAD samples, deconvoluted using the xCell algorithm. Correlations were analyzed via Pearson’s correlation test. (D) ELISA quantification of SPP1 secretion in CM from H1299 cells harboring KRAS mutation (H1299K), KRAS/TP53R175H (H1299KPR175H), and KRAS/TP53R248Q (H1299KPR248Q). Four-parameter logistic (4PL) standard curve in fig. S3C. Data are means ± SD (n = 3). ****P < 0.0001, one-way ANOVA. (E) ELISA quantification of SPP1 secretion in CM from KP cells harboring KRAS mutation (K), KRAS/TP53R172H (KPR172H), and KRAS/TP53R270H (KPR270H). 4PL standard curve in fig. S3D. Data are means ± SD (n = 3). ****P < 0.0001, one-way ANOVA. (F) Schematic diagram of the Transwell coculture system. Created in BioRender. Zhang, Z. (2026) https://BioRender.com/amdy7ga. (G) Thp-1 migration induced by coculture with H1299K, H1299KPR175H, and H1299KPR248Q cells or co-mutant H1299KPR175H/H1299KPR248Q cells supplemented with recombinant human SPP1 (rhSPP1). Data are means ± SD (n = 3). **P < 0.01; n.s., not significant, one-way ANOVA. (H) RAW264.7 migration induced by coculture with KP, KPR172H, and KPR270H cells or co-mutant KPR172H/KPR270H cells supplemented with recombinant mouse SPP1 (rmSPP1). Data are means ± SD (n = 3). ***P < 0.001; n.s., not significant, one-way ANOVA. (I) Schematic diagram of the in vivo xenograft model. Created in BioRender. Zhang, Z. (2026) https://BioRender.com/8cfrdz5. (J) Size of xenograft tumors from KP, KPR172H, and KPR270H cells. (K and L) Representative flow cytometry plots (left) and quantification (right) of tumor-infiltrating M-MDSC cells (K) and CD8+ T cells (L) [additional replicates, including representative images corresponding to those in (K) and (L), are provided in fig. S4, G and H]. Data are means ± SD (n = 6). *P < 0.05; ***P < 0.001; ****P < 0.0001, one-way ANOVA. KRAS up-regulates SPP1 expression via the ERK-STAT1 pathway We further investigated whether KRAS may mediate the expression of SPP1 expression. First, we analyzed transcriptomic data from H1299 cells with ectopic KRAS G12C expression. The results showed that SPP1 was up-regulated in the KRAS-mutant group, and gene set enrichment analysis (GSEA) (29) revealed that the MAPK pathway was enriched in this group (Fig. 5, A and B). We then validated these findings through cell-based experiments, which demonstrated that KRAS mutation not only up-regulates SPP1 expression at both the mRNA and protein levels but also enhances its secretion (Fig. 5, C and D, and fig. S3O). Because our results above identified STAT1 as a transcriptional activator of SPP1 (Figs. 3C and 5E). In addition, a previous study had indicated that the MAPK-ERK signaling pathway can induce STAT1 phosphorylation (30). Therefore, we hypothesized that KRAS mutation may up-regulate SPP1 expression via the ERK-STAT1 signal axis. To this end, we first overexpressed mutant KRAS in H1299 cells and found showed that KRAS enhance ERK phosphorylation, which was accompanied by increased STAT1 phosphorylation and up-regulated SPP1 expression (Fig. 5F). Conversely, knockdown of endogenous mutant KRAS in H460 cells resulted in decreased ERK and STAT1 phosphorylation, as well as reduced SPP1 expression (Fig. 5G). We further used U0126, a selective inhibitor of ERK phosphorylation (31), to confirm the pathway dependency. H1299K cells and H460 cells were treated with U0126 at gradient concentrations, this treatment dose-dependently reduced ERK and STAT1 phosphorylation, as well as down-regulated SPP1 expression (Fig. 5, H and I). However, overexpression mutant p53 in H1299K cells not alter the phosphorylation levels of ERK or STAT1 (fig. S4I). Consistent with our findings, the Transwell migration assay demonstrated that H1299K cells induced increased migration of THP-1 cells compared with H1299 