Genetic insights into colorectal cancer pathogenesis: a multi-omics and immunity perspective

Abstract

Background

Colorectal cancer (CRC) is a global health issue influenced by both genetic and environmental factors. Identifying key genes closely associated with CRC is crucial for understanding its pathological mechanisms and discovering therapeutic targets. This study aimed to integrate multi-omics datasets and Mendelian randomization (MR) approaches to identify CRC-related genes and to clarify their roles in tumor immunity and therapeutic potential.

Methods

This is a cross-sectional study. We utilized databases such as Gene Expression Omnibus (GEO), Finnish National Genome Project (FinnGen), expression Quantitative Trait Loci (eQTL), and The Cancer Genome Atlas (TCGA), and employed MR, summary-data-based MR (SMR), and gene expression analyses to screen genes related to CRC development. Through immune cell infiltration analysis and mediation MR, we explored the relationship between these genes, tumor immunity, and immunotherapy.

Results

Our study identified S100P and RIPK2 as risk factors for CRC, while FCGRT and VSIG2 were found to be protective factors. Additionally, the differential expression of these genes in CRC tissues was validated through immunohistochemistry (IHC) and reverse transcription quantitative polymerase chain reaction (RT-qPCR) experiments. Mediation MR analysis demonstrated that immune cell phenotypes could mediate the effect of these genes on CRC. These genes were also found to be associated with tumor mutational burden (TMB) and microsatellite instability (MSI) scores.

Conclusions

Our findings reveal how these genes contribute to CRC pathogenesis by modulating the immune microenvironment, providing important biomarkers and targets for the development of novel therapeutic strategies.

Highlight box.

Key findings

• This study identified four key genes associated with colorectal cancer (CRC): RIPK2 and S100P act as genetic risk factors, while FCGRT and VSIG2 serve as protective factors.

• Their expression patterns were validated through immunohistochemistry, reverse transcription quantitative polymerase chain reaction, and functional assays.

• These genes are linked to immune cell phenotypes, tumor mutational burden, and microsatellite instability, suggesting roles in shaping the tumor immune microenvironment.

What is known and what is new?

• Existing knowledge: CRC pathogenesis involves both genetic variants and immune regulation, but the causal genes and their immune mechanisms remain incompletely understood.

• This study: by integrating multi-omics data with Mendelian randomization and single-cell analysis, we provide causal evidence for four CRC-related genes and demonstrate their functional links to tumor immunity and immunotherapy response.

What is the implication, and what should change now?

• The identified genes may serve as biomarkers for prognosis and predictors of immunotherapy response in CRC.

• These findings support the development of targeted strategies that combine genetic and immune-based approaches, potentially improving personalized treatment for CRC patients.

Introduction

Colorectal cancer (CRC) stands as the fourth leading cause of cancer mortality worldwide, with notable variations in occurrence rates among diverse geographic regions, sexes, and age groups (1,2). Genetic and environmental factors are closely associated with the incidence of CRC, with approximately 20% of cases having a positive family history (3,4). In recent years, despite advancements in diagnostic and therapeutic technologies, the treatment outcomes for CRC continue to pose challenges. Surgical resection remains the primary treatment approach in the clinical setting; however, postoperative recurrence is common. Even with combined treatments, including surgery and chemoradiation, the 5-year survival rate for patients hovers around 60% (5).

The emergence of immunotherapeutic strategies, including the application of immune checkpoint inhibitors (ICIs) and the utilization of chimeric antigen receptor-modified T (CAR-T) cell treatments, has marked a significant advancement in cancer care, offering promising new avenues of therapy for individuals diagnosed with CRC (6). These therapies work by engaging and amplifying the body’s immune response to detect and destroy malignant cells, thus serving as a valuable complement to traditional treatment regimens. However, only a subset of patients, such as those with high microsatellite instability (MSI-H)/mismatch repair deficiency (dMMR) tumors, demonstrate favorable responses to immunotherapy, which is also associated with risks of resistance and recurrence (7). Recent multi-omics studies integrating genomic, transcriptomic, and microbiome data have revealed distinct CRC subtypes and provided biomarkers for predicting immunotherapy response (8). Consequently, it is imperative to discover novel therapeutic targets and enhance immunotherapeutic strategies for the treatment of CRC.

Genome-wide association studies (GWASs) have unveiled genetic variations underlying CRC, highlighting the role of single-nucleotide polymorphisms (SNPs) in the genetic landscape of CRC and providing valuable insights for the identification of disease susceptibility genes (9,10). While GWAS provide insights into CRC susceptibility, they are limited in explaining tumor heterogeneity and immune interactions (11). However, current exploration of the association between single SNPs and CRC remains insufficient, failing to fully exploit the potential value of these data (12-14). Mendelian randomization (MR) utilizes SNPs as instrumental variables (IVs) for assessing the causal effects of exposures on outcomes (15). In previous MR studies on CRC, the focus has predominantly been on investigating the causal relationships between CRC and various environmental or disease exposures (9,16), overlooking the regulation of CRC and tumor immunity by genes. The core concept of summary-data-based MR (SMR) methodology involves utilizing genetic variations as IVs to statistically analyze the association between aggregated data from GWAS and expression Quantitative Trait Loci (eQTL) studies. This analysis aims to infer the causal relationship between changes in the expression levels of specific genes (or other molecular phenotypes) and disease risk (17,18). This approach has been employed to elucidate causal associations between genes and diseases as well as identifying potential drug targets (19). Furthermore, it can be integrated with multi-omics studies to delve deeper into the mechanisms underlying disease onset (20).

