Multiple rare variants in NPC1L1 associated with reduced sterol absorption and plasma low-density lipoprotein levels

Abstract

An approach to understand quantitative traits was recently proposed based on the finding that nonsynonymous (NS) sequence variants in certain genes are preferentially enriched at one extreme of the population distribution. The NS variants, although individually rare, are cumulatively frequent and influence quantitative traits, such as plasma lipoprotein levels. Here, we use the NS variant technique to demonstrate that genetic variation in NPC1L1 contributes to variability in cholesterol absorption and plasma levels of low-density lipoproteins (LDLs). The ratio of plasma campesterol (a plant sterol) to lathosterol (a cholesterol precursor) was used to estimate relative cholesterol absorption in a population-based study. Nonsynonymous sequence variations in NPC1L1 were five times more common in low absorbers (n = 26 of 256) than in high absorbers (n = 5 of 256) (P < 0.001). The rare variants identified in low absorbers were found in 6% of 1,832 African-Americans and were associated with lower plasma levels of LDL cholesterol (LDL-C) (96 +/- 36 mg/dl vs. 105 +/- 36 mg/dl; P = 0.005). These data, together with prior findings, reveal a genetic architecture for LDL-C levels that does not conform to current models for quantitative traits and indicate that a significant fraction of genetic variance in LDL-C is due to multiple alleles with modest effects that are present at low frequencies in the population.

Fig. 1.

Plasma Ca:L ratios and mean plasma levels of LDL-C in African-American men and women in the Dallas Heart Study. Sequence variants found in only the low-absorber group were assayed by using 5′ nucleotidase assays (TaqMan, Applied Biosystems). All individuals in whom the NPC1L1 sequence variants were initially identified by sequencing and all study participants taking lipid-lowering medications were excluded from the analysis. Filled bars represent the median Ca:L ratios and mean (±SEM) levels of LDL-C in women (n = 66) and men (n = 39) who had at least one NS sequence variant in NPC1L1 identified in only the low absorber group. Hatched bars represent individuals who did not have one of these alleles (926 women and 674 men). ∗, P < 0.05; ∗∗, P < 0.01.

Fig. 2.

NS sequence variants in NPC1L1. (A) The membrane topology of NPC1L1 was predicted by using the program phd (31) and the model published for NPC1 (32). The positions of sequence variants found exclusively in low absorbers or in high absorbers (indicated with an asterisk) are provided. (B) Sequence conservation of NS sequence variants in NPC1L1 found in only low absorbers (Top), high absorbers (Middle), or both high and low absorbers (Bottom). The alignment includes human (hs), chimpanzee (pt), rhesus monkey (rm), dog (cf), cow (bt), mouse (mm), rat (rt), opossum (md), frog (xt), pufferfish (fr), and zebrafish (dr). The GenBank accession no. for the human sequence in this alignment is AAS56939.

Fig. 3.

Associations between common sequence variants in NPC1L1 and plasma Ca:L ratios. A total of 30 common sequence variants in NPC1L1 (minor allele frequency ≥0.01) were genotyped in the Dallas Heart Study population by using chip-based oligonucleotide hybridization and 5′ nucleotidase assays and tested for association with plasma Ca:L ratios by using ANOVA. Ca:L ratios were log transformed before analysis, and separate analyses were performed in white men (Upper , filled circles), white women (Upper, filled circles), black men (Lower, filled circles), and black women (Lower, open circles). SNPs associated with increased Ca:L ratios are shown above the solid lines, and SNPs associated with lower Ca:L rations are shown below the solid lines. Dotted lines represent P values of 0.05. Haplotype blocks (gray blocks) were constructed in whites (Upper) and blacks (Lower) by using the expectation-maximization algorithm as implemented in haploview (12) and tested for association with Ca:L ratios by using haplo.stats (17). Schematic of NPC1L1 is shown to scale, with boxes indicating exons and lines introns and flanking noncoding sequences. The identification numbers, minor allele frequencies, and the Hardy–Weinberg equilibrium P values are provided in Table 4.

Fig. 4.

Haplotype-sharing among NPC1L1 alleles. Haplotypes for the alleles bearing the NS variants were determined in the African-Americans in the Dallas Heart Study by using the partition-ligation method described in ref. and then manually curated to maximize the extent of haplotype-sharing among carriers. Yellow blocks represent the largest haplotype that could be shared among carriers. Numbers next to the amino acid substitutions indicate the number of individuals in each group. High and low Ca:L ratios denote sequence variants found in both the high and the low Ca:L groups. The identification numbers, minor allele frequencies, and the Hardy–Weinberg equilibrium P values are provided as Table 4.