Multi-trait genome-wide analysis identified novel risk loci and candidate drugs for heart failure

Zhengyang Yu, Maohuan Lin, Zhanyu Liang, Ying Yang, Wen Chen, Yonghua Wang, Yangxin Chen, Kaida Ning, Li C. Xia
doi: https://doi.org/10.1101/2024.03.24.24304812

Abstract

Heart failure (HF) is a common cardiovascular disease that poses significant morbidity and mortality risks. While genome-wide association studies reporting on HF abound, its genetic etiology is not well understood due to its inherent polygenic nature. Moreover, these genetic insights have not been completely translated into effective strategies for the primary treatment of HF. In this study, we conducted a large-scale integrated multi-trait analysis using European-ancestry GWAS summary statistics of coronary artery disease and HF, involving near 2 million samples to identify novel risk loci associated with HF. 72 loci were newly identified with HF, of which 44 were validated in the replication phase. Transcriptome association analysis revealed 215 HF risk genes, including EDNRA and FURIN. Pathway enrichment analysis of risk genes revealed their enrichment in pathways closely related to HF, such as response to endogenous stimulus (adjusted P = 8.83×10-3), phosphate-containing compound metabolic process (adjusted P = 1.91×10-2). Single-cell analysis indicated significant enrichments of these genes in smooth muscle cells, fibroblast of cardiac tissue, and cardiac endothelial cells. Additionally, our analysis of HF risk genes identified 74 potential drugs for further pharmacological evaluation. These findings provide novel insights into the genetic determinants of HF, highlighting new genetic loci as potential interventional targets to HF treatment, with significant implications for public health and clinical practice.

Introduction

Heart failure (HF) is a common cardiovascular syndrome that causes symptoms, such as shortness of breath, volume overload, and other functional limitations, resulting from structural or functional impairment of ventricular filling or ejection of blood[1]. It poses a significant global health challenge with high prevalence, morbidity, mortality, hospitalization rates, and healthcare costs[2]. Identifying HF risk loci and genes is highly meaningful for a full comprehension of the underlying molecular mechanisms and subsequent development of novel, early-intervention strategies for HF, albeit challenging due to its polygenic nature.

Genome-wide association studies (GWAS) have proven to be an effective tool to understand the genetic risk of HF, providing insights into its molecular mechanisms and potential therapeutic targets. Nevertheless, compared to other well-understood cardiovascular diseases, such as hypertension, coronary artery disease (CAD), and atrial fibrillation, risk loci for HF remained to be elucidated. Specifically, while hundreds, or even thousands, of risk loci have been pinpointed for hypertension, CAD, and atrial fibrillation, only 249 have been ascertained for HF. This calls for further studies aimed at understanding the genetic architecture of HF.

Fortunately, recent methodological advancements have shown that employing multi-ethnic and multi-trait approaches is effective in understanding the complex polygenic etiology. Multi-ethnic analysis aims to improve the generalizability of GWAS findings by leveraging the genetic diversity of study participants. The latest multi-ethnic analysis of HF has already reached a substantial sample size, comprising over 1.5 million individuals[3], making significant further augmentation in the short term challenging.

An alternative approach is the multi-trait analysis of GWAS (MTAG), which uses genetically correlated phenotypes to improve statistical power[4]. For instance, a recent MTAG analysis of cardioembolic stroke and AF identified 47 new loci associated with cardioembolic stroke[5]. Similarly, MTAG applied to primary sclerosing cholangitis and other clinical and epidemiological traits pinpointed seven novel loci associated with primary sclerosing cholangitis[6]. Another MTAG study of Lewy body dementia, Alzheimer’s disease, and Parkinson’s disease, unveiled eight novel genetic loci for Lewy body dementia and revealed the shared genetic etiology of these diseases[7]. These examples underscore the utility and efficacy of MTAG in further unraveling the genetic underpinnings of complex diseases.

In the case of HF, existing evidence has clearly demonstrated its clinical and genetic correlations with CAD. For example, in a longitudinal study on 892 CAD patients, it was found that 173 of these patients developed HF during a 13-year follow-up[8]. Another study used pairwise linkage disequilibrium score regression analysis (LDSC) to find a genetic correlation between HF and CAD of 0.67[9]. Publicly available GWAS summary statistics data have been progressively increasing in recent years. The primary repository that stores GWAS findings is the GWAS catalog[10], which currently includes full summary statistics of 65,590 studies. Notably, studies within it collected more than one million samples. Other platforms, such as IEU OpenGWAS[11] and PhenoScanner[12], also offer large repositories of GWAS summary statistics data.

Although GWAS has identified thousands of loci associated with complex diseases like HF, the presence of linkage disequilibrium often hinders the identification of potential causal genes from GWAS data. This limitation has prompted the development of methods that prioritize risk genes based on GWAS loci, among which the Transcriptome-Wide Association Study (TWAS) stands as one promising strategy. TWAS involves three main steps: training gene expression prediction models based on reference panels such as Genotype-Tissue Expression (GTEx), predicting individual expression in GWAS cohorts using these models, and conducting association analyses between predicted expression and phenotypes. Initially, methods like PrediXcan[13] focused on predicting gene expression levels at the individual level using GWAS data and performing association analyses. Recently, methods that leverage GWAS summary statistics data for predicting gene expression levels and conducting association analyses have emerged, such as FUSION[14] and S-PrediXcan[15].

Finally, drug-repurposing emerged as an efficient approach to translate approved drugs for new indications based on GWAS identified risk genes. Currently, although several medications exist for treating HF, most only alleviate symptoms and do not reverse the condition. Additionally, for HF with preserved ejection fraction, only SGLT2 inhibitors have demonstrated significant reductions in hospitalization and cardiovascular death for patients[16]. Therefore, there is an urgent need to discover new therapeutic drugs for HF. However, traditional drug development is time-consuming and labor-intensive. Drug repurposing, wherein already approved compounds are indicated for treating new conditions, offers the potential for faster deployment in clinical settings, with lower expenditures and higher safety assurance. In recent years, mapping of genome-wide significant GWAS loci to specific genes, thus enabling the reuse of drugs targeting these genes becomes an important repurposing method[17–20]. Another approach involves using TWAS[13, 15] to obtain gene information associated with the phenotype and conducting signature mapping. Specifically, this involves correlating expression level of genes associated with the disease with the pharmacological effects of compounds acting on these genes[21–23].

Illuminated by these latest methodological advances in GWAS and related fields, we designed this study. We collected and preprocessed the largest European GWAS summary statistics data for HF, and CAD to date (10/1/2022). Through MTAG analysis, we identified 99 HF risk loci, with 72 being newly reported in this study. Among these 72 loci, 44 were validated in the replication phase. We further conducted a TWAS study and identified 215 HF risk genes, including EDNRA and FURIN, newly associated with HF. We then conducted pathway enrichment analysis of TWAS risk genes revealing their enrichment in pathways closely related with HF, such as response to endogenous stimulus, phosphate-containing compound metabolic process, and cell population proliferation. Single-cell analysis showed that TWAS risk genes were significantly enriched in smooth muscle cells, fibroblast of cardiac tissue, and cardiac endothelial cells. Drug repurposing analysis based on MTAG genes and TWAS risk genes suggested 74 potential therapeutic drugs for HF.

Materials and Methods

Study design, data source and quality control

The complete workflow of this study is shown in Fig. 1. We obtained, by far (10/1/2022), the largest GWAS summary statistics of HF (47,309 cases and 930,014 controls), and the largest GWAS summary statistics of CAD (181,522 cases and 984,168 controls) from the GWAS Catalog (https://www.ebi.ac.uk/gwas/)[10, 24, 25] as the discovery phase. Additionally, we also obtained GWAS summary statistics for HF (19,350 cases and 288,996 controls) and CAD (33,628 cases and 275,526 controls) from the FinnGen repository (https://finngen.gitbook.io/documentation/v/r7/)[26], serving as an independent replication phase. These studies were all conducted on individuals of European ancestry and underwent stringent quality control procedures, as previously described[24–26].

Fig. 1
Fig. 1 Overall study design.

Employing the largest available GWAS summary statistics for heart failure (HF) encompassing 47,309 cases and 930,014 controls, as well as coronary artery disease (CAD) with 181,522 cases and 984,168 controls, we explored the polygenicity and genetic correlation between them using linkage disequilibrium score regression (LDSC), followed by multi-trait analysis of GWAS (MTAG) to identify significant SNPs. Functional annotation at the SNP level was performed to identify novel risk loci paired with MTAG on replication phase to confirm the novel identified risk loci. Additionally, transcriptome-wide association study (TWAS) was conducted to uncover HF risk genes. Subsequently, four gene-based analyses (pathway enrichment, single-cell enrichment, protein-protein interaction network, and animal knockout models) were performed to confirm the role of TWAS risk genes in the development of HF. Finally, drug repurposing was employed to identify potential therapeutic drugs for HF.