cells. H1299K cell–induced THP-1 migration was repressed by U0126 treatment, whereas this effect could be reversed by supplementation with rhSPP1 (Fig. 5J). Collectively, these results demonstrate that KRAS mutation drives ERK phosphorylation, which, in turn, leads to STAT1 phosphorylation and then the elevation of SPP1 expression and secretion, ultimately promoting monocyte migration. KRAS induces SPP1 expression via the ERK-STAT1 pathway. (A) Differential expression ranking plot of DEGs in KRAS-mutant versus wild-type H1299 cells. (B) GSEA plot showing the enrichment of MAPK signaling–related genes in KRAS-mutant cells. NES, normalized enrichment score; FDR, false discovery rate. (C) ELISA quantification of human SPP1 concentrations in CM from KRAS G12C mutant H1299 cells (H1299K) and H1299 cells (wild-type KRAS). Data are means ± SD (n = 3). ****P < 0.0001, two-tailed unpaired t test. (D) Western blot analysis of SPP1 and Flag-KRAS protein levels in H1299 cells transfected with pcDNA3.1-KRASG12C or pcDNA3.1-Ctrl. (E) Western blot analysis of p-STAT1, STAT1, and SPP1 protein levels in H460 cells transfected with siSTAT1 (#1/#2) or siNC. (F) Western blot analysis of p-STAT1, STAT1, SPP1, p-ERK, ERK, and Flag-KRAS protein levels in H1299 cells transfected with pcDNA3.1-KRASG12C or pcDNA3.1-Ctrl. (G) Western blot analysis of p-STAT1, STAT1, SPP1, p-ERK, ERK, and Kras protein levels in H460 cells transfected with siKRAS or siNC. (H and I) Western blot analysis of p-STAT1, STAT1, SPP1, p-ERK, and ERK protein levels in H1299K (H) and H460 (I) cells treated with increasing concentrations of U0126 (ERK inhibitor). (J) Thp-1 migration induced by coculture with H1299, H1299K cells, H1299K cells treated with U0126, or H1299K cells treated with U0126 and supplemented with recombinant human SPP1 (rhSPP1). Data are means ± SD (n = 3). *P < 0.05; ****P < 0.0001; n.s., not significant, one-way ANOVA. SPP1 blockade improves anti–PD-1 efficacy in KRAS-mutant LUAD Given that mutant p53 fosters an immunotherapy-favorable microenvironment by suppressing KRAS-induced SPP1, we hypothesized that SPP1 inhibition could enhance immunotherapy efficacy in KRAS-mutant LUAD in the absence of mutant p53. Accumulating preclinical evidence has validated the antitumor activity of SPP1-targeting antibodies in murine models (22, 32, 33). Thus, we sought to investigate whether anti-SPP1 could coordinate with anti–PD-1 to suppress KRAS-mutant LUAD growth in vivo. C57BL/6 mice were subcutaneously inoculated with KP cells (KRASG12D/TP53−/− mouse adenocarcinoma cell line) and randomly assigned to four groups, treated with anti–immunoglobulin G (IgG), anti–PD-1, anti-SPP1, or the combination of anti–PD-1 and anti-SPP1. All treatments were administered every 3 days for 2 weeks (Fig. 6A). Our results demonstrated that monotherapy with either agent reduced the tumor growth rate, volume, and weight relative to the control group (Fig. 6, B to D). The combination of anti-SPP1 and anti–PD-1 exerted a substantially greater antitumor effect than either monotherapy alone (Fig. 6, B to D). Notably, no significant body weight loss was observed across all treatment groups, indicating that the combination regimen was well tolerated in vivo (Fig. 6E). Flow cytometry analysis of tumor-infiltrating immune calls further revealed that combination therapy markedly increased the frequencies of GZMB+CD8+ cytotoxic T cells and IFN-γ+CD8+ effector T cells while concomitantly reducing the infiltration of M-MDSCs compared with the monotherapy groups (Fig. 6, F to H, and figs. S5, A and B, and S6, A to C). Collectively, our findings establish that the combination of anti-SPP1 and anti–PD-1 elicits synergistic antitumor immunity in KRAS-driven LUAD by remodeling the TME. SPP1 blockade