Apolipoprotein E4 (APOE4) is the strongest risk allele associated with the development of late onset Alzheimer’s disease (AD). Across the CNS, astrocytes are the predominant expressor of APOE while also being critical mediators of neuroinflammation and cerebral metabolism. APOE4 has been consistently linked with dysfunctional inflammation and metabolic processes, yet insights into the molecular constituents driving these responses remain unclear. Utilizing complementary approaches across humanized APOE mice and isogenic human iPSC astrocytes, we demonstrate that ApoE4 alters the astrocyte immunometabolic response to pro-inflammatory stimuli. Our findings show that ApoE4-expressing astrocytes acquire distinct transcriptional repertoires at single-cell and spatially-resolved domains, which are driven in-part by preferential utilization of the cRel transcription factor. Further, inhibiting cRel translocation in ApoE4 astrocytes abrogates inflammatory-induced glycolytic shifts and in tandem mitigates production of multiple pro-inflammatory cytokines. Altogether, our findings elucidate novel cellular underpinnings by which ApoE4 drives maladaptive immunometabolic responses of astrocytes.

Abstract Clonal hematopoiesis (CH) is the presence of clonal populations of hematopoietic stem cells, which arise when a cell acquires a somatic mutation and then undergoes clonal expansion. The presence of CH is strongly associated with age. Clonal hematopoiesis of indeterminate potential (CHIP) mutations is defined by having a mutation in a gene commonly altered in blood cancers. CHIP is associated with increased risk of illnesses such as blood cancer and cardiovascular diseases. In addition to CHIP mutations, other somatic mutations can drive CH. To comprehensively characterize CH dynamics, we conducted whole genome sequencing (WGS) of blood samples from a community-based brain aging cohort, sequencing multiple samples that were collected across multiple time points. The University of Kentucky (UK) Alzheimer's Disease Research Center clinical cohort comprises a continuously replenished group of approximately 800 participants followed longitudinally. We selected 12 participants for a pilot study using the following criteria: each individual had 3 blood draws, and a minimum of 2 years between draws. WGS for these samples was performed at the Genomics Core Laboratory at UK. For all participants the age at first blood draw ≤ 89 years (median age: 80.5, range: 66-89), age at last blood draw ≤ 95 (median age: 87, range: 80 - 95). The time in years between first and last blood draw ranged from 6 to 12 years. 75% of participants were female and 25% were male. The mean average coverage was 54.4X. Alignment and variant calling were performed using the Illumina DRAGEN Somatic pipeline. Post pipeline filtering steps include filtering out variants in low complexity and blacklisted regions and removal of putative germline variants. Using stringent quality control metrics, 3/12 (25%) individuals had CHIP in one of the 74 canonical CHIP genes. Of these participants, 2 had DNMT3A CHIP. One participant had DNMT3A CHIP at age 87 (last blood draw), but not at age 80 or 83. For the other participant with DNMT3A CHIP, the mutation was present in all 3 blood draws from age 68, 77, and 80 years and with mutant allele fraction (MAF) ranging between 10-16%. TET2 CHIP was detected in one participant at age 90 (MAF 10%) and 95 (MAF 15%), but not at age 84. Our observations support previous studies showing that CHIP is common in later life. This pilot study provides an opportunity to conduct more in-depth investigations of clonal dynamics in later life and expand these investigations to non-canonical CH driver mutations and mosaic chromosomal alterations. We are pursuing this opportunity and leveraging the longitudinal study design to develop novel approaches for distinguishing germline and somatic variants to improve methods for CHIP mutation calling. Citation Format: Rohini Chebbi, Steven Estus, Elif P. Coskun, David W. Fardo, Gregory A. Jicha, Peter T. Nelson, Erin L. Abner, Yasminka A. Jakubek, . Longitudinal whole genome sequencing to investigate clonal hematopoiesis dynamics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1511.

