Accurate prediction of the functional significance of single nucleotide polymorphisms and mutations in the ABCA1 gene

Abstract

The human genome contains an estimated 100,000 to 300,000 DNA variants that alter an amino acid in an encoded protein. However, our ability to predict which of these variants are functionally significant is limited. We used a bioinformatics approach to define the functional significance of genetic variation in the ABCA1 gene, a cholesterol transporter crucial for the metabolism of high density lipoprotein cholesterol. To predict the functional consequence of each coding single nucleotide polymorphism and mutation in this gene, we calculated a substitution position-specific evolutionary conservation score for each variant, which considers site-specific variation among evolutionarily related proteins. To test the bioinformatics predictions experimentally, we evaluated the biochemical consequence of these sequence variants by examining the ability of cell lines stably transfected with the ABCA1 alleles to elicit cholesterol efflux. Our bioinformatics approach correctly predicted the functional impact of greater than 94% of the naturally occurring variants we assessed. The bioinformatics predictions were significantly correlated with the degree of functional impairment of ABCA1 mutations (r2 = 0.62, p = 0.0008). These results have allowed us to define the impact of genetic variation on ABCA1 function and to suggest that the in silico evolutionary approach we used may be a useful tool in general for predicting the effects of DNA variation on gene function. In addition, our data suggest that considering patterns of positive selection, along with patterns of negative selection such as evolutionary conservation, may improve our ability to predict the functional effects of amino acid variation.

Figure 1. Comparison of subPSEC Scores for ABCA1 cSNPs, Mutations, Recently Described Variants in a Cohort of Individuals with Low HDL Cholesterol from the General Population [14], and a Random Distribution of Low Frequency Alleles

ABCA1 cSNPs (open circles) have significantly greater subPSEC scores than do mutations (filled squares) (p < 0.0001, Mann-Whitney U test). subPSEC scores for ABCA1 variants described in the general population (filled triangles) are significantly different from those of both ABCA1 cSNPs and mutations (p < 0.01, Mann-Whitney U test), as well as from the random distribution of ABCA1 variants (p < 0.001), indicating that this group of variants consists of both functional and neutral variants.

Figure 2. Conservation of ABCA1 Amino Acid Position 1091 in Related Proteins and Functional Effect of Mutation at This Site

(A) Multiple sequence alignment adapted from the view of family PTHR19229 available on the PANTHER Web site, showing the ABCA1, ABCA2, ABCA4, and ABCA7 subfamilies. Human ABCA1 position 1091 is highlighted in red; other conserved positions are highlighted in blue. (B) Cholesterol efflux was assessed in 293 cells stably transfected with wild-type, M1091T, M1091L, or M1091V ABCA1 alleles. *p < 0.001.

Figure 3. Correlation of Cholesterol Efflux Values with the Probability of a Functional Impairment (P_deleterious) for ABCA1 Mutations (filled squares) and SNPs (open circles)

PANTHER predictions are significantly correlated with the severity of impairment of ABCA1 mutations (r ² = 0.62, p = 0.0008) and of all ABCA1 variants (r ² = 0.56, p = 0.0004). The linear regression shown is for ABCA1 mutations.

Figure 4. Graphic Representation of the Evolutionary Relationship between Mouse, Human, and Chimpanzee ABCA1 Proteins

ABCA1 amino acid position 883 genotype is displayed under the species name. An “X” represents the likely point in evolutionary history at which the V883→M883 and M883→I883 mutation events occurred. The M883→I883 mutation likely occurred since the divergence of the last common ancestor between humans and chimpanzees, and the increased activity of the I883 allele suggests that this may have been one of the adaptive changes that occurred during the evolution of modern humans.