In the CAD summary statistics, base pair positions were converted to rs ID, employing the Genome Reference Consortium Human Reference 37 (GRCh37) (Index of /goldenPath/hg19/database (ucsc.edu). Furthermore, we excluded the major histocompatibility complex region (chromosome 6, 26–34 Mb) from our analyses due to its complex structure. SNPs with minor allele frequency less than 0.01 were filtered out, and the analysis was restricted to biallelic SNPs. SNPs with duplicated or missing rs IDs were removed from each GWAS summary dataset for subsequent analyses. A detailed description of all GWAS studies used in this research can be found in Additional file 1: Table S1.

Linkage disequilibrium score regression analysis

To estimate the polygenicity of HF and the genetic correlation between HF and CAD, we used LDSC[27, 28]. We first performed single-trait LDSC analysis on the GWAS summary statistics of HF and CAD to obtain several key parameters, including the mean χ2 statistic, genomic inflation factor (λGC), and intercept.

The mean χ2 statistic measures the association between genetic variants and a specific trait of interest in the GWAS summary statistics. We excluded traits with a mean χ2 statistic < 1.1 from further analysis. The factor λGC evaluates the presence of genetic inflation, which refers to an excess of test statistics over the expected null distribution. A value close to 1 indicates that the inflation in test statistics mostly results from polygenic effects. Similarly, the intercept obtained from LDSC analysis, subtracted by one, assesses the contribution of confounding biases to the inflation of test statistics. Again, a value close to 1 suggests that inflation is primarily driven by polygenic effects.

Subsequently, we applied pairwise LDSC analysis to investigate the genetic correlations between HF and CAD. We used the pre-computed linkage disequilibrium scores of European ancestry downloaded from the 1000 Genomes Project Phase 3 (https://alkesgroup.broadinstitute.org/LDSCORE/)[29]. Since that low imputation quality can potentially diminish test statistics, we focused on well-imputed HapMap3 SNPs.

Multi-trait meta-analysis with MTAG

To boost the power of HF-associated SNP discovery, we used MTAG v.1.0.8[4]. MTAG is a robust method designed to enhance trait-specific power by leveraging genetic correlations among multiple genetically related traits[4]. As input, it takes the GWAS summary statistics of correlated traits and incorporates the LDSC method to estimate the genetic covariance matrix. Subsequently, it generates weights and employs generalized inverse variance weighting to estimate trait-specific effects for a common set of SNPs. MTAG results can be similarly presented as the original GWAS summary statistics for individual traits, providing an easily understood view of genetic architecture.

To replicate new MTAG identified HF-associated loci, we also conducted MTAG on HF and CAD from FinnGen repository[26], which are independent from those in discovery phase. In our analyses, the summary statistics derived from single-trait GWAS are denoted as GWASHF and GWASCAD. Additionally, those acquired from the discovery phase of the MTAG analysis for HF are denoted as MTAGHF, while results from the replication phase are denoted as MTAGHF_R. The genome-wide significance level for MTAGHF was set at PMTAG_HF < 5 × 10-8.

Functional annotation by FUMA

To investigate the functional implications of HF associated SNPs, we used the Functional Mapping and Annotation (FUMA) platform v1.5.4[30]. Using its default settings, we annotated significant SNPs of MTAGHF, using the European ancestry data from the 1000 Genomes Project Phase 3 as a reference. We defined independent significant SNPs as those with PMTAG_HF < 5 × 10-8 and independent from each other at r2 < 0.6. Lead SNPs are independently significant SNPs that are independent from each other at r2 < 0.1. Genomic risk loci were identified by merging linkage disequilibrium blocks of independent significant SNPs in close proximity (< 250 kb). The top lead SNP was recognized as the SNP with the lowest PMTAG_HF[30] within a locus.

To gain further insight into the functional annotations of independent significant SNPs and their LD proxies, we employed tools including ANNOVAR categories[31], combined annotation-dependent depletion (CADD) scores[32], and Regulome DB scores[33] with their default parameters. For comparative purposes, we also annotated genome-wide significant SNPs from GWASHF using the same procedure.

Bayesian fine-mapping analysis

To facilitate the shortlisting of causal SNPs for each locus, we conducted Bayesian fine-mapping analysis to analyze risk loci determined in the FUMA analysis using the finemap.abf function of the coloc v5 R package (https://chr1swallace.github.io/coloc/)[34]. This function leverages bayesian methods to calculate the posterior probability that each SNP is a causal variant for the corresponding risk locus[34]. SNPs were incrementally included into the set based on their descending posterior probabilities, up to the point where cumulative posterior probability reached 0.90, resulting in a 90% credible SNP set for each locus.

Functional enrichment analysis

To investigate the functional enrichments of significant SNPs of MTAGHF, we used GWAS analysis of regulatory or functional information enrichment with LD correction (GARFIELD) [35]. GARFIELD uses GWAS summary statistics data, along with regulatory functional annotation data, including genic annotations, histone modifications, transcription factor binding sites, and chromatin segmentation states, across cell types and tissues[35]. Specifically, GARFIELD applies LD pruning based on LD and distance information to select independent SNPs. It then annotates SNPs with regulatory information if they, or highly correlated SNPs with them, overlap with such regulatory features. Finally, it computes odds ratios (OR) and enrichment p-values using logistic regression. We determined the enrichment to be significant at Bonferroni-corrected p < 4.98×10-5 (0.05/1005), where 1,005 is the number of feature annotations. We also applied GARFIELD to GWASHF for comparison.

Transcriptome-wide association analysis

To fully harness the information from SNPs for prioritizing risk genes associated with HF, we used the S-PrediXcan method[36] combined with the Joint Transcriptome Imputation (JTI) model[37] to perform TWAS analysis on MTAGHF. S-PrediXcan integrates summary-level GWAS data with gene expression prediction models, providing an estimate of the association between gene expression and HF[15]. To strengthen the accuracy and robustness of S-PrediXcan analysis, we applied the JTI model. This model combines information from multiple tissue-specific gene expression prediction models, allowing for more effective integration of gene expression data across tissues[37]. We used JTI models derived from the Genotype-Tissue Expression (GTEx) project version 8 transcriptome data[38], for eight different tissues associated with the heart. These tissues include subcutaneous adipose tissue, visceral omental adipose tissue, atrial appendage, left ventricle, kidney cortex, liver, lung, and whole blood. We applied a Bonferroni correction for multiple testing in each tissue. We also conducted S-PrediXcan on GWASHF as a comparison.

Functional characterization and contextual analysis for risk genes

To gain insight into the biological processes associated with risk genes identified from TWAS, we conducted pathway enrichment analysis using the web-based tool g:Profiler[39]. Pathways were considered significant if they reached a Bonferroni-corrected significance level (adjusted P < 0.05). For comparative purposes, we conducted separate enrichment analyses on the risk genes identified by TWAS on GWASHF.

We used the scDRS method[40] to explore the association between the expression levels of HF risk genes and cardiac-relevant cells. scDRS integrates single-cell RNA sequencing data with gene-disease association information derived from GWAS[40]. It associates individual cells with disease status by assessing the excess expression of HF risk genes within each cell. We also applied the scDRS method to risk genes identified by TWAS on GWASHF as a comparative analysis.

Furthermore, we investigated the functional enrichment of PPI networks among risk genes identified from TWAS using the STRING v12.0 database[41]. We then used the cytoHubba plugin in cytoscape to determine the top ten genes as hub genes[42].

Querying the MGI database

To confirm the roles of novel genes identified by MTAG and hub genes in our study within cardiovascular diseases, we queried the Mouse Genome Informatics (MGI, http://www.informatics.jax.org/) resource[43] for information on knockout models associated with them. MGI is a pivotal bioinformatics resource, providing comprehensive datasets pertaining to the mouse genome, and it is comprised of gene functionalities, expression patterns, mutation profiles, and data related to disease models. We focused on genes for which cardiovascular abnormalities became apparent after knockout experiments[43].

Drug repurposing analysis

In order to translate our genetic findings into therapeutic strategies for treating HF, we conducted drug repurposing, with the main steps outlined as follows: We first selected all TWAS risk genes associated with HF and genes identified by MTAG. Subsequently, we utilized the GEN2FUNC function within FUMA to map these genes onto the DrugBank database to generate a preliminary list of candidate drugs[44]. We then obtained a list of 26 FDA-approved HF drugs from Drug Bank[45] (Additional file 1: Table S2). For each candidate drug, we calculated its similarity value to all FDA-approved HF drugs, and subsequently averaged these values to obtain an overall similarity score. Here, similarity between two drugs is calculated using DICE score based their molecular fingerprints using drug structure data from DrugBank v5.1.10[45]. Finally, we identified drugs with an average similarity value greater than 0.2 as potential HF drugs. As a comparison, we also performed drug repurposing using 27 reported MTAG genes and the risk genes identified through TWAS results on GWASHF.