improves anti–PD-1 efficacy in KRAS-driven LUAD. (A) Schematic diagram of the in vivo therapeutic experiment. Created in BioRender. Zhang, Z. (2026) https://BioRender.com/8p64ru4. (B to D) Size (B), weight (C), and volume (D) of xenograft tumors from each treatment group. (E) Mouse body weight kinetics during the treatment period. Data are means ± SEM (n = 6). n.s., not significant. (F to H) Representative flow cytometry plots (left) and quantification (right) of tumor-infiltrating GZMB+CD8+ T cells (F), IFNγ+CD8+ T cells (G), and M-MDSC cells (H) [additional biological replicates, including representative images corresponding to those in (F) to (H), as well as full gating strategies, are provided in figs. S5, A and B, and S6, A to C] from each treatment group. Data are means ± SEM (n = 6). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001, one-way ANOVA. DISCUSSION Although ICB has transformed the treatment landscape for advanced LUAD, therapeutic resistance remains a major clinical challenge (34, 35). Accumulating evidence underscores that oncogene-defined LUAD subtypes exhibit considerable clinical heterogeneity in their response to immunotherapy, among which concurrent mutations with tumor suppressor genes emerging as a core determinant of this variability in oncogene-driven NSCLC subsets (36–38). Notably, for patients with KRAS-mutant NSCLC lacking access to first-line targeted therapies, co-mutation–based clinical subtyping is of particular significance for stratifying the efficacy of immunotherapy (39). Some study has classified KRAS-mutant LUAD into distinct subgroups, with the KRAS/TP53 (KP) subtype exhibiting an inflamed TME characterized by enhanced CD8+ T cell infiltration and elevated PD-L1 expression (19, 40, 41). Our clinical data extend this observation by directly linking KP co-mutation to improved survival in ICB-treated patients (Fig. 1B), consistent with the Heidelberg cohort (HD-ICI) where patients with KRASMUTTP53MUT derived greater benefits from ICB than those with KRAS mutation alone (18). In addition, our study delineates a regulatory axis, mutant p53–FOXA1–SPP1, that modulates the TME to promote antitumor immunity (Fig. 7). These findings resolve the heterogeneous responses to ICB observed in KRAS-mutant LUAD and provide a rational, actionable strategy of combining SPP1 inhibition with anti–PD-1 therapy (Fig. 6, A to E, and fig. S7). Schematic diagram of the study. Created in BioRender. Zhang, Z. (2026) https://BioRender.com/03dbtfi. SPP1, an extracellular matrix glycoprotein rich in arginine–glycine–aspartic acid (RGD) motifs, binds to integrin receptors to mediate diverse functions (20, 42, 43). Accumulating evidence has established SPP1 as a pivotal regulator of TME remodeling, such as recruiting MDSCs and polarizing macrophages toward an immunosuppressive M2 phenotype, thereby establishing an immunosuppressive TME (23, 44–47). Herein, we identify SPP1 as a convergent downstream target of TP53 and KRAS mutations in LUAD, which are regulated in opposing ways: Mutant p53 cooperates with FOXA1 to repress SPP1 transcription, whereas oncogenic KRAS activates the MAPK-ERK-STAT1 axis to up-regulate SPP1 expression (Figs. 3, E to L, and 5, E to I). This dual regulatory model explains the heterogeneous SPP1 levels observed in KRAS-mutant LUAD. Specifically, in KP co-mutant tumors where mutant p53 predominates, SPP1 repression creates a favorable TME, which is characterized by reduced secretion, diminished M-MDSC recruitment, and enhanced CD8+ T cell infiltration, thereby providing a molecular basis for the improved immunotherapy response. Furthermore, the interaction between mutant p53 and FOXA1 adds a previously uncharacterized dimension to our understanding of mutant p53’s GOF properties. Although