Association of rs2062323 in the TREM1 gene with Alzheimer's disease and cerebrospinal fluid‐soluble TREM2

Abstract Introduction and aims Genetic variations play a significant role in determining an individual's AD susceptibility. Research on the connection between AD and TREM1 gene polymorphisms (SNPs) remained lacking. We sought to examine the associations between TREM1 SNPs and AD. Methods Based on the 1000 Genomes Project data, linkage disequilibrium (LD) analyses were utilized to screen for candidate SNPs in the TREM1 gene. AD cases (1081) and healthy control subjects (870) were collected and genotyped, and the associations between candidate SNPs and AD risk were analyzed. We explored the associations between target SNP and AD biomarkers. Moreover, 842 individuals from ADNI were selected to verify these results. Linear mixed models were used to estimate associations between the target SNP and longitudinal cognitive changes. Results The rs2062323 was identified to be associated with AD risk in the Han population, and rs2062323T carriers had a lower AD risk (co‐dominant model: OR, 0.67, 95% CI, 0.51–0.88, p = 0.0037; additive model: OR, 0.82, 95% CI, 0.72–0.94, p = 0.0032). Cerebrospinal fluid (CSF) sTREM2 levels were significantly increased in middle‐aged rs2062323T carriers (additive model: β = 0.18, p = 0.0348). We also found significantly elevated levels of CSF sTREM2 in the ADNI. The rate of cognitive decline slowed down in rs2062323T carriers. Conclusions This study is the first to identify significant associations between TREM1 rs2062323 and AD risk. The rs2062323T may be involved in AD by regulating the expression of TREM1, TREML1, TREM2, and sTREM2. The TREM family is expected to be a potential therapeutic target for AD.


| INTRODUC TI ON
Previous studies indicated that the worldwide prevalence of dementia will increase ninefold by 2050. 1 Alzheimer's disease (AD) is a chronic disease with a long preclinical period (about 20 years), and it is the most common type of dementia. 2,3 About 95% of all AD patients worldwide belong to the late-onset AD (LOAD) subtype.
However, its pathogenesis is unclear, and the development of effective diagnostic and therapeutic tools has been hindered. Genetic studies have confirmed that LOAD is a polygenic and heterogeneous disease. And 60-80% of its risk depends on genetic factors. 1 Therefore, screening and verifying LOAD susceptibility or pathogenic genes using genetic techniques is one of the research priorities in this field.
The triggering receptor expressed on the myeloid cells 1 (TREM1) gene is located in human chromosome 6p21.1, and immediately adjacent to the TREM2 gene. 4,5 Monocytes/macrophages and blood neutrophils are the principal effector cells of human TREM1 in innate responses. 5 TREM1 and TREM2 activate myeloid cells by signaling through the adaptor protein DAP12. 5 The TREMs family is not only involved in the regulation of microglia phagocytosis and amyloid clearance, but also plays an indispensable role in the neuroinflammatory response of AD. [6][7][8] Replogle and colleagues reported one common variant of the TREM1 gene (rs6910730G) associated with AD pathology and aging-related cognitive decline. 9 One study found a significant decrease in TREM1 expression levels on the surface of peripheral blood monocytes from rs6910730G carriers. 10 Phagocytosis of Aβ pathology was also significantly diminished by peripheral blood monocytes from rs6910730G carriers. 10 In addition, the TREM1 can facilitate the microglial phagocytosis of Aβ, is associated with immune responses in AD. 10,11 These findings suggested that TREM1 can serve as a potential therapeutic target for AD. 12 However, one study investigated the association between rs6910730G and AD (17,008 AD cases and 37,154 controls) using the International Genomics of Alzheimer's Project (IGAP) dataset and did not find any significant association between rs6910730G and AD susceptibility. 13 Therefore, the association between TREM1 gene polymorphism and AD risk needs further investigation. Moreover, it is necessary to further explore the underlying mechanisms between TREM1 and AD susceptibility using human biological samples. The criteria for exclusion were: (1) family history of genetic diseases, (2)

