Single-cell transcriptomic analysis reveals the heterogeneity and functional characteristics of macrophage subpopulations in colon cancer

Abstract

Background

Colon cancer remains one of the leading causes of cancer-related mortality globally. Tumor-associated macrophages (TAMs) are key contributors to tumor progression within the tumor microenvironment (TME). However, the role of secreted phosphoprotein 1 (SPP1), a critical regulator of macrophage–tumor interactions, in specific macrophage subsets in colon cancer remains unclear.

Methods

We performed single-cell RNA sequencing (scRNA-seq) on tumor and adjacent normal tissues from three colon cancer patients. A comprehensive analysis integrating pseudotime trajectory, transcription factor network, cell–cell communication, and in silico SPP1 knockout modeling was conducted to characterize macrophage heterogeneity and function.

Results

Five macrophage subtypes were identified. Among them, the Macrophages_SPP1 was significantly enriched in tumors and exhibited enhanced glycolytic metabolism, lysosomal activity, angiogenesis, and immunosuppression functions. This subtype showed increased interactions with fibroblasts, particularly via FTL–SCARA5 and FTH1–SCARA5 ligand–receptor pairs, implicating roles in stromal remodeling. In silico SPP1 knockout identified 93 stable responsive genes enriched in MHC class II-related and immune regulatory pathways, highlighting the role of SPP1 in shaping an immunosuppressive TME.

Conclusions

The Macrophages_SPP1 subtype may contribute to colon cancer progression through metabolic reprogramming and stromal interactions, suggesting that SPP1 and the FTL–SCARA5 axis could represent potential therapeutic targets.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-04002-z.

Introduction

Colorectal cancer (CRC) ranks as the third most common malignancy and the second leading cause of cancer-related mortality worldwide [1]. The incidence of colon cancer, a major subtype of CRC, has steadily increased in recent years, highlighting the urgent need to elucidate its molecular and cellular mechanisms, particularly those involving the tumor microenvironment (TME) [2].

The TME plays a central role in tumor progression, with macrophages recognized as key regulators of tumor growth, invasion, metastasis, angiogenesis, and immunosuppression [3, 4]. Notably, secreted phosphoprotein 1 (SPP1/osteopontin), a multifunctional protein frequently overexpressed in various cancers, has been identified as a key modulator of macrophage-tumor interactions. SPP1 functions not only as a diagnostic and prognostic marker but also facilitates tumor proliferation, invasion, and resistance to therapy by modulating immune responses [5, 6]. Traditionally, macrophages have been classified into pro-inflammatory, anti-tumorigenic M1 or immunosuppressive, pro-tumorigenic M2 subtypes. However, this binary classification oversimplifies the functional complexity of macrophages, as tumor-associated macrophages (TAMs) exhibit substantial plasticity and heterogeneity, frequently displaying overlapping phenotypic and functional traits [7, 8].

Single-cell RNA sequencing (scRNA-seq) has revolutionized research on the TME by enabling high-resolution analysis of cellular heterogeneity. Unlike bulk RNA-seq, which obscures cellular diversity, scRNA-seq enables the identification of distinct macrophage subsets and elucidation of their functional roles across various solid tumors [9, 10]. Despite the recognize role of SPP1 in macrophage-tumor interactions, the specific contributions of SPP1 + macrophage subtypes in colon cancer remain poorly defined, which may limit the development of effective targeted therapies.

In this study, we performed scRNA-seq on tumor and adjacent normal tissues from colon cancer patients and identified five distinct macrophage subtypes, among which the Macrophages_SPP1 subtype was associated with tumor progression. We further identified interactions between this subtype and fibroblasts, particularly mediated by FTL-SCARA5 and FTH1-SCARA5 ligand–receptor pairs. These findings advance our understanding of the role of SPP1 + macrophage in colon cancer and highlight their potential as therapeutic targets, thereby providing a foundation for novel treatment strategies.

