The ɛ4 allele of the human apolipoprotein E gene (APOE) is a well-proven genetic risk factor for the late onset form of Alzheimer's disease (AD). However, the biological mechanisms through which the ɛ4 allele contributes to disease pathophysiology are incompletely understood. The three common alleles of APOE, ɛ2, ɛ3 and ɛ4, are defined by two single nucleotide polymorphisms (SNPs) that reside in the coding region of exon 4, which overlaps with a well-defined CpG island (CGI). Both SNPs change not only the protein codon but also the quantity of CpG dinucleotides, primary sites for DNA methylation. Thus, we hypothesize that the presence of an ɛ4 allele changes the DNA methylation landscape of the APOE CGI and that such epigenetic alteration contributes to AD susceptibility. To explore the relationship between APOE genotype, AD risk, and DNA methylation of the APOE CGI, we applied bisulfite pyrosequencing and evaluated methylation profiles of postmortem brain from 15 AD and 10 control subjects. We observed a tissue-specific decrease in DNA methylation with AD and identified two AD-specific differentially methylated regions (DMRs), which were also associated with APOE genotype. We further demonstrated that one DMR was completely un-methylated in a sub-population of genomes, possibly due to a subset of brain cells carrying deviated APOE methylation profiles. These data suggest that the APOE CGI is differentially methylated in AD brain in a tissue- and APOE-genotype-specific manner. Such epigenetic alteration might contribute to neural cell dysfunction in AD brain.

We performed a genome-wide linkage scan using highly polymorphic microsatellite markers. To minimize genetic heterogeneity, we focused on sibpairs meeting the strict diagnosis of autism. In our primary analyses, we observed a strong linkage signal (P = 0.0006, 133.16 cM) on chromosome 7q at a location coincident with other linkage studies. When a more relaxed diagnostic criteria was used, linkage evidence at this location was weaker (P = 0.01). The sample was stratified into families with only male affected subjects (MO) and families with at least one female affected subject (FC). The strongest signal unique to the MO group was on chromosome 11 (P = 0.0009, 83.82 cM), and for the FC group on chromosome 4 (P = 0.002, 111.41 cM). We also divided the sample into regression positive and regression negative families. The regression-positive group showed modest linkage signals on chromosomes 10 (P = 0.003, 0 cM) and 14 (P = 0.005, 104.2 cM). More significant peaks were seen in the regression negative group on chromosomes 3 (P = 0.0002, 140.06 cM) and 4 (P = 0.0005, 111.41 cM). Finally, we used language acquisition data as a quantitative trait in our linkage analysis and observed a chromosome 9 signal (149.01 cM) of P = 0.00006 and an empirical P-value of 0.0008 at the same location. Our work provides strong conformation for an autism locus on 7q and suggestive evidence for several other chromosomal locations. Diagnostic specificity and detailed analysis of the autism phenotype is critical for identifying autism loci.