Investigation of Functional Genes at Homologous Loci Identified Based
Total Page:16
File Type:pdf, Size:1020Kb
Journal of Atherosclerosis and Thrombosis Vol.22, No.5 455 Original Article Investigation of Functional Genes at Homologous Loci Identified Based on Genome-wide Association Studies of Blood Lipids via High-fat Diet Intervention in Rats using an in vivo Approach Koichi Akiyama, Yi-Qiang Liang, Masato Isono and Norihiro Kato Department of Gene Diagnostics and Therapeutics, Research Institute, National Center for Global Health and Medicine, Tokyo, Japan Aim: It is challenging to identify causal (or target) genes at individual loci detected using genome- wide association studies (GWAS). In order to follow up GWAS loci, we investigated functional genes at homologous loci identified using human lipid GWAS that responded to a high-fat, high-choles- terol diet (HFD) intervention in an animal model. Methods: The HFD intervention was carried out for four weeks in male rats of the spontaneously hypertensive rat strain. The liver and adipose tissues were subsequently excised for analyses of changes in the gene expression as compared to that observed in rats fed normal rat chow (n=8 per group). From 98 lipid-associated loci reported in previous GWAS, 280 genes with rat orthologs were initially selected as targets for the two-staged analysis involving screening with DNA microarray and validation with quantitative PCR (qPCR). Consequently, genes showing a differential expression due to HFD were examined for changes in the expression induced by atorvastatin, which was indepen- dently administered to the rats. Results: Using the HFD intervention in the rats, seven known (Abca1, Abcg5, Abcg8, Lpl, Nr1h3, Pcsk9 and Pltp) and three novel (Madd, Stac3 and Timd4) genes were identified as potential signifi- cant targets, with an additional list of 23 suggestive genes. Among these 33 genes, Stac3, Fads1 and six known genes exhibited nominally significant expression changes following treatment with atorv- astatin. Six (of 33) genes overlapped with those previously detected in the expression QTL studies. Conclusions: Our experimental in vivo approach increases the ability to identify target gene(s), when combined with other functional studies, thus improving understanding of the mechanisms by which GWAS variants act. See editorial vol. 22: 442-444 J Atheroscler Thromb, 2015; 22: 455-480. Key words: Lipid, mRNA, Gene expression, Genome-wide association study, Diet the plasma lipid concentrations1, 2). However, as most Introduction disease- and trait-associated variants lie within non- Genome-wide association studies (GWAS) have coding sequences3), it is difficult to identify causal (or succeeded in identifying a number of genomic loci target) genes at individual GWAS loci. Several lines of associated with complex diseases and traits, including evidence suggest that a substantial proportion of such variants are involved in transcriptional mechanisms, Address for correspondence: Norihiro Kato, Department of including the modulation of cis-regulatory elements Gene Diagnostics and Therapeutics, Research Institute, 3, 4) National Center for Global Health and Medicine, 1-21-1 (e.g., promoter and enhancer elements) . Toyama, Shinjuku-ku, Tokyo 162-8655, JAPAN One of the most popular approaches for finding E-mail: [email protected] associations between target genes and regulatory ele- Received: August 26, 2014 ments is to analyze SNPs. Information on SNP geno- Accepted for publication: September 30, 2014 types, which can be compared with trait phenotypes 456 Akiyama et al. using GWAS, is analyzed in conjunction with gene level. expression data for particular types of cells derived from individuals. This process enables researchers to Materials and Methods identify associations between specific genetic variants and the gene expression levels using expression quanti- Animals tative trait locus (eQTL) studies5). Therefore, disease- We used the spontaneously hypertensive rats of a and trait-associated variants have been shown to often Japanese colony (SHR/Izm, hereafter referred to as affect the expression of nearby genes (cis-eQTLs) and SHR) for the HFD intervention, as previously those located over long genetic distances (trans- reported7). SHR rats are characterized by prominent eQTLs). In general, authors are inclined to assume the hypertension with a conspicuous response to a high- enrichment of GWAS SNPs in regulatory elements lipid diet and are considered to constitute a rat model when lead SNPs (i.e., SNPs showing the strongest of metabolic syndrome. This model allowed us to association signal) lie within regulatory DNA or examine the lipid GWAS-identified gene loci in the exhibit strong LD (e.g., r2 ≥ 0.8) with SNPs in regula- context of metabolic syndrome. Male rats (n =8 each tory DNA3). That is, since eQTL studies by them- for the HFD and control groups) were weaned at 4 selves show possible relationships between disease- and weeks of age and placed on a diet of normal rat chow trait-associated variants and the gene expression levels, (SP diet, Funabashi