Single-cell sequencing and machine learning identify endoplasmic reticulum stress as a mechanism and therapeutic target in heart failure

Background and purpose: Understanding the intricate mechanisms underlying heart failure (HF) is paramount for the development of effective diagnostic toolkits and therapeutic strategies. In this study, single-cell sequencing and machine learning (ML) were used to reveal the mechanism and therapeutic target of HF. Methods: In this study, we employed an integrative approach combining single-cell ribonucleic acid (RNA) sequencing, bulk RNA sequencing, and ML to elucidate the potential role of endoplasmic reticulum (ER) stress in HF. Results: Our findings confirm that ER stress is involved across multiple cellular subtypes in HF. Compared with non-diseased donors, HF patients exhibited reduced expression of ER stress-related genes in cardiomyocytes, pericytes, fibroblasts, and myeloid cells. Specifically, among HF patients who underwent left ventricular assist device (LVAD) implantation, those who demonstrated recovery of cardiac function showed lower expression of ER stress-related genes across all cell subtypes compared with patients who did not recover. Utilizing ML algorithms, we identified the ER stress-associated genes HIF1A , TGM2 , PKD2 , and SEC31A as potential biomarkers for HF. Additionally, drug sensitivity analysis highlighted paclitaxel, PLX4720, and sorafenib as promising therapeutic agents targeting these biomarkers. Conclusion: These findings underscore the clinical significance of targeting ER stress pathways and provide novel insights into the diagnosis and treatment of HF.