Following stroke, B cells enter brain regions outside of the ischemic injury to mediate functional recovery. Although B cells produce neurotrophins that support remote plasticity, including brain-derived neurotrophic factor (BDNF), it remains unclear which signal(s) activate B cells in the absence of infarct-localized pro-inflammatory cues. Activation of N-methyl-d-aspartate (NMDA)-type receptor (NMDAR) subunits on neurons can upregulate mature BDNF (mBDNF) production from a pro-BDNF precursor, but whether this occurs in B cells is unknown. We identified GluN2A and GluN2B NMDAR subunits on B cells that respond to glutamate and mediate nearly half of the glutamate-induced Ca2+ responses in activated B cell subsets. Ischemic stroke recruits GluN2A+ B cells into the ipsilesional hemisphere and both stroke and neurophysiologic levels of glutamate regulate gene and surface expression. Regardless of injury, pro-BDNF+ B cells localize to spleen/circulation whereas mBDNF+ B cells localize to the brain, including in aged male and female mice. We confirmed B cell-derived BDNF was required for in vitro and in vivo B cell-mediated neuroprotection. Lastly, GluN2A, GluN2B, glutamate-induced Ca2+ responses, and BDNF expression were all clinically confirmed in B cells from healthy donors, with BDNF+ B cells present in post-stroke human parenchyma. These data suggest that B cells express functional NMDARs that respond to glutamate, enhance NMDAR signaling with activation, and upregulate mature BDNF expression within the brain. This study identifies potential glutamate-induced neurotrophic roles for B cells in the brain; an immune response to neurotransmitters unique from established pro-inflammatory stimuli and relevant to any CNS-localized injury or disease.

PLCG2 is associated with the risk of Alzheimer’s disease (AD) through a rare missense polymorphism, rs72824905 (P522R) as well as a common variant, rs12446759, within a long non-coding RNA adjacent to PLCG2. Elucidating the impact of genetics on PLCG2 expression and splicing will provide insights into the role of PLCG2 in AD risk and, potentially, treatments that might reduce AD risk. To evaluate PLCG2 expression and splicing as a function of AD genetics. PLCG2 isoform expression was detected by PCR and quantified by qRT-PCR in AD and non-AD brain samples and in blood buffy coat samples. The function of a genetic variant, rs1071644, was tested by using a minigene approach with both alleles in murine BV-2 microglial cells. The effects of ectopic splicing factor expression on PLCG2 minigene splicing were also compared in BV-2 cells. The extent that endogenous levels of a novel PLCG2 mRNA isoform lacking 65 bp within exon 28 (D65-PLCG2) was affected by nonsense mediated decay (NMD) was determined by using cycloheximide in vitro. Lastly, whether D65-PLCG2 manifested a Ca+2 response similar to PLCG2 was tested by comparing D65-PLCG2-GFP and PLCG2-GFP fusion proteins in transfected HEK293 cells. We report PLCG2 isoforms that include (i) a transcript that replaces PLCG2 exon 1 with exons from an adjacent long noncoding (LNC) RNA (LNC-PLCG2) and (ii) a transcript that lacks the initial 65 bp of exon 28 (D65-PLCG2). The ratio of LNC-PLCG2 to canonical PLCG2 was associated with rs12446759 genotype in both human brain and buffy coat samples. The proportion of PLCG2 expressed as D65-PLCG2 was increased by the T allele of rs1071644, a T/C SNP within the 65 bp variably spliced portion of exon 28. This SNP was demonstrated to be functional in a minigene splicing assay. Moreover, the rs1071644-T allele was found to be associated with increased AD risk, independent of rs72824905 (P522R) and rs12446759. D65-PLCG2 was susceptible to nonsense mediated RNA decay. In contrast to the canonical PLCG2 protein, D65-PLCG2 was not responsive to an increase in cytosolic Ca2⁺, likely due to the absence of the Ca2⁺-binding domain in this isoform. Hence, the rs1071644-T allele appears to increase AD risk by increasing the proportion of PLCG2 expressed as D65-PLCG2, representing a loss of PLCG2 function. We report that two AD genetic risk factors, rs12446759 and rs1071644, affect AD risk by impacting the LNC-PLCG2 to PLCG2 ratio and PLCG2 exon 28 splicing, respectively.

