Integrating multi-omics data to reveal the effect of genetic variant rs6430538 on Alzheimer's disease risk

Introduction Growing evidence highlights a potential genetic overlap between Alzheimer's disease (AD) and Parkinson's disease (PD); however, the role of the PD risk variant rs6430538 in AD remains unclear. Methods In Stage 1, we investigated the risk associated with the rs6430538 C allele in seven large-scale AD genome-wide association study (GWAS) cohorts. In Stage 2, we performed expression quantitative trait loci (eQTL) analysis to calculate the cis-regulated effect of rs6430538 on TMEM163 in both AD and neuropathologically normal samples. Stage 3 involved evaluating the differential expression of TMEM163 in 4 brain tissues from AD cases and controls. Finally, in Stage 4, we conducted a transcriptome-wide association study (TWAS) to identify any association between TMEM163 expression and AD. Results The results showed that genetic variant rs6430538 C allele might increase the risk of AD. eQTL analysis revealed that rs6430538 up-regulated TMEM163 expression in AD brain tissue, but down-regulated its expression in normal samples. Interestingly, TMEM163 showed differential expression in entorhinal cortex (EC) and temporal cortex (TCX). Furthermore, the TWAS analysis indicated strong associations between TMEM163 and AD in various tissues. Discussion In summary, our findings suggest that rs6430538 may influence AD by regulating TMEM163 expression. These discoveries may open up new opportunities for therapeutic strategies targeting AD.

Introduction: Growing evidence highlights a potential genetic overlap between Alzheimer's disease (AD) and Parkinson's disease (PD); however, the role of the PD risk variant rs in AD remains unclear.
Methods: In Stage , we investigated the risk associated with the rs C allele in seven large-scale AD genome-wide association study (GWAS) cohorts.In Stage , we performed expression quantitative trait loci (eQTL) analysis to calculate the cis-regulated e ect of rs on TMEM in both AD and neuropathologically normal samples.Stage involved evaluating the di erential expression of TMEM in brain tissues from AD cases and controls.Finally, in Stage , we conducted a transcriptome-wide association study (TWAS) to identify any association between TMEM expression and AD.

Results:
The results showed that genetic variant rs C allele might increase the risk of AD. eQTL analysis revealed that rs up-regulated TMEM expression in AD brain tissue, but down-regulated its expression in normal samples.Interestingly, TMEM showed di erential expression in entorhinal cortex (EC) and temporal cortex (TCX).Furthermore, the TWAS analysis indicated strong associations between TMEM and AD in various tissues.
Discussion: In summary, our findings suggest that rs may influence AD by regulating TMEM expression.These discoveries may open up new opportunities for therapeutic strategies targeting AD.

Background
Potential genetic association may exist between common neurodegenerative diseases (Small and Petsko, 2015;Brainstorm et al., 2018;Guo et al., 2022).Alzheimer's disease (AD) and Parkinson's disease (PD) are both classic age-related neurodegenerative diseases (Scheltens et al., 2016;Poewe et al., 2017;Hu et al., 2021;Qiu et al., 2022a).Although the pathogenesis of AD and PD differs, they share certain similar pathways in the development of neurodegeneration (Xie et al., 2014).One of the most important pathological features of AD is the deposition of amyloid-β (Aβ), which is also present in the brains of PD patients (Compta et al., 2011;Yu et al., 2014).Similarly, the pathological features of PD are also observed in cases of AD (Lippa et al., 1997).Currently, the shared susceptibility loci of AD and PD, such as APOE and MAPT, have been identified, indicating a possible overlap in genetic background and pathological characteristics (Laws et al., 2007;Zhu et al., 2017).However, further exploration of additional genetic links is necessary.
Members of the transmembrane (TMEM) protein family have been identified as being associated with various neurodegenerative disorders (Chang et al., 2017;Nalls et al., 2019;Li et al., 2020).Among these genes, TMEM163 serves as an important susceptibility locus for PD.It plays a regulatory role in ATP-evoked behavior in neurons, as well as binding to Zn 2+ and recruiting them to vesicular organelles (Chang et al., 2017;Salm et al., 2020;Kia et al., 2021;Mammadova-Bach and Braun, 2021).Notably, brain zinc has been found to be involved in the pathogenesis of AD by influencing amyloid metabolism (Sensi et al., 2018;Xu et al., 2019).It is plausible to hypothesize that rs6430538, through its impact on the expression level of zinc transporters, contributes to the aggravation of Aβ deposition and toxicity by regulating nearby genes.In an attempt to identify shared pathogenic genetic variants between PD and AD, Zhu et al. replicated the association of rs6430538 (ACMSD-TMEM163) with AD in a population study (Zhu et al., 2017).They investigated a sample comprising 992 sporadic AD patients and 1,358 controls from northern China and observed a protective effect of rs6430538 on AD (OR = 0.340, P = 0.015) (Zhu et al., 2017).Unfortunately, upon regression adjustment for confounding factors including age, gender, and APOE ε4 status, rs6430538 did not achieve statistical significance (P = 0.072) (Zhu et al., 2017).It is important to note, however, that the sample size of the independent case-control study was relatively small compared to genome-wide association studies (GWAS).Furthermore, previous investigations solely focused on exploring the potential risk association between rs6430538 and AD, neglecting the possibility that rs6430538 may influence AD through the regulation of gene expression.
In this study, we aimed to delve deeper into the association between the PD risk variant rs6430538 and AD.Specifically, we sought to investigate the impact of rs6430538 on AD by analyzing its influence on gene expression using comprehensive datasets, including large-scale GWAS, expression quantitative trait loci (eQTL), and RNA-seq datasets.By doing so, our findings would contribute to a better understanding of the functional role of noncoding region variants, such as rs6430538, and their implications for disease regulation.

