Rita Guimaraes

In this thesis, we investigate the role of repetitive DNA in the genetic architecture of Alzheimer's disease, with a focus on how structural variations in TRs and TEs may contribute to disease susceptibility. Although GWAS studies have identified numerous risk loci for AD, a significant portion of heritability remains unexplained. Our goal is to detect hidden genetic variations within AD-associated regions and assess their potential gene-regulatory functions by integrating genetic, epigenetic, and transcriptomic analyses. With this work, we hope to provide new insights into the genetic basis of AD.

In Chapter 1, we begin with a comprehensive overview of AD-associated loci, considering structural variants of repetitive DNA in each locus. Using a published GWAS, we performed a fine-mapping analysis and identified 29 genomic risk loci (GRL) associated with AD. Within the boundaries of the GRL, we integrated a dataset of structural variants and found 202 variants in 16 GRL. Notably, clusters of intergenic, GWAS-identified SNPs are located around structurally variable SVAs (SV-SVAs); therefore, we prioritize these loci for further functional analysis.

Chapter 2 investigates structural variation in TEs that have been shown to be active in a neuronal environment, and their association with neurological disorders. We used published long-read sequencing and PCR analysis to uncover the high degree of structural variation of SVA elements throughout the entire human genome. Given that a substantial number of SV-SVAs are located in loci associated with neurological disorders, we investigated the functional impact of these elements in 3 loci: BIN1 and CD2AP, associated with Alzheimer's, and BCKDK, associated with Parkinson's disease. Genetic deletions of the SV-SVAs were introduced in human embryonic stem cells using the CRISPR-Cas9 system, after which cortical organoids were generated as a model for a neuronal environment. Epigenetic features changed after deletion of the BCKDK-SV-SVA, in genes near the deletion site, and transcriptomic analysis showed differential expression of FUS, 75 kb away. No epigenetic or transcriptomic changes were observed in the BIN1-SVA-KO. Regarding the intronic CD2AP-SVA, RNA CaptureSeq was performed, which showed reduced expression levels for all CD2AP exons downstream of the SVA insertion site. In summary, our findings reveal a new layer of hidden genetic variation that must be taken into account when assessing the risk for complex diseases.

In Chapter 3, we continue the work from Chapter 1 by providing a functional assessment of two SV-SVAs located respectively in or upstream of the AD-associated genes PSEN1 and PICALM. Here we use the same method as described in Chapter 2: we generated SV-SVA KO cortical organoids and performed RNA CaptureSeq for a high resolution of the locus transcriptome. There was no evidence for a regulatory effect of the SV-SVA in the PICALM gene in the context of a neuronal environment. However, upon deletion of the intronic SV-SVA in PSEN1, preliminary findings showed increased expression levels of transcripts associated with nonsense-mediated decay.

In Chapter 4, we dive into a genetic association study to identify VNTRs associated with AD risk. We use a short-read sequencing dataset from the ADSP cohort to detect VNTRs whose length has increased or decreased in the majority of AD patients, or in a subgroup thereof, compared to controls. We discovered 52 VNTRs with a suggestive association with the disease, of which 5 were replicated in an independent long-read sequencing dataset consisting of AD patients and cognitively healthy centenarians. We highlight the most promising AD-associated VNTR, located in a regulatory element within the RTEL1 gene, in the promoter region of TNFRSF6B and in the fusion transcript RTEL1-TNFRSF6B. RTEL1 is an aging-related gene involved in cell homeostasis, while TNFRSF6B, which has an immunosuppressive function, is associated with AD. Where shorter lengths of the VNTR are associated with AD, longer lengths correlate with a protective GWAS SNP rs55765053 in TNFRSF6B, which in turn is in moderate LD with the main protective SNP rs6742. Further functional analysis is needed to validate our preliminary findings.