Integrative analysis of loss-of-function variants in clinical and genomic data reveals novel genes associated with cardiovascular traits

Abstract

Background: Genetic loss-of-function variants (LoFs) associated with disease traits are increasingly recognized as critical evidence for the selection of therapeutic targets. We integrated the analysis of genetic and clinical data from 10,511 individuals in the Mount Sinai BioMe Biobank to identify genes with loss-of-function variants (LoFs) significantly associated with cardiovascular disease (CVD) traits, and used RNA-sequence data of seven metabolic and vascular tissues isolated from 600 CVD patients in the Stockholm-Tartu Atherosclerosis Reverse Network Engineering Task (STARNET) study for validation. We also carried out in vitro functional studies of several candidate genes, and in vivo studies of one gene.

Results: We identified LoFs in 433 genes significantly associated with at least one of 10 major CVD traits. Next, we used RNA-sequence data from the STARNET study to validate 115 of the 433 LoF harboring-genes in that their expression levels were concordantly associated with corresponding CVD traits. Together with the documented hepatic lipid-lowering gene, APOC3, the expression levels of six additional liver LoF-genes were positively associated with levels of plasma lipids in STARNET. Candidate LoF-genes were subjected to gene silencing in HepG2 cells with marked overall effects on cellular LDLR, levels of triglycerides and on secreted APOB100 and PCSK9. In addition, we identified novel LoFs in DGAT2 associated with lower plasma cholesterol and glucose levels in BioMe that were also confirmed in STARNET, and showed a selective DGAT2-inhibitor in C57BL/6 mice not only significantly lowered fasting glucose levels but also affected body weight.

Conclusion: In sum, by integrating genetic and electronic medical record data, and leveraging one of the world's largest human RNA-sequence datasets (STARNET), we identified known and novel CVD-trait related genes that may serve as targets for CVD therapeutics and as such merit further investigation.

Keywords: Cardiovascular traits; Electronic Medical Records; Genetic association; Integrative data analysis; Loss-of-function variant; Target identification and validation.

Fig. 1

Overall workflow of the study. We first identified high-confidence LoFs in genotyping and the imputed data from the BioMe Biobank (a). Then for each gene with a LoF, we partitioned the BioMe individuals into LoF carriers and non-carriers (b) for comparison of 10 CVD-related traits obtained from the Mount Sinai Hospital Electronic Medical Records (MSH-EMR) (c). Next, we performed trait-specific quality-control (QC) by considering that the CVD-traits are affected by certain ongoing medications (d), followed by statistical analyses to robustly identify LoF-genes that were significantly associated with at least one of these CVD traits (e). In the next step (f), we assessed LoF-harboring genes associated with any of the CVD traits by exploring associations between RNA expression levels of these genes and corresponding CVD traits across seven tissues in STARNET [27]. We then selected genes with concordant CVD trait-associations in both BioMe and STARNET (i.e., when LoFs in a gene are associated with low values of a CVD trait, low expression of the same gene is also associated with low values of the trait) (g). For LoF genes associated with lower plasma cholesterol or triglyceride levels in BioMe Biobank and STARNET liver data, we carried out functional in vitro evaluation using HepG2 cells (H). Last, a knowledge-driven filtration approach was used for leveraging information in Gene Ontology (GO) to select potential therapeutic targets for validation in mice (i)

Fig. 2

The predicted LoF variants in the BioMe Biobank cohort. a Allele frequency of all LoF variants in BioMe Biobank population. b Distribution of the number of LoF variants per gene. c Carrier frequency of LoF genes. d Number of LoF genes carried per person