Farm, Japan). The rats were fed further investigation is warranted to clarify the mecha- the HFD diet (28.6% fat, Funabashi Farm, Japan) nistic basis in each case4). during the intervention period, which was restricted Recent large-scale GWAS and meta-analyses have to four weeks between 12 and 16 weeks of age, and identified >100 genomic loci contributing to inter- thereafter again fed the normal rat chow. The control individual variation in the plasma lipid concentra- rats (n =8) were fed the normal rat chow throughout tions1, 2). However, their practical value, in particular the study period. At two points, before and after the with regard to exact genetic elements and pathogenic dietary intervention, blood samples were drawn from events underlying associations, remains a subject of the tail vein of the rats in an overnight (16-hour) fast- debate. Focusing on human tissues relevant to lipo- ing state to measure the levels of glucose, total choles- protein metabolism (liver and fat—omental and sub- terol, triglycerides and free fatty acids. cutaneous fat), a total of 54 genes located at 35 unique Furthermore, in order to obtain clues regarding loci (i.e., 37% of 95 loci reported in a previous the involvement of uncharacterized genes in various study1)) have been shown to constitute SNP-to-gene lipid regulatory functions, we used male rats of a eQTLs, with partial overlap between tissues. Sepa- stroke-prone substrain of SHR (SHRSP) for pharma- rately, as an approach to systematically follow up cological intervention with atorvastatin, a lipid-lower- GWAS loci, RNAi-based functional profiling in vitro ing drug, as previously reported8). Briefly, the male has been performed in HeLa cells among 110 genes rats (n =5) were subjected to atorvastatin administra- derived from 43 lipid-associated loci (of which 33 tion (15 mg/kg/day via the gavage method) for four overlap with the list of 95 loci1)), subsequently identi- weeks between 12 and 16 weeks of age. The control fying an additional list of 45 genes from 25 lipid-asso- rats (n =5) were fed normal rat chow throughout the ciated loci, although overlap with eQTL data remains study period, similar to the HFD intervention. While rather limited6). a substantial portion of the predisposition to cardio- vascular disease is assumed to be shared between SHR and SHRSP animals, some features are relatively pro- Aim nounced or unique to each rat strain alone. SHR rats In order to provide evidence for disease-relevant are characterized by prominent hypertension as well as functions in vivo, we designed the present study with a conspicuous response to HFD. SHRSP rats, in con- two principal aims: [1] to investigate functional genes trast, are characterized by severe hypertension and car- at homologous loci identified in human lipid GWAS diovascular complications, including stroke and renal that respond to high-fat, high-cholesterol diet (HFD) dysfunction. Hence, we chose the individual strain to intervention in a rat model of metabolic syndrome; suit the nature of the intervention; that is, SHR for and [2] to examine the degree of overlap between the the dietary intervention and SHRSP for the pharma- list of genes with a differential expression induced by cological intervention, with a particular focus on end- HFD and those detectable using eQTL studies and/or organ protective effects7, 8). RNAi-based functional profiling, thereby integrating The rats were killed under pentobarbital anesthe- biological evidence for GWAS loci at the phenotypic sia (50 mg/kg via intraperitoneal infusion), and the Functional Genes for Lipid Regulation 457 organs were excised and immediately frozen at -70℃ tions. For both analyses, the probes were initially for subsequent RNA extraction. All rats were labora- selected if the fold induction was ≥ 1.3 with a nominal tory animals and treated in compliance with institu- P-value of <0.1. Such generous cut-off levels were set tional regulations. in order to avoid excluding probes for the functional All animal experiments were approved by the genes that exhibited relatively modest expression Animal Care and Use Committee of the National changes induced by HFD. In the first analysis, the Center for Global Health and Medicine (NCGM) corresponding probes were subjected to validation via Research Institute (permit number: 23-C-14) and qPCR, as described below. In the second analysis, the conducted in accordance with institutional proce- datasets were used for a pathway analysis with Ingenu- dures. ity iReport (Ingenuity Systems, Inc., CA). In this study, the gene annotation reports (Pathways, Biologi- Tissue Processing and RNA Isolation cal Processes and Disease) were based on insight from One gram each of the liver and retroperitoneal the Ingenuity® Knowledge Base (http://www.ingenu- adipose tissues were homogenized with a 0.5 inch- ity.com/). DAVID (ver. 6.7, http://david.abcc.ncifcrf. diameter generator shaft at a speed of 22,000 rpm for gov/)9) and g:Profiler (http://biit.cs.ut.ee/gprofiler/)10) 60 seconds. Total RNA was extracted using the were also used for functional profiling.