By integrating data from the Gene Expression Omnibus (GEO), Finnish National Genome Project (FinnGen), eQTL, and TCGA databases, this study aims to identify genes associated with the onset of CRC and to explore how these genes influence CRC through immune cell functions and their interactions with the tumor immune microenvironment. We present this article in accordance with the STREGA reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1072/rc).

Methods

Study design and data sources

Figure 1 provides a detailed description of the design of this study. The GWAS summary statistics for CRC were sourced from the Finnish R10 database (r10.finngen.fi), which is based on the European population (21). The sample comprised 6,847 CRC patients and 314,193 healthy controls. eQTL summary statistics were obtained from the eQTLGen Consortium (https://www.eqtlgen.org/phase1.html), which includes gene expression analysis results from blood samples of 31,684 individuals across 37 independent cohorts (22). This study utilized cis-acting eQTL (cis-eQTL) data, analyzing approximately 11 million SNPs with a window size of 1 Mb and a minimum allele frequency (MAF) of over 1%. We identified 16,989 cis-eQTL genes, representing 88.3% of the total analyzed. These cis-eQTLs may function by regulating gene expression, such as affecting promoters or enhancers, thereby enabling the exploration of genetic variations and their links to specific diseases. Additionally, the study investigates how these variations influence disease risk. RNA sequencing (RNA-seq) data for 620 CRC were obtained from The Cancer Genome Atlas (TCGA) dataset (https://portal.gdc.com). Immune cell data were sourced from the OpenGWAS website [IEU OpenGWAS project (mrcieu.ac.uk) (http://mrcieu.ac.uk)], which includes GWASs of immune cells encompassing 3,757 participants and 731 cell traits, covering 539 immune traits and 192 relative counts (ratios between cell levels) (23). Three gene expression datasets (GSE15781, GSE79793, GSE50117) were downloaded from the GEO database. The single-cell dataset GSE166555, available at the National Center for Biotechnology Information (NCBI)’s GEO repository (http://www.ncbi.nlm.nih.gov/geo), includes single-cell expression data from 12 patients with CRC (24). These datasets include data from 48 tumor tissue samples and 29 normal tissue samples.

Figure 1.

Study design flowchart for CRC-related gene research. This flowchart details the overall experimental design for studying CRC-related genes. Using MR and SMR analysis, eQTLs were used as exposures and CRC as the outcome. DEGs from GEO datasets GSE15781, GSE79793, and GSE50117 were identified. Upregulated DEGs intersected with MR genes having OR >1 and SMR b >0, while downregulated DEGs intersected with MR genes having OR <1 and SMR b <0, identifying candidate genes promoting or inhibiting CRC. The flowchart also shows analyses of gene expression correlation with TMB/MSI scores, immune cell infiltration, and mediation by immune cells, offering insights into the genetic mechanisms influencing CRC development and potential new therapeutic targets. CRC, colorectal cancer; DEG, differentially expressed gene; FC, fold change; eQTL, expression Quantitative Trait Loci; GEO, Gene Expression Omnibus; MR, Mendelian randomization; MSI, microsatellite instability; OR, odds ratio; SMR, summary-data-based Mendelian randomization; TMB, tumor mutational burden.

Differentially expressed genes analysis

In our research, the acquisition of data was chiefly from the GEO database, which is accessible to the public. We retrieved MINiML files from selected GEO series (GSE), encompassing detailed platform descriptions, specifics of the samples, and raw experimental data. The raw dataset underwent log₂transformation as part of a unified data preprocessing regimen. Non-normalized data were subjected to quantile normalization using the normalize.quantiles function within the preprocessCore package in R. We translated probe IDs into the corresponding gene names using the platform’s provided annotations. Probes corresponding to multiple genes were excluded, and for those associated with a single gene, we computed the mean expression level. Addressing batch effects present within and among the datasets was achieved using the removeBatchEffect function from the limma package. The success of the normalization and batch effect rectification was evaluated by box plots and principal component analysis (PCA). We used the limma package in R for the analysis of differentially expressed messenger RNAs (mRNAs), adopting a threshold for adjusted P value below 0.05 and an absolute value of log₂fold change (log₂FC) over 2, aiming to limit false discoveries. Visual representation of the findings was facilitated by PCA plots and expression heatmaps crafted with R’s ggord and pheatmap packages, respectively.

