2297 - An Integrative Multi-Omics Analysis of Biological and Clinical Impacts of Non-Histone Acetylation Enzymes and Substrates in Colon Cancer

Purpose/Objective(s):

Non-histone acetylation regulates protein stability, transcription, and immune signaling, yet its substrate-specific effects and clinical relevance remain unclear. We aimed to integrate multi-omics and digital pathology analyses to characterize acetylation imbalance in colon adenocarcinoma (COAD) and define its biological and spatial immune implications.

Materials/Methods:

Expression and prognostic relevance of 33 acetylation-related enzymes were analyzed using multi-platform pan-cancer datasets. Substrate-dependent effects were modeled through predictive enzyme–substrate interaction (ASI) networks. Based on validated and predicted substrates, an acetylation imbalance score (AIScore) was constructed for COAD. Single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (ST) were performed to examine tumor microenvironment differences between AIScore groups. A deep-learning pathomics workflow was applied to whole-slide images, including ResNet-50 feature extraction, PCA-based dimensionality reduction, unsupervised clustering of tumor regions, and Grad-CAM visualization to link AIScore biology with spatial morphologic patterns.

Results:

Acetylation enzymes were differentially expressed between tumor and normal tissues and associated with survival. Predictive modeling identified 864 acetyltransferase–substrate and 532 deacetylase–substrate high-probability interactions. A 15-gene substrate-derived AIScore stratified COAD patients into prognostically distinct groups, with significantly improved survival in the AIScore-low group. scRNA-seq and ST analyses showed that AIScore-high malignant cells preferentially interacted with SPP1? macrophages, forming immunosuppressive niches linked to reduced predicted immunotherapy response. Consistently, pathomics clustering identified tumor subregions enriched in AIScore-high signals, and Grad-CAM highlighted morphologic hotspots at immune–tumor interfaces, supporting a spatial tumor–macrophage interaction pattern.

Conclusion:

Non-histone acetylation imbalance in COAD is substrate-dependent and associated with poor prognosis and spatial immune suppression. Integrating multi-omics with AI-driven pathomics provides morphology-linked mechanistic insight into acetylation-driven tumor–immune interactions and may inform biomarker development and therapeutic stratification.