FOXA1 has been previously linked to LUAD progression, its role in immune regulation is poorly characterized (48–50). Our findings demonstrate that FOXA1 directly binds the SPP1 promoter to inhibit its transcription, and mutant p53 enhances this repression through protein-protein interaction with FOXA1. This mechanism expands the repertoire of mutant p53’s GOF activities in immune modulation. Concurrently, our study clarifies the pathway through which KRAS regulates SPP1. Constitutive MAPK activation by mutant KRAS induces ERK-dependent phosphorylation of STAT1, which subsequently transactivates the SPP1 promoter. This pathway links oncogenic KRAS signaling to TME remodeling, explaining why KRAS single-mutant tumors often exhibit high SPP1 levels and an immunosuppressive phenotype (Fig. 5J). Therefore, in KP co-mutant tumors, the SPP1-repressing effect of mutant p53 counteracts the SPP1-inducing activity of oncogenic KRAS. As a result, these tumors develop a more immune favorable TME than tumors with KRAS mutation alone. The therapeutic implications of our findings are substantial, particularly in KRAS-driven LUAD. Given the key role of SPP1 in maintaining immunosuppression, we evaluated SPP1 blockade in combination with anti–PD-1 therapy in KRAS-mutant LUAD xenograft models (Fig. 6, A to E, and fig. S7). The synergistic antitumor effect observed in these models is biologically plausible. Specifically, anti-SPP1 reduces MDSC recruitment to relieve immunosuppression, whereas anti–PD-1 reinvigorates exhausted CD8+ T cells. This combination addresses two distinct layers of immune evasion, making it particularly effective in KRAS-mutant LUAD where SPP1-driven immunosuppression and PD-1–mediated T cell exhaustion coexist. Notably, the absence of significant body weight loss in treated mice further supports the safety of this combination, supporting its potential translation to clinical trials. Although our study has yielded valuable insights into KRAS/TP53 co-mutant LUAD, we also have several limitations that warrant consideration. First, although our clinical data validate the prognostic value of KRAS/TP53 co-mutation, they are retrospective. Prospective studies with larger cohorts are needed to confirm the clinical utility of this biomarker and the efficacy of SPP1-targeted combination therapy. Second, although our study establishes the mutant p53–FOXA1–SPP1 axis as a key mediator of ICB sensitivity in KRAS-mutant LUAD through integrated clinical, cellular, and in vivo analyses, deeper mechanistic insight such as structural elucidation of the mutant p53–FOXA1–promoter complex and the link between the TP53 and KRAS regulatory pathways would further strengthen the model. Third, although we focused on SPP1 as a key mediator, mutant p53 likely regulates other immune-related genes that contribute to ICB sensitivity. Future transcriptomic and proteomic profiling could uncover additional components of the mutant p53-driven immunoregulatory network. Last, although our study focuses on the mutant p53–FOXA1–SPP1 axis driven by hotspot missense mutations, we recognize that TP53 mutation classes may have distinct biological consequences. Our analysis of the MSKCC immunotherapy cohort indicates that both missense and other loss-of-function (LOF) TP53 mutations are associated with favorable clinical outcomes in KRAS-mutant LUAD. However, our mechanistic model specifically addresses GOF properties of some hotspot missense mutations. Whether LOF mutations elicit similar TME remodeling through alternative pathways or share common downstream targets remains to be elucidated. Future studies are required to comprehensively map the immunoregulatory networks governed by different classes of TP53 alterations. In conclusion, our study identifies