Results

LDSC analysis revealed the polygenicity of HF and CAD, along with a strong genetic correlation between them

Single-trait LDSC analysis suggested that the inflation observed in the test statistics is likely a consequence of polygenicity, rather than population stratification (Additional file 1: Table S3). The mean χ2 statistics were both greater than 1.1. LDSC intercepts consistently remained close to 1. Further pairwise LDSC analysis revealed a strong positive genetic correlation between HF and CAD (rg = 0.67, se = 0.03; P = 8.03×10-112). These findings validated our rationale for conducting a multi-trait analysis of HF, CAD.

MTAG analysis identified new HF-associated risk loci with evidence of replication

We conducted MTAG on GWASHF and GWASCAD, and identified 4,899 SNPs significantly associated with HF (PMTAG_HF < 5 × 10-8), among which 4,749 SNPs did not reach genome-wide significance in the original GWASHF analysis (Fig. 2a, b).For instance, rs6841581, located within the exon of the EDNRA gene, demonstrated an increase in significance from GWASHF analysis (PGWAS = 1.41×10-2) to MTAG analysis (PMTAG_HF = 4.75×10-18). Similarly, rs3918226, located within the intron of the NOS3 gene, exhibited an increase in significance from GWASHF analysis (PGWAS = 5.98×10-5) to MTAGHF analysis (PMTAG_HF = 1.09×10-22). From the original GWAS results to multi-trait results, the mean χ2 statistics increased from 1.158 to 1.340. These results suggested MTAG analysis could be more powerful than single-trait analysis by leveraging the genetical correlation among multiple traits. And the maxFDR for MTAGHF is 0.03, suggesting no overall inflation due to violation of the homogeneous assumption.

Fig. 2
Fig. 2 MTAG analysis boosted SNP discovery power.

Manhattan plots for MTAGHF (a) and GWASHF (b). Horizontal red dashed lines indicate the genome-wide significance threshold (P < 5 × 10-8). SNPs are labeled in place on chromosome coordinates.

Through FUMA annotation of the 4,899 significant SNPs, we pinpointed 99 risk loci (Additional file 2: Fig. S1). Among these loci, 27 had been previously reported, while 72 were newly discovered. To verify these newly identified risk loci, we performed MTAG analysis on GWAS summary statistics data obtained from FinnGen repository[26]. Notably, these summary statistics were independent from the GWAS conducted in the discovery phase.

We replicated 44 HF-specific associations at the nominal significance level of 0.05 (Additional file 1: Table S4).Notable examples included the previously mentioned rs6841581 in EDNRA (PMTAG_HF_R = 8.01×10-9) and rs3918226 in NOS3 (PMTAG_HF_R = 4.22×10-5). Additionally, other associations such as rs7173743 in MORF4L1 (PMTAG_HF_R = 2.36×10-4), rs12509595 in FGF5 (PMTAG_HF_R = 3.34×10-3), rs2306556 in GUCY1A3 (PMTAG_HF_R = 5.60×10-3), and rs1250258 in FN1 (PMTAG_HF_R = 4.08×10-2) were also successfully replicated.

Functional annotation for significant SNPs of MTAGHF

We conducted FUMA to understand the functional characteristics of significant SNPs (Additional file 1: Table S5). For candidate SNPs located within risk loci, most (97.42%) were located in non-coding regions, including UTR3, UTR5, downstream, intergenic region, intron, ncRNA intron and upstream. Only a small proportion of SNPs were exon with 144 (1.41%) in coding RNA and 116 (1.17%) in non-coding RNA (Fig. 3a). Among exonic SNPs in coding RNA, rs10965215 was the most statistically significant (PMTAG_HF = 2.87 × 10-58) and was mapped to the CDKN2B-AS1 gene. This was followed by rs3798220 (PMTAG_HF = 6.74 × 10-51) mapped to the LPA gene. The most significant exonic SNP in non-coding RNA was rs564398 (PMTAG_HF = 1.95 × 10-54) also mapped to the CDKN2B-AS1 gene.

Fig. 3
Fig. 3 Functional annotations of HF-associated SNPs.

(a) Functional annotation of candidate SNPs. (b) Distribution of candidate SNPs across 15 categories of minimum chromatin state. Chromatin states with a value ≤ 7 are considered as open chromatin regions. (c) RegulomeDB scores of candidate SNPs. RegulomeDB scores from 1b to 2c are considered likely to affect transcription factor binding.

Furthermore, 94.89% of candidate SNPs were located in open chromatin regions (minimum chromatin state ⩽ 7) (Fig. 3b), 3.89% of candidate SNPs were likely to affect the binding of transcription factors (RegulomeDB scores from 1b to 2c) (Fig. 3c), and 3.62% of candidate SNPs were deemed potentially deleterious (Combined Annotation Dependent Depletion score > 12.37). Notably, the SNP with the highest CADD score was rs328 (CADD score 50) located in the 8p21.3 locus. It is an exonic SNP of gene LPL. All afore mentioned genes (LPA, CDKN2B-AS1, and LPL) have been associated with HF[46] and CAD[47, 48] in previous studies, now with higher-level of evidence offered in this study.

Defining SNP credible sets within risk loci

To facilitate the shortlisting of causal SNPs for each of the 99 risk loci, we identified 1,982 SNPs within the 90% credible set using Bayesian fine mapping test (see Methods, Additional file 1: Table S6). The credible set of 10 HF loci contained only the top SNP with posterior probability greater than 0.90. The credible set of 85 HF loci contained multiple SNPs. The credible set of the remaining four risk loci (1p36.32, 4q33, 5q31.3, 21q22.12) did not include any SNPs.

Among the 10 loci with exactly one top SNPs, two were particularly noticeable. One was the top SNP rs3918226 (PMTAG_HF = 1.09×10-22) in locus 7q36.1. This SNP was associated with systolic blood pressure[49] and myocardial infarction[50], and it was an intron of the NOS3 gene, with a CADD score of 12.89. The protein encoded by the NOS3 gene plays a crucial role in synthesizing endothelial nitric oxide and is involved in the modulation of vascular function and the maintenance of cardiovascular health. Research has also revealed that mice with a knockout of the NOS3 gene exhibit cardiovascular abnormalities, including increased heart weight and abnormal cardiac muscle relaxation[51].

The other one top SNP was rs7412 (PMTAG_HF = 6.95×10-24) in locus 19q13.32. It was within an exon of the APOE gene and possessed a high CADD score of 25.1. Additionally, it has been associated with low-density lipoprotein cholesterol measurement[52], triglycerides[53], and pulse pressure[49]. The APOE gene encodes a protein known as apolipoprotein E, which plays a crucial role in lipid metabolism, including cholesterol transport and lipoprotein clearance, thereby influencing cardiovascular health. Study has indicated that the knockout of APOE in mice increases the risk of atherosclerosis in arteries[54].

Functional enrichment of significant MTAGHF SNPs

We observed 780 significant enrichments of significant MTAGHF SNPs in regulatory and functional categories (Additional file 1: Table S7), As a comparison, among the significant SNPs of GWASHF results, only 3 reached the significance threshold (Additional file 1: Table S8). In terms of genic regions (Fig. 4a), the most significant enrichment of SNPs was found in exonic regions (OR = 3.76, P = 2.34×10-10). In terms of tissue specific DNase I hypersensitive sites (Fig. 4b), SNPs were significantly enriched in tissues, such as fetal heart, blood vessel, fetal large intestine, gingival, embryonic stem cell, and fetal small intestine, with fetal heart being the most significant (OR = 4.94, P = 6.14×10-18). In terms of tissue-specific chromatin states (Fig. 4c), SNPs were significantly enriched in the transcribed region, with the most significant enrichment found in the actively transcribed region of liver tissue (OR = 2.67, P = 1.00×10-8). In terms of tissue specific histone-modified regions, SNPs showed a significant enrichment across tissues, such as blood, liver, and cervix (Fig. 4d), with the H3K4me1 region of blood tissues (OR = 3.22, P = 2.45×10-11) being the most significant.

Fig. 4
Fig. 4 Functional enrichment of significant MTAGHF SNPs.

A total of 4,899 SNPs were analyzed with GARFIELD. Enrichment by (a) genic regions, (b) tissue-specific DNase I hypersensitive sites (DHS) regions (c) tissue-specific chromatin states, and (d) tissue-specific histone-modified regions. In all plots, enrichment odds ratios were plotted against −log10(P value). The dotted horizontal lines mark the statistical significance threshold of P = 0.05/1005.Size of markers is proportional to the number of independent significant SNPs in a specific annotation category, while in subplots (c) and (d), markers are also color-coded by tissue categories.

Uncovering HF risk genes through TWAS

To uncover HF risk genes for subsequent in-depth functional analysis, we conducted a comprehensive transcriptome-wide association analysis (TWAS) on MTAGHF (see Methods). After Bonferroni correction to adjust for multiple testing within each tissue, we identified 215 genes that displayed significant associations with HF in at least one tissue type (Additional file 1: Table S9).

