Integrating single-cell RNA sequencing and Mendelian randomization analysis to identify potential drug targets for dilated cardiomyopathy

Ruikang Liu, Yiying Liu, Chao Meng, Jun Li, Hui Wang, Xuanchun Huang, Shiyi Tao, Xia Xiao, Li Su, Yonghao Li, Weiming Fan, Chiyun SunHereditas 2025 · DOI

TL;DR

Researchers have integrated Mendelian randomization (MR) with single-cell RNA sequencing to identify IMPA1 and ITIH4 as high-priority drug targets for dilated cardiomyopathy (DCM). By bridging the gap between large-scale genetic data and cell-specific expression, the study reveals how dysregulated inositol metabolism in activated fibroblasts drives heart failure progression.

What Problem Does This Paper Address?

Dilated cardiomyopathy (DCM) is a leading cause of heart failure, characterized by the heart's ventricles thinning and stretching. While current treatments like beta-blockers manage symptoms, they don't effectively halt disease progression.

The challenge lies in translation. Genome-wide association studies (GWAS) have identified numerous genetic loci linked to DCM, but most of these variants don't change protein sequences—they change how genes are regulated. Determining which specific genes are "causal" (rather than just correlated) and which are "druggable" (amenable to pharmacological intervention) remains a major bottleneck in precision medicine.

Methods

The authors employed a "multi-omics" pipeline to filter 5,900 potentially druggable genes down to a few high-confidence candidates:

  1. Mendelian Randomization (MR): Used genetic variants as instrumental variables to mimic a randomized controlled trial, assessing whether changes in gene expression (eQTL) or protein levels (pQTL) causally influence DCM risk.
  2. Bayesian Co-localization: A statistical method used to determine if the genetic signal for gene expression and the genetic signal for DCM share the same underlying causal variant (PPH4 > 0.75).
  3. SMR & HEIDI Tests: Summary-data-based MR and "Heterogeneity in Dependent Instruments" tests were used to refine protein-level associations and rule out pleiotropy (where one variant affects multiple unrelated traits).
  4. Single-Cell RNA Sequencing (scRNA-seq): To move from "bulk" tissue data to cellular resolution, the team analyzed transcriptomes from individual cells in DCM and control hearts to see where and when these candidate genes are active.
  5. Pseudo-time & Cell Communication: Algorithms (Monocle2 and CellChat) were used to map how fibroblast subpopulations differentiate and interact during the disease process.

Datasets

The study leveraged several massive, independent datasets:

  • GWAS: DCM summary statistics from Jurgens et al. (9,365 cases and 946,368 controls).
  • eQTL/pQTL: Data from the eQTLGen Consortium (blood), GTEx V8 (heart and blood), and three large proteomic studies (Fenland, UKB-PPP, and deCODE).
  • scRNA-seq: Dataset GSE145154, containing transcriptomes from 5 control and 10 DCM heart samples.
  • Phenotypes: 2,514 traits from the UK Biobank were used for "Phe-MR" to check for potential side effects of targeting these genes.

Key Insights and Ideas

  • The "Protective" Candidates: IMPA1 and ITIH4 emerged as the strongest candidates. Higher genetically predicted expression of these genes was associated with a lower risk of DCM.
  • Fibroblast Subpopulations Matter: The study moved beyond seeing the heart as a simple muscle. It identified that IMPA1 expression is specifically reduced in activated fibroblasts and myofibroblasts in DCM hearts. This suggests that IMPA1 normally acts as a "brake" on the scarring process (fibrosis).
  • Metabolic-Structural Link: IMPA1 is involved in inositol signaling. Its downregulation suggests that impaired inositol recycling contributes to mitochondrial dysfunction and oxidative stress in the failing heart.
  • ITIH4 and the Matrix: While IMPA1 is a fibroblast story, ITIH4 appears to be a smooth muscle cell story, likely playing a role in stabilizing the extracellular matrix (ECM) and maintaining vascular structure.

Key Results

  • Causal Evidence: MR and co-localization identified 11 genes, but only IMPA1 and ITIH4 passed rigorous validation across both heart and blood tissues.
  • Effect Sizes: For IMPA1, the Odds Ratio (OR) was 0.66 in blood and 0.55 in the left ventricle, meaning higher expression significantly reduces DCM risk.
  • Single-Cell Validation: scRNA-seq confirmed that IMPA1 expression was significantly lower in DCM patients compared to controls, specifically within the fibroblast clusters.
  • Safety Profile: Phe-MR analysis showed that IMPA1 had no significant associations with adverse traits in the UK Biobank, suggesting it may be a safe therapeutic target. ITIH4 was linked to some metabolic traits but showed no adverse cardiac signals.

Limitations and Future Work

  • The "Cardiomyocyte" Gap: Due to their large size and elongated shape, adult cardiomyocytes are difficult to capture using standard 10x Genomics platforms. The authors suggest future studies should use single-nucleus RNA-seq (snRNA-seq) to better represent these cells.
  • Ancestry Bias: The genetic data used was predominantly from individuals of European ancestry, limiting the generalizability of the findings to other global populations.
  • Mechanism vs. Correlation: While the study provides strong causal evidence, it does not provide a "wet lab" proof of mechanism. Experimental validation in animal models or organ-on-a-chip systems is the necessary next step to confirm how IMPA1 modulation affects cardiac remodeling.

Summary updated

Abstract

Dilated cardiomyopathy (DCM), a leading cause of heart failure and sudden cardiac death, lacks therapies targeting disease progression. Genome-wide association studies (GWAS) have identified genetic loci linked to DCM, but translating these findings into actionable drug targets remains challenging. Integrating the druggable genome with multi-omics approaches offers a promising strategy for precision therapy. We combined Mendelian randomization (MR), Bayesian co-localization, and single-cell RNA sequencing to identify causal drug targets for DCM. Tissue-specific cis-eQTL and pQTL datasets from heart and blood tissues were analyzed using two-sample MR, Steiger filtering, and summary-data-based MR (SMR). Single-cell transcriptomic data (GSE145154) from DCM and control hearts were processed for cellular annotation, communication, and pseudo-time analysis. MR and co-localization identified IMPA1 and ITIH4 as protective candidates for DCM, with consistent evidence across cardiac and blood tissues (PPH4 > 0.75). SMR and HEIDI tests confirmed shared causal variants between protein expression and DCM. Single-cell analysis revealed reduced IMPA1 expression in activated fibroblasts of DCM hearts, implicating inositol metabolism dysregulation in fibrosis. ITIH4 showed associations with metabolic traits but no adverse cardiac effects. Fibroblast subpopulations exhibited altered communication and differentiation trajectories in DCM, highlighting their role in disease progression. This multi-omics study prioritizes IMPA1 and ITIH4 as transcriptomic candidates with suggestive causal associations to DCM, linking inositol signaling and extracellular matrix stability to disease mechanisms. These findings underscore the potential of integrating genomics and single-cell transcriptomics to accelerate drug discovery in cardiovascular diseases.
BibTeX
@inproceedings{liu2025integrating,
  author = {Ruikang Liu and Yiying Liu and Chao Meng and Jun Li and Hui Wang and Xuanchun Huang and Shiyi Tao and Xia Xiao and Li Su and Yonghao Li and Weiming Fan and Chiyun Sun},
  title = {{Integrating single-cell RNA sequencing and Mendelian randomization analysis to identify potential drug targets for dilated cardiomyopathy}},
  booktitle = {Hereditas},
  year = {2025},
  doi = {10.1186/s41065-025-00539-9},
}
Figure