Integrated multi-omics identifies a CD54+ iCAF-ITGAL+ macrophage niche driving immunosuppression via CXCL8-PDL1 axis in cervical cancer

Abstract

Cervical cancer (CC) remains a formidable clinical challenge, particularly in advanced stages where immune checkpoint blockade yields suboptimal responses. Despite the established role of the tumor microenvironment (TME) in fostering immunosuppression, the precise mechanisms of stroma-immune crosstalk in CC remain elusive. Leveraging single-cell RNA sequencing of 77,221 cells from CC and normal cervical tissues, we uncovered a tumor-enriched subpopulation of inflammatory cancer-associated fibroblasts (iCAFs) marked by elevated CD54 expression (CD54⁺ iCAFs), which independently predicted adverse clinical outcomes. Systematic dissection of intercellular communication networks revealed a tumor-specific alliance between CD54⁺ iCAFs and ITGAL⁺ macrophages, orchestrated through dysregulated ligand-receptor signaling. Spatial multi-omics approaches, including multiplex immunohistochemistry and spatial transcriptomics, confirmed their colocalization within an immunosuppressive niche. Mechanistically, CD54⁺ iCAFs promote immunosuppression by polarizing ITGAL⁺ macrophages toward an M2-like phenotype, primarily via CCL2 secretion. These fibroblasts further support immune evasion through two complementary pathways: direct CD54-ITGAL contact-dependent signaling and soluble CCL2-mediated macrophage reprogramming. The resulting macrophage activation stimulates autocrine CXCL8 secretion and subsequent PD-L1 upregulation, which ultimately suppresses CD8⁺ T cell functions, fostering an immune-tolerant microenvironment in CC. Therapeutic intervention using the CXCL8-CXCR1/2 inhibitor reparixin disrupted the CXCL8-PD-L1 axis, reduced PD-L1⁺ macrophage abundance and enhanced CD8⁺ T cell cytotoxicity. Notably, combination therapy with PD-L1 blockade demonstrated synergistic efficacy. Collectively, our findings reveal a stromal-immune checkpoint axis orchestrated by CD54⁺ iCAFs and ITGAL⁺ macrophages that underpins immunosuppression in CC, thereby providing a translational rationale for stroma-directed combination therapies that may overcome resistance to current immunotherapies.

Keywords: CXCL8; Cancer-associated fibroblasts; Cervical cancer; Immunosuppression; Macrophages; PD-L1.

Fig. 1

Single-cell characterization of cervical cancer and normal cervical tissues. A Left: UMAP projection of single-cell transcriptomes from cervical cancer (CC) and normal cervical tissues (NCT), colored by eight major cell types. Right: Dot plot displaying the average expression of established marker genes across annotated cell clusters. B Bar plot depicting the proportional distribution of cell types across samples. C Fibroblast abundance is significantly elevated in CC compared to normal tissue. D Left: Representative immunohistochemical (IHC) staining of α-SMA in CC (n = 10) and NCT (n = 6). Right: Quantification of α-SMA IHC intensity. E Kaplan–Meier survival analysis comparing patients with high versus low ACTA2 expression in the GEPIA2-CESC cohort. F Left: UMAP visualization of fibroblast subpopulations in tumor and normal tissues. Right: Dot plot showing mean expression of fibroblast subtype-defining markers. G Violin plots illustrating COL1A1 and COL1A2 expression across fibroblast subsets. H Volcano plot highlighting differentially expressed genes in fibroblasts between malignant and normal tissues. I Violin plot comparing CD54 expression levels in tumor versus normal tissue fibroblasts. J Aggregate CD54 expression in fibroblast subclusters, with iCAFs exhibiting predominant overexpression. Data in D (right panel) are presented as mean ± SD; statistical significance was determined by Student’s t-test. Survival analyses in E were performed using the log-rank test. *: P < 0.05, ***: P < 0.001