Materials and methods

Sample collection and scRNA-seq dataset reuse

This study was approved by the Ethics Committee of Hainan General Hospital (Approval No.: Med-Eth-Re [2024] 740). All tissue samples were collected from patients at Hainan General Hospital (Haikou, Hainan, China). The three enrolled patients were diagnosed with colon cancer and had not received any anti-tumor therapy prior to surgical resection. Detailed clinical characteristics, including age, sex, tumor location, histology, pathological stage, tumor budding, perineural invasion, and MSI status, are provided in Table S1. All participants provided written informed consent, and the study was conducted in accordance with the principles of the Declaration of Helsinki.

Single-cell RNA sequencing

Fresh tissue samples were preserved in sCelLive^® Tissue Preservation Solution (Singleron) on ice within 30 min after surgical resection. Single-cell suspensions were then prepared according to the Singleron protocol, and scRNA-seq libraries were generated using the GEXSCOPE^® Single Cell RNA Library Kit (Singleron) [11]. Each library was diluted to 4 nM, pooled, and sequenced on the Illumina NovaSeq 6000 platform with 150 bp paired-end reads.

Raw data underwent quality control, dimensionality reduction, and clustering analyses using Scanpy v1.8.2 [12] under Python 3.9.10. For each dataset, cells were filtered based on the following criteria: (1) cells with fewer than 200 detected genes or within the top 2% of gene counts were excluded; (2) cells within the top 2% of UMI counts were excluded; (3) cells with mitochondrial gene content exceeding 50% were excluded [13]; and (4) genes expressed in fewer than five cells were excluded. To address potential inter-patient batch effects, we adopted a selective strategy. No batch correction was applied for global integration, in order to preserve genuine patient-specific variability, particularly for epithelial (tumor) cells. For macrophages and certain immune populations, batch correction was performed using Harmony to reduce technical variation and highlight biologically relevant subcluster differences. After filtering, a total of 46,528 cells were retained for downstream analyses, including 1,697 mononuclear phagocytes and 752 macrophages. Cell type annotation was performed using canonical lineage marker genes, for example: Epithelial cells (EPCAM, CDH1, KRT8), Endothelial cells (PECAM1, CLDN5, CDH5), Fibroblasts (COL1A1, DCN, LUM), Mural cells (RGS5, ACTA2, TAGLN), B cells (MS4A1, CD79A), Plasma cells (MZB1, IGHG1), T/NK cells (CD3D, CD3E, CD3G), Neutrophils (CSF3R, CXCR2, FCGR3B), Mast cells (TPSAB1, TPSB2, CPA3), and Mononuclear phagocytes (LYZ, CD14, C1QC).

Differentially expressed gene (DEG) analysis

DEGs were identified using the scanpy.tl.rank_genes_groups() function with the Wilcoxon rank-sum test and default parameters. Genes expressed in > 10% of cells in either group with a log₂ fold change > 0.25 and adjusted p-value < 0.05 (Benjamini-Hochberg correction) were considered significant.

Pathway enrichment analysis

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were conducted using the “clusterProfiler” R package (v4.0.0) [14]. Pathways with an adjusted p-value (p_adj) < 0.05 were considered significantly enriched and were visualized using lollipop plots. For Gene Set Variation Analysis (GSVA), the mean expression profile of each cell type was used as input [15].

Pseudotime trajectory analysis (Monocle 2)

To delineate the differentiation landscape of mononuclear phagocytes (MPs), Monocle2 (v2.22.0) was used for pseudotime trajectory analysis [16]. The top 2,000 highly variable genes were identified using Seurat’s FindVariableFeatures() function and utilized for trajectory inference. Dimensionality reduction was conducted using the DDRTree() method, and the trajectory was visualized with plot_cell_trajectory().

Transcription factor regulatory network analysis (pySCENIC)

Transcription factor regulatory networks were constructed using pySCENIC (v0.11.0) based on the scRNA-seq expression matrix and transcription factor annotations from AnimalTFDB [17]. The GRNBoost2 algorithm was used to infer co-expression networks, and CisTarget was employed to identify direct transcriptional targets and binding motifs. Regulon activity in each cell was quantified using AUCell. Cluster-specific regulons were identified based on the Regulon Specificity Score (RSS), and their activity patterns were visualized using heatmaps.