| Construction of linkage disequilibrium (LD) blocks and selection of tag-SNPs
The 1000 Genomes Project is a transnational cooperation project that contains whole-genome sequencing information for 26 populations from the Americas, Europe, South Asia, East Asia, and Africa. We selected 1000 Genomes Project Han Chinese in Bejing Data (reference: Human Genome grch37) and used haploview 4.1 (Hardy-Weinberg equilibrium p-value > 0.05, MAF > 0.1, LD threshold r 2 > 0.8) to analyze the SNPs locus on TREM1 gene (6:41267385-41286692). 18 Finally, 10 tag-SNPs inside TREM1 were chosen based on the collected SNP information and LD blocks created by Haploview software (Table S2 and Figure S1). In addition, we also performed LD analysis using the Southern Chinese Han Chinese data ( Figure S2) and European data ( Figure S3) to verify the selected SNPs.

| Measurement of AD core biomarkers
CSF was collected from participants in a standardized manner via lumbar puncture. 19 Within 2 h, the CSF was transported to the lab and centrifuged at 2000g for 10 min. Prior to testing, the thaw/ freeze cycle was limited to 2 cycles. Enzyme-linked immunosorbent assay (ELISA) was used to quantify CSF Aβ42 (INNOTEST®; are from a single batch to exclude variability between batches. In addition, the intra-batch coefficient of variation (CV) was <5%, and the inter-batch CV was <15%.
Alzheimer's Disease Neuroimaging Initiative provides sTREM2 data based on two platforms. One of the sTREM2 data is based on the MSD platform and has been comprehensively described in previous publications. [20][21][22] The corrected values were used and are available in the ADNI database as variables "MSD_sTREM2corrected". Furthermore, partial sTREM2 was tested at Washington University in St Louis. (developed in-house: WU platform, Piccio group). The corrected values were used and are available in the ADNI database as variables "WU_sTREM2corrected." The raw values are provided as pg/mL. The intraplate CV for CSF sTREM2 was <10%.

| Statistical analysis
The baseline demographic characteristics (age, gender, APOE4 carriage status, and MMSE score) were compared using ANOVA and chi-square tests. Logistic regression was utilized to analyze the relationship between target SNP and AD risk. Three genetic models, including co-dominant, dominant, and additive models, were applied.
All risk analysis models were corrected for age, sex, and APOE4 carriage status. Finally, Bonferroni correction was employed in the setting of multiple comparisons: p-value < 0.005 was considered significant.
All continuous variables were normalized using the Box-Cox transformations (R software "car" package) and standardized using the z-scale transformations (R software "scale" package). We used multiple linear regression models to analyze the association between the target SNP and AD CSF biomarkers. The analysis used three genetic models (co-dominant, dominant, and additive). All models were corrected for age, sex, and APOE4 status. In addition, we investigated the interaction of age, sex, and APOE4 carriage status on the association of rs2062323 with AD CSF biomarkers. Based on the results of the interaction, we further performed a subgroup analysis.
We included only non-Hispanic white individuals in the ADNI replication cohort. The CSF sTREM2 (MSD and WU platform) data were depolarized (mean ± 3-fold standard deviation), and the data were normalized and standardized using the R software "car" and "scale" packages. We used multiple linear regression models to analyze the association between the target SNP and CSF sTREM2.
Three genetic models (co-dominant, dominant, and additive) were utilized in the analysis. All models were corrected for age, sex, and APOE4 status. In addition, we investigated the interaction of age, sex, and APOE4 carrying status on the relationship between rs2062323 and sTREM2. Finally, we used linear mixed models to explore the effects of target SNP on longitudinal changes in cognition.
The above data were analyzed using SPSS software (version 22.0) and R software (version 3.6.1). We used the International Genomics of Alzheimer's Project (IGAP) to query the information related to our target SNP. In addition, expression quantitative trait loci (eQTL) analyses were conducted using multiple publicly available datasets (Genotype-Tissue Expression (GTEx), https://gtexp ortal.org/home/ snp/rs206 2323) in human brain tissues and the whole blood. 23

| Characteristics of the study population
We included 870 cognitively normal individuals and 1081 AD patients to study the association between candidate SNPs and AD risk. Table 1 shows the demographic characteristics of these Han Chinese subjects. We studied the relationship between target SNP loci and AD biomarkers in 503 cognitively normal Han Chinese populations. The demographic characteristics of these 503 individuals are presented in Table 3. In addition, we included 676 non-demented and 166 AD participants in the ADNI replication cohort. The demographic characteristics are shown in Tables S5 and S7.