Fig. 2

CD54⁺ iCAFs promote migration and predict poor prognosis in cervical cancer. A Multiplex immunofluorescence staining shows co-localization of CD54 and α-SMA in cervical cancer (CC) and normal cervical tissues (NCT) (Scale bar: 100 μm). B Flow cytometry analysis of isolated CD54⁺ iCAF subtypes from CC. C Transwell migration assay of HeLa cells co-cultured with CD54-knockdown or CD54-overexpression iCAFs and their corresponding control fibroblasts for 48 h. D Quantification of migrated HeLa cells from three independent experiments. E Left: Representative immunohistochemical staining of CD54 in CC stromal tissue (n = 105) and normal cervical stroma (n = 19). Right: Quantitative analysis of stromal CD54 IHC staining intensity. F Left: Representative IHC images of high and low stromal CD54 expression in CC tissues. Right: Kaplan–Meier survival analysis of patients stratified by high (n = 39) and low (n = 66) stromal CD54 IHC scores. G Kaplan–Meier survival analysis comparing CC patients with high vs. low CD54⁺ iCAF infiltration (GEPIA2-CESC dataset). Data in A (right panel), D, and E (right panel) are presented as mean ± SD; statistical significance was determined by Student’s t-test. Survival analyses in F and G were performed using the log-rank test. **: P < 0.01, ***: P < 0.001

Fig. 3

Interaction between fibroblasts and mononuclear phagocytes in cervical cancer. A Heatmap displaying Spearman correlation coefficients between cell types, with clustering analysis revealing the shortest Euclidean distance between mononuclear phagocytes (MPs) and fibroblasts. B CellChat analysis demonstrating robust fibroblast–MP interactions at both Counts and Weight levels. C UMAP projection of re-clustered MP subsets following dimensionality reduction. D Dot plot depicting average expression of marker genes across MP subpopulations. E CellChat analysis identified iCAFs as the fibroblast subtype with the strongest interaction (Counts and Weight) with macrophages. F Ligand–receptor analysis reveals ITGAL as the top predicted receptor for CD54 (ligand). G Multiplex immunofluorescence staining confirmed the presence of ITGAL⁺ macrophage clusters co-expressing the M2 marker CD163 in CC tissues (Scale bar, 100 μm). H Multiplex immunofluorescence revealed spatial co-localization of CD54⁺ iCAFs and ITGAL⁺ macrophages in cervical cancer tissues (Scale bar, 100 μm)

Fig. 4

Co-localization of CD54⁺ iCAFs and ITGAL⁺ macrophages revealed by spatial transcriptomics. A Left: Hematoxylin and eosin (H&E) staining of spatial transcriptomic sections from two representative cervical cancer patients (Scale bar: 500 μm). Right: Spatial feature plots showing signature scores of CD54⁺ iCAFs, ITGAL⁺ macrophages, and their co-localization regions in matched tissue sections. B Distribution of signature scores for CD54⁺ iCAFs, ITGAL⁺ macrophages, and ITGAL⁻ macrophages across all spatial clusters. CD54⁺ iCAFs show strong co-localization with ITGAL⁺ macrophages, but not with ITGAL⁻ macrophages. C Pearson correlation analysis between signature scores of CD54⁺ iCAFs and ITGAL⁺ macrophages within co-localized spatial clusters across six patients. D GO analysis of biological processes significantly enriched in regions where CD54⁺ iCAFs and ITGAL⁺ macrophages co-localize. E Top eight KEGG pathways enriched among predicted target genes derived from co-localized CD54⁺ iCAF and ITGAL⁺ macrophage clusters