Cell–cell communication analysis: cellphone DB

Cell–cell communication (CCI) between macrophages and other cell types was inferring using Cellphone DB (v2.1.0) [18], based on curated ligand–receptor interaction databases. A total of 1000 permutations were used to established the null distribution, and ligand–receptor pairs were filtered based on average log-transformed expression values. Interaction pairs with p < 0.05 and average log expression >0.1 were considered significant and visualized using the net_plot and dot_plot functions.

Online database

GEPIA2 (Gene Expression Profiling Interactive Analysis 2, https://gepia2.cancer-pku.cn/) was used to analyze cancer gene expression by integrating TCGA and GTEx datasets, thereby providing comprehensive expression profiles across normal and tumor samples [19].

Immune cell fraction and functional activity analysis using CIBERSORT and SsGSEA

Pseudo-bulk gene expression profiles for each macrophage subtype were generated by averaging the normalized scRNA-seq data. CIBERSORT (LM22 gene signature matrix) with 1,000 permutations and quantile normalization estimated immune feature profiles [20]. Single-sample GSEA (ssGSEA) using the GSVA R package assessed pathway activity for immune-related gene sets (e.g., M1/M2 macrophage, hypoxia, angiogenesis). Spearman correlation analysis explored associations between CIBERSORT and ssGSEA scores.

In Silico knockout analysis (scTenifoldKnk)

In silico SPP1 knockout was performed using scTenifoldKnk on merged scRNA-seq data [21]. Following PCA normalization, 20 independent simulations were conducted, each involving 1,500 randomly sampled cells (nc_nNet = 10, nc_nCells = 1,500, and nc_nComp = 3). The top 100 genes ranked by |fold change| in each simulation were extracted, and genes appearing in ≥ 5 were defined as stable KO-responsive genes.

TCGA data acquisition and differential expression analysis

RNA-seq data for colon adenocarcinoma (COAD) were obtained from The Cancer Genome Atlas (TCGA) using the TCGAbiolinks (v2.36.0) R package [22]. Data were filtered for “Transcriptome Profiling” with “Gene Expression Quantification” (STAR - Counts). Gene annotations were converted to gene symbols using org.Hs.eg.db, retaining only protein-coding genes. Samples were grouped into SPP1-high and SPP1-low based on median expression, and DEGs were identified using DESeq2, with significance thresholds set at padj < 0.05 and |log₂FC| >0.5 considered.

Results

Identification and pseudotime trajectory analysis of mononuclear phagocyte subtypes