Traumatic brain injury (TBI) increases gastrointestinal morbidity and associated mortality. Clinical and preclinical studies implicate gut dysbiosis as a consequence of TBI and an amplifier of brain damage. However, little is known about the association of gut dysbiosis with structural and functional changes of the gastrointestinal tract after an isolated TBI. To assess gastrointestinal dysfunction, mice received a controlled cortical impact or sham brain injury and intestinal permeability was assessed at 4 h, 8 h, 1 d, and 3 d after injury by oral administration of 4 kDa FITC Dextran prior to euthanasia. Quantification of serum fluorescence revealed an acute, short-lived increase in permeability 4 h after TBI. Despite transient intestinal dysfunction, no overt morphological changes were evident in the ileum or colon across timepoints from 4 h to 4 wks post-injury. To elucidate the timeline of microbiome changes after TBI, 16 s gene sequencing was performed on DNA extracted from fecal samples collected prior to and over the first month after TBI. Differential abundance analysis revealed that the phylum Verrucomicrobiota was increased at 1, 2, and 3 d after TBI. The Verrucomicrobiota species was identified by qPCR as Akkermansia muciniphila, an obligate anaerobe that resides in the intestinal mucus bilayer and produces short chain fatty acids (e.g. butyrate) utilized by intestinal epithelial cells. We postulated that TBI promotes intestinal changes favorable for the bloom of A. muciniphila. Consistent with this premise, the relative area of mucus-producing goblet cells in the medial colon was significantly increased at 1 d after injury, while colon hypoxia was significantly increased at 3 d. Our findings reveal acute gastrointestinal functional changes coupled with an increase of beneficial bacteria suggesting a potential compensatory response to systemic stress after TBI.

The single nucleotide polymorphisms rs35349669 and rs10933431 within Inositol Polyphosphate-5-Phosphatase D (INPP5D) are strongly associated with Alzheimer's Disease risk. To better understand INPP5D expression in the brain, we investigated INPP5D isoform expression as a function of rs35349669 and rs10933431, as well as Alzheimer's disease neuropathology, by qPCR and isoform-specific primers. In addition, INPP5D allelic expression imbalance was evaluated relative to rs1141328 within exon 1. Expression of INPP5D isoforms associated with transcription start sites in exon 1 and intron 14 was increased in individuals with high Alzheimer's disease neuropathology. In addition, a novel variant with 47bp lacking from exon 12 increased expression in Alzheimer's Disease brains, accounting for 13% of total INPP5D expression, and was found to undergo nonsense-mediated decay. Although inter-individual variation obscured a possible polymorphism effect on INPP5D isoform expression as measured by qPCR, rs35349669 was associated with rs1141328 allelic expression imbalance, suggesting that rs35349669 is significantly associated with full-length INPP5D isoform expression. In summary, expression of INPP5D isoforms with start sites in exon 1 and intron 14 are increased in brains with high Alzheimer's Disease neuropathology, a novel isoform lacking the phosphatase domain was significantly increased with the disease, and the polymorphism rs35349669 correlates with allele-specific full-length INPP5D expression.

Background:Understanding the mechanisms whereby genetic variants influence the risk of Alzheimer's disease (AD) may provide insights into treatments that could reduce AD risk.Objective:Here, we sought to test the hypothesis that a single nucleotide polymorphism (SNP) associated with AD risk, rs2070902, influences splicing of FCER1G exon 2.Methods:AD and non-AD brain samples were analyzed for FCER1G expression by genotyping, immunohistochemistry, immunofluorescence, and qPCR.Results:The protein encoded by FCER1G, FcRγ, is robustly expressed in microglia in both AD and non-AD brain. The FCER1G isoform lacking exon 2 (D2-FCER1G) was readily detectable. Moreover, the proportion of FCER1G expressed as this isoform was increased in brains with high AD neuropathology. However, the proportion of FCER1G expressed as the D2-FCER1G isoform was not associated with rs2070902 genotype.Conclusions:In summary, the proportion of FCER1G expressed as the D2-FCER1G isoform is increased with AD neuropathology but is not associated with rs2070902.

