Genetic Determinants of Human Serum Sterol Levels

Dozens of different cholesterol metabolites are synthesized by mammalian cells and are known to play important physiological roles in the liver, brain, and immune system. These metabolites are ultimately inactivated by conversion into bile acids in the liver and are thereafter excreted from the body. Mutations in several genes encoding enzymes that metabolize cholesterol have been identified and the clinical consequences of these mutations range from progressive central nervous system neuropathy to spastic paraplegia to liver failure in children. There are several genes in the cholesterol metabolic pathway in which human mutations and their clinical consequences have not yet been identified. For example, the metabolite 24-hydroxycholesterol is produced in the brain by cholesterol 24-hydroxylase (encoded by the CYP46A1 gene). Disruption of this gene in the mouse causes severe learning defects, but it is not known whether mutations in the human gene cause the same phenotype. To overcome the uncertainty inherent in guessing what the phenotype arising from mutations in the human CYP46A1 gene might be and to detect mutations in other genes specifying cholesterol metabolic enzymes, we developed mass-spectrometry based methods to measure >60 different sterol metabolites in small volumes (200 µl) of human serum. We applied these methods to quantify sterols in 3,230 serum samples derived from clinically well-characterized subjects participating in the Dallas Heart Study. Large intra-individual variation was detected in the serum levels of approximately 22 of the >60 sterols assessed. To identify the genes underlying the observed variation, we took both targeted and untargeted approaches. In the targeted approach, advantage was taken of the fact that the substrates and products of many sterol metabolizing enzymes are known and we thus sequenced the corresponding genes in subjects with very high or low levels of the sterol in question. In the untargeted approach, the 3,230 individuals whose sterol levels were measured were genotyped for 9,229 non-synonymous (amino acid-changing) variations (SNPs) in multiple human genes. For 24-hydroxycholesterol, the metabolite mentioned above, both approaches yielded concordant results. Direct sequencing of the oxysterol 7α-hydroxylase gene (CYP39A1), which uses 24-hydroxycholesterol as a substrate, identified several mutations that decrease enzyme activity and thus lead to increased serum levels of this sterol substrate. In the genetic linkage analysis, a SNP in CYP39A1 was strongly (P=2.1 x 10-27) correlated with serum 24-hydroxycholesterol levels. Expression analysis indicated that this SNP also decreased CYP39A1 enzyme activity. Together, these results illustrate that our approach has the ability to identify genetic determinants of serum sterol levels. Additional correlations were identified between a SNP in EPHX2 and levels of 24,25-epoxycholesterol (P=9.6 x 10-25) and a SNP in SDR42E1 and levels of 8-dehydrocholesterol levels (P=1.5 x 10-15). Biochemical assays were performed to characterize these previously unidentified genetic associations and to reveal their effect on oxysterol and sterol metabolism. Simultaneously, SNP genotypes and sterol levels were correlated with specific clinical phenotypes in an effort to shed light on the role of non-cholesterol sterols in disease. The next step is to apply these methods to the other 19 sterols that are routinely detected in serum and to correlate the analytical and genetic findings with the >100 clinical phenotypes of the 3,230 subjects whose sterol levels we have measured.