To obtain an overview of the cellular composition, we performed clustering and annotation of all 46,528 cells. Based on canonical lineage markers, ten major cell types were identified, including epithelial cells, endothelial cells, fibroblasts, mural cells, B cells, plasma cells, T/NK cells, neutrophils, mast cells, and mononuclear phagocytes (MPs) (Fig. 1A). The accuracy of this annotation was confirmed by expression of representative marker genes consistent with canonical lineage markers (Fig. S1A), and further supported by global quality control metrics across cell types (Fig. S1B). On the UMAP plot, non-MP lineages were displayed as broad categories to provide global context, whereas MPs were further resolved into five transcriptionally distinct subpopulations: proliferating MPs, macrophages, monocytes, mature dendritic cells (mature DCs), and conventional dendritic cells (cDCs) (Fig. 1B). Among these subsets, macrophages constituted the largest fraction, indicating their predominance within the MP compartment (Fig. 1C). We subsequently performed pseudotime trajectory analysis using Monocle 2 to investigate their developmental dynamics. The trajectory revealed a clear progression from monocytes toward macrophages, with the majority differentiating into macrophages, while a subset remained a proliferative state (Fig. 1D). It should be noted that pseudotime in this context represents a continuum of transcriptional states rather than canonical hematopoietic lineages. The trajectory does not imply that dendritic cells are direct precursors of macrophages; instead, their intermediate positioning reflects partial sharing of transcriptional programs with both monocytes and macrophages. Highly variable genes were grouped into five dynamic modules, and their dynamic expression patterns were visualized using a heatmap (Fig. 1E). Cluster 1 genes (e.g., EREG, CD300E, S100A12, NAMPT) peaked at early pseudotime stages, corresponding to ProliferatingMPs/Monocytes. Clusters 2–3 (e.g., S100A8/A9, CXCL8, IL1RN, PLAUR, TIMP1) dominated at intermediate pseudotime, mapping to inflammatory/activated macrophages. Clusters 4–5 (e.g., STAT1, LGMN, APOE, C1QA/B/C) were upregulated at late pseudotime stages, corresponding to dendritic cells and homeostatic macrophages. We independently validated the trajectory on the same embedding using Monocle3 [23]: pseudotime proceeds from monocytes along the trunk toward macrophages, with macrophages occupying the terminal state (Fig. 1F). Together, these results suggest a transcriptional continuum from monocytes toward distinct macrophage and dendritic cell states, highlighting the plasticity of mononuclear phagocytes within the TME.

Fig. 1.

Identification and pseudotime trajectory analysis of mononuclear phagocytes in colon cancer. A UMAP plot of all 46,528 cells colored by cell type annotation. B UMAP plot of MPs clustered by annotated cell types (left) and tissue origin (right). C Bar plots showing the proportions of MP subtypes in tumor and adjacent normal tissues. D Pseudotime trajectory of MPs inferred using Monocle2, colored by clusters (left) and pseudotime progression (right). E Heatmap showing pseudotime-ordered expression of highly variable genes grouped into five co-expression modules. F Monocle 3–based pseudotime trajectory showing dynamic transcriptional changes in MPs

Identification and characterization of macrophage subtypes

Based on these results, we further subclustered macrophages and identified five subtypes with distinct molecular features: Macrophages_LRMDA, Macrophages_DNAJB1, Macrophages_SPP1, Macrophages_FOLR2, and Macrophages_CXCL3 (Fig. 2A). To validate the annotation, UMAP feature plots of the defining markers (SPP1, FOLR2, CXCL3, DNAJB1, and LRMDA) showed that each gene was enriched the corresponding cluster, supporting the robustness of subtype identification (Fig. S1C). Among these, the Macrophages_SPP1 subtype appeared more prevalent in tumor samples compared to non-tumor samples (Fig. 2B). Although macrophages are traditionally categorized into M1 and M2 subtypes, analysis of canonical marker expression revealed that multiple subtypes co-expressed M1 and M2 markers to varying degrees (Fig. 2C). Notably, the Macrophages_SPP1 subtype exhibited higher expression of M2 markers, suggesting its potential involvement in tumor progression and warranting further characterization. Consistently, analysis of an independent scRNA-seq dataset (GSE132465) [24] using the original authors’ cell type annotations also demonstrated that SPP1 + Macrophages were more enriched in tumor tissues than in normal controls (Fig. S1D, E), further supporting the robustness of our findings.

Fig. 2.

Heterogeneity and functional characterization of macrophage subpopulations in colon cancer. A A UMAP plot of Macrophages clustered by annotated cell types (left) and tissue origin (right). B Bar plots illustrating the distribution of Macrophages subtypes in tumor and adjacent normal tissues. C Violin plots displaying the expression of M1 and M2 signature genes across macrophage subtypes. D Volcano plot showing DEGs of the Macrophages_SPP1 subtype between tumor and control groups. E Lolly plot illustrating enrichment analysis of genes upregulated in the Macrophages_SPP1 subtype in tumors. F Bar plot displaying GSEA results of pathways upregulated in the Macrophages_SPP1 subtype within the tumor group. H Dot plot showing SPP1 expression across 33 TCGA tumor types