MR analysis

We established criteria to filter SNPs that qualify as IVs for MR. Through associative analysis, SNPs strongly correlated with the exposure factor were selected as IVs. The filtering criteria in the eQTL database are: (I) A P value less than 5e−08. (II) Within a 10,000 kb window, the linkage disequilibrium (LD) R² value between IVs should be less than 0.001. (III) The F-statistic should be greater than 10 [F = beta²/standard error (SE)²] to eliminate bias from weak IVs. In the assessment of causal relationships, the inverse variance-weighted (IVW) method was employed as the primary tool to estimate the relationship between exposure and outcome. Heterogeneity was assessed using Cochran’s Q test, with P<0.05 considered indicative of heterogeneity. In the MR analysis, SNPs with palindromic structures and ambiguous information were excluded. Additionally, this study employed MR-Egger regression to detect pleiotropy issues, with particular attention to whether the intercept significantly differs from zero, identifying potential horizontal pleiotropic confounders. The significance level was set at a P value less than 0.05. This method aids in assessing and correcting biases caused by pleiotropy. Two-step MR analysis was utilized to explore the pathway from genes to CRC mediated by immune cells (25). Firstly, the overall effect (beta all) of the genes on CRC was obtained, followed by the calculation of beta1 from the genes to immune cells and beta2 from immune cells to CRC through two-sample MR analysis. The mediating effect of immune cells (beta mediate) is calculated as beta1 × beta2. The direct effect (beta direct) is derived from beta all to beta mediate. To include a broader set of SNPs as IVs in the mediation MR analysis, the selection criteria for exposure SNPs are relaxed to a P value threshold of 1e−05 (26). All two-sample MR analyses were conducted using the “TwoSampleMR” package in R software version 4.0.3.

SMR analysis

In the SMR analysis, genes were treated as the exposure, while cis-eQTLs were used as IVs, with CRC as the outcome. A Heterogeneity in Dependent Instruments (HEIDI) test was conducted to distinguish pleiotropy from linkage models by identifying heterogeneity in the IVs. In the HEIDI test, a P value less than 0.05 was indicative of a potential pleiotropic association. Analysis was carried out using version 1.3.1 of the SMR software, and the results were visualized using code obtained from the official website (17) (https://cnsgenomics.com/software/smr/#Overview).

Single-cell analysis

Single-cell analysis involved downloading the corresponding .h5 formatted data file and annotations from TISCH22 (27). The data were initially processed using the MAESTRO tool in R, which included quality control of cells and filtering of genes. Subsequently, the data were normalized, and PCA was conducted using the Seurat package to identify key variational features within the data. Based on the PCA results, cells were re-clustered and visualized using Seurat’s clustering tools and the t-distributed stochastic neighbor embedding (t-SNE) method to explore the relationships between cell populations. Furthermore, differential gene expression analysis was performed to identify and validate the biological characteristics of each cell subgroup using the statistical functions within Seurat, enhancing the understanding of the diverse cellular subpopulations present in CRC.

Immune infiltration analysis

To conduct a reliable assessment of immune relevance, we employed the immunedeconv R package, which integrates multiple algorithms including CIBERSORT and EPIC. CIBERSORT is a computational tool designed for quantifying the relative abundances of diverse cell types within complex tissue samples. In our study, we applied this algorithm to assess the abundance of 22 infiltrating immune cell subpopulations (28). We extracted the expression values of the genes S100P, FCGRT, RIPK2, and VSIG2 from the TCGA database to observe the expression patterns of genes associated with immune checkpoints. Statistical analyses were performed using R software version 4.0.3. Unless otherwise stated, we employed the Wilcoxon rank-sum test to detect differences between two sets of data, considering P values <0.05 as statistically significant. To further evaluate the relationship between gene expression and the CRC immune microenvironment, we utilized EPIC to analyze the infiltration proportions of 8 types of immune cells [B cells, cancer-associated fibroblasts, CD4⁺ T cells, CD8⁺ T cells, endothelial cells, macrophages, and natural killer (NK) cells] based on the expression data (29). CIBERSORT (https://cibersortx.stanford.edu; accessed on August 10, 2024) and EPIC (http://epic.gfellerlab.org; accessed on August 12, 2024) were used via the immunedeconv R package to estimate immune cell infiltration from bulk transcriptome data.

Tumor mutational burden (TMB) and MSI score correlation analysis

TMB refers to the total count of somatic mutations for each megabase (Mb) of the genome assessed, usually spanning a scope of 38 Mb, in the context of targeted sequencing (30). TMB and programmed death-ligand 1 (PD-L1) serve as significant biomarkers for predicting responses to programmed death-1 (PD-1) antibody therapy in immunotherapy. MSI refers to changes in the length of microsatellite sequences due to insertions or deletions of repeats during DNA replication, playing a crucial role in CRC (31,32). The TMB data were derived from a study published in 2018 by Thorsson and colleagues (33). The MSI data was sourced from a publication by Bonneville and others in 2017 (34). MSI data can be obtained using the R package “tcgabiolinks”. To investigate the association between gene expression, TMB, and MSI in cases where the quantitative variables are not normally distributed, we employ Spearman’s rank correlation coefficient. We acknowledge a P value below 0.05 as indicative of statistical significance. This approach facilitates an in-depth understanding of the interplay between gene expression levels and factors such as TMB and MSI.