| Allele frequencies and genotype distributions of the tag-SNPs
The genotype distribution of the target SNPs is shown in Tables S1 and S2. We identified a separate genetic locus rs2062323 on the TREM1 that was significantly associated with AD risk in the Han Chinese population.

| Association between tag-SNPs within the TREM1 gene and the risk of AD
People carrying the rs2062323T allele had a significantly lower risk of AD (

| Association between tag-SNP and CSF AD biomarkers
In the total population, no significant associations between the rs2062323 and AD CSF biomarkers were found in either the codominant, dominant, or additive models (Table S3). In addition, interaction analysis revealed an interaction between the rs2062323 and CSF sTREM2 with age (β = −0.0308, p = 0.0436) (Table S4). CSF sTREM2 levels were significantly higher in middle-aged (≤65 years) Han Chinese carrying rs2062323T compared to non-carriers in age-based subgroup analyses ( Figure 1 and Table 4). We found no significant correlations in the elderly population (>65 years) (Table 4).

TA B L E 2
Associations between tag-SNPs and the risk of sporadic Alzheimer's disease.

| Replication in the ADNI cohort
We found significantly higher CSF sTREM2 (MSD and WU platform) levels in people carrying rs2062323T in the ADNI database. For MSD CSF sTREM2, significant associations existed in the non-demented population ( Figure 2A, Figure 2B, and Table 5) and AD patients ( Figure 2C, Figure 2D, and Table 5). For WU CSF sTREM2, significant associations existed both in the non-demented population and AD patients (Table S8).
In addition, we performed interaction analyses in both diagnostic groups. Age, sex, and APOE4 genotype had no significant effect on the relationship between the rs2062323 and CSF sTREM2    (Tables S6 and S9). We also investigated longitudinal cognitive changes in a non-demented population. The results showed that the rate of cognitive decline was significantly slower in rs2062323T carriers compared to non-carriers ( Figure 3).

| Functional annotation of rs2062323
The IGAP did a meta-analysis using the existing Genome-Wide Association Studies (GWAS) dataset in 2013. The stage 1 metaanalysis included 17,008 AD cases and 37,154 controls of European ancestry. Using these summary results, we identified that the rs2062323 was significantly associated with AD risk (effect allele, T, β = −0.0381, p = 0.03086). 24 Moreover, multi-tissue eQTL analyses showed that rs2062323 could significantly influence TREM1, TREM2, and TREML1 expression in the whole blood and specific brain regions (including basal ganglia, hypothalamus, and hippocampus) ( Figures S4-S6). TREM1, TREM2, and TREML1 expression levels are elevated in rs2062323T carriers ( Figures S4-S6).

