Interaction effects between variants in TOMM40 and PVRL2 with plasma amyloid-β and Alzheimer's disease among Chinese older adults: a population-based study

Emerging evidence has linked TOMM40 and PVRL2 polymorphisms with Alzheimer’s disease (AD). We examined their associations with AD and plasma AD biomarkers and interaction on AD risk among Chinese older adults. Methods This population-based study included 4876 participants (age ≥ 65 years, 57.2% women) from MIND-China. TOMM40(rs2075650) and PVRL2(rs6859) polymorphisms were detected using multiple-polymerase chain reaction amplication. Plasma Aβ40, Aβ42, and t-tau were measured using SIMOA in a subsample (n = 1257). Data was analyzed using multiple logistic and general linear regression models. The number of participants with missing values was 39 for hypertension. As a covariate in the subsequence analysis, a dummy variable was created for hypertension to represent these with missing values.


Background
Alzheimer's disease (AD) is the most common neurodegenerative disease among elderly people [1]. The etiology and pathogenesis of AD remain poorly understood. The heritability of late-onset AD was estimated to be approximately 56%-79%, suggesting that genetic components play an important role in the pathogenesis [2]. Thus, exploring genetic factors may help increase our understanding of the etiopathogenesis of AD. Indeed, the genome-wide association studies (GWAS) have identi ed a large number of susceptibility loci, of which, the chromosome 19q13 region has been the strongest genetic loci for AD [3]. In addition to the well-known APOE gene within this region, translocase of outer mitochondrial membrane 40 homolog (TOMM40) gene and poliovirus receptor-related 2 (PVRL2) gene have been identi ed to be associated with AD [4,5].
TOMM40, an APOE nearby gene [6], encodes a subunit of the translocase of the outer mitochondrial membrane complex, which is involved in AD-related pathology [7]. Evidence has shown that multiple single nucleotide polymorphisms (SNPs) within TOMM40 are associated with AD risk, of which, the SNP rs2075650 has been frequently reported in various ethnic populations [8][9][10][11]. PVRL2, which locates near the TOMM40, also serves as a risk variant that is involved in AD pathology [12]. In addition, TOMM40 and PVRL2 genes have been implicated in common biological pathways, such as lipid metabolism [13] and immunity [14,15]. Thus, it is biologically plausible to hypothesize that TOMM40 and PVRL2 genes may confer an interaction effect in the pathogenesis of AD. However, this hypothesis has yet to be explored.
Although GWAS studies have identi ed the associations of TOMM40 and PVRL2 with AD risk, the neuropathological pathways underlying their associations remain unclear. Aggregation of the β-amyloid (Aβ) peptide and neuro brillary tangles in the brain are the neuropathological hallmarks of AD [16]. Data from the Framingham Heart Study showed that lower plasma Aβ and higher plasma total tau (t-tau) proteins were associated with an increased risk of AD [17,18] However, whether TOMM40 and PVRL2 are associated with AD via in uencing plasma Aβ and tau proteins remains largely unknown. Therefore, in this population-based study of rural-dwelling Chinese older adults, we aimed to investigate the associations of TOMM40 rs2075650 and PVRL2 rs6859 as well as their interactions with AD risk, and to further explore the associations of these two genes with plasma AD biomarkers (Aβ and t-tau proteins).

Study population
This population-based study involved the participants from the ongoing Multimodal Interventions to Delay Dementia and Disability in Rural China (MIND-China) [19][20][21], a participating project in the World-Wide FINGERS Network [22]. Brie y, all registered residents who were aged ≥ 60 years and living in 52 villages of Yanlou Town, Yanggu County, western Shandong Province, China, were eligible for the MIND-China study. From March to September 2018, 5765 participants were enrolled. Of these, 519 participants aged 60-64 years were excluded, because we aimed to investigate the late-onset sporadic AD. Of the 5246 participants who were aged ≥65 years, 370 were excluded due to missing genotyping information (n = 230), insu cient information for dementia diagnosis (n = 46), and other types of dementia than AD (n = 94). Thus, the analytical sample included 4876 participants. Compared with the excluded participants (n = 370), individuals included in the analytical sample (n = 4876) were younger (mean age, 71.6 vs. 73.6 years, p < 0.001) and more educated (mean years of education, 3.14 vs. 2.60, p < 0.001). However, the two groups did not differ signi cantly in sex distribution (female, 57.2% vs. 56.2%, p = 0.701).
In addition, data on plasma AD biomarkers were available in a subsample of 1257 individuals that were derived from the participants of MIND-China. This subsample was used for the analysis involving plasma AD biomarkers. Figure 1 shows a owchart of the study participants.
The MIND-China project was approved by the Ethics Committee of Shandong Provincial Hospital a liated to Shandong University, Jinan, Shandong. All participants provided written informed consent for both study enrollment and blood sample collection, or in case of persons with severe cognitive impairment, their informants provided informed consent. Research within MIND-China has been conducted in accordance with the ethical principles expressed in the Declaration of Helsinki as well as relevant national guidelines and regulations. MIND-China was registered in the Chinese Clinical Trial Registry (registration no.: ChiCTR1800017758).

