The SORL1 gene encodes for the protein SorLA, a sorting receptor involved in retromer-related endosomal traffic. Many SORL1 genetic variants increase Alzheimer’s disease (AD) risk, and rare loss-of-function truncation mutations have been found to be causal of late-onset AD. SORL1 is expressed in neurons and glia of the central nervous system and loss of SORL1 has been reported in AD tissue. To model the causal loss-of-function mutations, we used CRISPR/Cas9 technology to deplete SORL1 in human induced pluripotent stem cells (hiPSCs) to test the hypothesis that loss of SORL1 contributes to AD pathogenesis by leading to endosome dysfunction. We report that loss of SORL1 in hiPSC-derived neurons leads to early endosome enlargement, a cellular phenotype that is indicative of ‘traffic jams’ and is now considered a hallmark cytopathology AD. We validate defects in neuronal endosomal traffic by showing decreased localization of amyloid-precursor protein (APP) in the trans-Golgi network (TGN), and increased localization of APP in early endosomes, a site of APP cleavage by the β secretase BACE1. Microglia, immune cells of the CNS, which play a role in AD pathology also express SORL1. We therefore tested and found no effect of SORL1 depletion on endosome size or morphology in hiPSC-derived microglia, suggesting a selective effect on neuronal endosomal trafficking. Finally, because BACE1 dependent APP fragments can cause endosome enlargement, we treated SORL1 deficient hiPSC-derived neurons with BACE1 inhibitors and demonstrate that endosome enlargement occurs independent of amyloidogenic APP fragments. Collectively, these findings clarify where and how SORL1 links to AD. Moreover, our data, together with recent findings, underscores how sporadic AD pathways that regulate endosomal trafficking, and autosomal-dominant AD pathways that regulate APP cleavage, independently converge on the defining cytopathology of AD.

Alzheimer's disease (AD) is genetically, pathogenically and clinically heterogenous and likely involves multiple pathways and cell types. Many genetic variants associated with the development of AD map to genomic regions associated with genes known to function in the formation and transfer of endosomes. The molecular mechanisms by which the risk alleles influence AD pathogenesis is not known. We hypothesize that a higher load of endosomal pathway variants results in altered biology of neurons and microglia. Single nuclei RNA-seq and ATAC-seq analyses can be employed to identify epigenomic and transcriptomic dysregulation associated with high endosomal pathway genetic risk. Establishing a method to incorporate inherited genetic risk with cellular phenotype may help identify biological networks involved in AD pathogenesis. We developed an endosome function specific polygenic risk score (ePRS) for established AD risk alleles. Rapid autopsy brain tissue (post-mortem interval <8hrs) was collected through the University of Washington ADRC and samples with an ePRS from the highest and lowest quartiles were taken for analysis. Neuronal and myeloid cell nuclei were isolated using fluorescence activated cells sorting. The nuclei preparations were used to generate single nuclei transcriptomes and chromatin landscape maps for each cell type. Leptomeningeal cells from the same brain samples will be either trans-differentiated into neurons or reprogrammed to induced pluripotent stem cells followed by differentiation into neurons or microglia for functional studies. The ePRS correlates with presence of AD and allows the stratification of cases investigated with epigenomic and transcriptomic analysis, iPSC reprogramming, and subsequent neural cell differentiation. Integration of epigenetic and transcriptomic can identify cell type specific patterns associated with the presence of high or low ePRS. Identifying cellular phenotypes associated with high ePRS in neuropathologically confirmed AD is an important step in the development of a platform to screen potential therapies aimed at tailored modulation of the pathogenic impact of high ePRS. Leptomeningeal cell derived neurons and microglia can be used to interrogate AD associated cellular phenotypes identified by cell type specific epigenomic and transcriptomic analysis. Assessing impact of genetic risk loci on these measures may help establish mechanisms by which genetic variants confer increased risk for AD.

