Amyloid precursor protein (APP) processing genes and cerebrospinal fluid APP cleavage product levels in Alzheimer's disease

Abstract

The aim of this exploratory investigation was to determine if genetic variation within amyloid precursor protein (APP) or its processing enzymes correlates with APP cleavage product levels: APPα, APPβ or Aβ42, in cerebrospinal fluid (CSF) of cognitively normal subjects or Alzheimer's disease (AD) patients. Cognitively normal control subjects (n = 170) and AD patients (n = 92) were genotyped for 19 putative regulatory tagging SNPs within 9 genes (APP, ADAM10, BACE1, BACE2, PSEN1, PSEN2, PEN2, NCSTN and APH1B) involved in the APP processing pathway. SNP genotypes were tested for their association with CSF APPα, APPβ, and Aβ42, AD risk and age-at-onset while taking into account age, gender, race and APOE ε4. After adjusting for multiple comparisons, a significant association was found between ADAM10 SNP rs514049 and APPα levels. In controls, the rs514049 CC genotype had higher APPα levels than the CA, AA collapsed genotype, whereas the opposite effect was seen in AD patients. These results suggest that genetic variation within ADAM10, an APP processing gene, influences CSF APPα levels in an AD specific manner.

Figure 1. Mean CSF APP Cleavage Product Levels

Each graph shows CSF APPα, APPβ, or Aβ42 mean level by group (cognitively normal controls versus AD patients), along with 95% confidence intervals. Each dot represents the CSF APP cleavage product level for one subject. AD patients have a significantly lower mean level than controls for CSF APPα (p = 0.049; 95%CI [0.01, 6.7]) (panel A), CSF APPβ levels (p = 0.028; 95% CI [0.2, 4.0]) (panel B) and CSF Aβ42 levels (p < 0.001; 95% CI [36, 54]) (panel C).

Figure 2. SNP Effect on CSF APP cleavage product levels

The effect of a SNP within APP processing genes on CSF APP cleavage product level for each disease group (AD versus controls) was analyzed using linear regression, where CSF APP cleavage product level was the outcome variable and the independent variables were age, gender, race, APOE ε4 status, disease status, SNP collapsed genotype (presence or absence of the minor allele; i.e., EF and FF vs. EE), and a disease status by SNP interaction term to allow the effect of SNP to possibly vary between disease groups. A confidence interval which does not cross the vertical line at zero indicates that the difference in genotypes is significant (p < 0.05) before adjustment for multiple comparisons for that group. A beta coefficient (solid square for AD, circle for controls) to the right of the vertical line represents higher CSF APP cleavage product levels for the collapsed genotype group that contains minor alleles (EF and FF). After correcting for multiple comparisons, the only significant SNP effect was for the ADAM10 rs514049 SNP, which showed a differential effect on CSF APPα levels in AD patients versus control subjects (asterisk; p = 0.014).

Figure 3. Effect of rs514049 Genotype on CSF APPα levels

Dots represent CSF APPα level for each subject and are grouped into rs514049 major genotype (CC) or collapsed genotype (CA, AA) groups for both controls and AD. For AD patients, the CA, AA collapsed genotype group has a mean CSF APPα level that is 6.0 nM greater than the mean for the CC genotype, whereas the CC genotype has higher levels in the control subjects. Significant differences between group CSF APPα level were tested with and without adjusting for gender, race, age and APOE ε4. Adjusted p-values are in parentheses.