Reverse transcription quantitative polymerase chain reaction (RT-qPCR)

Six pairs of tumor and corresponding normal tissue samples were collected from CRC patients. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The First People’s Hospital of Yancheng (No. 2025-K-192) and informed consent was obtained from all individual participants. The samples were immediately frozen at −80 ℃ after surgery to preserve RNA integrity. Total RNA was extracted from CRC and normal tissues using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA), following the manufacturer’s instructions. The RNA concentration and purity were measured using NanoDrop (Thermo Fisher Scientific), ensuring that the A260/A280 ratio was between 1.8 and 2.0 to confirm the quality of RNA for downstream experiments. Complementary DNA (cDNA) was synthesized from the extracted RNA using a reverse transcription kit (Takara, Dalian, China). The reverse transcription reaction was performed at 37 ℃ for 15 minutes, followed by inactivation at 85 ℃ for 5 seconds. qPCR was conducted using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) to detect the expression of target genes S100P, RIPK2, FCGRT, and VSIG2. The expression levels were normalized to 18S ribosomal RNA (rRNA) as an internal control. The qPCR conditions were as follows: initial denaturation at 95 ℃ for 10 minutes, followed by 40 cycles of denaturation at 95 ℃ for 15 seconds, and annealing at 60 ℃ for 1 minute. All reactions were performed on the ABI Prism 7500 system (Applied Biosystems). Relative gene expression levels were calculated using the 2^−ΔΔCt method.

Immunohistochemistry (IHC)

Ten pairs of CRC tissue and corresponding normal tissue slices were selected. IHC was performed on tissue sections to detect the expression of target proteins. The sections were fixed in 10% formalin, followed by dehydration and paraffin embedding. The slides were stained with specific antibodies against S100P, RIPK2, FCGRT, and VSIG2 (dilution 1:250, antibodies sourced from BIOSS, Beijing, China). Visualization was achieved using the 3,3'-diaminobenzidine (DAB) staining system. After counterstaining with hematoxylin, the slides were observed under a microscope. The IHC results were evaluated using a scoring system to assess protein expression levels, comparing the differences between tumor and normal tissues.

Plasmids and cell transfection

The human CRC cell line HCT116 was used for functional validation experiments. The plasmid pCMV-VSIG2(human)-3xFLAGNeo for VSIG2 overexpression and pLKO.1-U6-S100P(human)-shRNA1-Puro for S100P knockdown were purchased from Miaoling Plasmid Company (Wuhan, China). The corresponding empty vector (pCMV-3xFLAGNeo) and non-targeting control plasmid (pLKO.1-shNC) were used as negative controls (NCs). HCT116 cells were transfected using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. Stable transfectants were selected with G418 (for pCMV vector) or puromycin (for pLKO.1 vector).

MTT assay

Cell proliferation was assessed using the MTT Cell Proliferation and Cytotoxicity Assay Kit (KeyGEN Biotech, Nanjing, China; Cat. No. KGA9301-500). Briefly, transfected cells were seeded into 96-well plates at a density of 3×10³ cells/well and cultured for 24, 48, and 72 h. At each time point, 20 µL of MTT solution was added to each well and incubated for 4 h at 37 ℃. The medium was then removed, and 150 µL of dimethyl sulfoxide (DMSO) was added to dissolve the formazan crystals. Absorbance was measured at 570 nm using a microplate reader (BioTek Instruments, Winooski, VT, USA).

Statistical analysis

All statistical analyses were performed using R software (version 4.0.3; R Foundation for Statistical Computing, Vienna, Austria). Continuous variables were compared using the Wilcoxon rank-sum test, and correlations were assessed with Spearman’s rank correlation coefficient. Differential gene expression was analyzed using the limma package, with adjusted P<0.05 and |log₂FC| >2 considered significant. MR analyses were conducted with the “TwoSampleMR” package. For all analyses, P values <0.05 were considered statistically significant unless otherwise stated.