Differential expression analysis of the Macrophages_SPP1 subtype between tumor and non-tumor groups (Fig. 2D) was followed by GO, KEGG, Reactome, Wikipathways, and GSEA enrichment analyses (Fig. 2E, F). GO analysis indicated enrichment in lysosomal activity, immune regulation, and cell adhesion while KEGG analysis highlighted the lysosome pathway, suggesting enhanced phagocytic and degradative capacity. Reactome pathways enriched included platelet activation and degranulation, neutrophil degranulation, cytoprotection by HMOX1, and SLIT-ROBO signaling, indicating roles in inflammation, immune modulation, and migration. Wikipathways analysis revealed enrichment in aerobic glycolysis, glycolysis and gluconeogenesis, and complement pathways, suggesting increased glycolytic activity and immune regulatory potential. GSEA further revealed enrichment in IL-17, TNF, cytokine-cytokine receptor interaction, and Toll-like receptor signaling pathway, supporting the activated immune phenotype of this subtype.

Collectively, these findings suggest that the Macrophages_SPP1 subtype may contribute to colon cancer progression and is associated with glycolytic metabolism, lysosomal function, immune modulation, and pro-migratory activities within the TME. Additionally, GEPIA database showed SPP1upregulation in 22 of 33 cancer types, suggesting potential therapeutic relevance across multiple malignancies, pending further investigation.

Functional characterization of the macrophages_SPP1 subtype

To further compare functional states across macrophage subtypes, we performed gene set scoring analyses. The Macrophages_SPP1 subtype exhibited significantly higher scores in gene sets related to lipid metabolism, extracellular matrix remodeling, and pro-inflammatory pathways compared to other macrophage subtypes (Fig. 3A–C). We subsequently conducted scGSVA analysis for pathway activity profiling. Hallmark pathway analysis revealed strong associations with epithelial-mesenchymal transition (EMT), angiogenesis, and apical surface pathways (Fig. 3D), while KEGG analysis indicated enrichment in ECM-receptor interaction, phenylalanine metabolism, and the amino acid biosynthesis pathways (Fig. 3E). These results suggest that the Macrophages_SPP1 subtype is characterized by enhanced ECM remodeling, pro-angiogenic potential, and metabolic activity within the TME, which may contribute to colon cancer progression.

Fig. 3.

Functional signature and transcriptional regulatory analysis of macrophage subpopulations. A Box plots displaying lipid metabolism signature scores across macrophage subtypes. B Box plots displaying extracellular matrix remodeling signature scores across macrophage subtypes. C Box plots displaying pro-angiogenic signature scores across macrophage subtypes. D Heatmap of hallmark pathway activity across macrophage subtypes based on ssGSVA analysis. E Heatmap of KEGG pathway activity across macrophage subtypes based on ssGSVA analysis. F Dot plot showing the top five regulons identified in the Macrophages_SPP1 subtype using pySCENIC analysis. G Lolly plot displaying pathway enrichment analysis of target genes regulated by the top five transcription factors in the Macrophages_SPP1 subtype within the tumor group

We further performed pySCENIC-based transcriptional regulatory network analysis of the Macrophages_SPP1 subtype in both tumor and control groups (Fig. 3F). GO, KEGG, and Reactome enrichment analyses of the top five transcription factors identified in the tumor group revealed significant enrichment in TNF signaling, TGF-β receptor signaling, Hippo signaling, and cell cycle pathways, as well as colorectal cancer-related pathways in KEGG (Fig. 3G). These findings suggest that the identified regulons may regulate tumor-associated pathway in colon cancer.

Immune feature profiles and functional pathway activities across macrophage subtypes

We applied CIBERSORT with the LM22 signature matrix on pseudo-bulk gene expression profiles generated from scRNA-seq data to evaluate the immune feature of five macrophage subtypes. Stacked bar plots revealed distinct immune profiles, with the Macophages_SPP1 subtype showing higher similarity to M2 macrophage and activated mast cell (Fig. 4A). Consistently, heatmap analysis highlighted the heterogeneity of immune features, with the Macrophages_SPP1 subtype strongly association with activated mast cell signatures, suggesting a phenotype linked to angiogenesis and immune modulation (Fig. 4B). PCA based on the CIBERSORT estimates further demonstrated clear segregation among the subtypes (Fig. 4C).

