TREM2 coding variants in Slovak Alzheimer’s disease patients

Background: Triggering receptor expressed on myeloid cells 2 (TREM2) is an important modulator of innate immune responses. In the human brain, TREM2 is primarily expressed on microglia and is involved in cell survival, phagocytosis, and regulation of inflammation. TREM2 dysfunction has been linked to the pathogenesis of various neurodegenerative diseases including Alzheimer’s disease (AD). Rare coding variants of the TREM2 gene have been reported to modulate AD risk in several populations, however, data on their association with susceptibility to AD in the Slovak population have been missing. Methods: We have analyzed 10 non-synonymous coding variants located in TREM2 exon 2 by direct sequencing in 270 late-onset Alzheimer’s disease (LOAD) patients and 331 controls. Results: Four out of 10 TREM2 mutant variants have been identified in the analyzed groups, namely rs75932628 C > T (R47H), rs142232675 C > T (D87N), rs143332484 C > T (R62H), and rs2234253 G > T (T96K). R47H was found only in the AD group, while T96K was present only in the controls. Although no significant association between TREM2 coding variants and LOAD susceptibility has been detected, the observed odds ratio (OR) of 3.69 for R47H carriers suggests an increased risk of LOAD for this variant in the Slovak population. Moreover, we also found a higher OR for the rs143332484-T allele in APOE ε4 non-carriers (1.99) when compared to APOE ε4 carriers (0.62). Conclusions: Our results suggest an impact of specific TREM2 rare coding variants on AD risk in the Slovak population.

In the human brain, TREM2 is primarily expressed on microglia and is involved in the phagocytosis of apoptotic neurons, and modulation of inflammation [10,[14][15][16]. As revealed in microglia, the reduction of TREM2 signaling increases TNF and NO synthase-2 (NOS2) transcription, while overexpression of TREM2 decreases transcription of TNF, IL1β, and NOS2 [14].
Several TREM2 genetic variants have been identi-fied that increase the risk of late-onset Alzheimer's disease (LOAD). The most-well studied variant is rs75932628 C > T, a single nucleotide polymorphism (SNP) in exon 2 encoding an arginine to histidine missense substitution at amino acid 47 (R47H). Two independent studies in 2013 found for the first time that heterozygous form of the TREM2 R47H is associated with LOAD risk [34,35]. This rare TREM2 variant can increase AD risk by 2-4 fold which is comparable to the increased risk associated with one copy of the known APOE ε4 risk allele [36,37]. The association of the TREM2 R47H variant with AD has been observed in European [38][39][40][41][42][43][44][45][46], and Afro-American populations [47,48]. However, the R47H variant does not seem to be associated with LOAD risk in the Asian population [49][50][51][52][53][54]. Other TREM2 variants suggested as risk factors for developing LOAD include D87N, R62H, T96K, Y38C, L211P, and H157Y [34,38,47,[55][56][57][58][59][60]. Most variants are associated with decreased cell-surface expression of TREM2 (R47H, Y38C, R136Q) or by impaired in vitro interactions with ligands like APOE, LDL, and clusterin (R47H, R62H and D87N) [27,61,62]. R47H carriers in AD patients also demonstrated upregulation of IFN type I response and proinflammatory cytokines [63]. Moreover, the R47H variant of sTREM2 is less able to bind and disaggregate oligomeric Aβ, which leads to Aβ protofibril formation and neurotoxicity [64]. The aim of our study was to perform an association analysis between TREM2 coding variants and risk of LOAD in the Slovak population. We have analyzed 10 nonsynonymous rare variants in exon 2 of the TREM2 by means of direct sequencing. To our knowledge, no such analysis has been performed in the Slovak population until now.

Study groups
The case-control study involved 270 late-onset AD patients (99 men and 171 women, mean age: 78.56 ± 6.18 years). The diagnosis of probably AD was established by physicians according to the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA) diagnostic criteria [65]. The clinical examination included personal medical and family history assessment, neurologic examination, neuropsychological assessment, and neuroimaging. The clinicians estimated age at onset (AAO) using standardized methodology; it is the age, at which the patient started to show significant symptoms of memory loss and cognitive impairment with a progressive clinical course. Unrelated cases with earlyonset AD (age at onset <65 years) and having other neurologic or psychiatric diseases were excluded from the study. All AD patients have been recruited in the period from 2016 until 2020 via several psychiatric clinics throughout Slovakia. The mean age at disease onset was 75.50 ± 6.38 years. The reference cohort in our case-control study com-prised 331 unrelated volunteers (133 males and 198 females with mean age of 76.05 ± 7.64 years). Montreal Cognitive Assessment (MoCA) was selected as the screening test for cognitive impairment in this study [66]. The cut-off score of 26 from 30 has been considered for normal cognition. All control subjects were without any personal or family history of AD and they were recruited from a matched group of a larger population sample. All patients and controls were Caucasians of Slovak descent. APOE ε4 allele as a known genetic risk factor for AD was evaluated in both study groups and implemented as a stratification factor in further analyses. Detailed parameters of the study groups are summarized in Table 1. This case-control study was approved by the Independent Ethical Committee of the Bratislava Municipality under the No. 05440/2021/HF. All the investigations were carried out in accordance with the International Ethical Guidelines and the Declaration of Helsinki. From all patients or their representatives, as well as from control subjects, informed written consent for enrolling in the study and for personal data management was obtained.

