Fig 2: Sex-specific long-range linkage disequilibrium.Genotype correlation between the loci identified in CETP in Figure 3a and all SNPs with MAF >5% in ADCY9 for (a,b) the PEL population and (c,d) LIMAA cohort in males (a,c) and in females (b,d). Genotype frequencies per sex are shown in Figure 4—figure supplement 1 and sex-specific PBS values in Figure 4—figure supplement 2. The horizontal line shows the threshold for the 99th percentile of all comparisons of SNPs (MAF >5%) between ADCY9 and CETP. The vertical dotted line represents the position of rs1967309. Blue dots represent the rs158477 SNPs and pink represents the other three SNPs identified in Figure 3a (rs158480, rs158617, and rs12447620), which are in near-perfect LD. Figure 4—figure supplement 3 shows the same analysis in Andeans from NAGD. Gene plots for ADCY9 showing location of its exons are presented in blue below each plot. Figure 4—source data 1.R2 values of all SNPs between ADCY9 and CETP genes in the PEL population from 1000G and LIMAA cohort in male and female.This zip archive contains all files of r2 values obtained from the geno-r2 command of the vcftools software for all SNPs (MAF >5%) of the PEL population (files beginning by F4a [male] and F4b [female]) and the LIMAA cohort (files beginning by F4c [male] and F4d [female]) between ADCY9 and CETP genes stratified by sex. Scripts to create those figures can be found here: Gamache, 2021.

Fig 3: Sex-specific epistatic effects between rs1967309 and rs158477 on CETP expression depending on the number of sPEER factors in GTEx by tissue.P-values are presented on a -log10 scale. For all models, rs158477 is coded as additive (GG = 0, GA = 1, AA = 2). In the additive model (green triangle), rs1967309 is coded as additive (AA = 0, AG = 1, GG = 2), p-values are obtained using a linear regression in R. In the genotypic model (orange circle), rs1967309 is coded as a genotypic variable and p-values are obtained from a likelihood ratio test comparing models with and without the interaction term between the SNPs. The orange, red and pink lines represent p-values of 0.1, 0.05 and 0.01 respectively. The tissue type and the number of samples for each, used in the analysis, are reported in the titles of the subgraphs. The color of lines represents the sex label. Only tissues with at least one value under 0.10 are showed. Tissues with an asterisk (*) next to their title are tissues showing a the suggestive/significant effect in the sex-combined analysis.

Fig 4: Liver ABCA1 and ACADM protein amounts are increased in huCETP mice. (A) Immunoblots for protein expression of liver CETP, ABCA1, HMGCS2, and ACADM in huCETP and wild type littermates. β-ACTIN was used as the loading control and protein markers were indicated on the right side of the blots. Quantification of western blots for each protein shown in (B–D) Data shown are mean ± SEM, n = 8–9. Significant differences were determined by the Student’s t-test.

Fig 5: Epistatic effects between rs1967309 and rs158477 on CETP expression in GEUVADIS (LCL, N = 287) and CARTaGENE (Whole blood samples, N = 728).P-values are presented on a -log10 scale and are reported in function of the number of PEER/sPEER factors in GEUVADIS (LCL) (a,c) and CARTaGENE (b,d) in sex-combined (a,b) and sex-stratified (c,d) analyses. For all models, rs158477 is coded as additive (GG = 0, GA = 1, AA = 2). In the additive model (green triangle), rs1967309 is coded as additive (AA = 0, AG = 1, GG = 2), p-values are obtained using a linear regression in R. In the genotypic model (orange circle), rs1967309 is coded as a genotypic variable and p-values are obtained from a likelihood ratio test comparing models with and without the interaction term between the SNPs. The orange, red, and pink lines represent p-values of 0.1, 0.05, and 0.01 respectively. The sample sizes reported are the number of individuals left after removing participants with missing genotypes for rs1967309 and/or rs158477. In (c,d) the color of the lines represents the sex label.