Fig. 3

In vitro validation of candidate genes for lowering plasma cholesterol and triglycerides. a Schematic illustration of lipoprotein metabolism in vivo, and b An in vitro HepG2 cell model to validate three hepatic plasma cholesterol and three hepatic triglyceride candidate genes. I) Cholesterol- and triglyceride-containing very low density lipoprotein (VLDL) particles are synthesized in, and secreted from, the liver to circulation elevating plasma levels of cholesterol and triglycerides. II) The VLDL particles then travel in blood to microcirculation in peripheral tissues, such as the skeletal muscle and adipose tissue, where lipoprotein lipase (LPL) anchored to the endothelium mediates hydrolysis of VLDL-triglycerides forming free fatty acids that are taken up by the local tissue. This extra-hepatic process lowers plasma triglyceride levels. III) The LPL-mediated hydrolysis of VLDL particles results in the formation of smaller cholesterol-rich low-density lipoproteins (LDL) particles. Some LDL particles are taken up by LDL receptors in extra-hepatic tissues, a process that lowers plasma levels of LDL and cholesterol. IV) The most important regulatory process of plasma LDL and cholesterol levels is, however, the uptake of LDL by hepatic LDL receptors. V) Last, uptake of LDL by the LDL receptor in the liver is inhibited by hepatic synthesis of PCSK9 that binds to the LDL receptors and permits their recirculation away from the hepatocyte cell surface (and LDL receptor degradation) effectively lowering the uptake of LDL particles. Thus, high levels of hepatic PCKS9 lead to reduced LDL uptake and higher plasma cholesterol levels. The HepG2 in vitro model of the liver was chosen as it is the most important organ to control plasma triglyceride and cholesterol levels (a) and since all lipid-associated candidate genes were identified in STARNET liver RNA-seq data. However as illustrated in panels a and b, an in vitro model of the liver in the form of HepG2 cells cannot fully model lipid metabolism and the extrahepatic tissue contribution in vivo. c For the plasma cholesterol-lowering candidate genes (RNMTL1, SCRN2 and PCK2), Apolipoprotein B-100 (APOB-100), Proprotein Convertase Subtilisin/Kexin type 9 (PCSK9) protein levels were measured in the cell media whereas LDL-receptor (LDLR) was measured in cell lysates. d For the plasma triglyceride-lowering candidate genes (APOC3, SLC39A5, NMRAL1, ABHD14B), Apolipoprotein B-100 (APOB-100), triglycerides (TG) were measured in cell media and lysates. In all experiments, the silencing efficiency of each gene resulted in more than 90% decrease in gene expression measured after 72 h incubation, including 24 h treatment with oleic acid performed before cell harvesting. Values are means ± SEM. Results are based on 3 biological replicates. p < 0.0332 (*), p < 0.0021 (**), p < 0.0002 (***). IDL, intermediate-density lipoprotein. Computational analysis results of the genes tested in (c) and (d) are shown in Additional file 1: Figure S9

Fig. 4

In silico analysis of DGAT2 and in vivo effect of selective DGAT1 and DGAT2 inhibitors on bodyweight and glucose levels in male C57BL/6 mice. a Significant association within the BioMe Biobank cohort between non-carriers (n = 5469) and carriers (n = 8) of LoF DGAT2 mutation on glucose level (logistic regression, see Methods; p = 0.049, β = − 19, t_β = − 1.4). b Association between non-carriers (n = 4768) and carriers (n = 8) of LoF DGAT2 mutation on cholesterol levels (logistic regression, see Methods; p = 0.078, β = − 24, t_β = − 1.8). c Association between non-carriers (n = 4721) and carriers (n = 8) of LoF DGAT2 mutation on triglyceride levels (logistic regression, see Methods; p = 0.34, β = − 26, t_β = − 1.0). d Association from STARNET RNA-clinical trait analysis comparing DGAT2 expression levels with glucose levels in liver tissue (p = 0.024, β = 0.56). e Association from STARNET RNA-clinical trait analysis comparing DGAT2 expression levels with cholesterol levels in mammary artery (p = 0.036, β = 0.089). f Association from STARNET RNA-clinical trait analysis comparing DGAT2 expression levels with triglyceride levels in liver tissue (p = 0.0030, β = 0.27). g Body weight loss (ΔBW) of chow-fed mice treated for 3 days with vehicle, DGAT1 inhibitor (PF3), and DGAT2 inhibitor (PF9). h Short fasting blood glucose of chow-fed mice treated for 3 days with vehicle, DGAT1 inhibitor (PF3), and DGAT2 inhibitor (PF9) (n = 5). p < 0.05 (*), p < 0.01 (**), versus vehicle (student t-test)