Heritability and family-based GWAS analyses of the N-acyl ethanolamine and ceramide plasma lipidome

Abstract

Signalling lipids of the N-acyl ethanolamine (NAE) and ceramide (CER) classes have emerged as potential biomarkers of cardiovascular disease (CVD). We sought to establish the heritability of plasma NAEs (including the endocannabinoid anandamide) and CERs, to identify common DNA variants influencing the circulating concentrations of the heritable lipids, and assess causality of these lipids in CVD using 2-sample Mendelian randomization (2SMR). Nine NAEs and 16 CERs were analyzed in plasma samples from 999 members of 196 British Caucasian families, using targeted ultra-performance liquid chromatography with tandem mass spectrometry. All lipids were significantly heritable (h2 = 36-62%). A missense variant (rs324420) in the gene encoding the enzyme fatty acid amide hydrolase (FAAH), which degrades NAEs, associated at genome-wide association study (GWAS) significance (P < 5 × 10-8) with four NAEs (DHEA, PEA, LEA and VEA). For CERs, rs680379 in the SPTLC3 gene, which encodes a subunit of the rate-limiting enzyme in CER biosynthesis, associated with a range of species (e.g. CER[N(24)S(19)]; P = 4.82 × 10-27). We observed three novel associations between SNPs at the CD83, SGPP1 and DEGS1 loci, and plasma CER traits (P < 5 × 10-8). 2SMR in the CARDIoGRAMplusC4D cohorts (60 801 cases; 123 504 controls) and in the DIAGRAM cohort (26 488 cases; 83 964 controls), using the genetic instruments from our family-based GWAS, did not reveal association between genetically determined differences in CER levels and CVD or diabetes. Two of the novel GWAS loci, SGPP1 and DEGS1, suggested a casual association between CERs and a range of haematological phenotypes, through 2SMR in the UK Biobank, INTERVAL and UKBiLEVE cohorts (n = 110 000-350 000).

Graphical Abstract

Figure 1

Schematic overview of the biosynthetic pathways for (A) N-acyl ethanolamines (NAEs) and (B) ceramides (CERs). (A) NAE species, including the endocannabinoid anandamide (AEA), are produced through four independent enzymatic pathways from the membrane phospholipid precursor (N-acyl phosphatidylethanolamine; NAPE). Fatty acid amide hydrolase (FAAH) degrades NAEs to free fatty acids (such as arachidonic acid for AEA) and ethanolamine. (B) CER species are biosynthesised via the enzyme serine palmitoyltransferase (SPTLC 1-3) that converts palmitoyl-CoA and L-serine to 3-keto dihydrosphingosine, in the rate-limiting step of the sphingolipid de novo pathway. The resulting dihydrosphingosine is coupled to various fatty acids via ceramide synthases (CERS) to generate dihydroceramides CER[NDS] that are further converted to CER[NS] via the enzyme delta 4-desaturase (DEGS1). Conversion of these pro-apoptotic CER[NS] species to sphingosine and sphingosine 1-phosphate, with roles in cell survival, degrades ceramides through reversible reactions. CER[NS] are also reversibly converted to sphingomyelin or further metabolized to ceramide 1-phosphate). Measured lipid species are in bold; genes encoding enzymes are in italics; genes identified through SNPs that associated at GWAS with circulating lipid levels are in red.

Figure 2

Heritability estimates of NAEs and CERs found in human plasma. This figure depicts the heritability estimated for each of 25 lipid species in 999 plasma samples using SNP-based GCTA software and reported pedigree-based QTDT software. These data are presented in detail in Supplementary Material, Table S8.

Figure 3

Family-based GWAS results for NAEs and the lead SNP in FAAH. (A) The forest plot depicts the effect (beta) and significance (GWAS P-value) for association between the lead SNP and eQTL of FAAH (rs324420) and each of 9 NAE species in 999 plasma samples. The P-values are grouped into ‘P < 5 × 10⁻⁸’, ‘P = 5 × 10⁻⁴–5.1 × 10⁻⁸’ and ‘P > 5 × 10⁻⁴’. (B) Trend in concentrations of plasma NAE species separated by FAAH rs324420 genotype. The figure depicts the mean standardized residuals of the NAE species in participants with the three genotypes at rs324420. The mean is shown with standard error. Fifty-one participants had the AA genotype, 310 had the AC genotype and 638 had the CC genotype in the cohort. (C) LocusZoom plot of the association of PEA with FAAH SNP rs324420. The LocusZoom plot depicts the association of NAE lipid species PEA with FAAH SNP rs324420 on chromosome 1 in 993 plasma samples. The r² for each SNP is depicted in colour. The plot was created using the LocusZoom plot tools at http://locuszoom.sph.umich.edu/.