ABSTRACT Apolipoprotein E (ApoE) is a lipid transport protein that is hypothesized to suppress proinflammatory cytokine production, particularly after stimulation with Toll-like receptor (TLR) ligands such as lipopolysaccharide (LPS). Studies using transgenic ApoE human replacement mice (APOE) expressing one of three different allelic variants suggest that there is a hierarchy in terms of responsiveness to proinflammatory stimuli such as APOE4/E4 > APOE3/E3 > APOE2/E2. In this study, we test the hypothesis that APOE genotype can also predict susceptibility to infection with the facultative intracellular gram-positive bacterium Listeria monocytogenes . We found that bone-marrow-derived macrophages isolated from aged APOE4/E4 mice expressed elevated levels of nitric oxide synthase 2 and were highly resistant to in vitro infection with L. monocytogenes compared to APOE3/E3 and APOE2/E2 mice. However, we did not find statistically significant differences in cytokine or chemokine output from either macrophages or whole splenocytes isolated from APOE2/E2, APOE3/E3, or APOE4/E4 mice following L. monocytogenes infection. In vivo , overall susceptibility to foodborne listeriosis also did not differ by APOE genotype in either young (2 mo old) or aged (15 mo old) C57BL/6 mice. However, we observed a sex-dependent susceptibility to infection in aged APOE2/E2 male mice and a sex-dependent resistance to infection in aged APOE4/E4 male mice that was not present in female mice. Thus, these results suggest that APOE genotype does not play an important role in innate resistance to infection with L. monocytogenes but may be linked to sex-dependent changes that occur during immune senescence.

Background: Genome-wide association studies have identified BIN1 within the second most significant susceptibility locus in late-onset Alzheimer's disease (AD). BIN1 undergoes complex alternative splicing to generate multiple isoforms with diverse functions in multiple cellular processes including endocytosis and membrane remodeling. An increase in BIN1 expression in AD and an interaction between BIN1 and Tau have been reported. However, disparate descriptions of BIN1 expression and localization in the brain previously reported in the literature and the lack of clarity on brain BIN1 isoforms present formidable challenges to our understanding of how genetic variants in BIN1 increase the risk for AD.Methods: In this study, we analyzed BIN1 mRNA and protein levels in human brain samples from individuals with or without AD. In addition, we characterized the BIN1 expression and isoform diversity in human and rodent tissue by immunohistochemistry and immunoblotting using a panel of BIN1 antibodies.Results: Here, we report on BIN1 isoform diversity in the human brain and document alterations in the levels of select BIN1 isoforms in individuals with AD. In addition, we report striking BIN1 localization to white matter tracts in rodent and the human brain, and document that the large majority of BIN1 is expressed in mature oligodendrocytes whereas neuronal BIN1 represents a minor fraction. This predominant non-neuronal BIN1 localization contrasts with the strict neuronal expression and presynaptic localization of the BIN1 paralog, Amphiphysin 1. We also observe upregulation of BIN1 at the onset of postnatal myelination in the brain and during differentiation of cultured oligodendrocytes. Finally, we document that the loss of BIN1 significantly correlates with the extent of demyelination in multiple sclerosis lesions.Conclusion: Our study provides new insights into the brain distribution and cellular expression of an important risk factor associated with late-onset AD. We propose that efforts to define how genetic variants in BIN1 elevate the risk for AD would behoove to consider BIN1 function in the context of its main expression in mature oligodendrocytes and the potential for a role of BIN1 in the membrane remodeling that accompanies the process of myelination.

Background: Understanding the mechanisms whereby genetic variants influence the risk of Alzheimer’s disease (AD) may provide insights into treatments that could reduce AD risk. Objective: Here, we sought to test the hypothesis that a single nucleotide polymorphism (SNP) associated with AD risk, rs2070902, influences splicing of FCER1G exon 2. Methods: AD and non-AD brain samples were analyzed for FCER1G expression by genotyping, immunohistochemistry, immunofluorescence, and qPCR. Results: The protein encoded by FCER1G, FcR γ , is robustly expressed in microglia in both AD and non-AD brain. The FCER1G isoform lacking exon 2 ( D2-FCER1G ) was readily detectable. Moreover, the proportion of FCER1G expressed as this isoform was increased in brains with high AD neuropathology. However, the proportion of FCER1G expressed as the D2-FCER1G isoform was not associated with rs2070902 genotype. Conclusions: In summary, the proportion of FCER1G expressed as the D2-FCER1G isoform is increased with AD neuropathology but is not associated with rs2070902.