Results

Screening for genes associated with the onset of CRC

Differential gene expression analysis was performed on genes from the GEO dataset, applying thresholds of log₂FC >2 and P<0.05 (Figure S1). This analysis identified 206 upregulated and 377 downregulated genes (Figure 2A, table available at https://cdn.amegroups.cn/static/public/tcr-2025-1072-1.xlsx). Subsequently, these genes were subjected to MR and SMR analyses, using CRC as the outcome. The analyses identified 295 genes causally linked to CRC, with MR results showing 131 genes with odds ratio (OR) >1 and 164 genes with OR <1 (Figure 2B, table available at https://cdn.amegroups.cn/static/public/tcr-2025-1072-2.xlsx). SMR results indicated 472 risk factors and 496 protective factors for CRC (Figure 2C, table available at https://cdn.amegroups.cn/static/public/tcr-2025-1072-3.xlsx). Intersection analysis of MR and SMR identified genes overlapping between the upregulated risk genes and downregulated protective genes. Ultimately, two upregulated and two downregulated intersecting genes were identified (Figure 2D). The results of the MR analysis between these genes and CRC are displayed using forest plots (Figure 2E). The SMR results for these genes are presented in Table 1, and the effects of individual genes and single variant cis-eQTLs on CRC are shown using scatter plots (Figure S2).

Figure 2.

Screening of genes associated with CRC onset. (A) Heatmap of differential gene expression analysis. This heatmap shows differential gene expression between normal and tumor groups from GEO data (GSE15781, GSE79793, and GSE50117). Genes are on the rows, samples on the columns, with expression levels color-coded from blue (low) to red (high). Clustering dendrograms indicate patterns of similarity. (B) Volcano plot of MR results. This plot shows MR analysis results. (C) Manhattan plot of SMR analysis. This plot displays −log₁₀(P values) from SMR analysis across chromosomes (Chr 1–22). (D) Venn diagram of intersecting genes. The Venn diagram shows that the upregulated DEGs and risk genes from MR and SMR analyses intersect at RIPK2 and S100P, while downregulated DEGs and protective genes intersect at FCGRT and VSIG2. (E) Forest plot of MR analysis. The forest plot shows MR results using cis-eQTLs as exposures and CRC as the outcome, displaying OR values and 95% CIs for the weighted median and IVW methods. CI, confidence interval; CRC, colorectal cancer; DEG, differentially expressed gene; eQTL, expression Quantitative Trait Loci; GEO, Gene Expression Omnibus; IVW, inverse variance-weighted; MR, Mendelian randomization; OR, odds ratio; SMR, summary-data-based Mendelian randomization.

Table 1. SMR analysis results.

Gene	Top SNP	b_SMR	se_SMR	p_HEIDI	P value
FCGRT	rs59774409	−0.3098	0.091119	0.470449	0.000674
RIPK2	rs428484	0.176431	0.068775	0.257167	0.010307
S100P	rs3822263	0.044107	0.019641	0.955024	0.024728
VSIG2	rs12293621	−0.48699	0.21202	0.952495	0.021624

The table includes gene names, top SNP markers, SMR effect sizes (b_SMR), SEs (se_SMR), HEIDI test P values (p_HEIDI), and SMR analysis P values, exemplified by FCGRT with significant SMR effect size −0.3098, SE 0.091119, and P value 0.000674. HEIDI, Heterogeneity in Dependent Instruments; SE, standard error; SMR, summary-data-based Mendelian randomization; SNP, single-nucleotide polymorphism.

Single-cell analysis

Subsequently, we further investigated the single-cell expression profiles of these key genes within the CRC microenvironment. Single-cell analysis revealed the expression differences of CRC-related genes FCGRT, RIPK2, S100P, and VSIG2 across various primary cell types using t-SNE visualization techniques. Cells were categorized based on surface markers, and the distribution of major cell types was displayed on t-SNE plots. Expression levels of each gene were visualized and analyzed across different cell types: FCGRT exhibited the highest expression in monocytes/macrophages, followed by epithelial and tumor cells (Figure 3A). RIPK2 showed higher expression in monocytes/macrophages and lower in other cell types (Figure 3B). S100P was predominantly expressed in tumor cells, followed by epithelial cells (Figure 3C). VSIG2 was significantly expressed in epithelial and tumor cells, with lower expression in immune cells (Figure 3D).

Figure 3.

Single-cell analysis reveals expression patterns of FCGRT, RIPK2, S100P, and VSIG2. (A) t-SNE plot of major cell types with FCGRT expression and average expression levels in a bar plot. (B) RIPK2 expression t-SNE plot and bar plot. (C) S100P expression t-SNE plot and bar plot. (D) VSIG2 expression t-SNE plot and bar plot. DC, dendritic cell; t-SNE, t-distributed stochastic neighbor embedding.

The connection between genes and tumor immunity

The regulation of tumor immune microenvironment by genes represents a critical determinant in the pathogenesis, progression, and therapeutic response of CRC (35). Using the CIBERSORT algorithm, we thoroughly analyzed the associations between four genes—FCGRT, RIPK2, S100P, and VSIG2—and 22 immune cell subtypes (Figure 4A). FCGRT exhibits significant differences in immune infiltration between colon and rectal cancers, showing negative correlations with various T cells, resting mast cells, and activated NK cells in colon cancer, and a positive correlation with regulatory T (Treg) cells in rectal cancer. RIPK2 is negatively correlated with Treg, naïve CD4⁺ T cells, and memory B cells in CRC, and positively correlated with neutrophils, activated dendritic cells, and M1 and M2 macrophages. S100P is positively correlated with resting mast cells and negatively with M0 macrophages. VSIG2 shows negative correlations with activated CD4⁺ T cells, M1, and M0 macrophages, and positive correlations with plasma B cells and Treg cells; in colon cancer, it is also positively correlated with resting CD4⁺ T cells.

