DNA methylation regulates TREM1 expression to modulate immune responses and drive progression in colorectal neuroendocrine neoplasm as a potential therapeutic target

Abstract

Background: Colorectal neuroendocrine neoplasms (CrNENs) are rare malignancies with limited therapeutic options and poorly understood molecular mechanisms. The roles of genetic, epigenetic, and immune factors in CrNEN progression remain largely unknown.

Methods: We employed an integrative multi-omics approach combining two-sample Mendelian randomization, Bayesian colocalization, methylation quantitative trait loci (mQTLs), cis-expression QTLs (cis-eQTLs), protein QTLs (pQTLs), summary data-based Mendelian randomization, mediation analyses, immunohistochemistry validation, and pan-cancer validation using TCGA and GTEx data, including DNA methylation profiling using the SMART, UALCAN, UCSC Xena and MethSurv platforms to provide integrative insights into potential epigenetic regulation in oncogenesis. Molecular docking was performed to identify candidate therapeutic compounds that targeted the genes.

Results: Our analyses identified TREM1 as a robust therapeutic candidate, with elevated TREM1 expression promoting the progression of CrNEN. Epigenetic analysis revealed that hypomethylation at the cg04451353 locus was associated with increased TREM1 expression, mediating approximately 74% of the CrNEN risk attributable to this epigenetic mechanism. Immune mediation analysis suggested that increased TREM1 expression may influence the infiltration of specific immune subsets (CD14 + CD16- monocytes, CD25 + + CD8br Tregs, CD3 on Tregs, CD3 on CD39 + secreting Tregs, CD3 on CD8br Tregs, and CD28 on CD39 + CD8br Tregs), with each subset contributing approximately 1-4% to the total 16.65% increase in CrNEN risk. Pan-cancer validation underscored the oncogenic potential and prognostic significance of TREM1 across various malignancies, with particular relevance in the colorectal cancer. Molecular docking analysis suggested favorable binding affinities between TREM1 and bioactive natural compounds, such as artemisinin and quercetin, which may support their potential as therapeutic candidates.

Conclusions: Multi-omics analysis suggests that TREM1 may play a role in CrNEN pathogenesis, with potential implications as a therapeutic target. Further validation and research are needed to confirm its clinical relevance and therapeutic potential.

Keywords: TREM1; Colorectal neuroendocrine neoplasms; Druggable genes; Mendelian randomization.

Fig. 1

Overview of the multi-omics analysis pipeline. This figure presents the workflow used to investigate the role of TREM1 in CrNEN pathogenesis. The pipeline integrates MR, colocalization, SMR, HEIDI testing, and immune cell mediation analysis, following the central dogma of DNA variation to phenotypic traits. The steps include identifying candidate druggable genes using cis-eQTL, pQTL, and mQTL data, followed by enrichment and PPI network analyses to explore associated biological functions and pathways. Colocalization analysis identified shared causal variants between eQTL and CrNEN signals, and SMR analysis was performed for gene prioritization, with protein-level validation using pQTL data. Immune cell traits were analyzed through MR to evaluate their effect on CrNEN risk, and molecular docking was employed to screen therapeutic compounds targeting TREM1. Tissue-level validation was conducted through IHC and pan-cancer expression analysis across 33 tumor types. The findings highlight TREM1 as a promising biomarker and therapeutic target in CrNEN

Fig. 2

Identification and characterization of candidate druggable genes associated with CrNEN. (A) Forest plot of MR results showing the genes significantly associated with CrNEN risk (IVW, FDR < 0.05). (B) Circular plot of the genomic distribution of significant genes. (C) GO (BP, CC, and MF) and KEGG pathway enrichment of the identified genes. (D) PPI network using GeneMANIA, showing co-expression and functional links. (E) Heatmap of Bayesian colocalization (PP.H4) between the eQTL and CrNEN GWAS loci. (F) SMR locus and effect plots for TREM1 and RAMP3: top cis-eQTLs (red triangles) and supporting variants (blue circles). (G) Forest plot integrating multi-omics evidence (eQTL, pQTL, mQTL, SMR) supporting TREM1 association with CrNEN risk

Fig. 3

Mediation analysis of DNA methylation and immune traits in the relationship between TREM1 and CrNEN risk. (A) Circos plot of the immune cell traits causally associated with CrNEN. (B) Circos plot showing MR-based associations between TREM1 expression and specific immune cell subsets. The outer heatmap displays p-values from various statistical analyses, with red indicating stronger (lower p-values) and blue indicating weaker (higher p-values) associations. The statistical methods of MR are represented by the outer rings. (C) Methylation mediation analysis showing the effect of cg04451353 on CrNEN risk via TREM1 expression. (D–J) Immune-cell–mediated MR analysis quantifying mediation effects (1.22% – 3.51%) via specific immune subsets: (D) CD14 + CD16- monocyte, (E) CD25 + + CD8br Treg, (F) CD3 on T cell, (G) CD3 on CD39 + secreting Treg, (H) CD3 + CD8br Treg, (I) CD28 on CD39 + CD8br Treg, and (J) CX3CR1on CD14 + CD16 + monocytes. Beta_all: total effect; Beta_dir: direct effect