Endosomal abnormalities are documented in post-mortem AD brain tissue and multiple endocytic regulatory genes are associated with increased AD risk in population studies. SORL1 is a vesicular trafficking gene that functions in transporting cargo between endosomes, Golgi, lysosomes, and the plasma membrane. SORL1 plays an integral in trafficking amyloid beta and the amyloid precursor protein through the endocytic network and loss of SORL1 is documented in AD brain tissue. Previously, we have used human induced pluripotent stem cell (hiPSC)-derived neurons to show that deficiencies in SORL1 expression induction correlate with the presence of AD-associated variants in non-coding regions of SORL1. We are deriving neural cell types from hiPSCs with loss of SORL1 alleles or introduced SORL1 AD-associated risk variants using CRISPR/Cas9 gene-editing technology. We are also generating a cohort of hiPSC lines with high endosomal risk burden from autopsy confirmed AD cases. We assay endosome size, morphology, endosomal network function (recycling and degradation), and cellular AD phenotypes: Amyloid beta and phospho-tau. Autopsy-confirmed AD tissue from which we are generating hiPSCs shows increased genetic risk burden in endososmal genes. In gene-edited cell lines deficient in SORL1 we observed enlarged early endosomes and deficiencies in endocytic recycling, indicative of imbalance in endosomal trafficking. From SORL1 deficient and SORL1 risk variant cell lines we document increases in Ab peptides secreted by hiPSC-derived neurons. This work will investigate a functional genotype-phenotype relationship of genetic variants in the endosomal network, which is known to be disrupted early in AD pathogenesis. Investigating this driver of disease pathogenesis and how it relates to human genetic variation is critical in the development of new and precision treatments for AD.

OBJECTIVE:Autosomal-dominant familial Alzheimer disease (AD) is caused by by variants in presenilin 1 (PSEN1), presenilin 2 (PSEN2), and amyloid precursor protein (APP). Previously, we reported a rare PSEN2 frameshift variant in an early-onset AD case (PSEN2 p.K115Efs*11). In this study, we characterize a second family with the same variant and analyze cellular transcripts from both patient fibroblasts and brain lysates. METHODS:We combined genomic, neuropathological, clinical, and molecular techniques to characterize the PSEN2 K115Efs*11 variant in two families. RESULTS:Neuropathological and clinical evaluation confirmed the AD diagnosis in two individuals carrying the PSEN2 K115Efs*11 variant. A truncated transcript from the variant allele is detectable in patient fibroblasts while levels of wild-type PSEN2 transcript and protein are reduced compared to controls. Functional studies to assess biological consequences of the variant demonstrated that PSEN2 K115Efs*11 fibroblasts secrete less Aβ 1-40 compared to controls, indicating abnormal γ-secretase activity. Analysis of PSEN2 transcript levels in brain tissue revealed alternatively spliced PSEN2 products in patient brain as well as in sporadic AD and age-matched control brain. INTERPRETATION:These data suggest that PSEN2 K115Efs*11 is a likely pathogenic variant associated with AD. We uncovered novel PSEN2 alternative transcripts in addition to previously reported PSEN2 splice isoforms associated with sporadic AD. In the context of a frameshift, these alternative transcripts return to the canonical reading frame with potential to generate deleterious protein products. Our findings suggest novel potential mechanisms by which PSEN variants may influence AD pathogenesis, highlighting the complexity underlying genetic contribution to disease risk.

Heterozygous mutations in Presenilin 2 (PSEN2) cause nearly fully penetrant autosomal dominant Alzheimer Disease (AD). PSEN2, an essential component of the gamma-secretase complex of proteins, catalyzes a crucial step in the amyloidogenic amyloid precursor protein (APP) cleavage cascade to generate amyloid-beta (Aβ) peptides. Accumulation of aberrantly-produced insoluble Aβ isoforms is an important component of AD pathogenesis. While most driver PSEN2 mutations are missense, we have discovered a novel heterozygous PSEN2 two-basepair deletion frameshift mutation (PSEN2K115fsX) in two unrelated individuals with AD. To determine whether this mutation exhibits molecular hallmarks of AD, we generated induced pluripotent stem cells (iPSCs) from patient fibroblasts to study cell type-specific mutational effects. Following differentiation of control and mutant iPSCs into cortical neurons, we measured secreted Aβ isoforms in conditioned media and found that the ratio of insoluble to soluble Aβ is increased from mutant cells. To validate this PSEN2 mutation as a molecular driver of these observations, we are testing whether correction of the mutation in genome-edited isogenic iPSC lines rescues the Aβ phenotype. This frameshift mutation offers a unique way to probe the impact of major biochemical changes to the expression and function of one PSEN2 allele. We are analyzing the impact of mutation on PSEN2 RNA splice isoforms as well as exploring the greater role of PSEN2 AD mutations in human microglia through iPSC differentiation.