Among these genes, two were noteworthy. The first was EDNRA (PTWAS = 1.72×10-17 in adipose visceral omentum tissue), which was previously linked to ischemic stroke[55], pulse pressure[56], but newly associated with HF in this study. Importantly, it did not appear in the TWAS results using only GWASHF (Additional file 1: Table S10). EDNRA encodes the Endothelin Receptor Type A, which is a G protein-coupled receptor predominantly expressed in endothelial cells and smooth muscle cells. It plays a pivotal role in mediating the signaling of Endothelin-1 hormone. Research suggested that the deficiency in this gene could result in cardiovascular anomalies in mice, such as elevated blood pressure, abnormal cardiovascular system morphology, and the dilation of the right ventricle[57].

The second gene of significance is FURIN. This was also the first study to report the association between FURIN and HF. FURIN (PTWAS = 3.09×10-15 in liver tissue) is a gene encoding a protein known as furin, which belongs to the family of serine proteases. This protease primarily functions within the cell’s Golgi apparatus and vesicles, playing a pivotal role in the post-translational modification and processing of proteins. Research has indicated that the absence of furin leads to severe cardiovascular anomalies, including anomalous cardinal vein morphology, the absence of vitelline blood vessels, and abnormal heart development in developing mouse embryos[58].

Contextual analysis of TWAS risk genes

We obtained 10 significantly enriched pathways of TWAS risk genes (Fig. 5a, Additional file 1: Table S11). These pathways primarily involve metabolic processes of important compounds, and cell proliferation. such as response to endogenous stimulus (adjusted P = 8.83×10-3), phosphate-containing compound metabolic process (adjusted P = 1.91×10-2), and cell population proliferation (adjusted P = 2.41×10-2). In comparison, risk genes identified by GWASHF showed no enrichment in any pathway.

Fig. 5
Fig. 5 Contextual analysis of TWAS risk genes.

(a) Bubble plot of 10 enriched pathways. The x-axis represents pathway names, and the y-axis represents −log10(padj). Size of the bubbles indicates the number of genes enriched in the corresponding pathway, while the color of the bubbles changes according to the −log10(padj). (b) Enrichment of risk genes at the cellular level. Node color represents the risk score on the cells with red indicating high risk and blue indicating low risk. (c) Protein-protein interaction (PPI) sub-network of 10 hub genes. Gene nodes are arranged from left to right based on their degrees, with round rectangle representing the hub genes.

Using single-cell enrichment analysis, we found that HF risk genes were enriched in smooth muscle cells, fibroblast of cardiac tissue, cardiac endothelial cells (Fig. 5b, Additional file 1: Table S12). Risk genes identified by GWASHF showed no enrichment in any cardiac cells (Additional file 1: Table S13).

We also constructed a protein-protein interaction (PPI) network on the risk genes, resulting in a network with 204 nodes and 402 edges (Additional file 1: Fig. 2). PPI network demonstrated significant enrichment (P < 1×10-16), indicating that the proteins in this network are biologically interconnected. Among all genes, ten hub genes were identified: APOE, APOB, PCSK9, LPL, SCARB1, LIPC, CETP, ABCG8, SREBF1, ANGPTL4 (Fig. 5c).

Mouse knock-out models for novel MTAG genes and hub genes

We conducted queries for in silico knock out mouse models using the Mouse Genomics resource to find evidence of target gene modifications that could produce phenotypes associated with HF (Additional file 1: Table S14). A total of 29 gene knock-out models yielded evidence associated with cardiovascular abnormalities. To be specific, knock-out models of PRDM16, ABCG8, FN1, FGD5, EDNRA, GUCY1A3, NOS3, PLCE1, CNNM2, MORF4L1, FES showed evidence for congestive heart failure, myocardial abnormalities, cardiac hypertrophy, and abnormal myocardial fiber morphology, all were related to the development of HF.

Drug repurposing analysis

We identified 74 potential therapeutic drugs based on MTAG genes and HF risk genes identified by TWAS (see Methods, Additional file 1: Table S15), out of which 21 are approved by the U.S. FDA (Table 1). Notably, five (DB06403, DB00945, DB00559, DB08932, and DB06268) have undergone clinical trials for the treatment of HF[59–63]. Another three (DB09237, DB05676, and DB12548) have been subjected to clinical trials for diseases related to HF, such as hypertension[64, 65] and vascular inflammation[66]. The remaining 53 candidate drugs are investigational compounds that have not obtained approval from the U.S. FDA for the treatment of any disease. In contrast, using 27 reported MTAG genes and the risk genes identified through TWAS results on GWASHF, we identified only seven drug compounds. However, their overall similarity score to FDA-approved HF drugs were all below 0.2 and did not pass the threshold for repurposing (Additional file 1: Table S16).

View this table:
Table 1 Twenty-one FDA-approved drugs were identified as candidate drugs for HF.

Discussion

In this study, we conducted a multi-trait analysis by combining HF and CAD, leading to the identification of 72 novel HF risk loci and 215 HF risk genes through TWAS. Based on these genetic discoveries, we conducted drug repurposing and identified 74 potential therapeutic drugs for HF, among which 21 were U.S. FDA-approved.

Using summary statistics data from the largest HF and CAD GWAS, we proved a highly positive genetic correlation between HF and CAD, which was corroborated by previously findings[3]. This connection led us hypothesize that these cardiovascular diseases with high genetic correlation could be analyzed with the multi-trait approach to expand novel genetic discoveries by integrating large-scale data[5, 67, 68].

Our multi-trait analysis significantly improved the identification of HF risk loci. Specifically, we identified 72 novel loci associated with HF. One noteworthy risk locus is the 2p21, with the top lead SNP rs4245791 reaching significance in GWASCAD[24] and demonstrating a consistent effect direction with MTAGHF. The locus mapped to the gene ABCG8, also referred to as ATP binding cassette subfamily G member 8. ABCG8 encodes a protein crucially involved in cholesterol transport and homeostasis. Research has revealed that its deficiency in mice lead to abnormalities in heart left ventricle morphology, along with an increase in cardiac muscle contractility[69]. Another noteworthy risk locus is 15q25.1, with the top lead SNP being rs7173743. The SNP was an intronic of gene MORF4L1. MORF4L1, also known as Mortality Factor 4 Like 1, is a gene responsible for encoding a protein engaged in cellular processes, such as chromatin remodeling and DNA repair. Research has additionally demonstrated that its deficiency in mice resulted in cardiovascular abnormalities, including myocardial fiber disarray and cardiac hypertrophy[70].

Since the replication analysis is recommended to evaluate the credibility of each SNP association when applying MTAG to low-powered GWAS or to GWAS characterized by significant heterogeneity in statistical power. We conducted MTAG analysis on the FinnGen cohort serving as a replication phase. Our findings revealed that out of the 99 loci, 68 exhibited replicated HF-specific associations. Additionally, among the 72 novel loci identified, 44 were successfully replicated. Notably, several genes mapped with replicated loci showed direct relevance to cardiovascular abnormalities, such as myocardial abnormalities, cardiac hypertrophy, and abnormal myocardial fiber morphology, in mouse knockout models. These genes include EDNRA, NOS3, MORF4L1, FGF5, GUCY1A3, and FN1.

We further conducted a TWAS analysis and identified 215 HF risk genes and validated the roles of these genes in the progression of HF. Pathway enrichment analysis identified several key pathways closely associated with HF, including response to endogenous stimulus, phosphate-containing compound metabolic process, and cell population proliferation. Response to endogenous stimulus refers to any process that induces a modification in the state or function of a cell or organism, encompassing alterations in movement, secretion, enzyme production, gene expression, and other physiological activities. Further, a close association of endogenous hemodynamic[71] and endogenous nitric oxide[72] with HF has been reported. The phosphate-containing compound metabolic process pathway encompasses chemical reactions related to the phosphate group, which is the anion or salt derived from any phosphoric acid. This includes processes such as the synthesis of adenosine triphosphate. Notably, a reduction in the energy reserve available for adenosine triphosphate synthesis can render the heart more susceptible to systolic and diastolic failure[73]. The cell population proliferation pathway refers to the multiplication or reproduction of cells, leading to the expansion of a cell population. The proliferation of cardiac myocytes is intricately linked to the development of HF[74–76], and contemporary researches have proposed various approaches aimed at inducing cardiac myocyte proliferation as a therapeutic strategy for HF[77–79].

In the context of single-cell enrichment, it is noteworthy that risk genes demonstrated enrichment in cell types, such as smooth muscle cells, fibroblasts of cardiac tissue, and cardiac endothelial cells. Additionally, an earlier study has demonstrated that the deletion of the mineralocorticoid receptor specifically from smooth muscle cells can mitigate HF induced by transverse aortic constriction[80]. Furthermore, contemporary research indicated that endothelial cells play a crucial role in both coronary microvascular dysfunction and cardiac remodeling, ultimately contributing to the development of HF[81]. Moreover, investigations have highlighted the involvement of cardiac fibroblasts in myocardial stiffening through collagen production, as well as their active participation in modulating cardiac inflammation by synthesizing chemoattractive substances[82].