Data collection and assessments
The procedure of baseline data collection was described elsewhere [19,20]. Brie y, all data were collected by trained staff via face-to-face interviews, clinical examinations, cognitive testing, and laboratory tests. The in-person interview was performed following a structured questionnaire covering demographic data (e.g., age, sex, and education), lifestyle factors (e.g., smoking and alcohol drinking), health conditions (e.g., hypertension, diabetes, and hyperlipidemia), medications, and cognitive function. Alcohol consumption and smoking status were categorized as current, former, and never drinking or smoking, respectively. After a 5-minute rest, seated arterial blood pressure was measured on the right upper arm using an electronic sphygmomanometer (HEM-7127J, Omron Corporation, Kyoto, Japan). After an overnight fast, venous blood samples were collected and blood glucose, total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) were measured using an Automatic Biochemical Analyzer (CS-600B, DIRUI Corporation, Changchun, China). Hypertension was de ned as arterial blood pressure ≥ 140/90 mmHg or taking antihypertensive agents. Diabetes was de ned as the fasting blood glucose ≥ 7.0 mmol/L or current use of antidiabetic agents. Hyperlipoidemia was de ned as TC ≥ 6.2 mmol/L or TG ≥ 2.3 mmol/L or LDL-C ≥ 4.1mmol/L or HDL-C < 1.0 mmol/L or taking hypolipidemic agents.

Genomic DNA extraction and genotyping
Genomic DNA was extracted from venous blood leukocytes using the TIANamp blood DNA kit (Tiangen, Beijing, China) according to manufacturer's instructions. Then, DNA was quanti ed using Nanodrop 3300 spectrometry and a total amount of 100 ng genomic DNA per sample was used for the DNA sample preparation. Subsequently, sequencing libraries were generated using MultipSeqCustom Panel (iGeneTech, Beijing, China) following manufacturer's recommendations and index codes were added to each sample. Finally, quali ed libraries were subjected to next-generation sequencing (NGS) on a Novaseq system (Illumina), and raw data were ltered to remove low-quality reads using FastQC.
Genotyping was conducted by an operator who was blinded to the clinical data.

Measurements of Plasma Aβ42, Aβ40 and Tau
Blood samples were collected in EDTA-coated vacutainers, followed by centrifugation to obtain plasma. The samples were stored at -80℃ and thawed immediately before component quanti cation. Levels of plasma Aβ42, Aβ40, and t-tau were measured using a Human Neurology 3-Plex A assay (N3PA) Kit run on the fully automated single molecule array (SIMOA) instrument (Quanterix Corp, MA, USA) according to the manufacturer's protocol. Two quality control samples were run in duplicates on each plate for each analyte. The upper and lower detection limits for quality sample of each index were within the range indicated in the manual, and the inter-assay coe cient of variability (CV) and inter-plate inter-assay CV were controlled within 13%.

Diagnosis of Alzheimer's disease
We employed a 3-step procedure for the clinical diagnosis of dementia. In brief, clinicians and trained junior interviewers conducted comprehensive assessments for each participant following the structured questionnaires. The assessments included health-related factors, medical history, a neurocognitive assessment battery, the Chinese version of activities of daily living (ADLs), and the Clinical Dementia Rating Scale (CDR). Then, the neurologists specialized in dementia diagnosis and care reviewed all the data collected from clinical examination, in-person interviews, and neurocognitive testing, and made a preliminary judgement for participants who were suspected to have dementia. In addition, ~ 9.5% had incomplete data for the assessment of dementia status. Finally, all these persons were contacted by neurologists to conduct the second interview and reassess their medical history, cognitive status, ADLs, and whenever available, neuroimaging data. If the participants were not able to take the interview due to severe cognitive impairment or not available for the face-to-face interviews (~ 13%), the neurologists interviewed their family members, neighbors, or village doctors (primary care providers to local residents).
Based on all the assessments, a diagnosis of dementia was made according to the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, criteria [23]. In case of uncertainty, a senior neurologist was consulted and discussed, and a consensus diagnosis of dementia was reached. The diagnosis of AD was made according to the National Institute on Aging-Alzheimer's Association criteria for probable AD [24] .