Fig. 4.

Immune feature profiling and functional pathway activities across macrophage subtypes. A Stacked bar plots of CIBERSORT-estimated immune signature profiles across five macrophage subtypes. B Heatmap of relative immune feature profiles. C PCA showing subtypes clustering based on CIBERSORT signatures. D Heatmap of ssGSEA-derived pathway enrichment scores across subtypes. E PCA of ssGSEA scores indicating functional state separation. F Spearman correlation heatmap between immune signature and ssGSEA pathway activities

To assess functional states, we performed ssGSEA using curated immune-related gene sets. The Macrophages_SPP1 subtype showed high enrichment in hypoxia and angiogenesis pathways and lower enrichment in M1 macrophage and inflammatory response pathways, indicating a pro-tumoral functional state (Fig. 4D). PCA of ssGSEA scores confirmed functional segregation across subtypes (Fig. 4E).

Spearman correlation analysis revealed a positive association between M2-like signatures and angiogenesis, and a negative association between activated NK signatures and hypoxia (Fig. 4F). Notably, the activated mast cell-like feature of the Macrophages_SPP1 subtype correlated positively with angiogenesis activity, suggesting a possible role in angiogenesis and in shaping an immunosuppressive microenvironment in colon cancer.

Together, these findings demonstrate that the Macrophages_SPP1 subtype exhibits a distinct immune and functional profile characterized by hypoxia and angiogenesis pathway activity, M2-like features, and transcriptional similarity to activated mast cells, which may collectively contribute to its pro-tumoral functions within the TME. To exclude misclassification, we examined canonical mast cell markers, which were absent from macrophage subtypes but enriched in mast cells. Moreover, macrophage lineage module scores were consistently elevated across all macrophage subtypes, including Macrophages_SPP1. These results confirm its macrophage identity and indicate that the mast cell-like features reflect convergent functional programs rather than misannotation (Fig. S2).

Macrophages exhibit enhanced interactions with stromal and cancer cells in colon cancer

To elucidate interactions between macrophages and other TME components, we conducted cell–cell communication analysis. Compared to control (Fig. 5A), macrophages, particularly the Macrophages_SPP1 subtype, showed an increased number of interactions with fibroblasts, mural cells, and cancer cells in tumor samples, while interactions with endothelial cells remained unchanged (Fig. 5B). Ligand–receptor analysis revealed elevated interactions, including FTL–SCARA5 and FTH1–SCARA5 between macrophages and fibroblasts and CCL3–CCR1 among macrophages (Fig. 5C, D). The upregulated FTL/FTH–SCARA5 axis suggests enhanced iron metabolism crosstalk between macrophages and fibroblasts, which may promote fibroblasts activation and ECM remodeling within the TME. Collectively, these interactions may influence immune modulation and potentially support tumor progression.

Fig. 5.

Cell–cell communication analysis between Macrophages_SPP1 subtype and stromal cells. A Net plot showing predicted interactions between the Macrophages_SPP1 subtype and fibroblasts or other cell types in control group. B Net plot showing predicted interactions between the Macrophages_SPP1 subtype and fibroblasts or other cell types in tumor group. C Dot plot displaying the top 30 predicted ligand–receptor interaction pairs between the Macrophages_SPP1 subtype and fibroblasts or other cell types in control group. D Dot plot displaying the top 30 predicted ligand–receptor interaction pairs between the Macrophages_SPP1 subtype and fibroblasts or other cell types in tumor group