Drug repurposing facilitated by GWAS methods has gained popularity in recent years[83]. Our study extensively leveraged the genetic discoveries including those loci identified by MTAG and the risk genes identified through TWAS to explore potential drugs for primary prevention of HF. We identified 74 potential therapeutic drugs for HF treatment. Notably, five of these approved drugs (DB06403, DB00945, DB00559, DB08932, and DB06268) have undergone clinical trials for the treatment of HF[59–63], thus boosting the credibility of our findings. Moreover, the target genes of these potential medications exhibited substantial evidence of associations with cardiovascular system abnormalities in knock out models. Notably, ALDH2, NOS3, MIF, FURIN, and PDGFD, demonstrated direct associations with such pathological features as congestive heart failure, myocardial abnormalities, cardiac hypertrophy, and abnormal myocardial fiber morphology[51, 58, 84–86]. Furthermore, ALDH2, EDNRA, and FURIN, which exhibited expression levels positively correlated with HF risk, may represent more favorable targets for the development of novel therapies, since therapeutics aimed at downregulating a gene are typically easier to develop than those aimed at upregulation. It is noteworthy that using solely previously reported loci information and the original GWASHF data, we failed to identify any drugs.

Despite these novel findings, our study did have limitation. Our data were derived from European populations, thus potentially limiting the generalizability of our study to other ethnicities. However, recent advancements have introduced methods for transferring genetic findings from European populations to others, increasing the applicability of our research[87, 88].

Conclusions

In conclusion, our study identified 72 novel HF risk loci. Subsequently, we employed MTAG on replication phase, Bayesian fine mapping, GARFIELD, and MGI queries to establish the biological and statistical credibility of these loci. We also identified 215 significant HF risk genes through TWAS analysis. Pathway enrichment, single-cell enrichment, PPI network, and MGI queries confirmed the roles of these genes in the progression of HF. Based on these novel HF genes, we identified 74 drugs suitable for repurposing for HF. These findings have translational value in future efforts aimed toward the discovery of HF prevention and treatment regimens.

Data Availability

The datasets analyzed during the current study are publicly available. GWAS summary statistics data for discovery phase are available from GWAS Catalog under study accessions GCST009541 and GCST90132314 at https://www.ebi.ac.uk/gwas/. GWAS summary statistics data for replication phase are available at https://storage.googleapis.com/finngen-public-data-r7/summary_stats/finngen_R7_I9_HEARTFAIL_ALLCAUSE.gz and https://storage.googleapis.com/finngen-public-data-r7/summary_stats/finngen_R7_I9_CHD.gz. Data used in LDSC analysis can be obtained at alkesgroup.broadinstitute.org/LDSCORE/. JTI models of gene expression in 8 heart tissues are available from Zenodo at http://doi.org/10.5281/zenodo.3842289. Single-cell data used in scDRS analysis are available from Tabula Sapiens Single-Cell Dataset at https://doi.org/10.6084/m9.figshare.14267219.v5. All drug information is available from Drugbank v.5 at https://go.drugbank.com/. The information on all mouse knockout models can be obtained from Mouse Genome Informatics resources at https://www.informatics.jax.org/

https://www.ebi.ac.uk/gwas/

https://alkesgroup.broadinstitute.org/LDSCORE/

http://doi.org/10.5281/zenodo.3842289

https://doi.org/10.6084/m9.figshare.14267219.v5

https://go.drugbank.com/

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and materials

The datasets analyzed during the current study are publicly available. GWAS summary statistics data for discovery phase are available from GWAS Catalog under study accessions GCST009541[89] and GCST90132314[90] at https://www.ebi.ac.uk/gwas/. GWAS summary statistics data for replication phase are available at https://storage.googleapis.com/finngen-public-data-r7/summary_stats/finngen_R7_I9_HEARTFAIL_ALLCAUSE.gz and https://storage.googleapis.com/finngen-public-data-r7/summary_stats/finngen_R7_I9_CHD.gz[26].

Data used in LDSC analysis can be obtained at https://alkesgroup.broadinstitute. org/LDSCORE/[29]. JTI models of gene expression in 8 heart tissues are available from Zenodo at http://doi.org/10.5281/zenodo.3842289[91]. Single-cell data used in scDRS analysis are available from Tabula Sapiens Single-Cell Dataset at https://doi.org/10.6084/m9.figshare.14267219.v5[92]. All drug information is available from Drugbank v.5 at https://go.drugbank.com/[93]. The information on all mouse knockout models can be obtained from Mouse Genome Informatics resources at https://www.informatics.jax.org/[94].

Competing interests

The authors declare that they have no competing interests.

Fundings

This work was supported by the Guangdong Basic and Applied Basic Research Foundation under Grant 2022A1515-011426, in part by the National Natural Science Foundation of China under Grant 61873027, Grant 81970200 and Grant 82271609, in part by the Guangzhou Municipal Science and Technology Project under Grant 2023B01J1011, and in part by the Shenzhen Science and Technology Program under Grant JCYJ20190808100817047 and Grant RCBS20200714114909234.

Authors’ contributions

Z.Y., Y.C., K.N., and L.C.X. conceived and designed the study. Z.Y. and Z.L. collected the data. Z.Y. and Z.L. performed the data analysis. Z.Y., Y.Y., and M.L. conducted the interpretation of analysis results. Z.Y., L.C.X, and K.N. wrote the manuscript. Y.C., Y.Y., M.L., W.C., and Y.W. substantively revised the manuscript during the writing process. L.C.X supervised the overall process. All authors read and approved the submitted version.

Acknowledgements

We acknowledge GWAS catalog and FinnGen study for providing the GWAS summary statistics data of HF and CAD, the Tabula Sapiens Single-Cell Dataset for providing the single-cell data, and the Drugbank for providing drug data, we are grateful to all the participants in these studies.

List of abbreviations

HF
Heart failure
GWAS
Genome-wide association study
CAD
Coronary artery disease
LDSC
Linkage disequilibrium score regression
MTAG
Multi-trait analysis of GWAS
SNP
Single nucleotide polymorphism
TWAS
Transcriptome-wide association study
GTEx
Genotype-Tissue Expression
GRCh37
Genome Reference Consortium Human Reference 37
FUMA
Functional mapping and annotation of genetic associations
CADD
Combined annotation-dependent depletion
GARFIELD
GWAS analysis of regulatory or functional information enrichment with linkage disequilibrium correction
JTI
Joint-tissue imputation
MGI
Mouse Genome Informatics
OR
Odds ratio
PPI
Protein-protein interaction