Materials and methods . AD GWAS datasets
We investigated the risk of rs6430538 in seven largescale GWAS for AD, including two GWAS meta-analyses by International Genomics of Alzheimer's Project (IGAP), three genome-wide association studies by proxy (GWAX) by UK Biobank, and two meta-analyses of GWAS and GWAX (Lambert et al., 2013;Marioni et al., 2018;Buniello et al., 2019;Jansen et al., 2019;Kunkle et al., 2019;Schwartzentruber et al., 2021).UK Biobank used self-reported family history of Alzheimer's disease as the standard of cases, while IGAP used Aβ and tau levels as the standard of cases (the NINCDS-ADRDA criteria or DSM-IV guidelines) (Lambert et al., 2013).Detailed GWAS information was shown in the original studies and Table 1.
We obtained eQTL datasets of AD and non-AD samples from Mayo RNAseq Study (MAYO) and Religious Orders Study and Memory and Aging Project (ROSMAP) (Bennett et al., 2012a,b;Zou et al., 2012;GTEx Consortium, 2017;Ng et al., 2017).Mayo RNAseq Study collected the cerebellar (CER) (197 cases and 177 controls) and temporal cortex (TCX) (202 cases and 197 controls) tissues of AD individuals and non-AD individuals, respectively (Zou et al., 2012).Non-AD subjects contained several brain pathologies, such as frontotemporal dementia (FTD), multiple system atrophy and vascular dementia (Zou et al., 2012).ROSMAP collected samples from the dorsolateral prefrontal cortex (DLPFC) of 494 participants (Ng et al., 2017).97% of the samples were diagnosed with pathological or clinical AD (Ng et al., 2017).The donors were all of European descent.

. Statistical analysis . . Genetic association of rs with AD
We used seven GWAS datasets for AD to evaluate the genetic association between rs6430538 C allele and AD.In addition, we carried out gender stratification analysis using GWAX datasets diagnosed by paternal history and maternal history from UK Biobank (Marioni et al., 2018).The statistically significant association is defined to be P < 5E-08 after adjusting for multiple testing.

. . eQTL analysis of rs on TMEM
Disease states are known to alter the expression of specific genes (Gratuze et al., 2018;Liu et al., 2019;Hu et al., 2020;Ma and Qiu, 2022;Qiu et al., 2022b,c).Here, we calculated the regulated effect of rs6430538 C allele on TMEM163 in neuropathologically Beta is the regression coefficient based on the effect allele.Beta > 0 and beta < 0 mean the effect allele could increase and reduce AD risk, respectively.SE, standard error.GWAX, genome-wide association studies by proxy.GWAS, genome-wide association studies.The statistically significant association is defined to be P < 0.05/7 = 7.14E-03.
normal and neuropathologically disease individuals, respectively, by applying linear regression based on an additive model (Hu et al., 2020;Qiu et al., 2022b).According to the additive model, each allele has an independent effect on the trait.We coded the possible genotypes of rs6430538 (CC = 2, CT = 1, TT = 0), where C is an effect allele and T is a non-effect allele.The statistically significant association is defined to be P < 0.05/18 = 2.78E-03.

. . Gene expression analysis of TMEM
We evaluated the differential expression of TMEM163 in four brain tissues between AD cases and controls.All differential expression results were adjusted for sex with an FDR of 0.05.

. . Transcriptome-wide association study analysis
TWAS is a test for a significant association between cis components of gene expression and traits (Gusev et al., 2016).TWAS integrates GWAS and eQTL datasets, not only mining significant correlations between eQTLs and SNPs, but also capturing complete cis-SNP signals.We looked for evidence of the association between TMEM163 and AD in TWAS using FUSION software (Gusev et al., 2016).Herein, we integrated AD GWAS with gene expression dataset from GTEx and ROSMAP for TWAS analysis (Gusev et al., 2016;GTEx Consortium, 2017;Kunkle et al., 2019).The statistically significant association is defined to be P < 5E-08 after multiple testing.