Simulated SPP1 knockout reveals gene expression changes in macrophages

To investigate downstream effects of SPP1 depletion, we performed 20 independent in silico knockout simulations using scTenifoldKnk, identifying 93 stable KO-responsive genes appearing in ≥ 5 runs (Fig. 6A). Using TCGA-COAD RNA-seq data, samples were stratified into SPP1-high and SPP1-low groups based on median expression. Differential expression analysis revealed 1,470 upregulated and 808 downregulated genes in the SPP1-high group (padj < 0.05, |log2FC| >0.5) (Fig. 6B). Intersecting the KO-responsive gene set with the upregulated DEGs in SPP1-high samples identified 13 overlapping genes (Fig. 6C). Enrichment analyses revealed significant involvement in MHC class II protein complex-related processes (Fig. 6D) and KEGG pathway including antigen processing, IgA immune networks, and cell adhesion molecules (Fig. 6E). A STRING-based PPI network revealed clear modular structure, with Louvain community detection identifying distinct modules and hub genes that may serve as potential mediators of SPP1-related immune and microenvironmental regulation in colon cancer. These findings are based on the computational inference and will require experimental validation in large cohorts to confirm their biological relevance (Fig. 6F).

Fig. 6.

Identification and functional analysis of SPP1 KO-responsive genes in colon cancer. A Frequency distribution of KO-responsive genes identified from 20 in silico scTenifoldKnk simulations, highlighting 93 stable core genes. B Volcano plot showing DEGs between SPP1-high and SPP1-low TCGA-COAD samples. C Venn diagram illustrating the overlap between KO-responsive genes and upregulated DEGs in SPP1-high samples (13 genes). D GO enrichment analysis of the overlapping genes. E KEGG enrichment analysis of the overlapping genes F STRING-based PPI network displaying modular structure, hub genes, and connectivity of KO-upregulated genes

Discussion

Tumor-associated macrophages (TAMs) exhibit substantial heterogeneity within the colon cancer TME, influencing tumor progression, immune modulation, and therapeutic response [25, 26]. In this study, we performed in-depth characterization of macrophage subpopulations in colon cancer using scRNA-seq, focusing on the Macrophages_SPP1 subtype. This subtype displayed enhanced glycolytic metabolism, lysosomal activity, and gene signatures associated with inflammation and migration, suggesting its potential role in promoting colon cancer progression.

Previous studies have established that TAMs promote tumor growth and metastasis through angiogenesis, ECM remodeling, and immune suppression in various cancers [27, 28]. In colon cancer, our enrichment analyses align with these roles, revealing that the Macrophages_SPP1 subtype is associated with EMT, TNF, and TGF-β signaling pathways, which are pivotal in metastasis and progression [29]. The identification of the Macrophages_SPP1 as displaying a transcriptional resemblance to activated mast cells and high enrichment in hypoxia and angiogenesis pathways underscores its potential pro-tumoral role within the colon cancer microenvironment. The resemblance to activated mast cell signatures suggests that the Macrophages_SPP1 subtype may be associated with angiogenesis and immune suppression, in line with previously described functional roles of activated mast cells. Furthermore, we identified enhanced macrophages–fibroblasts interactions in tumor samples, particularly through the FTL–SCARA5 and FTH1–SCARA5 ligand–receptor pairs, which represent iron metabolism-related crosstalk. FTL and FTH1, as subunits of ferritin, are critical for intracellular iron storage, while SCARA5 is a scavenger receptor that mediates cellular uptake of ferritin. The upregulation of these interactions implies that macrophages may serve as a source of iron-loaded ferritin, taken up by fibroblasts via SCARA5, thereby supporting fibroblast metabolic activity, activation, and ECM remodeling. Such iron transfer may also contribute to oxidative stress and immune suppression within the TME, collectively fostering a pro-tumorigenic niche [30, 31]. These findings highlight the significance of the Macrophages_SPP1 subtype as a potential target for therapeutic strategies aimed at modulating the tumor immune microenvironment.

Using scTenifoldKnk, we simulated SPP1 knockout to predict downstream regulatory gene changes, emphasizing the role of SPP1 in modulating MHC class II molecule expression and thereby influencing macrophage function within the TME. Such computational perturbation approaches are valuable for identifying key regulatory factors and potential therapeutic targets in colon cancer.