Background Genetic variants in TREM2 are strongly associated with Alzheimer’s Disease (AD) risk but alternative splicing in TREM2 transcripts has not been comprehensively described. Objective Recognizing that alternative splice variants can result in reduced gene expression and/or altered function, we sought to fully characterize splice variation in TREM2. Methods Human blood and anterior cingulate autopsy tissue from 61 donors were used for end-point and quantitative PCR and Western blotting to identify and quantify novel TREM2 isoforms. Results In addition to previously described transcripts lacking exon 3 or exon 4, or retaining part of intron 3, we identified novel isoforms lacking exon 2, along with isoforms lacking multiple exons. Isoforms lacking exon 2 were predominant at approximately 10% of TREM2 mRNA in the brain. Expression of TREM2 and frequency of exon 2 skipping did not differ between AD samples and non-AD controls (p = 0.1268 and p = 0.4909, respectively). Further, these novel splice isoforms were also observed across multiple tissues with similar frequency (range 5.3 – 13.0%). We found that the exon 2 skipped isoform D2-TREM2 is translated to protein and localizes similarly to full-length TREM2 protein, that both proteins are primarily retained in the Golgi complex, and that D2-TREM2 is expressed in AD and non-AD brain. Conclusion Since the TREM2 ligand binding domain is encoded by exon 2, and skipping this exon retains reading frame while conserving localization, we hypothesize that D2-TREM2 acts as an inhibitor of TREM2 and targeting TREM2 splicing may be a novel therapeutic pathway for AD.

Apolipoprotein E ( APOE ) alleles impact pathogenesis and risk for multiple human diseases, making them primary targets for disease treatment and prevention. Previously, we and others reported an association between APOE alleles and the gut microbiome. Here, we evaluated effects of APOE heterozygosity and tested whether these overall results extended to mice maintained under ideal conditions for microbiome analyses. To model human APOE alleles, this study used APOE targeted replacement (TR) mice on a C57Bl/6 background. To minimize genetic drift, homozygous APOE3 mice were crossed to homozygous APOE2 or homozygous APOE4 mice prior to the study, and the resulting heterozygous progeny crossed further to generate the study mice. To maximize environmental homogeneity, mice with mixed genotypes were housed together and used bedding from the cages was mixed and added back as a portion of new bedding. Fecal samples were obtained from mice at 3-, 5- and 7-months of age, and microbiota analyzed by 16S ribosomal RNA gene amplicon sequencing. Linear discriminant analysis of effect size (LefSe) identified taxa associated with APOE status, depicted as cladograms to show phylogenetic relatedness. The influence of APOE status was tested on alpha-diversity (Shannon H index) and beta-diversity (principal coordinate analyses and PERMANOVA). Individual taxa associated with APOE status were identified by classical univariate analysis. Whether findings in the APOE mice were replicated in humans was evaluated by using published microbiome genome wide association data. Cladograms revealed robust differences with APOE in male mice and limited differences in female mice. The richness and evenness (alpha-diversity) and microbial community composition (beta-diversity) of the fecal microbiome was robustly associated with APOE status in male but not female mice. Classical univariate analysis revealed individual taxa that were significantly increased or decreased with APOE, illustrating a stepwise APOE2-APOE3–APOE4 pattern of association with heterozygous animals trending as intermediate in the stepwise pattern. The relative abundance of bacteria from the class Clostridia, order Clostridiales, family Ruminococacceae and related genera increased with APOE2 status. The relative abundance of Erysipelotrichia increased with APOE4 status, a finding that extended to humans. In this study, wherein mice were maintained in an ideal fashion for microbiome studies, gut microbiome profiles were strongly and significantly associated with APOE status in male APOE- TR mice. Erysipelotrichia are increased with APOE4 in both mice and humans. APOE allelic effects appeared generally intermediate in heterozygous animals. Further evaluation of these findings in humans, as well as studies evaluating the impact of the APOE -associated microbiota on disease-relevant phenotypes, will be necessary to determine if alterations in the gut microbiome represent a novel mechanism whereby APOE alleles impact disease.

BackgroundEthnic discrimination is frequently experienced among U.S. Latinx communities, and is linked to CVD risk factors, such as depression. Genetic variants may influence this relationship.ObjectivesThe objectives of this study were to examine associations between experiences of discrimination, rs4680 genotype, and depressive symptoms in Latinx adults.MethodsWe analyzed data from 124 Latinx adults with two or more CVD risk factors, and conducted hierarchical linear regression, adjusting for sex, age, income, education, and acculturation.ResultsParticipants were predominately female (74.2%) and aged 40.2 ± 9.3 years. More experiences of discrimination were associated with higher depressive symptoms (p = 0.041). Those with Met-Met-and Val-Met-genotypes had increased depressive symptoms than those with Val-Val-genotype (p = 0.049). Rs4680 was not a moderator.ConclusionFindings suggest discrimination and rs4680 genotype are associated with depressive symptoms in Latinx adults, which may increase CVD risk. Further research is needed to better understand biological mechanisms of these relationships.

