A missense variant in CST3 exerts a recessive effect on susceptibility to age-related macular degeneration resembling its association with Alzheimer's disease

Abstract

Age-related macular degeneration (AMD) and Alzheimer's disease (AD) are degenerative, multifactorial diseases involving age-related accumulation of extracellular deposits linked to dysregulation of protein homeostasis. Here, we strengthen the evidence that an nsSNP (p.Ala25Thr) in the cysteine proteinase inhibitor cystatin C gene CST3, previously confirmed by meta-analysis to be associated with AD, is associated with exudative AMD. To our knowledge, this is the first report highlighting a genetic variant that increases the risk of developing both AD and AMD. Furthermore, we demonstrate that the risk associated with the mutant allele follows a recessive model for both diseases. We perform an AMD-CST3 case-control study genotyping 350 exudative AMD Caucasian individuals. Bringing together our data with the previously reported AMD-CST3 association study, the evidence of a recessive effect on AMD risk is strengthened (OR = 1.89, P = 0.005). This effect closely resembles the AD-CST3 recessive effect (OR = 1.73, P = 0.005) previously established by meta-analysis. This resemblance is substantiated by the high correlation between CST3 genotype and effect size across the two diseases (R(2) = 0.978). A recessive effect is in line with the known function of cystatin C, a potent enzyme inhibitor. Its potency means that, in heterozygous individuals, a single functional allele is sufficient to maintain its inhibitory function; only homozygous individuals will lack this form of proteolytic regulation. Our findings support the hypothesis that recessively acting variants account for some of the missing heritability of multifactorial diseases. Replacement therapy represents a translational opportunity for individuals homozygous for the mutant allele.

Fig. 1

Odds ratios for CST3 genotypes at rs1064039 estimated for AD by meta-analysis and for AMD by a single association study. ORs are measured relative to the “GG” genotype, by definition this baseline genotype has an OR of 1. Error bars represent 95 % CIs

Fig. 2

The per-allele odds ratio (OR_N) decreases linearly with decreasing allele frequency (f _N) when the true model is recessive; the elevated risk of homozygotes is kept constant (OR_NN specified) and heterozygotes are at baseline risk (OR_NX = 1). This relationship can be expressed as: OR_N = f _N(OR_NN − 1) + 1. The single point represents CST3 rs1064039 with respect to AD

Fig. 3

Forest plots for the meta-analysis of CST3 rs1064039 with respect to exudative AMD in the Caucasian population using a fixed effects model. Size of the squares represents the weight of the study and horizontal bars represent 95 % CI of the OR. Applied to a “AA” genotype versus “GG” genotype and b “AG” genotype versus “GG” genotype

Fig. 4

Odds ratios for CST3 genotypes at rs1064039 estimated for AD and AMD meta-analyses. Note that the odds ratios are measured relative to the “GG” genotype, by definition this baseline genotype has an odds ratio of 1. Error bars represent 95 % CIs

Fig. 5

Pairwise linkage disequilibrium map of CST3 SNPs (maf >0.05) from a Caucasian sample (n = 503, from Phase 3 of the 1000 Genomes Project). Solid black squares represent pairs of SNPs in high LD (R ² > 0.9) as depicted by Haploview. Missense SNP highlighted in red, the two other SNPs in the PCR product highlighted in blue, and the SNP associated with plasma level of cystatin C highlighted in green (colour figure online)