Ultrasound-Responsive Serotonylation and Hallmark Pathway Gene Signatures Reveal Tumor Microenvironment Vulnerabilities and Prognostic Subtypes in Colorectal Cancer

Abstract

Background:

Focused ultrasound, low-intensity focused ultrasound, and microbubble-enhanced sonoporation are examples of ultrasound-based cancer therapies that have shown promise as biophysical modalities for enhancing drug penetration, immunogenic cell death, and targeted delivery of radiopharmaceuticals in solid tumors. The molecular factors controlling ultrasonic therapy receptivity, however, are still not well understood. Because of the significant variability of the tumor microenvironment (TME), colorectal cancer (CRC) necessitates biomarker-guided techniques to enhance ultrasound-based therapy regimens.

Methods:

To investigate serotonylation-related and hallmark-pathway-related genes that might influence ultrasound-responsive cellular pathways, such as extracellular matrix (ECM) remodeling, mechanotransduction, and immune activation, the authors combined bulk RNA-sequencing (RNS-seq) (TCGA-COAD), single-cell RNA-seq (GSE132465), and spatial transcriptomics (GSE280313) datasets. Nine prognostic genes were found using survival analysis and differential expression screening. To create a prognostic classifier with translational relevance for ultrasonic therapies, the authors used non-negative matrix factorization clustering, single-cell functional scoring, spatial deconvolution, and 101 machine-learning models.

Results:

Of the 2475 serotonylation–hallmark genes found, 784 exhibited differential expression in tumor and normal tissues. CRC was divided into six molecular subgroups with different TME symptoms and survival patterns by nine important prognostic genes (PCOLCE2, TIMP1, FJX1, FABP4, CALB2, NAT1, CDKN2A, FSTL3, and INHBB). Elevated stromal activation, epithelial–mesenchymal transition signals, macrophage infiltration, and ECM stiffness were seen in high-risk clusters; these variables are known to affect cavitation thresholds, ultrasonic energy absorption, and treatment response. Strong prognostic accuracy was demonstrated by the final RSF–SuperPC model (concordance index 0.72–0.85 across validation cohorts). Strong enrichment in mechanotransduction, oxygen metabolism, and immune chemotaxis pathways—pathways previously demonstrated to regulate ultrasound-triggered drug delivery and immune activation—was revealed by functional studies.

Conclusions:

This multiomics integration reveals a serotonylation–hallmark gene signature that represents microenvironmental characteristics, such as matrix stiffness, stromal density, and immune infiltration, that are pertinent to ultrasound-based CRC therapy. These biomarkers could direct patient classification for radiopharmaceutical, immunotherapy, and ultrasound-enhanced medication delivery. This work supports future clinical trial stratification frameworks and offers a mechanistic basis for precision ultrasound oncology.