Elucidating the impact of the gut microbiome on Alzheimer's Disease (AD) is an area of intense interest. Short chain fatty acids (SCFAs) are major microbiota metabolites that have been implicated as a mediator of gut microbiome effects in the brain. Here, we tested the effects of SCFA-treated water vs. saline-treated water on APPswe/PSEN1dE9 mice maintained under standard laboratory conditions. Mice were treated with SCFAs from five months of age until ten months of age, when they were evaluated for microbiome profile, impaired spatial memory as evaluated with the radial arm water maze, astrocyte activation as measured by Gfap expression and amyloid burden as assessed by histochemistry and MSD ELISA. We report that SCFA treatment increased alpha-diversity and impacted the gut microbiome profile by increasing, in part, the relative abundance of several bacteria that typically produce SCFAs. However, SCFA treatment did not significantly affect behavior. Similarly, SCFAs did not affect cortical or hippocampal astrocyte activation observed in the APP/PS1 mice. Lastly, although robust levels of soluble and insoluble amyloid were present in the APP/PS1 mice, SCFA treatment had no effect on these indices. Overall, our findings are that SCFA treatment modifies the microbiome in a fashion that may increase further SCFA production. However, SCFA treatment did not alter behavior, astrocyte activation, nor amyloid neuropathology in APP/PS1 mice maintained with a conventional microbiome.

Elucidating the actions of genetic polymorphisms associated with the risk of Alzheimer's disease (AD) may provide novel insights into underlying mechanisms. Two polymorphisms have implicated ABI3 as a modulator of AD risk. Here, we sought to identify ABI3 isoforms expressed in human AD and non-AD brain, quantify the more abundant isoforms as a function of AD genetics and neuropathology, and provide an initial in vitro characterization of the proteins produced by these novel isoforms. We report that ABI3 expression is increased with AD neuropathology but not associated with AD genetics. Single-cell RNAseq of APP/PS1 mice showed that Abi3 is primarily expressed by microglia, including disease-associated microglia. In human brain, several novel ABI3 isoforms were identified, including isoforms with partial or complete loss of exon 6. Expression of these isoforms correlated tightly with total ABI3 expression but were not influenced by AD genetics. Lastly, we performed an initial characterization of these isoforms in transfected cells and found that, while full-length ABI3 was expressed in a dispersed punctate fashion within the cytosol, isoforms lacking most or all of exon six tended to form extensive protein aggregates. In summary, ABI3 expression is restricted to microglia, is increased with Alzheimer's neuropathology, and includes several isoforms that display a variable tendency to aggregate when expressed in vitro.

Although gene editing workflows typically consider the possibility of off-target editing, pseudogene-directed homology repair has not, to our knowledge, been reported previously. Here, we employed a CRISPR-Cas9 strategy for targeted excision of exon 2 in CD33 in U937 human monocyte cell line. Candidate clonal cell lines were screened by using a clinically relevant antibody known to label the IgV domain encoded by exon 2 (P67.6, gemtuzumab). In addition to the anticipated deletion of exon 2, we also found unexpected P67.6-negative cell lines which had apparently retained CD33 exon 2. Sequencing revealed that these lines underwent gene conversion from the nearby SIGLEC22P pseudogene during homology repair that resulted in three missense mutations relative to CD33. Ectopic expression studies confirmed that the P67.6 epitope is dependent upon these amino acids. In summation, we report that pseudogene-directed homology repair can lead to aberrant CRISPR gene editing.