Statistical Analysis
Characteristics of study participants were present and compared using the non-parametric test for continuous variables or chi-square test for categorical variables. We used chi-square test to assess whether the allele frequencies agreed with expectations in Hardy-Weinberg equilibrium (HEW). The Haploview software was used to explore the strati ed linkage disequilibrium (LD) patterns of SNPs [25]. Multivariable logistic regression models were used to calculate the odds ratio and 95% con dence interval (95% CI) of AD associated with TOMM40 and PVRL2 genotypes. Furthermore, we evaluated the interaction and joint exposures of TOMM40 and PVRL2 genotypes associated with AD risk. We reported the main results from two models: Model 1 was controlled for age, sex, and education, and Model 2 was additionally controlled for current smoking, current drinking, diabetes, hyperlipoidemia, hypertension, and APOE genotype. In the plasma biomarker subsample, we used the multiple general linear models to examine the associations of TOMM40 and PVRL2 genotypes with plasma AD biomarkers. The statistical signi cance was set to be two-tailed p < 0.05. All analyses were performed using IBM SPSS Statistics version 26.0 (IBM SPSS Inc., Chicago, Illinois, USA) and Haploview version 4.2 (Cambridge, MA02141, USA).

Characteristics of study participants (n = 4876)
Of the 4876 participants, the mean age was 71.6 (SD, 5.4) years, 57.2% were women, and the average years of formal schooling was 3.1 (SD, 3.4; 39.9% were illiterate) years. Among all the participants, 182 (3.7%) were ascertained to have AD. Compared with people without dementia, participants with AD were older, more likely to be women, and less educated (p < 0.001). The distribution of the TOMM40 and PVRL2 genotypes conformed to Hardy-Weinberg equilibrium (HWE). In addition, there was a signi cant difference in the distributions of current smoking, alcohol drinking, and TOMM40 and PVRL2 genotypes between the two groups. However, the two groups did not signi cantly differ in the distributions of diabetes, hyperlipoidemia, hypertension, and carrying APOE ε4 allele (Table 1). To dissect AD-associated genetic structure among TOMM40 rs2075650 and PVRL2 rs6859, we used Haploview software to investigate strati ed LD patterns of the two SNPs among AD cases and nondemented subjects respectively( Figure S1). People with AD manifested a diverse genomic structure relative to people who were free of dementia, represented by a stronger D' value among rs2075650 and rs6859 variants (0.91 vs. 0.79).
Controlling for age, sex, and education, TOMM40 GG genotype was signi cantly associated with an increased likelihood of AD (  We detected a statistical interaction of TOMM40 and PVRL2 genotypes on the likelihood of AD (p for interactions = 0.0003). Further analysis strati ed by both genotypes showed that TOMM40 GG genotype in combination with PVRL2 AA genotype was signi cantly associated with a substantially increased likelihood of AD compared with TOMM40 AA + AG genotype in combination with PVRL2 GG + AG genotype (

Associations of TOMM40 and PVRL2 SNPs with plasma AD biomarkers (n = 1257)
Multiple general linear regression analysis showed that TOMM40 AG and GG genotypes (vs. AA genotype) were signi cantly associated with a lower level of plasma Aβ42 after controlling for age, sex, and education (Table 3, Model 1). When additionally adjusting for smoking, drinking, diabetes, hyperlipoidemia, hypertension, and APOE genotype, the association between TOMM40 GG genotype and lower plasma Aβ42 was still signi cant (Table 3, Model 2). The associations of TOMM40 genotype with the Aβ42/Aβ40 ratio were similar to those with plasma Aβ42 (Table 3). There were no signi cant associations of the PVRL2 genotype with either plasma Aβ42 or the Aβ42/Aβ40 ratio. Abbreviations: Aβ, amyloid-β; t-tau, total tau protein; TOMM40, Translocase of outer mitochondrial membrane 40 homolog gene; PVRL2, poliovirus receptor-related 2 gene.

