# m6AHD: a new framework for identifying abnormal N6-methyladenosine (m6A) in heart diseases based on sequencing features

## TL;DR

This paper introduces m6AHD, a new framework for predicting abnormal m6A RNA modifications in heart diseases, offering insights into potential drug targets.

## Contribution

The first predictive framework for multi-condition cardiac m6A dysregulation using epitranscriptome data and validated in zebrafish models.

## Key findings

- m6AHD achieved AUROC scores between 0.728 and 0.880 in predicting m6A dysregulation across cardiac pathologies.
- Aberrant m6A patterns are consistent across different heart diseases and linked to similar regulatory factors.
- Zebrafish experiments confirmed m6A dysregulation in drug-induced cardiotoxicity, showing methyltransferase complex downregulation.

## Abstract

Cardiovascular disease (CVD) is a major threat to health, with high incidence rates and a trend toward younger age groups. RNA modifications are an important component of epigenetics, widely present and indispensable in cells. Increasing evidence suggests that RNA modifications are key regulatory factors involved in cardiac physiological and pathological changes. Understanding the role of RNA modifications in heart-related diseases can help us to identify new drug targets.

To systematically investigate the role of m6A modification in different cardiac diseases, we integrated m6A epitranscriptome profiles from five cardiac pathological conditions (three drug-induced cardiac toxicity models—Evodiamine, Matrine, and TKI, hypertrophy, and heart calcification) and their control groups to construct the first predictive model for abnormal m6A modification in cardiac diseases. We constructed separate models for upregulated and downregulated modifications under different pathological conditions, performed feature selection and parameter optimization, and validated the performance of our models using an independent test set.

m6AHD demonstrated excellent performance on the independent test set, with AUROC scores ranging from 0.728 to 0.880 across various pathological conditions. Cross-validation across different conditions and model interpretability demonstrated that m6A modifications exhibit similar patterns under different pathological conditions and are potentially regulated by similar factors, providing new clues for identifying targets in cardiovascular diseases at the epitranscriptome level. Furthermore, we validated our findings using a zebrafish model of Evodiamine-induced cardiotoxicity. The experimental results revealed significant morphological defects and a broad downregulation of m6A methyltransferase complex components, confirming the involvement of aberrant m6A machinery in the pathology of cardiotoxicity.

m6AHD is the first dedicated framework for predicting multi-condition cardiac m6A dysregulation. Our findings underscore the critical role of m6A homeostasis in cardiomyocyte function and demonstrate that aberrant methylation patterns can serve as reliable indicators of cardiac pathology. This framework provides a robust computational tool for identifying potential therapeutic targets at the epitranscriptome level for cardiovascular diseases.