Brain arteriolosclerosis (B-ASC) is a small-vessel cerebrovascular pathologic change involving sclerotic thickening of arterioles in the brain. B-ASC pathology is common in aged autopsy cohorts, is associated with worse cognitive functioning, and is also associated with other neuropathologies. The genetics of dementia and other pathologic changes have been studied extensively, but to date no in-depth study has been conducted on genetic risk of autopsy-proven B-ASC. We performed the first genome-wide association study (GWAS) focusing on B-ASC using multiple cohorts. We then followed up on identified risk variants with functional analysis to investigate potential biological mechanisms.Individual GWAS and mega-analyses were conducted using data from participants in the National Alzheimer's Coordinating Center (NACC) (N = 3318) and the Religious Orders Study and Memory and Aging Project (ROSMAP) (N=1190). In total, 4,577,283 variants (MAF > 5%) were present in both datasets and passed quality control. Variants identified in GWAS were then checked for quantitative trait loci (QTL) associations in the Genotype Tissue Expression (GTEx) project. Colocalization analysis was then performed on identified QTL in brain and vascular tissues. Variants most highly associated with B-ASC were also examined for association with neuroimaging correlates of B-ASC in the Alzheimer's Disease Neuroimaging Initiative.One intergenic locus on Chromosome 6 was significantly associated with B-ASC in NACC (P = 1.4e-08) and colocalized with ELOVL4 gene expression in the brain. Other loci in both cohorts achieved less stringent thresholds of association and colocalized with multiple QTL in GTEx. One intergenic locus on Chromosome 10 near SORCS3 was suggestively associated with B-ASC in NACC and was replicated in ROSMAP.Genetic loci associated with B-ASC pathology were identified using multiple cohorts. Most loci associated with B-ASC in one cohort were not associated in the other. Multiple identified risk loci colocalized with gene expression or splicing QTL, providing evidence for potential biological mechanisms through which variants may influence disease risk. Due to the relatively small sample sizes and heterogeneity between the cohorts used, additional studies are merited to better characterize genetic risk of B-ASC.

Genome-wide association studies (GWAS) have identified immune-related genes as risk factors for Alzheimer's disease (AD), including TREM2 and CD33, frequently passing a stringent false-discovery rate. These genes either encode or signal through immunomodulatory tyrosine-phosphorylated inhibitory motifs (ITIMs) or activation motifs (ITAMs) and govern processes critical to AD pathology, such as inflammation and amyloid phagocytosis. To investigate whether additional ITIM and ITAM-containing family members may contribute to AD risk and be overlooked due to the stringent multiple testing in GWAS, we combined protein quantitative trait loci (pQTL) data from a recent plasma proteomics study with AD associations in a recent GWAS. We found that pQTLs for genes encoding ITIM/ITAM family members were more frequently associated with AD than those for non-ITIM/ITAM genes. Further testing of one family member, SIGLEC14 which encodes an ITAM, uncovered substantial copy number variations, identified an SNP as a proxy for gene deletion, and found that gene expression correlates significantly with gene deletion. We also found that SIGLEC14 deletion increases the expression of SIGLEC5, an ITIM. We conclude that many genes in this ITIM/ITAM family likely impact AD risk, and that complex genetics including copy number variation, opposing function of encoded proteins, and coupled gene expression may mask these AD risk associations at the genome-wide level.

Background Genetic variants in TREM2 are strongly associated with Alzheimer’s Disease (AD) risk but alternative splicing in TREM2 transcripts has not been comprehensively described. Objective Recognizing that alternative splice variants can result in reduced gene expression and/or altered function, we sought to fully characterize splice variation in TREM2. Methods Human blood and anterior cingulate autopsy tissue from 61 donors were used for genotyping and cDNA synthesis followed by both end-point and quantitative PCR to identify and quantify novel TREM2 isoforms. Results In addition to previously described transcripts lacking exon 3 or exon 4, or retaining part of intron 3, we identified novel isoforms lacking exon 2, along with isoforms lacking multiple exons. Isoforms lacking exon 2 were predominant at approximately 10% of TREM2 mRNA in the brain. Expression of TREM2 and frequency of exon 2 skipping did not differ between AD samples and non-AD controls (p = 0.1268 and p = 0.4909, respectively). Further, these novel splice isoforms were also observed across multiple tissues with similar frequency (range 5.3 – 13.0%). Using ectopic expression, we found that the exon 2 skipped isoform D2-TREM2 is translated to protein and localizes similarly to full-length TREM2 protein, and that both proteins are primarily retained in the Golgi complex. Conclusion Since the TREM2 ligand binding domain is encoded by exon 2, and skipping this exon retains reading frame while conserving localization, we hypothesize that D2-TREM2 acts as an inhibitor of TREM2 and that targeting TREM2 splicing may be a novel therapeutic pathway for AD.