Discussion
In this population-based study of rural-dwelling Chinese older adults, we found that TOMM40 GG genotype was associated with an increased likelihood of AD. Whereas, there was a marginal association of PVRL2 AA genotype with AD. An interaction between TOMM40 and PVRL2 on the likelihood of AD was detected such that carrying both TOMM40 GG and PVRL2 AA genotypes was associated with an over 12fold increased likelihood of AD. In our subsample, the TOMM40 GG genotype, but not PVRL2 AA genotype, was associated with a reduced level of plasma Aβ42. Taken together, these results suggest that TOMM40 and PVRL2 risk alleles may act interactively to increase the likelihood of AD, possibly through in uencing Aβ metabolism.
The GWAS analyses suggested that TOMM40 SNPs may play some role in AD [14,26]. In addition, casecontrol studies from Italy [9], Colombia [10], and China [11] also revealed that the TOMM40 G allele serves as a risk variant for AD. In line with previous studies, our community-based study con rmed that the TOMM40 G allele could confer a substantial risk to AD. However, the mechanism underlying the association of TOMM40 with AD is not well established. The neuroimaging and neuropathological data showed that the TOMM40 was associated with amyloid burden in the brain parenchyma and vessels [27]. Coincidentally, reports from ADNI database demonstrated that TOMM40 was a susceptible putative locus associated with cerebrospinal uid AD biomarkers (e.g., Aβ42, p-tau181p/Aβ42 ratio, and t-tau/Aβ42 ratio) [28,29]. In the current population-based study, we found that TOMM40 GG genotype was associated with a lower level of plasma Aβ42 and the Aβ42/Aβ40 ratio. Taken together, these results suggested that TOMM40 might be involved in the Aβ metabolism, which in turn may be linked with AD.
The GWAS analyses among Chinese or European populations showed that PVRL2 rs6859 was associated with AD, with OR ranging from 1.46 to 1.61 [4,5,26]. In line with these GWAS analyses, our community-based study showed that PVRL2 AA genotype was associated with a 56% increased likelihood of AD after controlling for demographic factors (OR = 1.56; 95% CI 0.96-2.51), although the association was not statistically evident, possibly owing to limited statistical power. Thus, large-scale population-based studies are warranted to replicate the association of PVRL2 genotype with AD risk.
Our study revealed a synergistic interaction between TOMM40 and PVRL2 variants on AD, such that having both TOMM40 GG and PVRL2 AA genotypes was associated with an over 12-fold likelihood of AD.
To the best of our knowledge, this is the rst population-based study that identi ed the interaction between TOMM40 and PVRL2 genes on AD risk. The neurobiological mechanisms underlying their interaction effect on AD are not clear. The expression quantitative trait loci (eQTL) analysis indicated that the TOMM40 rs2075650 G allele could increase the binding a nity of transcription factors (TFs), thus, increasing PVRL2 gene expression [15]. In this context, the synergistic interaction of TOMM40 GG and PVRL2 AA genotypes with AD may partly attribute to the mechanism that TOMM40 G allele might enhance the effect of PVRL2 risk allele. Alternatively, the synergistic effects of TOMM40 and PVRL2 risk variants on AD might be resulted from their shared molecular pathways involved in AD pathological mechanisms such as Aβ metabolism [30,31] and in ammation [14,15].
Previous studies reported that TOMM40 and PVRL2 might be indirectly involved in the accumulation of Aβ[32, 33]. Our results showed that TOMM40 GG carriers had the lower level of plasma Aβ42 than carriers of other genotypes, supporting a possible link of TOMM40 with metabolism of Aβ42, but not Aβ40. However, we did not nd associations between PVRL2 genotype with any of the three plasma AD biomarkers (Aβ40, Aβ42, and t-tau), which is in line with the lack of evident association between PVRL2 genotype and AD risk. In addition, the lack of associations between plasma t-tau and TOMM40 or PVRL2 genotype may be due partly to the facts that plasma t-tau is not associated with dementia and AD[34]. Alternatively, TOMM40 and PVRL2 genes may be not involved in the process of tau pathology.
Our study was based on a large-scale sample of rural-dwelling Chinese older adults, where epidemiological, clinical, neuropsychological, and genetic data were integrated with plasma AD biomarkers. Thus, we are able to explore the associations and interaction effects of TOMM40 and PVRL2 genes with AD risk as well as the possible neuropathological mechanisms.

Limitation
Our study also has limitations. First, due to the cross-sectional nature of the study design, the observed associations might be subject to selective survival bias, which usually leads to an underestimation of the true association. Furthermore, the study participants were recruited from a single rural area in northern China, which should be kept in mind when generalizing our ndings to other populations.

Conclusions
In conclusion, our study con rms the association of TOMM40 GG genotype with AD among rural-dwelling Chinese older adults and further reveals a potential genetic interaction between TOMM40 and PVRL2 risk alleles on AD risk and the association of TOMM40 GG genotype with low plasma Aβ42. These results increase our understanding of the roles of TOMM40 and PVRL2 genes in AD and the potential neuropathological mechanisms.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.