Fig. 5

CD54⁺ iCAFs promote monocyte migration and M2-like polarization through CCL2 secretion. A Left: Transwell migration assay of THP-1 cells co-cultured with CD54⁺ iCAFs or normal fibroblasts (NFs) for 48 h. Right: Quantification of migrated cells (n = 3 independent experiments). Scale bar: 100 μm. B-C THP-1 cells co-cultured with CD54⁺ iCAFs show upregulated expression of M2-like macrophage biomarkers and cytokines. D CD54⁺ iCAFs were transfected with CD54-targeting siRNA (si-CD54) or negative control siRNA (si-NC). Volcano plot of differentially expressed genes (DEGs) identified by RNA-seq analysis between si-CD54 and si-NC groups. E CD54⁺ iCAFs were transfected with si-CD54 or si-NC. Then, CD54 and CCL2 protein levels were assessed by Western blot. Left: Representative images. Right: Quantification from n = 3 independent experiments. F ELISA quantification of CCL2 secretion from CD54⁺ iCAFs and NFs. G Left: THP-1 cell migration in response to recombinant CCL2 treatment (50 ng/ml, 48 h) or PBS (Control group). Right: Quantification of migrated cells. All quantitative data are presented as mean ± SD. Statistical significance was determined using Student’s t-test for panels A (right), B, C, E (right), F, and G (right). *: P < 0.05, **: P < 0.01, ***: P < 0.001; ns, not significant

Fig. 6

CD54⁺ iCAF-driven macrophage reprogramming promotes CXCL8 expression and tumor progression in vivo. A THP-1-derived macrophages, transfected with control siRNA (Control group) or ITGAL-targeting siRNA (Treatment group), were co-cultured with CD54⁺ iCAFs for 48 h and subsequently subjected to RNA-seq analysis. Volcano plot shows differentially expressed genes (DEGs) between control and treatment groups. B Left: Representative flow cytometry plots of CD54⁺ iCAF abundance. Middle: Quantitative analysis of CXCL8⁺ cell frequencies in CD54⁺ iCAF-high tumors (n = 3). Right: quantitative analysis of CXCL8 expression in CD54⁺ iCAF-high versus CD54⁺ iCAF-low (Displaying in Supplementary Fig. 5F) tumors. C Flow cytometric quantification of CXCL8-producing immune cell subsets in cervical cancer tissues. D Multiplex immunohistochemistry images showing macrophage-specific CXCL8 expression in cervical cancer tissue (Scale bar: 100 μm). E NIH/3T3 fibroblasts transfected with CD54 overexpression plasmid (OE-CD54) or empty vector control (OE-NC). CD54 and CCL2 secretion levels were quantified by ELISA. F Representative tumor images from C57BL/6 mice co-injected with TC-1 cells and NIH/3T3 fibroblasts expressing CD54 or empty vector (Control). G Tumor growth curves measured every 3 days (n = 3 mice/group). H-I Flow cytometry analysis of CD206⁺ macrophage infiltration in tumors from (F). I (Right): Quantification of CD206⁺ macrophage proportions. J Serum MIP-2 levels measured by ELISA in experimental groups from (F). All quantitative data are presented as mean ± SD. Statistical significance was determined using Student’s t-test for panels B (right), E, G(right), I (right), and J. *: P < 0.05, **: P < 0.001, ***: P <0.001