Reference

  1. 1.↵
    Bozkurt B, Coats AJS, Tsutsui H, Abdelhamid CM, Adamopoulos S, Albert N, Anker SD, Atherton J, Böhm M, Butler J et al: Universal definition and classification of heart failure: a report of the Heart Failure Society of America, Heart Failure Association of the European Society of Cardiology, Japanese Heart Failure Society and Writing Committee of the Universal Definition of Heart Failure: Endorsed by the Canadian Heart Failure Society, Heart Failure Association of India, Cardiac Society of Australia and New Zealand, and Chinese Heart Failure Association. European Journal of Heart Failure 2021, 23(3):352–380.
    OpenUrlCrossRefPubMed
  2. 2.↵
    Savarese G, Becher PM, Lund LH, Seferovic P, Rosano GMC, Coats AJS: Global burden of heart failure: a comprehensive and updated review of epidemiology. Cardiovascular Research 2022, 118(17):3272–3287.
    OpenUrl
  3. 3.↵
    Levin MG, Tsao NL, Singhal P, Liu C, Vy HMT, Paranjpe I, Backman JD, Bellomo TR, Bone WP, Biddinger KJ et al: Genome-wide association and multi-trait analyses characterize the common genetic architecture of heart failure. Nature Communications 2022, 13(1):6914.
    OpenUrl
  4. 4.↵
    Team aR, Consortium SSGA, Turley P, Walters RK, Maghzian O, Okbay A, Lee JJ, Fontana MA, Nguyen-Viet TA, Wedow R et al: Multi-trait analysis of genome-wide association summary statistics using MTAG. Nature Genetics 2018, 50(2):229–237.
    OpenUrlCrossRefPubMed
  5. 5.↵
    Meseguer Monfort L, Ahlberg G, Andreasen L, Ghouse J, Haunso S, Bundgaard H, Svendsen JH, Olesen MS: Genome-wide multi-trait analysis on cardioembolic stroke identifies 47 novel loci. Eur Heart J 2022, 43(Supplement_2):ehac544.2855.
    OpenUrl
  6. 6.↵
    Han Y, Byun J, Zhu C, Sun R, Roh JY, Cordell HJ, Lee H-S, Shaw VR, Kang SW, Razjouyan J et al: Multitrait genome-wide analyses identify new susceptibility loci and candidate drugs to primary sclerosing cholangitis. Nature Communications 2023, 14:1069.
    OpenUrl
  7. 7.↵
    Guo P, Gong W, Li Y, Liu L, Yan R, Wang Y, Zhang Y, Yuan Z: Pinpointing novel risk loci for Lewy body dementia and the shared genetic etiology with Alzheimer’s disease and Parkinson’s disease: a large-scale multi-trait association analysis. BMC Medicine 2022, 20(1):214.
    OpenUrl
  8. 8.↵
    John JE, Claggett B, Skali H, Solomon SD, Cunningham JW, Matsushita K, Konety SH, Kitzman DW, Mosley TH, Clark D et al: Coronary Artery Disease and Heart Failure With Preserved Ejection Fraction: The ARIC Study. Journal of the American Heart Association 2022, 11(17):e021660.
    OpenUrl
  9. 9.↵
    Shah S, Henry A, Roselli C, Lin H, Sveinbjörnsson G, Fatemifar G, Hedman ÅK, Wilk JB, Morley MP, Chaffin MD et al: Genome-wide association and Mendelian randomisation analysis provide insights into the pathogenesis of heart failure. Nature Communications 2020, 11:163.
  10. 10.↵
    Sollis E, Mosaku A, Abid A, Buniello A, Cerezo M, Gil L, Groza T, Güneş O, Hall P, Hayhurst J et al: The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Research 2023, 51(D1):D977–D985.
    OpenUrlCrossRef
  11. 11.↵
    Lyon M, Andrews SJ, Elsworth B, Gaunt TR, Hemani G, Marcora E: The variant call format provides efficient and robust storage of GWAS summary statistics. 2020.
  12. 12.↵
    Staley JR, Blackshaw J, Kamat MA, Ellis S, Surendran P, Sun BB, Paul DS, Freitag D, Burgess S, Danesh J et al: PhenoScanner: a database of human genotype-phenotype associations. Bioinformatics (Oxford, England) 2016, 32(20):3207–3209.
    OpenUrlCrossRefPubMed
  13. 13.↵
    Gamazon ER, Wheeler HE, Shah KP, Mozaffari SV, Aquino-Michaels K, Carroll RJ, Eyler AE, Denny JC, Nicolae DL, Cox NJ et al: A gene-based association method for mapping traits using reference transcriptome data. Nature Genetics 2015, 47(9):1091–1098.
    OpenUrlCrossRefPubMed
  14. 14.↵
    Gusev A, Ko A, Shi H, Bhatia G, Chung W, Penninx BWJH, Jansen R, Geus EJd, Boomsma DI, Wright FA et al: Integrative approaches for large-scale transcriptome-wide association studies. Nature genetics 2016, 48(3):245.
    OpenUrlCrossRefPubMed
  15. 15.↵
    Barbeira AN, Dickinson SP, Bonazzola R, Zheng J, Wheeler HE, Torres JM, Torstenson ES, Shah KP, Garcia T, Edwards TL et al: Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nat Commun 2018, 9(1):1825.
    OpenUrlCrossRefPubMed
  16. 16.↵
    McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, Burri H, Butler J, Čelutkienė J, Chioncel O et al: 2023 Focused Update of the 2021 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: Developed by the task force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC) With the special contribution of the Heart Failure Association (HFA) of the ESC. European Heart Journal 2023, 44(37):3627–3639.
    OpenUrl
  17. 17.↵
    Biological insights from 108 schizophrenia-associated genetic loci. Nature 2014, 511(7510).
  18. 18.
    A X, Y W, Z Z, F Z, Ke K, Z Z, L Y, Lr L-J, J S, Y W et al: Genome-wide association analyses identify 143 risk variants and putative regulatory mechanisms for type 2 diabetes. Nature communications 2018, 9(1).
  19. 19.
    Jb N, Rb T, Lg F, W Z, Mw S, Se G, Tj H, S M, Em S, G S et al: Biobank-driven genomic discovery yields new insight into atrial fibrillation biology. Nature genetics 2018, 50(9).
  20. 20.↵
    Li H, Zhang Z, Qiu Y, Weng H, Yuan S, Zhang Y, Zhang Y, Xi L, Xu F, Ji X et al: Proteome-wide mendelian randomization identifies causal plasma proteins in venous thromboembolism development. Journal of Human Genetics 2023, 68(12):805–812.
    OpenUrl
  21. 21.↵
    Geb W, Ns C, B N, L C, W C, X X, J O, Ma P, S M, Cjd R et al: Gene expression profiles complement the analysis of genomic modifiers of the clinical onset of Huntington disease. Human molecular genetics 2020, 29(16).
  22. 22.
    Zf G, Er G, A W, Em D: Integrative Network-Based Analysis Reveals Gene Networks and Novel Drug Repositioning Candidates for Alzheimer Disease. Neurology Genetics 2021, 7(5).
  23. 23.↵
    Zhang W, Voloudakis G, Rajagopal VM, Readhead B, Dudley JT, Schadt EE, Björkegren JLM, Kim Y, Fullard JF, Hoffman GE et al: Integrative transcriptome imputation reveals tissue-specific and shared biological mechanisms mediating susceptibility to complex traits. Nature Communications 2019, 10(1):3834.
    OpenUrl
  24. 24.↵
    Aragam KG, Jiang T, Goel A, Kanoni S, Wolford BN, Atri DS, Weeks EM, Wang M, Hindy G, Zhou W et al: Discovery and systematic characterization of risk variants and genes for coronary artery disease in over a million participants. Nature Genetics 2022, 54(12):1803–1815.
    OpenUrl
  25. 25.↵
    Center RG, Shah S, Henry A, Roselli C, Lin H, Sveinbjörnsson G, Fatemifar G, Hedman ÅK, Wilk JB, Morley MP et al: Genome-wide association and Mendelian randomisation analysis provide insights into the pathogenesis of heart failure. Nature Communications 2020, 11(1):163.
    OpenUrl
  26. 26.↵
    Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, Reeve MP, Laivuori H, Aavikko M, Kaunisto MA et al: FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 2023, 613(7944):508–518.
    OpenUrl
  27. 27.↵
    Bulik-Sullivan B, Finucane HK, Anttila V, Gusev A, Day FR, Loh P-R, Duncan L, Perry JRB, Patterson N, Robinson EB et al: An atlas of genetic correlations across human diseases and traits. Nature Genetics 2015, 47(11):1236–1241.
    OpenUrlCrossRefPubMed
  28. 28.↵
    Bulik-Sullivan BK, Loh P-R, Finucane HK, Ripke S, Yang J, Patterson N, Daly MJ, Price AL, Neale BM: LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nature Genetics 2015, 47(3):291–295.
    OpenUrlCrossRefPubMed
  29. 29.↵
  30. 30.↵
    Watanabe K, Taskesen E, Van Bochoven A, Posthuma D: Functional mapping and annotation of genetic associations with FUMA. Nat Commun 2017, 8(1):1826.
