The identification of high-penetrance variants in late-onset Alzheimer disease by whole-exome sequencing in extended families. M. A. Kohli1, B. W. Kunkle1, A. C. Naj2, L.-S. Wang2, K. L. Hamilton1, R. M. Carney1, W. R. Perry1, P. L. Whitehead1, J. R. Gilbert1, E. R. Martin1, G. W. Beecham1, R. P. Mayeux3, J. L. Haines4, L. A. Farrer5, G. D. Schellenberg2, S. Zuchner1, M. A. Pericak-Vance1, Alzheimers Disease Genetics Consortium (ADGC) 1) John P. Hussman Institute for Human Genomics, University of Miami, Miami, FL, USA; 2) Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; 3) Taub Institute of Research on Alzheimers Disease, Columbia University, New York, NY, USA; 4) Center for Human Genetics Research, Vanderbilt University, Nashville, TN, USA; 5) School of Medicine, Boston University, Boston, MA, USA.

   We chose 6 multi-generational, late-onset Alzheimer disease (LOAD) families to identify rare LOAD risk variants. The pedigrees are consistent with dominant inheritance, have an average of 6 AD-affected subjects, and are free of known AD mutations. We performed whole-exome sequencing (WES) on 4-9 affected subjects per pedigree, comprising 2-6 cousin pairs. For validation, we used the Alzheimers Disease Genetics Consortiums (ADGC) case-control dataset composed of 13,748 individuals (7,652 cases) genotyped on the Illumina Exome SNP Array (195,039 SNPs with Minor Allele Frequencies (MAF) <2%).
   We filtered WES-identified variants for potential high-penetrance AD variants by applying the following criteria: (1) segregation with LOAD allowing for one phenocopy per pedigree, (2) non-synonymous variants at evolutionarily conserved sites predicted to be deleterious, (3) MAF<1%, and (4) allelic case-control association odds ratio (OR) >2, if available, in the ADGC dataset. Nineteen variants passed these criteria, among which only 2 had ADGC data (MYO3B, PPP1R12A). We further genotyped 4 of the remaining variants in the ADGC dataset. Of these six variants, 3 were exclusively found in cases (in DYM, LONRF3 and TBC1D8B). The fourth reached an OR of 1.54 in an analysis excluding controls with age <80 (RTTN). Known functions of mentioned genes are actin-binding, regulation of actin-myosin interaction, brain development, and Rab GTPase activation. Of the remaining 13 genes, the androgen receptor (AR) has most often been related to AD. In a second approach, we did not require an OR>2 and lowered the segregation criterion to two phenocopies per pedigree. We obtained a total of 60 genes, of which five (CAPZA3, LONRF3, LRP1B, MYO3B, TBC1D17) showed nominally significant associations in gene-based tests in the ADGC data. Interestingly, TBC1D17 is another Rab GTPase-activating protein implicated in endocytosis, and LRP1B is a low-density lipoprotein receptor that allows endocytosis of a complex of ApoE-containing lipoprotein particles bound to -amyloid.
   These variants are extremely rare and may well qualify as high-penetrance AD variants. Most promising are completely segregating variants in genes that also obtained gene-based nominal significance (LONRF3, MYO3B), or have been functionally linked to AD (LRP1B).

You may contact the first author (during and after the meeting) at