Integrative analysis identifies copy number variations and their controlled causal molecular networks in Alzheimer’s disease

• Published in: Alzheimer's & dementia Vol. 16
• Main Author: Zhang, Bin
• Format: Journal Article
• Language: English
• Published: 01.12.2020
• DOI: 10.1002/alz.043341
• ISSN: 1552-5260

Background The pathogenic mechanism of late‐onset Alzheimer’s disease (LOAD) remains elusive despite intensive studies in the past few decades. Even though ‐27 AD risk loci have been identified through genome‐wide association studies, they explain only a small fraction of the heritability of LOAD. LOAD is such a complex disease that more types of genetic variations need to be studied. Copy number variation (CNV), as one primary type of genomic variation, has been found to play a significant role in many neurological diseases such as Parkinson’s disease, schizophrenia, mental retardation, and AD. A systematic functional study of CNVs through omics‐levels can provide valuable insights into the mechanism of LOAD. Method In this study, we employed four complementary CNV calling approaches including CNVnator, Pindel, MetaSV and Delly2 to comprehensively identify CNVs in the WGS data from 1,662 individuals (including 827 LOAD, 330 mild cognitive impairment, and 505 control subjects) across three LOAD cohorts including the Mount Sinai/JJ Peters VA Medical Center Brain Bank (MSBB) AD cohort, the ROSMAP cohort, and the Mayo Clinic cohort. Integrative analysis of the identified CNVs and the matched transcriptomic, proteomic, clinical and pathological data was then performed to determine the functional impacts of these CNVs. Result Genes associated with CNVs at the transcriptomic level in seven different brain regions were consistently enriched for glutathione metabolism and antigen processing and presentation pathways. The mRNA and protein expression levels of two genes (ACOT2, and GSTM1) in the frontal pole were significantly correlated with cis‐regulatory CNVs in AD cases. Rare LOAD specific CNVs were involved in axonogenesis and neurite morphogenesis. The LOAD cases had a higher disease‐specific CNV burden compared with the normal controls. Moreover, the causal networks based on the integration of CNV, transcriptomic, proteomic, and clinicopathological data demonstrated brain region‐specificity. Conclusion To our knowledge, this is the first genomic CNV study focused on LOAD by integrating genomic, transcriptomic, proteomic, and clinicopathologic data. The identified CNVs and their downstream molecular networks provide a blueprint for studying the pathogenic mechanisms of LOAD.