Rs regulated the expression of TMEM and increased AD risk
We inquired proxy SNPs of rs6430538 and the annotations information on noncoding genome using HaploReg4.1 (Ward and Kellis, 2016).Rs6430538 is located in the intergenic region on chromosome 2, with eight genetic variants highly linked to rs6430538 (r 2 > 0.8) (Figure 1, Supplementary Table 1) (Pruim et al., 2010).In addition to the GWAX studies by UK Biobank, rs6430538 C allele was suggestively associated with the risk of AD in four large-scale GWAS meta-analyses (Table 1).Moreover, rs6430538 showed no sex-specific differences in AD risk (Table 1).
Based on all brain eQTL studies, rs6430538 involved in regulating the expression of TMEM163, CCNT2, VDAC2P4, CCNT2-AS1.Herein, we focused on TMEM163, the shared regulated gene, in all the eQTL studies.The eQTL analysis showed that rs6430538 regulated TMEM163 overexpression in AD patients (Beta = 0.325, P = 1.31E-13), while rs6430538 inhibited TMEM163 expression in non-AD individuals (Table 2).The opposing effects of the same variant may be explained if there is dysregulation of TMEM163 in disease conditions.

Discussion
Neurodegenerative diseases, such as AD and PD, share certain genetic associations and pathological characteristics, despite differences in their pathogenesis (Majd et al., 2015;Zhu et al., 2017; Frontiers in Neuroscience frontiersin.org

FIGURE
Regional visualization of genome-wide association scan results near variant rs .Han et al., 2018).The identification of shared susceptibility loci, such as APOE and MAPT, suggests a potential overlap in genetic background and pathological mechanisms between these diseases.However, further exploration is required to uncover additional genetic links.
Based on the hypothesis that rs6430538, a risk variant associated with PD, may impact AD through the regulation of nearby genes, we aimed to investigate the influence of rs6430538 on AD by analyzing its effect on gene expression using largescale datasets.To examine the genetic association of rs6430538 with AD, we analyzed seven GWAS datasets, including those from the IGAP and UK Biobank.Our findings indicated that the rs6430538 C allele was associated with an increased risk of AD.Additionally, gender stratification analysis revealed no sexspecific differences in the association between rs6430538 and AD risk.Understanding gender-specific and nongender-specific effects could guide the development of sex-specific therapeutic strategies, taking into account potential differences in disease mechanisms and responses to treatment between males and females.
Although GWAS analysis indicated only suggestive association between rs6430538 and AD, stringent GWAS thresholds might obscure some genetic variants that affect AD through regulating gene expression.Therefore, we investigated the impact of rs6430538 on the expression of TMEM163 and other nearby genes using eQTL datasets.The results demonstrated that rs6430538 played a role in regulating the expression of TMEM163, CCNT2, VDAC2P4, and CCNT2-AS1.Specifically, rs6430538 exhibited a regulatory effect on TMEM163 expression, with overexpression observed in AD patients and inhibited expression in non-AD individuals.As revealed in previous studies, genetic variants linked to AD risk that modulate the expression of genes in human tissues, emphasizing the importance of non-coding regions in disease pathogenesis (Raj et al., 2014).To prioritize potential susceptibility genes for AD, we conducted a TWAS using GWAS and eQTL datasets.Our results indicated that TMEM163 was strongly associated with AD in multiple tissues, including brain tissues and whole blood.
However, our study had certain limitations.First of all, gender, age (or age of death), disease status (or cause of death) and ancestry, might lead to different gene expression levels in different eQTL datasets.We could not control all the variables.For example, 1.2% of the donors in GTEx died of neurological diseases, and these samples with abnormal neuropathology might have some effects.Similarly, due to the lack of sex stratified eQTLs and original genotype data, it was difficult to deeply explore the contribution of female genotype and gene expression in AD.Furthermore, although the PD susceptibility locus TMEM163 significantly affected AD in brain tissue and blood tissue, to our knowledge, none of the known methods claimed to find causal gene (Gusev et al., 2016;Wainberg et al., 2019;Zhu et al., 2021).Novel methods of exploring causal genes may be more powerful evidence for our research.Beta is the regression coefficient based on the effect allele.Beta > 0 and beta < 0 mean that this effect allele could increase and reduce gene expression, respectively.The statistically significant association is defined to be P < 0.05/18 = 2.78E-03.

FIGURE
Box plots about di erential expression analysis of TMEM .The y-axis represents the expression level of TMEM , while the x-axis represents di erent brain tissues.

Conclusions
Genetic variants located in non-coding region may affect traits by regulating gene expression.Here, we comprehensively analyzed the effect of a PD risk variant rs6430538 on AD through regulating gene expression.We demonstrated that different neuropathological samples played different roles in the regulation of gene expression by genetic variants.These findings highlighted the possible genetic overlap between neurodegenerative diseases.