SPP1 as a critical mediator of ICB efficacy in KRAS-mutant LUAD. Mechanistically, mutant p53 cooperates with FOXA1 to repress SPP1 transcription, whereas KRAS activates the MAPK-ERK-STAT1 pathway to up-regulate SPP1 expression. Our findings further indicate that, although oncogenic KRAS promotes an immunosuppressive TME by up-regulating SPP1, mutant p53 counteracts this effect to induce a more favorable TME, thereby enhancing response to immunotherapy. Last, our results demonstrate that SPP1 blockade can synergistically enhance anti–PD-1 therapy efficacy in KRAS-driven LUAD in the absence of mutant p53, providing a therapeutic strategy for this clinically challenging subset of lung cancer. MATERIALS AND METHODS Study cohorts and efficacy assessment The FUSCC cohort consisted of 120 patients diagnosed with LUAD based on pathology from the FUSCC between July 2019 and August 2024. Inclusion criteria for patients included histologically confirmed diagnosis of stage III to IV LUAD, an absence of targetable EGFR/ALK/ROS1 mutations, with KRAS mutation or KRAS/TP53 co-mutation, treatment-naïve status, and a baseline Eastern Cooperative Oncology Group (ECOG) performance status score of 0 or 1. Clinical stage was determined according to the eighth edition of the American Joint Committee on Cancer (AJCC) TNM staging system (51). All patients received ICIs in combination with chemotherapy every 3 weeks for four cycles, followed by ICI maintenance therapy every 3 weeks until disease progression or unacceptable toxicity occurred. PFS was defined as the time from the start of immunochemotherapy until either disease progression or death from any cause. OS was defined as the time from the start of immunochemotherapy until death from any cause. In addition to the FUSCC cohort, we also included data from the MSKCC cohort (52–55). The MSKCC cohort datasets are available on the cBioPortal website (www.cbioportal.org/) and 211 patients with LUAD receiving immunotherapy with KRAS mutation or KRAS/TP53 co-mutation are included in our study. Cell culture and transient transfection Mutant p53–harboring cancer cell lines, PC9, H1975, p53-null cell lines, H1299 and A549p53−/−, KRAS mutant cell lines H460, H1299K, and the murine immune cell line RAW264.7 and the human embryonic kidney cell line HEK293T were cultured in Dulbecco’s modified Eagle’s medium (DMEM). The p53-null cell line KP and the human immune cell line THP-1 were cultured in RPMI 1640 medium. Both media were supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/ml), and streptomycin (0.1 mg/ml). All cells were maintained at 37°C in a humidified incubator with 5% CO2. All cell lines were regularly tested and confirmed free of mycoplasma contamination. For transient transfection, plasmids or siRNAs were delivered into cells using the Hieff Trans Liposomal Transfection Reagent (Yeasen, Shanghai, China) in accordance with the manufacturer’s protocol. Plasmids, antibodies, and reagents The p53-encoding plasmids were described previously. Human FOXA1 and KRAS were cloned into the vectors of Myc-pcDNA3.1 and Flag-pcDNA3.1, respectively. For the luciferase reporter assay, sequences upstream of the transcription initiation site of SPP1 were cloned into the vector pGL3-basic. The primers used for construction are listed in table S2. The anti-Flag (catalog no. F1805, Sigma-Aldrich, St Louis, MO, USA), anti-Myc (catalog no. 10828-1-AP, Proteintech), anti-p53 (catalog no. sc-126, DO-1, Santa Cruz Biotechnology), anti-SPP1 (catalog no. 22952-1-AP, Proteintech), anti-FOXA1 (catalog no. 20411-1-AP, Proteintech), anti-KRAS (catalog no. 12063-1-AP, Proteintech), anti-STAT1 (catalog no. T55227, Abmart), anti-Phospho-Stat1-Tyr701 (catalog no. 9167T, Cell Signaling Technology), anti-ERK (catalog no. 