This study has limitations. The small sample size may limit the generalizability of our findings across diverse colon cancer subtypes and patient populations. Additionally, the lack of in vivo validation of the Macrophages_SPP1 subtype and FTL/FTH1–SCARA5 interactions restricts mechanistic insights. Future studies should validate these findings in larger clinical cohorts, employ in vivo models to confirm the functional roles of Macrophages_SPP1, and explore multi-omics approaches to dissect SPP1-mediated pathways or their role in immunotherapy resistance. Moreover, as our analyses heavily relied on computational inference, these findings should be interpreted with caution. Future work will benefit from experimental validation of the Macrophages_SPP1 subtype and its fibroblast interactions, as well as integration of spatial transcriptomics and multi-omics datasets to capture the spatial context and regulatory mechanisms of SPP1-mediated crosstalk. Such efforts may also clarify the implications of the Macrophages_SPP1 subtype in immunotherapy response and resistance.

In summary, our findings suggest that the Macrophages_SPP1 subtype may play an important role in colon cancer progression through enhanced glycolytic metabolism, lysosomal activity, immune modulation, and stromal interactions via the FTL/FTH1–SCARA5 axis. Targeting the SPP1 pathway, including the FTL/FTH1–SCARA5 axis, could be considered as a potential therapeutic approach to complement existing immunotherapies; however, these hypotheses require validation in larger patient cohorts and in vivo experimental models.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CCI

Cell–cell communication

COAD

colon adenocarcinoma

CRC

Colorectal cancer

DEG

Differentially Expressed Genes

ECM

Extracellular matrix

EMT

Epithelial-mesenchymal transition

GO

Gene Ontology

GSVA

Gene Set Variation Analysis

KEGG

Kyoto Encyclopedia of Genes and Genomes

MPs

Mononuclear phagocytes

MSI

Microsatellite instability

MSS

Microsatellite stable

PCA

Principal component analysis

PPI

Protein–protein interaction

RSS

Regulon specificity score

scRNA-seq

Single-cell RNA sequencing

SPP1

Secreted phosphoprotein 1

ssGSEA

Single-sample gene set enrichment analysis

TAM

Tumor-associated macrophages

TCGA

The cancer genome atlas

TME

Tumor microenvironment

Author contributions

Jiazheng Zhao: Writing – original draft, Formal analysis, Investigation, Methodology; Xiao Li: Resources; Chunting Wei: Investigation; Baochun Wang: Funding acquisition, Resources, Formal analysis; Yi Min: Funding acquisition, Resources, Supervision; Dayong Wang: Writing – review and editing, Conceptualization, Data curation, Funding acquisition, Project administration, Supervision. All authors have read and approved the final manuscript.

Funding

This research was supported by the Hainan Provincial Graduate Student Innovation Research Project (Grant No. Qhyb2024-43 to Jiazheng Zhao), the National Natural Science Foundation of China (Grant Nos. 32160214 and 31760246 to Dayong Wang), the Natural Science Foundation of Hainan Province (Grant Nos. 821RC1053, 822RC651 and 823RC562 to Dayong Wang, Yi Min, and Baochun Wang), and the Cooperative Innovation Center of Hainan University (Grant No. XTCX2022JKB07 to Dayong Wang).

Data availability

The raw sequence data have been deposited in the GSA-Human database (HRA009194) [32, 33] and are publicly available at https://ngdc.cncb.ac.cn/gsa-human.

Declarations

Ethics approval and consent to participate

Sample collection and study procedures were carried out in accordance with the Declaration of Helsinki. Any related ethical issues were approved by the Ethics Committee of Hainan General Hospital (Ethics Approval No.: Med-Eth-Re [2024] 740). Every patient and/or their legal guardians signed informed consent after receiving oral and written information.

Consent for publication

All authors agree to publish.

Competing interests

The authors declare no competing interests.