Figure 4.

Association between genes and tumor immunity. (A) Heatmap of gene expression and immune infiltration scores. The heatmap shows Spearman correlation between gene expression and immune infiltration scores in CRC. (B) Mediation MR analysis of immune phenotypes on CRC risk. The forest plot shows OR values, CIs, and P values for each step of the two-step MR analysis. CD19 on IgD⁺CD24⁻: this denotes B cells that express CD19 and IgD on their surface but do not express CD24. CD24 is typically associated with B cell activation and maturation, so CD24⁻ may indicate a more primitive or less fully activated state of B cells. CD19 on IgD⁺CD38⁻: this denotes B cells that express CD19 and IgD on their surface but do not express CD38. CD19 on IgD⁺CD38-naïve: these cells express CD19, IgD, and low levels of CD38, typically remaining in an inactive state and not participating in immune responses. CD19 on IgD⁺: this denotes B cells that express both CD19 and IgD, which are usually naïve, non-activated B cells. CD33 on CD33br HLA DR⁺ CD14⁻: this denotes a cell population that expresses high levels of CD33 and HLA-DR but does not express CD14. CI, confidence interval; COAD, colon adenocarcinoma; CRC, colorectal cancer; MR, Mendelian randomization; OR, odds ratio; READ, rectum adenocarcinoma.

To further explore how gene expression relates to tumor immunity in CRC, we conducted a two-step MR analysis using 731 immune cell subtypes as exposures with CRC as the outcome. This analysis did not identify any immune cells that could mediate the effects of FCGRT, S100P, or VSIG2 on CRC. However, several immune cell types and phenotypes were identified as mediators for RIPK2 (Figure 4B). We then calculated the mediation effects of MR (Table 2). RIPK2 expression, a risk factor for CRC, is mediated through immune cell phenotypes associated with B cells: CD19 on IgD⁺CD24⁻, CD19 on IgD⁺CD38⁻, CD19 on IgD⁺CD38-naïve, CD19 on IgD⁺, and CD33 on CD33br HLA DR⁺ CD14⁻.

Table 2. Mediation MR results.

Exposure	Mediator	Outcome	Mediated effect	Proportion
RIPK2	CD19 on IgD+CD24−	CRC	0.00419	3.75%
	CD19 on IgD+CD38−	CRC	0.00569	5.09%
	CD19 on IgD+CD38-naïve	CRC	0.00575	5.14%
	CD19 on IgD+	CRC	0.00333	2.98%
	CD33 on CD33br HLA DR+ CD14−	CRC	0.00623	5.57%

The table shows mediation effects of various immune cell phenotypes for RIPK2 on CRC, with CD19 on IgD⁺CD24⁻ cells showing a mediation effect of 0.00419, accounting for 3.75% of the total effect. CRC, colorectal cancer; MR, Mendelian randomization.

Correlation between CRC-related genes and TMB/MSI scores

We obtained gene expression data for CRC-related genes from colon/rectal cancer and matched normal tissues from the TCGA database. FCGRT is underexpressed in colon cancer, and VSIG2 is underexpressed in both colon and rectal cancers, while RIPK2 and S100P are overexpressed in these cancers. Gene expression trends for FCGRT (except in rectal cancer), RIPK2, S100P, and VSIG2 aligned with results from differential gene analysis and MR analysis (Figure S3). We also analyzed the correlation between gene expression (log-transformed) and tumor MSI as well as TMB scores. FCGRT showed a significant negative correlation with TMB and MSI scores, while S100P showed a positive correlation. RIPK2 and VSIG2 did not show strong associations with TMB and MSI scores (Table 3, Figure S4). These findings suggest that different gene expression patterns may influence tumor MSI and TMB differently, reflecting heterogeneity in tumor biology and treatment response.

Table 3. Correlation between gene expression and TMB/MSI scores.