4695T, Cell Signaling Technology), anti-GAPDH (catalog no. 60004-1-Ig, Proteintech), and anti-Phospho-ERK-Thr202/Tyr204 (catalog no. 4370T, Cell Signaling Technology) were commercially purchased. The secondary antibodies used were horseradish peroxidase (HRP)-conjugated AffiniPure goat anti-rabbit IgG (catalog no.SA00001-2, Proteintech) and anti-mouse IgG (catalog no. SA00001-1, Proteintech). Proteins were visualized with the ECL chemiluminescence reagent (Yeasen). U0126 and the proteasome inhibitor were purchased from Sangon Biotech and MedChemExpress, respectively. Establishment of KRAS-mutant and KRAS/TP53 co-mutant cell lines and RNA interference We established our experimental models using two distinct genetic background cell lines: the human p53-null H1299 cell line and the murine KP cell line (derived from KrasG12D/Trpnull LUAD). For the human H1299 model, lentiviral particles were produced by cotransfecting HEK293T cells with the Flag-tagged KRASG12C plasmid and packaging plasmids psPAX2 and pMD2.G(5:3:2) using polyethylenimine (PEI; Sigma-Aldrich, USA). Viral supernatants were collected 48 hours posttransfection, filtered (0.45 μm), mixed 1:1 with fresh DMEM, and added to H1299 cells. At 24 hours postinfection, stable cells were selected using puromycin (2 μg/ml; MedChemExpress, Shanghai, China) for 48 hours. Expression of KRASG12C was validated by immunoblotting (IB). To generate KRAS/TP53 co-mutant lines, these KRASG12C-mutant H1299 cells were subsequently infected with lentiviruses carrying Flag-tagged control, Flag-tagged p53-R175H, or Flag-tagged p53-R248Q plasmids, following the same packaging and infection protocol. After infection, cells were selected with blasticidin (5 μg/ml) to obtain cells stably expressing both mutant transgenes. Coexpression of KRAS and TP53 mutants was validated by IB. For the murine KP model, Flag-tagged p53-R172H or Flag-tagged p53-R270H plasmids were introduced via lentiviral transduction using the same protocol as described above. Transduced cells were selected using puromycin (2 μg/ml) to generate stable cell lines stably expressing respective p53 mutants. Confirmation of transgene expression was verified by IB. The siRNAs used in this study were synthesized and purified by GenePharma (Shanghai, China), and sequences are listed in table S3. siRNAs were introduced into cells using the Hieff Trans Liposomal transfection reagent following the manufacturer’s protocol (Yeasen). Cells were harvested 48 to 72 hours after transfection for IB or RT-qPCR. RNA sequencing and bioinformatic analysis Total RNA was extracted from H1299 cells, KRAS-mutant cells, and KRAS/TP53 co-mutant H1299 cells using RNAiso Plus (Takara, Japan) according to the manufacturer’s protocol. RNA sequencing (RNA-seq) and subsequent bioinformatic analysis was performed by OE Biotech (Shanghai, China). Differentially expressed mRNAs were identified using the limma package in the R (56), with thresholds set at an adjusted P value of <0.05 and an absolute log2(FC) of > 1. The ggplot2 package in the R software was used to draw the volcano plot. Functional annotation and pathway enrichment analyses were conducted using the ClusterProfiler package (version: 3.18.0) for Gene Ontology (GO) and KEGG pathways (57). Reverse transcription and quantitative real-time PCR Total RNA was extracted from cells using RNAiso Plus (Takara, Japan) according to the manufacturer’s protocol. Genomic DNA (gDNA) was extracted from cells using the TIANamp Genomic DNA kit (TIANGEN, China) according to the provided protocol. cDNA was synthesized from the extracted RNA using the HiScriptIII RT SuperMix for qPCR (+gDNA wiper) (Vazyme, China). RT-qPCR analysis was performed using the SYBR qPCR Master Mix as directed (Vazyme, China). The relative expression levels of target genes were calculated using the comparative ΔΔCt method. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was used as an internal control. The primers for RT-qPCR are listed in table S4. Immunoblotting Proteins were extracted using radioimmunoprecipitation assay (RIPA) buffer. The extracts were separated by SDS-polyacrylamide gel electrophoresis (PAGE) and subsequently transferred onto polyvinylidene difluoride membranes. After blocking with 5% nonfat milk for 1 hour at room temperature, the membranes were incubated with primary antibodies overnight at 4°C. Following three washes with TBST, the membranes were incubated with HRP-conjugated secondary antibodies diluted at 1:10,000 for 1 hour at room temperature. Proteins were visualized using the ECL chemiluminescence reagent (Yeasen, China) and imaged with a chemiluminescence detection system. Determine the secretion of endogenous SPP1 To determine the endogenous SPP1 secretion, KRAS-mutant cells and KRAS/TP53 co-mutant cells were seeded in 10-cm dishes and cultured under standard conditions until they reached 80% confluence. The culture medium was then replaced with serum-free conditioned medium (SFCM) for 48 hours. Subsequently, the supernatant was collected and filtered through a 0.45-μm filter. Levels of secreted SPP1 in SFCM were quantified using ELISA kits (catalog no. KE10118, KE00233, Proteintech, China) according to the manufacturer’s protocols. In vitro coculture with tumor cell and RAW264.7 or THP-1 Coculture assays were performed using Transwell chambers to evaluate the recruitment of RAW264.7 or THP-1 cells. Briefly, 1 × 105 KRAS-mutant or KRAS/TP53 co-mutant cells were plated in the lower chamber of a 24-well plate in 800 μL of culture medium containing 20% FBS. Then, 5 × 104 RAW264.7 or PMA-treated (catalog no. 50601ES03, Yeasen, China) THP-1 cells were seeded into the upper chambers (pore size, 8 μm; Corning, NY, USA). After incubation at 37°C for 36–48 hours, nonmigratory cells on the upper surface of the membrane were gently removed with a cotton swab. Cells that migrated to the lower surface were fixed with methanol and stained with 0.2% crystal violet. The number of migrated cells was quantified by counting at least three randomly selected fields under an optical microscope and analyzed using ImageJ. IP assay For exogenous co-IP, HEK293T cells were transfected with the indicated plasmids using PEI for 36 to 48 hours and treated with MG132 for 4 to 6 hours before being harvested. For endogenous IP, PC9 cells were transfected with Flag-pcDNA3.1-FOXA1 for 36 to 48 hours and treated with MG132 for 4 to 6 hours before being harvested. The IP assay was conducted using antibodies as indicated in the figure legends. Briefly, 500 to 1000 μg of protein lysates was incubated with specified antibody at 4°C for 5 hours. Protein A or G beads (Santa Cruz Biotechnology) were then added, and the mixture was incubated for additional 2 hours at 4°C. The beads were washed six to eight times with lysis buffer. Protein interactions were detected by IB as described above. Luciferase reporter assay Cells were seeded in 24-well plates and cotransfected with pGL3-basic, pGL3-SPP1, or pGL3-SPP1-DEL1/2/3 plasmids, along with plasmids encoding Renilla luciferase and FOXA1. After 48 hours of culture, the cells were harvested and luciferase activity was measured using a Dual-Luciferase Reporter Assay System (Vazyme, China) according to the manufacturer’s protocol. Chromatin immunoprecipitation Cells were seeded in 10-cm dishes and cultured for 24 hours. Cross-linking was performed using 37% formaldehyde for 10 min at room temperature, neutralized with 800 μl of 10× glycine for 5 min, and washed three times with cold phosphate-buffered saline (PBS). Cells