Fig. 7

CXCL8 correlates with CD8⁺ T cell exclusion and promotes PD-L1 expression on macrophages through cell-contact and soluble-factor dependent mechanisms. A CIBERSORT analysis showing an inverse correlation between CXCL8 mRNA levels and CD8⁺ T cell infiltration. B Representative immunohistochemistry (IHC) images showing CD8⁺ T cell density in high- versus low-CXCL8 expressing tumors (Scale bar: 100 μm). C Negative correlation between protein levels of CXCL8 and PD-L1 based on IHC scoring (Pearson correlation). D Left: Flow cytometry plots of PD-L1⁺ cells. Right: Quantification of PD-L1 expression in high- versus low-CXCL8 tumors (n = 3 per group). E Representative IHC staining confirming macrophage-specific PD-L1 expression (Scale bar: 100 μm). F Left: Gating strategy for identifying PD-L1⁺ cells. Right: Quantification of PD-L1 expression across immune cell subtypes, showing macrophage dominance (n = 6). G Left: Flow cytometry profiles of PD-L1 expression. Right: PD-L1 levels in high- versus low-CXCL8 tumors. H Flow cytometry analysis of CXCL8 and PD-L1 expression on CD68+ macrophages co-cultured with CD54⁺ iCAFs under direct contact or Transwell conditions, with normal fibroblasts (NFs) and macrophage-only cultures as controls. The bar graph shows geometric mean fluorescence intensity from three independent experiments. Corresponding representative flow cytometry plots are shown in Supplementary Fig. 6D. I CXCL8 and PD-L1 expression on macrophages after co-culture with CD54⁺ iCAFs and treatment with IgG control, anti-CD54, or anti-ITGAL blocking antibodies. See Supplementary Fig. 6E for flow plots. J CXCL8 and PD-L1 expression on macrophages transfected with control siRNA (si-NC) or ITGAL-targeting siRNA (si-ITGAL), with or without CD54⁺ iCAF co-culture. Corresponding representative flow cytometry plots are shown in Supplementary Fig. 6F. K CXCL8 and PD-L1 expression on macrophages co-cultured with CD54⁺ iCAFs and treated with IgG control or anti-CCL2 neutralizing antibody. Representative flow plots are provided in Supplementary Fig. 6G. Data are presented as mean ± SD. Statistical tests used: two-tailed Student’s t-test (D, right; G, right; K); one-way ANOVA with Tukey's multiple comparisons test (F, right; H, I, J). **: P < 0.01, ***: P < 0.001; ns, not significant

Fig. 8

Therapeutic targeting of the CXCL8 pathway enhances anti-tumor immunity in cervical cancer. A Frequencies of PD-L1⁺ cells and PD-L1⁺ macrophages (gated on CD68⁺ cells) in primary cervical cancer tissues after ex vivo treatment with reparixin or DMSO control (n = 9 independent patient samples with sufficient cell yield for analysis). B-C Frequencies of CD45⁺CD3⁺CD8⁺ tumor-infiltrating lymphocytes (TILs) and percentages of cytokine-producing (TNF-α, IFN-γ) and cytolytic (CD107a, PRF1, GZMB) CD8⁺ TILs in patient tissues following reparixin or DMSO treatment (n = 6 patients with adequate cell numbers for full T cell immunophenotyping). D Experimental timeline: C57BL/6 mice were subcutaneously inoculated with TC-1 tumor cells (5 × 10⁵) and CD54-overexpressing NIH/3T3 fibroblasts (5 ×10⁶) on day 0, followed by treatment with anti-PD-1 (αPD-1, every three days) and/or reparixin (50 µg/mouse, every two days) (n = 3 per group). E Left: Representative tumor images at endpoint. Right: Tumor growth curves. Treatments were initiated when tumor volumes reached approximately 50 mm³. F-G Functional profiles of CD8⁺ TILs showing frequencies of cells expressing effector cytokines (IFN-γ, TNF-α) and cytolytic markers (GZMB, CD107a) following treatment with αPD-L1, reparixin, or their combination. Data in A–C, E (right), F, and G are presented as mean ± SD. Statistical analyses: paired two-tailed Student’s t-test (reparixin vs. DMSO control for each patient); one-way ANOVA with Dunnett’s multiple comparisons test (E, right, F, G). *: P < 0.05, **: P < 0.01, ***: P < 0.001; ns, not significant

Fig. 9

CD54⁺ iCAFs-ITGAL⁺ macrophage drive immunosuppression through CXCL8-PD-L1 mechanisms, and targeting this axis with reparixin and PD-L1 represents a promising strategy to overcome immune resistance in cervical cancer. This image was created with the help of BioRender

Fig. 10

The overall flowchart and main conclusion of this experiment. Comprehensive flowchart illustrating the experimental workflow from initial setup to final analysis. The process involves scRNA-seq and spatial-omics (mIHC and spatial transcriptomics), integration analysis, in vitro experiment and in vivo experimental validation. The main conclusion is presented at the bottom of each section. This image was created with the help of BioRender