Gene	Correlation coefficient	P value
Correlation between gene expression and TMB score
S100P	1.39e−01	2.55e−03
RIPK2	3.03e−03	9.48e−01
VSIG2	−9.12e−02	4.77e−02
FCGRT	−3.07e−01	1.22e−11
Correlation between gene expression and MSI score
S100P	1.43e−01	5.94e−04
RIPK2	9.41e−02	2.38e−02
VSIG2	−7.81e−02	6.10e−02
FCGRT	−3.47e−01	8.30e−18

The table presents correlations between gene expression levels and TMB/MSI scores, such as FCGRT’s negative correlation with TMB (−0.307) and a highly significant P value (1.22e−11). MSI, microsatellite instability; TMB, tumor mutational burden.

mRNA expression and IHC analysis of RIPK2, S100P, FCGRT, and VSIG2 in normal and tumor tissues

To further validate the expression patterns of these genes, we performed experimental analyses using clinical specimens. In tumor tissues, the mRNA expression levels of RIPK2 and S100P were significantly higher compared to normal tissues (Figure 5A). Conversely, the mRNA levels of FCGRT and VSIG2 were significantly reduced in tumor tissues. Figure 5B illustrates the results of hematoxylin and eosin (HE) staining and IHC analysis, showing a marked difference in the protein expression of RIPK2, S100P, FCGRT, and VSIG2 between tumor and normal tissues. Statistical analysis demonstrated that RIPK2 and S100P were significantly overexpressed in tumor tissues, whereas FCGRT and VSIG2 exhibited significantly lower expression in tumors.

Figure 5.

Expression analysis of RIPK2, S100P, FCGRT, and VSIG2 in normal and tumor tissues. (A) Relative mRNA expression levels of RIPK2, S100P, FCGRT, and VSIG2 in normal and tumor tissues. (B) HE and IHC staining analysis showing the differential expression of RIPK2, S100P, FCGRT, and VSIG2 in normal and tumor tissues (original magnification ×40). Statistical analysis indicates that the IHC scores of RIPK2 and S100P are significantly elevated in tumor tissues, while the scores of FCGRT and VSIG2 are significantly reduced in tumors. **, P<0.01; ***, P<0.001; ****, P<0.0001. CRC, colorectal cancer; HE, hematoxylin and eosin; IHC, immunohistochemistry; mRNA, messenger RNA; NT, normal tissue.

Functional validation of S100P and VSIG2

To further validate the functional roles of the identified genes, we conducted in vitro experiments. Western blotting confirmed efficient knockdown of S100P and successful overexpression of VSIG2 in CRC cells (Figure 6A). Subsequent MTT assays demonstrated that S100P knockdown significantly inhibited CRC cell proliferation at 48 and 72 h compared with the short hairpin RNA (shRNA) NC (shNC), while VSIG2 overexpression suppressed cell proliferation at all tested time points (24, 48, and 72 h) compared with vector control (Figure 6B). These experimental findings corroborate our bioinformatics analyses, providing direct evidence that S100P exerts pro-carcinogenic effects, whereas VSIG2 acts as a tumor suppressor in CRC.

Figure 6.

Functional validation of S100P and VSIG2 in CRC cells. (A) Western blot confirmed S100P knockdown (10 kDa) and VSIG2 overexpression (34 kDa). β-actin was used as a loading control. (B) MTT assays showed reduced proliferation after S100P knockdown and VSIG2 overexpression. Data are mean ± SD (n=3). *, P<0.05; **, P<0.01; ***, P<0.001; ns, not significant (P>0.05). CRC, colorectal cancer; NC, negative control; OD, optical density; OE, overexpression; SD, standard deviation; sh, short hairpin RNA.

Discussion

This study integrated cis-eQTL data with MR results related to CRC and differential gene expression analyses from the GEO database to explore genes associated with CRC onset from genetic and transcriptomic levels. Four genes linked to CRC development were identified. Additionally, the study examined the relationship between gene expression and the immune microenvironment through immune infiltration analysis and further investigated how these genes affect CRC onset through subtypes of immune cells using intermediary MR. Finally, the expression of these genes and their correlation with TMB/MSI scores were validated using the TCGA database.

Previous studies on CRC have demonstrated that S100P enhances the invasive and migratory capabilities of CRC cells by mediating the demethylation and transcriptional activation of the SLC2A5 promoter (36,37). Additionally, targeting RIPK2 with small molecule inhibitors in mouse models has been shown to ameliorate colitis and reduce the incidence of CRC (38). This is consistent with our findings that S100P and RIPK2 are risk factors for CRC. Previous studies primarily relied on functional assays to demonstrate the oncogenic roles of these genes, without deeply exploring the mechanisms or considering the potential links between gene expression and tumor immunity. Our study supports previous findings from a genetic variation perspective and further elucidates the relationship between these genes and tumor immunity in CRC. The other identified genes also present research potential, providing direction for future studies.

Furthermore, our single-cell transcriptomic analysis revealed distinct expression patterns of these genes within the CRC tumor microenvironment. Specifically, S100P and VSIG2 were predominantly expressed in tumor and epithelial cells, suggesting their direct involvement in tumor-intrinsic processes. In contrast, FCGRT and RIPK2 showed higher expression in monocytes/macrophages, indicating a potential role in shaping the immune landscape. These findings highlight that the functional impact of the identified genes may be mediated through both tumor cell-intrinsic mechanisms and immune cell-associated pathways, thereby underscoring the complexity of CRC pathogenesis.