Figure 3

Family-based GWAS results for NAEs and the lead SNP in FAAH. (A) The forest plot depicts the effect (beta) and significance (GWAS P-value) for association between the lead SNP and eQTL of FAAH (rs324420) and each of 9 NAE species in 999 plasma samples. The P-values are grouped into ‘P < 5 × 10⁻⁸’, ‘P = 5 × 10⁻⁴–5.1 × 10⁻⁸’ and ‘P > 5 × 10⁻⁴’. (B) Trend in concentrations of plasma NAE species separated by FAAH rs324420 genotype. The figure depicts the mean standardized residuals of the NAE species in participants with the three genotypes at rs324420. The mean is shown with standard error. Fifty-one participants had the AA genotype, 310 had the AC genotype and 638 had the CC genotype in the cohort. (C) LocusZoom plot of the association of PEA with FAAH SNP rs324420. The LocusZoom plot depicts the association of NAE lipid species PEA with FAAH SNP rs324420 on chromosome 1 in 993 plasma samples. The r² for each SNP is depicted in colour. The plot was created using the LocusZoom plot tools at http://locuszoom.sph.umich.edu/.

Figure 4

Family-based GWAS results for CER[NS] and precursor CER[NDS] with an exemplar SNP in serine palmitoyltransferase (SPTLC3). (A) The forest plot depicts the effect (beta) and significance (GWAS P-value) for association between the lead SNP and liver eQTL of SPTLC3 (rs680379) with the 13 CER[NS] and 3 CER[NDS] species in 999 plasma samples. The P-values are grouped into ‘P < 5 × 10⁻⁸’, ‘P = 5 × 10⁻⁴–5.1 × 10⁻⁸’ and ‘P > 5 × 10⁻⁴’. (B) Trend in concentrations of plasma CER species separated by SPTLC3 rs680379 genotype. The figure depicts the mean standardized residuals of the ceramide species in participants with the three genotypes at rs680379. The mean is shown with standard error. Four hundred nine participants had the GG genotype, 442 had the AG genotype and 148 had the AA genotype in the cohort. (C) LocusZoom plot of the association of CER[N(24)S(19)] with SPTLC3 SNP rs680379. The LocusZoom plot depicts the association of CER[N(24)S(19)] with SPTLC3 SNP rs680379 on chromosome 20 in 991 plasma samples. Although there is a group of lead SNPs, this SNP was depicted as it has been identified previously to associate at GWAS with sphingolipid species. The r² for each SNP is depicted in colour. The plot was created using the LocusZoom plot tools at http://locuszoom.sph.umich.edu/.

Figure 4

Family-based GWAS results for CER[NS] and precursor CER[NDS] with an exemplar SNP in serine palmitoyltransferase (SPTLC3). (A) The forest plot depicts the effect (beta) and significance (GWAS P-value) for association between the lead SNP and liver eQTL of SPTLC3 (rs680379) with the 13 CER[NS] and 3 CER[NDS] species in 999 plasma samples. The P-values are grouped into ‘P < 5 × 10⁻⁸’, ‘P = 5 × 10⁻⁴–5.1 × 10⁻⁸’ and ‘P > 5 × 10⁻⁴’. (B) Trend in concentrations of plasma CER species separated by SPTLC3 rs680379 genotype. The figure depicts the mean standardized residuals of the ceramide species in participants with the three genotypes at rs680379. The mean is shown with standard error. Four hundred nine participants had the GG genotype, 442 had the AG genotype and 148 had the AA genotype in the cohort. (C) LocusZoom plot of the association of CER[N(24)S(19)] with SPTLC3 SNP rs680379. The LocusZoom plot depicts the association of CER[N(24)S(19)] with SPTLC3 SNP rs680379 on chromosome 20 in 991 plasma samples. Although there is a group of lead SNPs, this SNP was depicted as it has been identified previously to associate at GWAS with sphingolipid species. The r² for each SNP is depicted in colour. The plot was created using the LocusZoom plot tools at http://locuszoom.sph.umich.edu/.