Sequence analysis
Genomic DNA was isolated from ethylenediamine tetraacetic acid (EDTA)-treated whole blood samples (2 mL) by a modified salting-out procedure [67]. TREM2 exon 2 region from 6472 to 7004 bp was amplified using forward primer 5 ′ -TCCTTCAGGGCAGGATTTTT-3 ′ and reverse primer 5 ′ -AGTGGGTGGTTCTGCACAC-3 ′ . A 25 µL PCR reaction mixture contained 50 ng of template DNA, 0.2 mM of each dNTP (Thermo Fisher Scientific Inc., Lot. Nr. 00654321, MA, USA), 1 U of Taq DNA polymerase (Thermo Fisher Scientific Inc., Lot Nr. 1817560, MA, USA), 1.5 mmol MgCl2 (Thermo Fisher Scientific Inc., Lot Nr. 1817560, MA, USA), and 10 pmol of each specific primer. The PCR conditions consisted of initial denaturation at 94°C for 3 min, followed 30 cycles of denaturation at 94°C for 1 min, annealing at 56°C for 30 s, and elongation at 72°C for 1 min. The final elongation at 72°C for 10 min completed the reaction. The PCR products were run in 1.0% agarose gel for 20 min and then visualized under UV light. A fragment size of 533 bp was confirmed using the 100 bp DNA ladder (Solis BioDyne Inc., Lot. Nr. 07110000, Estonia, Europe). To perform sequence analysis, the PCR products were purified using an EXO SAP-IT kit according to the manufacturer's recommendations (USB Inc., Lot. Nr. 00123457, WI, USA) and then sequenced (approximately 100 ng) using BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific Inc., Lot Nr. 00567001, MA, USA). Direct sequencing was performed individually on both strands using forward and reverse PCR primers. The sequencing reaction was run on a 3130 ABI PRISM Genetic analyzer (Applied Biosystems Inc., MA, USA) and the sequence data were analyzed by SeqScape software (version 3.0, Applied Biosys- Differences in age and MoCA score between the two groups were examined by the Mann-Whitney unparametric test. Differences in sex and APOE ε4 status were assessed using the Pearson χ 2 test. p < 0.05 is considered statistically significant. CI, confidence interval; LOAD, late-onset Alzheimer's disease; MoCA, Montreal Cognitive Assessment; n, number; OR, odds ratio; SD, standard deviation; y, years. tems Inc., MA, USA). The reference sequence was reported on the website http://www.ebi.ac.uk/imgt/hla/align.html. Genotyping of the APOE ε4 risk allele was performed by the determination of rs429358 (C > T) and rs7412 (T > C) polymorphisms in the fourth exon using direct sequencing as described previously [68].

Statistical analysis
Allele and genotype frequencies were determined by direct counting. Genotypes were tested for their fit to Hardy-Weinberg equilibrium using the chi-squared goodness-of-fit test. Statistical differences in allele and genotype frequencies between AD patients and the control group were evaluated by the Pearson chi-squared test using the InStat statistical software (version 3.10, Graph-Pad Software Inc., CA, USA). The p values, odds ratios (OR), and 95% confidence intervals (95% CI) were calculated in the codominant inheritance model. The multivariate logistic regression analysis adjusted for sex, age (age at onset in patients and age at the examination in controls), and APOE ε4 carrier status as possible modifying factors was performed by the SNPstats web software available at http://bioinfo.iconcologia.net/SNPstats [69]. Regression analysis and synergy factor (SF) measurement were also performed to assess the significance and size of interaction between TREM2 variants and the APOE ε4 allele, as previously described [70]. The p value of <0.05 was considered as statistically significant.

Characteristics of the study groups
The demographic and clinical characteristics of the study groups are shown in Table 1. The study included 270 AD patients and 331 unrelated controls. There was no statistically significant difference between the AD group and controls in relation to gender (p = 0.38), with females having a higher prevalence in both AD patients (63.33%) and controls (59.82%). The mean age at examination was significantly higher in the AD group than in controls (78.56 versus 76.05 years; p < 0.0001). Concerning the MoCA assessment, there was a significantly lower MoCA score in AD patients compared to controls (14.53 versus 27.53; p < 0.0001). A significantly higher prevalence of the APOE ε4 risk allele was found in the AD group compared to controls (49.63% vs 19.03%, p < 0.0001).