Fig. 4

Validation of TREM1 expression in CrNEN tissues by IHC. (A–C) Representative IHC images showing strong cytoplasmic and membranous TREM1 staining in tumor tissues (top) and minimal expression in paracancerous tissues (bottom). Scale bars: 100 μm. (D) Quantitative analysis of AOD confirmed significantly higher TREM1 expression in tumors than in adjacent tissues (0.294 vs. 0.164, paired t-test, *P < 0.01)

Fig. 5

Expression landscape and prognostic significance of TREM1 in human cancers. (A–B) Differential expression analysis of TREM1 in tumor versus normal tissues across pan-cancer datasets. (A) Combined TCGA and GTEx cohorts (XENA platform); (B) TCGA paired tumor–normal matched samples. Red and blue represent the tumor and normal tissues, respectively. (C) Radar plots showing the expression trends of TREM1 in tumors compared to the corresponding normal tissues. (D) Pan-cancer summary of TREM1 expression associations with OS, DSS, and PFI. (E) Kaplan–Meier survival analyses illustrating the prognostic impact of TREM1 expression in representative cancers. Patients were stratified into high and low TREM1 expression groups based on the median cutoff, and survival curves were plotted for OS, DSS, and PFI endpoints. P-values were calculated using the log-rank test, and HRs are displayed for each model. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant

Fig. 6

Integrated analysis of the TREM1 methylation landscape and its clinical relevance in human cancers. (A) Chromosomal locations of CpG probes mapped to the TREM1 locus. (B) Genomic architecture of TREM1, including transcript isoforms and seven annotated CpG sites within a 1000-bp region. (C) Average methylation levels aggregated across all TREM1-associated CpG probes in the pan-cancer cohorts. (D) Methylation level (β-value) of the specific CpG probe cg04451353 across 33 tumor types and matched normal tissues. (E) Kaplan–Meier survival curves depicting the association between the methylation status of cg04451353 and overall survival in COAD, READ, KICH, KIRC, LUAD, and LUSC. (F) Heatmaps showing the methylation patterns of TREM1-related CpG sites across individual tumor samples in the above six tumor types, stratified by clinical and molecular features. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant

Fig. 7

Methylation levels of TREM1 in COAD and READ tissues, and association with tumor characteristics. (A) Box plots showing the distribution of methylation levels of TREM1 in various cancer stages, tumor types, and other tumor characteristics (e.g., TP53 mutation status, nodal metastasis). Significant differences are highlighted with corresponding statistical tests. The scatter plot in the lower right of each graph shows the correlation between methylation levels and clinical features. (B) Bar plots and violin plots representing the distribution of TREM1 methylation levels in different molecular subtypes of colorectal cancer (MSI-H, MSI-L, and MSS)

Fig. 8

Correlation analysis between TREM1 expression and immune-related features across different cancer types. (A) Heatmap illustrating the Spearman correlation between TREM1 expression and the infiltration levels of 24 immune cell subsets estimated by the ssGSEA algorithm across 33 TCGA cancer types. (B) Correlation matrix based on CIBERSORT deconvolution results, depicting the associations between TREM1 and the relative abundance of 22 immune cell types. (C) Association between TREM1 expression and immune microenvironment scores calculated by the ESTIMATE algorithm, including StromalScore, ImmuneScore, and ESTIMATE Score. The color scales represent the correlation coefficients (red = positive; blue = negative), and the significance levels are indicated by stars. P-values were computed using Spearman’s correlation and visualized using corresponding grayscale triangles. *P < 0.05

Fig. 9

Molecular docking of TCM monomers with TREM1. (A–F) Binding interaction between TREM1 and six TCM compounds: Artemisinin (A), Ginsenoside Rg3 (B), Geniposide (C), Matrine (D), Quercetin (E), and Icariin (F). Left: 3D protein-ligand binding mode; middle: surface interaction highlighting hydrophobic (green) and hydrophilic (purple) regions; right: 2D interaction map showing van der Waals (light green), water hydrogen bonds (blue), conventional hydrogen bonds (green), π-σ (purple), π-sulfur (yellow), and π-alkyl (pink) interactions