MSI molecular characteristics are considered significant genetic markers with considerable clinical value in the screening, typing, diagnosis, treatment, and prognostic assessment of CRC (39). Patients with high levels of MSI respond better to immunotherapy because the cancer cells in MSI patients harbor more mutations, making them easier for the immune system to recognize (7,40). Additionally, a higher TMB typically indicates the generation of more neoantigens, which can be recognized by the immune system, thereby triggering an immune response against the tumor and suggesting a better efficacy of ICIs (41). Analyzing the relationship between gene expression and MSI/TMB can provide insights into cancer molecular mechanisms and personalized treatment strategies. The downregulation of FCGRT is associated with increased TMB and MSI, suggesting that FCGRT might help tumors evade immune surveillance, contributing to tumor aggressiveness and progression. Thus, FCGRT could serve as a potential prognostic marker or therapeutic target. Given the positive correlation of S100P with TMB/MSI, targeting S100P may modulate the tumor immune microenvironment and enhance immunotherapy efficacy. Developing inhibitors or neutralizing antibodies against S100P could reduce tumor mutation rates, improving treatment sensitivity.

Through single-cell analysis, we discovered that S100P and VSIG2 are primarily expressed in tumor cells and epithelial cells, while FCGRT and RIPK2 are highly enriched in monocytes/macrophages among the tumor cell populations. In addition, the tumor microenvironment, which is characteristically immunosuppressive, plays a pivotal role in the effectiveness of tumor immunotherapy (42). We delved deeper into the dynamics between gene expression patterns and immune cell presence within the tumor’s immune environment via in-depth analysis of immune infiltration. In the analysis of gene expression and immune infiltration, we found that the expression of individual genes showed a broadly consistent trend in both colorectal tumor and rectal tumor. The association between gene expression patterns and the penetration of immune cells might indicate their dual function in regulating both the immune reaction and the advancement of the tumor. Further mediating MR analyses continued to dissect the phenotypes of immune cells, delving deeper into how immune phenotypes mediate the mechanisms by which gene expression contributes to the pathogenesis of CRC. Notably, several immune cells expressing IgD and CD19 can mediate the risk effect of RIPK2 on CRC. These cells are typically mature B cells. MR results show that this type of cell acts as a protective factor against CRC, while RIPK2 expression decreases their abundance. The integrated analysis of these results provides a comprehensive view of how these genes function within different immune cell subtypes, influencing the tumor microenvironment and potentially participating in tumor progression and immune evasion. Interestingly, the expression trend of FCGRT in rectal cancer did not fully align with the differential expression and MR results. This discrepancy may reflect the heterogeneity between colon and rectal cancers, including differences in anatomical location, immune microenvironment, and sample size in TCGA rectal cohorts, which may contribute to variability in gene expression patterns.

Despite this research offering fresh perspectives on the genetic underpinnings of CRC pathogenesis, it does possess certain constraints. Firstly, the diversity of the sample population is limited, and future studies should include a broader range of ethnicities and geographical backgrounds to enhance the generalizability of the results. Secondly, despite the integration of data from multiple databases, differences in data collection standards and methodologies between databases may have introduced a degree of bias. Furthermore, although our analysis suggests that these genes are associated with the onset of CRC, the precise biological pathways through which these genes influence the development of CRC remain unclear. Future research is required to further explore these potential biological mechanisms.

Overall, the findings of this study not only reveal new genes associated with the onset of CRC and potential therapeutic targets but also deepen our understanding of the mechanisms of CRC pathogenesis, especially in terms of immune regulation. These accomplishments furnish a vital scientific foundation for devising innovative immunotherapeutic approaches to combat CRC. They underscore the significant promise of integrative analysis techniques, utilizing multiple databases to uncover the intricate mechanisms underlying the disease. As future research further elucidates the functions of these genes and their roles in the immune microenvironment, we hope to improve the efficacy of CRC treatments, reduce the risks of drug resistance and recurrence. Moreover, this study highlights the importance of integrating genetic and immunological data in cancer research, offering a powerful tool for uncovering the complex biology of cancer. As personalized medicine and precision treatment strategies continue to evolve, our findings may offer more personalized and effective treatment options for CRC patients.

Conclusions

This study identifies RIPK2 and S100P as genetic risk factors for CRC, while FCGRT and VSIG2 act as protective factors. Through integrating gene expression, MR analysis, and immune infiltration assessments, we have elucidated the significant roles of these genes in CRC pathogenesis and their interactions with the immune system. Notably, the expression of these genes correlates with TMB/MSI scores, suggesting their relevance in predicting immunotherapy outcomes. The findings offer critical perspectives for the creation of focused treatments and prognostic instruments for CRC, underscoring the importance of genotypic and immunologic investigations in enhancing the prognosis for patients.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The First People’s Hospital of Yancheng (No. 2025-K-192) and informed consent was obtained from all individual participants.

Data Sharing Statement

Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1072/dss

tcr-14-11-7669-dss.pdf ^{(69.8KB, pdf)}

DOI: 10.21037/tcr-2025-1072