were scraped and resuspended in cell lysis buffer [50 mM tris-HCl (pH 7.5), 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP-40, 0.25% Triton X-100, and protease inhibitor cocktail] and incubated on ice for 20 min with vortexing every 5 min. Nuclei were collected and resuspended in 400 μL nuclear lysis buffer [50 mM tris-HCl (pH 8.0), 10 mM EDTA, 1% SDS, and protease inhibitor cocktail]. After sonication (200 W, 60 cycles with 30-s on and 30-s off) and centrifuged at 12,000g for 5 min at 4°C, and the supernatants were mixed with anti-p53 or IgG at 4°C overnight and then with protein A/G beads at 4°C for 2 hours. Beads were washed sequentially with Low Salt Wash Buffer [50 mM tris-HCl (pH 8.0), 0.1% SDS, 0.5% deoxycholate, 1 mM EDTA, 1% NP-40, and 150 mM NaCl], High Salt Wash Buffer [50 mM tris-HCl (pH 8.0), 0.1% SDS, 0.5% deoxycholate, 1 mM EDTA, 1% NP-40, and 500 mM NaCl], LiCl Wash Buffer [50 mM tris-HCl (pH 8.0), 250 mM LiCl, 0.1% SDS, 0.5% deoxycholate, 1 mM EDTA, and 1% NP-40], and TE Buffer [10 mM tris-HCl (pH 8.0) and 1 mM EDTA]. Protein-DNA complexes were eluted using ChIP elution buffer (1% SDS and 0.1 M NaHCO3). After decross-linking at 62°C for 2 hours, DNA was purified and analyzed by qPCR and agarose gel electrophoresis. The primers used are as follows: ChIP-F: (5′-3′); ChIP-R: (5′-3′). Mouse xenograft study Five-week-old female C57BL/6 mice were purchased from and housed in the Laboratory Animal Science of Fudan University Shanghai Cancer Center. To establish orthotopic tumor xenograft models, 2 × 106 KP cells or KP cells stably expressing mutant p53–R172H or p53-R270H suspended in RPMI 1640 medium were injected into the right flanks of the mice. Tumor growth was monitored using an electronic digital caliper by measuring in two dimensions. Tumor volume was calculated according to the formula: tumor volume (mm3) = (length × width2) × 0.52. For drug treatment, the mice were randomized assigned to treatment groups after the tumor volume reached 50 to 100 mm3. Each mouse received intraperitoneal injections of 200 μg of anti-IgG, anti–PD-1, anti-SPP1, or anti-SPP1 plus anti–PD-1 antibody (BioXcell, USA) every 3 days, and the tumor size was measured after each administration of the therapeutic agents. Last, the mice were euthanized, and tumors were collected for analysis. All animal procedures were conducted in compliance with ethical guidelines and were approved by the Animal Welfare Committee of FUSCC. Flow cytometry assay Tumor tissues were collected from mice and digested into single-cell suspensions using the Gentle Enzymatic Digestion Kit for Mouse Tumor Tissue (RWD, China) according to the manufacturer’s protocol. Cell suspension was sequentially filtered through 100- and 40-μm cell strainers. Following erythrocytes lysis, the cells were incubated with fluorochrome-conjugated antibodies (table S4) for 30 min at 4°C in the dark. Data were acquired on a flow cytometer and analyzed using the CytExpert software (Beckman Coulter). Statistical analysis All in vitro data were obtained from three independent replicate experiments. Data were shown as means ± SD. A two-tailed unpaired t test or one-way analysis of variance (ANOVA) was used to compare the differences between two or more groups. The PFS and OS of patients were analyzed using the Kaplan-Meier method. P < 0.05 was considered statistically significant. ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Ethics statements The study was approved by the Institutional Review Board of the Fudan University Shanghai Cancer Center (approval no. 2004216-20-2005) and was performed in compliance with the Helsinki declaration. The animal study was approved by the ethics committee of the Fudan University Shanghai Cancer Center (grant no. FUSCC-IACUC-2024492).