| DISCUSS ION
In conclusion, we identified a genetic variant in TREM1 is associated with the risk of AD in the Han Chinese population. The rs2062323T carriers had a significantly reduced risk of AD. In addition, the protective effect of rs2062323T on AD also applies to the European the association between rs2062323 and sTREM2. In addition, associations between rs2062323T and sTREM2 were significant in AD and non-AD populations, suggesting that these associations may not be affected by diagnostic status.
The GTEx databases showed that rs2062323T was associated with high expression of the TREM1, TREM2, and TREML1 genes in parietal brain regions. The TREM1, TREM2, and TREML1 belong to the TREM receptor family and can regulate inflammation by magnifying or inhibiting the toll-like receptor-induced signaling. 25 Human TREM1 was an amplifier of acute inflammation and can link innate and adaptive immunity. 5 TREM family genes were located at 6p21.1 in humans. Previous studies suggested that the 6p21.1 region is closely related to AD. 26 Neuroinflammation is one of the critical pathogenic mechanisms of AD and is associated with selective neuronal vulnerability. 27,28 Microglia are immune cells residing in the central nervous system and play a critical regulatory role in maintaining brain homeostasis. Microglia (namely macrophages) are important mediators of neuroinflammation in AD. 29 Previous studies have revealed associations between the TREM family genes (mainly TREM1 and TREM2) and AD risk. Given the homology among the TREM family genes, the TREM region likely contains a number of distinct variations and genes that affect various aspects of AD risk. 9 Previous research has demonstrated that the adaptor protein DAP12 facilitates the signaling of TREM1 and TREM2, which in turn activates myeloid cells. 5 The TREM2 protein is present throughout the brain's white matter and is more abundant in the neocortex and hippocampus, but is absent from the cerebellum. 30 Additionally, the TREM2 protein is primarily expressed in myeloid cells, including microglia, dendritic cells generated from monocytes, osteoblasts, and macrophages produced from bone marrow. 5 Previous studies have shown that mutations and polymorphisms in the TREM2 gene (rs75932628, rs143332484, rs142232675) are significantly associated with the risk of developing AD. 31,32 These AD-associated risk polymorphisms have pathogenic consequences due to reduced TREM2 function. 30

F I G U R E 1
Associations of rs2062323 genotypes with CSF sTREM2 in middle-aged (≤65 years) Han Chinese population. We categorized the rs2062323 into four subgroups: CC, CT, TT, and CT/TT. We found that CSF sTREM2 was significantly higher in rs2062323T carriers.
In AD patients, microglia processing of Aβ is an expansion and acti- protein. 34 Suárez-Calvet et al. showed that CSF sTREM2 levels were significantly elevated in the early stages of AD, which may reflect a corresponding change in microglia activation status after neuronal degeneration. 22 Heslegrave et al.' study also revealed significantly elevated levels of CSF sTREM2 in AD, which may imply that CSF sTREM2 can be utilized to quantify the activation of glial cells. 35 Notably, as AD progresses, the level of sTREM2 varies dynamically.
CSF sTREM2 was correlated with markers of neuronal damage (CSF T-tau and CSF p-tau), suggesting that it could be used as a biomarker of neurodegeneration. In addition, sTREM2 was found to co-localize with neurons and plaques in vivo, and sTREM2 may play a chemoattractant role in recruiting microglia to the vicinity of plaques. 36,37 The sTREM2 can also trigger microglia activation and induce an inflammatory response to protect microglia from apoptosis. 38 In addition, sTREM2 has a protective effect against amyloid pathology and associated toxicity. The sTREM2 not only reduces a load of amyloid plaques to rescue functional defects in spatial memory, but also enhances microglia proliferation, migration, aggregation, and uptake and degradation of Aβ in the vicinity of amyloid plaques. 36,38 The F I G U R E 2 Associations of rs2062323 genotypes with CSF sTREM2 in ADNI. We categorized the rs2062323 into four subgroups: CC, CT, TT, and CT/TT. We found that CSF sTREM2 (MSD platform) were significantly higher in CT and CT/TT groups in the non-demented population (A and B). In addition, CSF sTREM2 was significantly higher in CT/TT groups in AD patients (C and D).

| CON CLUS ION
In conclusion, TREM1 rs2062323 is an AD-protective SNP that is linked to higher levels of CSF sTREM2. The rs2062323 may be involved in AD by regulating the expression of TREM1, TREML1,

ACK N OWLED G M ENTS
This study was supported by grants from the National Natural Science

FU N D I N G I N FO R M ATI O N
This study was supported by grants from the National Natural Science Foundation of China (81771148, 82071201).

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare that they have no conflict of interest, financial, or otherwise.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

E TH I C S A PPROVA L
All data sources used in this study received approval from an ethics standards committee on human experimentation and the procedures used in this study adhere to the tenets of the Declaration of Helsinki. Written consent for genetic screening were obtained from all participants or their legal representatives. Their confidentiality was preserved according to the guidelines for studies of human subjects. About ADNI, data involved in the study came from the publicly open Alzheimer's disease neuroimaging initiative (ADNI) database.

CO N S E NT TO PA RTI CI PATE
Informed written consent was obtained for all participants.