    OpenUrlCrossRefPubMed
  31. 31.↵
    Wang K, Li M, Hakonarson H: ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Research 2010, 38(16):e164–e164.
    OpenUrlCrossRefPubMed
  32. 32.↵
    Kircher M, Witten DM, Jain P, O’Roak BJ, Cooper GM, Shendure J: A general framework for estimating the relative pathogenicity of human genetic variants. Nature Genetics 2014, 46(3):310–315.
    OpenUrlCrossRefPubMed
  33. 33.↵
    Boyle AP, Hong EL, Hariharan M, Cheng Y, Schaub MA, Kasowski M, Karczewski KJ, Park J, Hitz BC, Weng S et al: Annotation of functional variation in personal genomes using RegulomeDB. Genome Research 2012, 22(9):1790–1797.
    OpenUrlAbstract/FREE Full Text
  34. 34.↵
    Giambartolomei C, Vukcevic D, Schadt EE, Franke L, Hingorani AD, Wallace C, Plagnol V: Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS genetics 2014, 10(5):e1004383.
    OpenUrl
  35. 35.↵
    Iotchkova V, Ritchie GRS, Geihs M, Morganella S, Min JL, Walter K, Timpson NJ, Dunham I, Birney E, Soranzo N: GARFIELD classifies disease-relevant genomic features through integration of functional annotations with association signals. Nature Genetics 2019, 51(2):343–353.
    OpenUrlCrossRefPubMed
  36. 36.↵
    Barbeira AN, Dickinson SP, Bonazzola R, Zheng J, Wheeler HE, Torres JM, Torstenson ES, Shah KP, Garcia T, Edwards TL et al: Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nat Commun 2018, 9:1825.
    OpenUrlCrossRefPubMed
  37. 37.↵
    Zhou D, Jiang Y, Zhong X, Cox NJ, Liu C, Gamazon ER: A unified framework for joint-tissue transcriptome-wide association and Mendelian randomization analysis. Nature Genetics 2020, 52(11):1239–1246.
    OpenUrlCrossRefPubMed
  38. 38.↵
    Barbeira AN, Bonazzola R, Gamazon ER, Liang Y, Park Y, Kim-Hellmuth S, Wang G, Jiang Z, Zhou D, Hormozdiari F et al: Exploiting the GTEx resources to decipher the mechanisms at GWAS loci. Genome Biology 2021, 22(1):49.
    OpenUrl
  39. 39.↵
    Raudvere U, Kolberg L, Kuzmin I, Arak T, Adler P, Peterson H, Vilo J: g:Profiler: a web server for functional enrichment analysis and conversions of gene lists (2019 update). Nucleic Acids Research 2019, 47(W1):W191–W198.
    OpenUrlCrossRefPubMed
  40. 40.↵
    Zhang MJ, Hou K, Dey KK, Sakaue S, Jagadeesh KA, Weinand K, Taychameekiatchai A, Rao P, Pisco AO, Zou J et al: Polygenic enrichment distinguishes disease associations of individual cells in single-cell RNA-seq data. Nature Genetics 2022, 54(10):1572–1580.
    OpenUrl
  41. 41.↵
    Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, Simonovic M, Doncheva NT, Morris JH, Bork P et al: STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Research 2019, 47(D1):D607–D613.
    OpenUrlCrossRefPubMed
  42. 42.↵
    Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T: Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Research 2003, 13(11):2498–2504.
    OpenUrlAbstract/FREE Full Text
  43. 43.↵
    Shaw D: Searching the Mouse Genome Informatics (MGI) resources for information on mouse biology from genotype to phenotype. Current protocols in bioinformatics 2016, 56:1.7.1–1.7.16.
    OpenUrl
  44. 44.↵
    de Leeuw CA, Mooij JM, Heskes T, Posthuma D: MAGMA: Generalized Gene-Set Analysis of GWAS Data. PLoS Computational Biology 2015, 11(4):e1004219.
    OpenUrl
  45. 45.↵
    Wishart DS, Feunang YD, Guo AC, Lo EJ, Marcu A, Grant JR, Sajed T, Johnson D, Li C, Sayeeda Z et al: DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Research 2018, 46(D1):D1074–D1082.
    OpenUrlCrossRefPubMed
  46. 46.↵
    Wang H-P, Zhang N, Liu Y-J, Xia T-L, Chen G-C, Yang J, Li F-R: Lipoprotein(a), family history of cardiovascular disease, and incidence of heart failure. Journal of Lipid Research 2023, 64(7):100398.
    OpenUrl
  47. 47.↵
    Li Y-Y, Wang H, Zhang Y-Y: CDKN2B-AS1 gene rs4977574 A/G polymorphism and coronary heart disease: A meta-analysis of 40,979 subjects. Journal of Cellular and Molecular Medicine 2021, 25(18):8877–8889.
    OpenUrl
  48. 48.↵
    Xie L, Li Y-M: Lipoprotein Lipase (LPL) Polymorphism and the Risk of Coronary Artery Disease: A Meta-Analysis. International Journal of Environmental Research and Public Health 2017, 14(1):84.
    OpenUrl
  49. 49.↵
    Giri A, Hellwege JN, Keaton JM, Park J, Qiu C, Warren HR, Torstenson ES, Kovesdy CP, Sun YV, Wilson OD et al: Trans-ethnic association study of blood pressure determinants in over 750,000 individuals. Nature Genetics 2019, 51(1):51–62.
    OpenUrlCrossRefPubMed
  50. 50.↵
    Hartiala JA, Han Y, Jia Q, Hilser JR, Huang P, Gukasyan J, Schwartzman WS, Cai Z, Biswas S, Trégouët D-A et al: Genome-wide analysis identifies novel susceptibility loci for myocardial infarction. European Heart Journal 2021, 42(9):919–933.
    OpenUrl
  51. 51.↵
    Gregg AR, Schauer A, Shi O, Liu Z, Lee CG, O’Brien WE: Limb reduction defects in endothelial nitric oxide synthase-deficient mice. The American Journal of Physiology 1998, 275(6):H2319–2324.
    OpenUrlPubMed
  52. 52.↵
    Wu Y, Marvelle AF, Li J, Croteau-Chonka DC, Feranil AB, Kuzawa CW, Li Y, Adair LS, Mohlke KL: Genetic association with lipids in Filipinos: Waist circumference modifies an APOA5 effect on triglyceride levels. Journal of lipid research 2013, 54(11):3198–3205.
    OpenUrlAbstract/FREE Full Text
  53. 53.↵
    Hoffmann TJ, Theusch E, Haldar T, Ranatunga DK, Jorgenson E, Medina MW, Kvale MN, Kwok P-Y, Schaefer C, Krauss RM et al: A large electronic-health-record-based genome-wide study of serum lipids. Nature Genetics 2018, 50(3):401–413.
    OpenUrlCrossRef
  54. 54.↵
    van Ree JH, van den Broek WJ, Dahlmans VE, Groot PH, Vidgeon-Hart M, Frants RR, Wieringa B, Havekes LM, Hofker MH: Diet-induced hypercholesterolemia and atherosclerosis in heterozygous apolipoprotein E-deficient mice. Atherosclerosis 1994, 111(1):25–37.
    OpenUrlCrossRefPubMedWeb of Science
  55. 55.↵
    Dichgans M, Malik R, König IR, Rosand J, Clarke R, Gretarsdottir S, Thorleifsson G, Mitchell BD, Assimes TL, Levi C et al: Shared genetic susceptibility to ischemic stroke and coronary artery disease: a genome-wide analysis of common variants. Stroke 2014, 45(1):24–36.
    OpenUrlAbstract/FREE Full Text
  56. 56.↵
    Plotnikov D, Huang Y, Khawaja AP, Foster PJ, Zhu Z, Guggenheim JA, He M: High Blood Pressure and Intraocular Pressure: A Mendelian Randomization Study. Investigative Ophthalmology & Visual Science 2022, 63(6):29.
    OpenUrlCrossRef
  57. 57.↵
    Kurihara Y, Kurihara H, Suzuki H, Kodama T, Maemura K, Nagai R, Oda H, Kuwaki T, Cao W-H, Kamada N et al: Elevated blood pressure and craniofaclal abnormalities in mice deficient in endothelin-1. Nature 1994, 368(6473):703-710.
    OpenUrlCrossRefPubMed
  58. 58.↵
    Roebroek AJ, Umans L, Pauli IG, Robertson EJ, van Leuven F, Van de Ven WJ, Constam DB: Failure of ventral closure and axial rotation in embryos lacking the proprotein convertase Furin. Development (Cambridge, England) 1998, 125(24):4863–4876.
    OpenUrlAbstract
  59. 59.↵
    Actelion: A Long-term, Multicenter, Single-arm, Open-label Extension of the SERENADE Study, to Assess the Safety and Efficacy of Macitentan in Subjects With Heart Failure With Preserved Ejection Fraction and Pulmonary Vascular Disease. In.: clinicaltrials.gov; 2022.
  60. 60.
    Center UoTSM: Safety and Efficacy Trial Using Ambrisentan for Pulmonary Hypertension Associated With Congestive Heart Failure With Preserved Left Ventricular Ejection Fraction. In.: clinicaltrials.gov; 2020.
  61. 61.
    Grander W: Endothelin Receptor Blockade in Heart Failure With Diastolic Dysfunction and Pulmonary Hypertension. In.: clinicaltrials.gov; 2014.
  62. 62.
    Homma S, Thompson JLP, Pullicino PM, Levin B, Freudenberger RS, Teerlink JR, Ammon SE, Graham S, Sacco RL, Mann DL et al: Warfarin and aspirin in patients with heart failure and sinus rhythm. The New England Journal of Medicine 2012, 366(20):1859–1869.