Association of TREM2 variants with LOAD risk
Ten non-synonymous rare variants in TREM2 exon 2 have been analyzed in 270 AD patients and 331 controls: , and rs374851046 C > G (R52H). Allele and genotype frequencies of the TREM2 variants observed in the analyzed groups are shown in Table 2. Four out of 10 TREM2 coding variants have been identified in the analyzed groups, namely R47H, D87N, R62H and T96K. Genotype frequencies of TREM2 variants fit the Hardy-Weinberg equilibrium in AD patients (p = 0.98 for R47H; p = 0.93 for D87N, p = 0.80 for R62H, p = NA for T96K) as well as in controls (p = NA for R47H, p = 0.87 for D87N, p = 0.82 for R62H, p = 0.98 for T96K).
The carriage of the rs75932628-T allele (R47H) was identified in one of the AD cases (0.19%), while this allele was missing in the control group. Univariate chi-square analysis revealed that the carrier of the minor T allele had a 3.69-fold increased risk to develop AD compared to non-T carriers. Regarding other TREM2 variants, the D87N variant was identified in 3 AD cases (0.56%) and 6 controls (0.91%), R62H was identified in 8 AD cases (1.48%) and 7 controls (1.06%) and T96K variant was identified in one control subject (0.15%). No statistically significant differences in either TREM2 mutant allele or genotype frequencies were found between the AD group and controls (p > 0.05, Table 2). Multivariate analysis of the association between the TREM2 variants and LOAD risk adjusted for age, sex and APOE ε4 status as potential confounding variables revealed no changes in comparison with the univariate analysis (p > 0.05, Table 2). Logistic regression analysis also revealed no statistically significant p value for the interaction of TREM2 variants with APOE ε4 allele dosage (p > 0.05, Table 2). NA * Logistic regression analysis adjusted for age (age at onset in patients and age at examination in controls), sex and APOE ε4 status. † Crude/adjusted analysis (for age and sex) of interaction between TREM-2 mutant allele and APOE ε4 allele. p < 0.05 is considered statistically significant. -: OR values couldn't be calculated by software program. a, mutant allele; A, wild-type allele; CI, confidence interval; LOAD, late-onset Alzheimer's disease; n, number; NA, not available; OR, odds ratio.

Table 3. Association between TREM2 rs143332484 C > T (R62H) variant and LOAD risk in subjects stratified by APOE ε4
status.

Analysis of TREM2 variants in relation to APOE ε4 carrier status
Association between the TREM2 variants and LOAD risk in subjects stratified by APOE ε4 carrier status was also performed. Analyses in APOE ε4-positive and APOE ε4negative groups revealed no statistically significant differences in the distribution of TREM2 variants between AD patients and the control group (p > 0.05 for R62H variant, Table 3). However, we found a somewhat higher odds ratio for rs143332484-T (R62H) in APOE ε4 non-carriers (1.99) than in APOE ε4 carriers (0.62).
To further determine the genetic interaction between the APOE ε4 allele and the TREM2 variants, we assessed the combined gene effects of TREM2 rs143332484-T (R62H) and the APOE ε4 alleles on LOAD risk (Table 4). When compared to subjects without both alleles, carriers of rs143332484-T, but no APOE ε4 were at two-fold increased odds of developing LOAD (p = 0.32), while the OR in subjects with at least one copy of APOE ε4, but no TREM2 rs143332484-T (R62H) was 4.33 (p < 0.0001). In subjects with both alleles, the OR was reduced to 2.67 (p = 0.19). This observed combined effect size of the two alleles was markedly lower than the predicted joint OR assuming independent effects of both rs143332484-T and APOE ε4 (OR = 8.66), however, the difference was not significant (p = 0.27). The calculated SF value of 0.31 indicates antagonism between TREM2 rs143332484-T (R62H) and APOE ε4 alleles in LOAD risk (Table 4).

