Metabolomic Quantitative Trait Loci (mQTL) Mapping Implicates the Ubiquitin Proteasome System in Cardiovascular Disease Pathogenesis

Abstract

Levels of certain circulating short-chain dicarboxylacylcarnitine (SCDA), long-chain dicarboxylacylcarnitine (LCDA) and medium chain acylcarnitine (MCA) metabolites are heritable and predict cardiovascular disease (CVD) events. Little is known about the biological pathways that influence levels of most of these metabolites. Here, we analyzed genetics, epigenetics, and transcriptomics with metabolomics in samples from a large CVD cohort to identify novel genetic markers for CVD and to better understand the role of metabolites in CVD pathogenesis. Using genomewide association in the CATHGEN cohort (N = 1490), we observed associations of several metabolites with genetic loci. Our strongest findings were for SCDA metabolite levels with variants in genes that regulate components of endoplasmic reticulum (ER) stress (USP3, HERC1, STIM1, SEL1L, FBXO25, SUGT1) These findings were validated in a second cohort of CATHGEN subjects (N = 2022, combined p = 8.4x10-6-2.3x10-10). Importantly, variants in these genes independently predicted CVD events. Association of genomewide methylation profiles with SCDA metabolites identified two ER stress genes as differentially methylated (BRSK2 and HOOK2). Expression quantitative trait loci (eQTL) pathway analyses driven by gene variants and SCDA metabolites corroborated perturbations in ER stress and highlighted the ubiquitin proteasome system (UPS) arm. Moreover, culture of human kidney cells in the presence of levels of fatty acids found in individuals with cardiometabolic disease, induced accumulation of SCDA metabolites in parallel with increases in the ER stress marker BiP. Thus, our integrative strategy implicates the UPS arm of the ER stress pathway in CVD pathogenesis, and identifies novel genetic loci associated with CVD event risk.

Fig 1. Manhattan plots of GWAS results.

Displayed are Manhattan plots of the association results for GWAS (discovery cohort, whites only) with (A) factor 1 additive model, (B) factor 1 dominant model, (C) factor 2 additive model, (D) factor 2 dominant model, (E) factor 3 additive model and (F) factor 3 dominant model.

Fig 2. Genomic region plots for significant mQTL associated with SCDA levels.

Displayed are LocusZoom plots with -log10(p-value) (left Y-axis) and LD (right Y-axis), additive model, discovery cohort: (A) USP3|HERC1, whites only; (B) STON2|SEL1L, race meta-analysis; (C) RRM1|STIM1, race meta-analysis; (D) OLFM4|SUGT1, whites only.

Fig 3

Dicarboxylic (DC) acylcarnitines measured in HEK 293 cell lysates (A) and conditioned medium (B) after 24 h exposure to BSA alone or in complex with 500 uM fatty acids (FA, oleate:palmitate, 1:1). C) Representative Western blot analysis of the ER stress protein, BiP, in HEK 293 cells treated 24 h with 500 uM FA ± increasing doses of tunicamycin (NT; no treatment, Vehicle (DMSO), 8 ng/mL and 32 ng/mL tunicamycin). High dose tunicamycin (500 ng/mL) served as a positive control. Asterisks indicate significant difference between BSA and FA experiments (p<0.05).

Fig 4. Representation of metabolomics, GWAS, eQTL, and methylation leading to convergence on ER stress as a pathway for CVD event pathogenesis.