    OpenUrlCrossRefPubMedWeb of Science
  63. 63.↵
    Pfizer: A PHASE 2 RANDOMISED, DOUBLE-BLIND, PLACEBO-CONTROLLED EXPLORATORY EFFICACY STUDY OF SITAXSENTAN SODIUM TO IMPROVE IMPAIRED EXERCISE TOLERANCE IN SUBJECTS WITH DIASTOLIC HEART FAILURE. In.: clinicaltrials.gov; 2022.
  64. 64.↵
    A.Ş NA-GSvT: Efficacy and Safety of S-amlodipine 2,5 mg and 5 mg in Hypertension Patients: Open-label, Local, Phase IV Study. In.: clinicaltrials.gov; 2018.
  65. 65.↵
    Pharmaceuticals L: Prospective, Randomized, Double-Blind, Placebo-Controlled, Parallel-Group Study to Evaluate the Safety and Efficacy of a Novel Dual Angiotensin and Endothelin Receptor Antagonist (PS433540) in Subjects With Stage I and II Hypertension. In.: clinicaltrials.gov; 2011.
  66. 66.↵
    Pennsylvania Uo: A Phase IV, Open Label Study of the Effects of Apremilast on Vascular Inflammation and Cardiometabolic Function in Psoriasis. In.: clinicaltrials.gov; 2022.
  67. 67.↵
    Ortiz-Fernández L, Carmona EG, Kerick M, Lyons P, Carmona FD, Mejías RL, Khor CC, Grayson PC, Tombetti E, Jiang L et al: Identification of new risk loci shared across systemic vasculitides points towards potential target genes for drug repurposing. Annals of the Rheumatic Diseases 2023.
  68. 68.↵
    Temprano-Sagrera G, Sitlani CM, Bone WP, Martin-Bornez M, Voight BF, Morrison AC, Damrauer SM, de Vries PS, Smith NL, Sabater-Lleal M et al: Multi-phenotype analyses of hemostatic traits with cardiovascular events reveal novel genetic associations. Journal of Thrombosis and Haemostasis 2022, 20(6):1331–1349.
    OpenUrl
  69. 69.↵
    References
  70. 70.↵
    Tominaga K, Kirtane B, Jackson JG, Ikeno Y, Ikeda T, Hawks C, Smith JR, Matzuk MM, Pereira-Smith OM: MRG15 regulates embryonic development and cell proliferation. Molecular and Cellular Biology 2005, 25(8):2924–2937.
    OpenUrlAbstract/FREE Full Text
  71. 71.↵
    Packer M: New concepts in the pathophysiology of heart failure: beneficial and deleterious interaction of endogenous haemodynamic and neurohormonal mechanisms. Journal of Internal Medicine 1996, 239(4):327–333.
    OpenUrlCrossRefPubMedWeb of Science
  72. 72.↵
    Xie Y-W, Shen W, Zhao G, Xu X, Wolin MS, Hintze TH: Role of Endothelium-Derived Nitric Oxide in the Modulation of Canine Myocardial Mitochondrial Respiration In Vitro. Circulation Research 1996, 79(3):381–387.
    OpenUrlAbstract/FREE Full Text
  73. 73.↵
    Ingwall JS, Atkinson DE, Clarke K, Fetters JK: Energetic correlates of cardiac failure: Changes in the creatine kinase system in the failing myocardium. Eur Heart J 1990, 11(suppl_B):108–115.
    OpenUrlCrossRefPubMedWeb of Science
  74. 74.↵
    Engel FB: Cardiomyocyte Proliferation: A Platform for Mammalian Cardiac Repair. Cell Cycle 2005, 4(10):1360–1363.
    OpenUrlCrossRefPubMedWeb of Science
  75. 75.
    Sager HB, Hulsmans M, Lavine KJ, Moreira MB, Heidt T, Courties G, Sun Y, Iwamoto Y, Tricot B, Khan OF et al: Proliferation and Recruitment Contribute to Myocardial Macrophage Expansion in Chronic Heart Failure. Circulation Research 2016, 119(7):853–864.
    OpenUrlAbstract/FREE Full Text
  76. 76.↵
    Wilsbacher L, McNally EM: Genetics of Cardiac Developmental Disorders: Cardiomyocyte Proliferation and Growth and Relevance to Heart Failure. Annual Review of Pathology: Mechanisms of Disease 2016, 11(1):395–419.
    OpenUrl
  77. 77.↵
    Abouleisa RRE, Salama ABM, Ou Q, Tang X-L, Solanki M, Guo Y, Nong Y, McNally L, Lorkiewicz PK, Kassem KM et al: Transient Cell Cycle Induction in Cardiomyocytes to Treat Subacute Ischemic Heart Failure. Circulation 2022, 145(17):1339–1355.
    OpenUrlCrossRef
  78. 78.
    Cheng RK, Asai T, Tang H, Dashoush NH, Kara RJ, Costa KD, Naka Y, Wu EX, Wolgemuth DJ, Chaudhry HW: Cyclin A2 Induces Cardiac Regeneration After Myocardial Infarction and Prevents Heart Failure. Circulation Research 2007, 100(12):1741–1748.
    OpenUrlAbstract/FREE Full Text
  79. 79.↵
    Woo YJ, Panlilio CM, Cheng RK, Liao GP, Atluri P, Hsu VM, Cohen JE, Chaudhry HW: Therapeutic delivery of cyclin A2 induces myocardial regeneration and enhances cardiac function in ischemic heart failure. Circulation 2006, 114(1 Suppl):I206–213.
    OpenUrlCrossRefPubMedWeb of Science
  80. 80.↵
    Kim SK, Biwer LA, Moss ME, Man JJ, Aronovitz MJ, Martin GL, Carrillo-Salinas FJ, Salvador AM, Alcaide P, Jaffe IZ: Mineralocorticoid Receptor in Smooth Muscle Contributes to Pressure Overload–Induced Heart Failure. Circulation: Heart Failure 2021, 14(2):e007279.
    OpenUrl
  81. 81.↵
    Lindner D, Zietsch C, Tank J, Sossalla S, Fluschnik N, Hinrichs S, Maier L, Poller W, Blankenberg S, Schultheiss H-P et al: Cardiac fibroblasts support cardiac inflammation in heart failure. Basic Research in Cardiology 2014, 109(5):428.
    OpenUrlCrossRefPubMed
  82. 82.↵
    Wang Y, Zhang J, Wang Z, Wang C, Ma D: Endothelial-cell-mediated mechanism of coronary microvascular dysfunction leading to heart failure with preserved ejection fraction. Heart Failure Reviews 2023, 28(1):169–178.
    OpenUrl
  83. 83.↵
    Reay WR, Cairns MJ: Advancing the use of genome-wide association studies for drug repurposing. Nature Reviews Genetics 2021, 22(10):658–671.
    OpenUrlCrossRefPubMed
  84. 84.↵
    Gladh H, Folestad EB, Muhl L, Ehnman M, Tannenberg P, Lawrence A-L, Betsholtz C, Eriksson U: Mice Lacking Platelet-Derived Growth Factor D Display a Mild Vascular Phenotype. PloS One 2016, 11(3):e0152276.
    OpenUrlCrossRef
  85. 85.
    Qi D, Atsina K, Qu L, Hu X, Wu X, Xu B, Piecychna M, Leng L, Fingerle-Rowson G, Zhang J et al: The vestigial enzyme D-dopachrome tautomerase protects the heart against ischemic injury. The Journal of Clinical Investigation 2014, 124(8):3540–3550.
    OpenUrlCrossRefPubMedWeb of Science
  86. 86.↵
    Zambelli VO, Gross ER, Chen C-H, Gutierrez VP, Cury Y, Mochly-Rosen D: Aldehyde dehydrogenase-2 regulates nociception in rodent models of acute inflammatory pain. Science Translational Medicine 2014, 6(251):251ra118.
    OpenUrlAbstract/FREE Full Text
  87. 87.↵
    Miao J, Guo H, Song G, Zhao Z, Hou L, Lu Q: Quantifying portable genetic effects and improving cross-ancestry genetic prediction with GWAS summary statistics. Nature Communications 2023, 14(1):832.
    OpenUrl
  88. 88.↵
    Tian P, Chan TH, Wang Y-F, Yang W, Yin G, Zhang YD: Multiethnic polygenic risk prediction in diverse populations through transfer learning. Frontiers in Genetics 2022, 13.
  89. 89.↵
    Shah S HA, Roselli C, Lin H, Sveinbjörnsson G, Fatemifar G, Hedman ÅK, Wilk JB, Morley MP, Chaffin MD et al:: Heart failure. GWAS catalog. 2020.
  90. 90.↵
    Tcheandjieu C ZX, Hilliard AT, Clarke SL, Napolioni V, Ma S, Lee KM, Fang H, Chen F, Lu Y et al: Coronary artery disease. GWAS catalog. 2022.
  91. 91.↵
    Gamazon ERZ, D.: JTI (Version 1.0) [Data set]. Zenodo. doi:10.5281/zenodo.3842289. 2020.
    OpenUrlCrossRef
  92. 92.↵
    Pisco AC, Tabula Sapiens Tabula Sapiens Single-Cell Dataset. figshare. Dataset. doi:10.6084/m9.figshare.14267219.v5 2021.
    OpenUrlCrossRef
  93. 93.↵
    Wishart DS FY, Guo AC, Lo EJ, Marcu A, Grant JR, Sajed T, Johnson D, Li C, Sayeeda Z et al: Drugbank 5.0 https://go.drugbank.com/. 2018.
  94. 94.↵
    D S: Mouse Genome Informatics resources https://www.informatics.jax.org/. 2016.
Back to top
Posted March 27, 2024.
Download PDF

Supplementary Material

Data/Code
Email
Multi-trait genome-wide analysis identified novel risk loci and candidate drugs for heart failure
Zhengyang Yu, Maohuan Lin, Zhanyu Liang, Ying Yang, Wen Chen, Yonghua Wang, Yangxin Chen, Kaida Ning, Li C. Xia
medRxiv 2024.03.24.24304812; doi: https://doi.org/10.1101/2024.03.24.24304812
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area