Discussion
TREM2 is a pattern recognition receptor expressed on myeloid cells involved in the modulation of the innate immune response [3]. In the human brain, TREM2 is primarily expressed on microglia, and is involved in cell survival, chemotaxis, phagocytosis, and regulation of inflammation [10,[14][15][16]. It was found that TREM2 can protect against AD by binding with Aβ40 and Aβ42 monomers [27][28][29][30][31]33]. Moreover, the soluble form of TREM2 is involved in the inhibition of Aβ aggregation thus it prevents the formation of pathological amyloid plaques [64].
Rare coding variants in TREM2 were identified as risk factors for Alzheimer's disease in several populations. In 2013 two independent studies described for the first time the association of R47H with LOAD risk [34,35]. Since then the R47H variant has been consistently reported to increase the risk for AD across ethnicities as stated in meta-analyses [37,46,54,59,60] and large-scale GWAS analyses [71,72].
In this study, we examined the contribution of TREM2 rare variants on risk for LOAD in the Slovak population. We have analyzed 10 non-synonymous coding variants located in TREM2 exon 2 encoding the ectodomain. Four out of 10 TREM2 mutant variants have been identified in analyzed groups, namely R47H, D87N, R62H and T96K. No statistically significant differences in the distribution of the TREM2 R47H variant were found between AD patients and the control group. This finding is due to the low frequency of the risk allele in the study population and agrees with other reports [39,55]. In our study, the R47H variant was identified in one AD case (0.19%), but it was absent in the control group. The OR for R47H was 3.69 what is comparable with previous studies on this variant in Caucasian populations with pooled OR of 3.93 [37].
The distribution of R47H variant in AD patients seems to differ across ethnicities. The rs75932628-T was found in 0.6 to 1.4% of AD cases in the Spanish population [38,40], 1.6 to 2.1% in the French population [73,74], 0.74% of AD cases in the Belgian population [55], 0.4% in the Polish population [58] [48], 0.98% in the Argentinian population [59], and in 0.05% in the Japanese population [51].
Regarding other TREM2 variants in the Slovak population, D87N and R62H were found in both AD patients and controls, while T96K was identified in one control subject. No statistically significant differences in the distribution of TREM2 mutant variants were found between AD patients and the control group, as reported in other studies [34,47,55].
We are aware of several limitations of the current study. First, the relatively small sample sizes may reduce the power of the study to detect associations between TREM2 gene variants and the risk of late-onset Alzheimer's disease. A larger number of LOAD patients would be desirable for replication, especially with regard to the low prevalence of most TREM variants. We assume our present-day results as useful for sample size planning in future investigations on this topic. Secondly, the estimation of age at onset may be biased by different factors such as the significance of clinical symptoms as experienced by patients.
The diminished effect of TREM2 rare coding variants on protein function has been confirmed by biochemical analyses. The most studied R47H variant showed decreased cell-surface expression and impaired ligand-binding [9,77]. R47H variant also affects TREM2 maturation [82]. Finally, the R47H variant of soluble TREM2 is less capable of binding and disaggregating oligomeric Aβ, thus promoting Aβinduced neurotoxicity [64]. As a consequence of loss of TREM2 function, R47H carriers showed reduced myeloid cell responses to amyloid deposition and more frequent αsynucleinopathy [45,83]. Some studies found that R47H carriers had a trend toward a decrease in the age of AD onset [41,45], however, others revealed no significant association [40,43].
In our study we also calculated the synergy factor value that has predicted an antagonism between TREM2 rs143332484-T (R62H) and APOE ε4 alleles in LOAD risk. It is not clear whether the observed antagonistic interaction between TREM2 rs143332484-T (R62H) and APOE ε4 alleles correlates with the biological events in AD subjects. APOE ε4 has been associated with impaired binding and clearance of Aβ leading to deposition of amyloid plaques in the brain [89,90]. Moreover, APOE ε4 has been involved in impaired synaptic integrity and plasticity [91]. Carriers of at least one APOE ε4 allele exhibited an earlier disease onset, faster disease progression, and increased brain atrophy [36,92,93].
The APOE-TREM2 relationship has been studied by several authors [6][7][8]94]. It was found that TREM2 is binding to APOE to enhance the phagocytosis of apoptotic neurons [6]. Regarding the TREM2 R62H variant its decreased ligand affinity has been observed [9]. It can be hypothesized that impaired ligand affinity of R62H affects phagocytosis of APOE-bound apoptotic cells by microglia contributing to AD pathology.

Conclusions
In our case-control study, we assessed the contribution of TREM2 rare variants on risk for LOAD in the Slovak population. We have analyzed 10 non-synonymous coding variants located in TREM2 exon 2 encoding for the extracellular domain. Four out of 10 TREM2 mutant variants have been identified in both analyzed groups, namely R47H, D87N, R62H, and T96K. R47H substitution was found only in the AD group, while T96K was present only in the controls. The OR of 3.69 in TREM2 R47H carriers suggests an increased risk of this variant for LOAD also in the Slovak population.

Author contributions
VD and IS-study design and manuscript writing. ZP, BV and IK-sample collection. GM and RP-analysis by direct sequencing. JJ-data analysis and interpretation. VD and AO-PCR assays. SS-data collection.

Ethics approval and consent to participate
The study was approved by the Independent Ethical Committee of the Bratislava Municipality under the No. 05440/2021/HF. Informed written consent was obtained from all participants.