Association of Frontotemporal Dementia GWAS Loci with Late-Onset Alzheimer’s Disease in a Northern Han Chinese Population

INTRODUCTION

Alzheimer’s disease (AD) is one of the most common causes of dementia in the elderly, which is neuropathologically characterized by the formation of extracellular senile plaques from amyloid-β (Aβ) and intracellular neurofibrillary tangles containing hyperphosphorylated tau protein [1, 2]. Similarly, a subset of frontotemporal dementia (FTD), the second most common form of dementia, is a class of neurodegenerative diseases associated with the pathologic aggregation of tau protein in the human brain [3, 4]. Strong similarities in cerebrospinal fluid (CSF) biomarkers, imaging markers, and disease progression profiles suggest that some or most of the pathophysiology is shared between AD and FTD [5, 6]. AD and FTD are the two common age-related neurodegenerative diseases. Clinically, both AD and FTD are characterized by progressive deficits in cognitive function [7]. Neuropathologically, AD is characterized by the formation of Aβ and tau protein, whereas FTD involves deposition of tau, transactive response DNA binding protein 43, fused-in-sarcoma, and dipeptide-repeat proteins [3]. Importantly, although tau-associated pathology is considered a hallmark of AD, primary tauopathies also belong to the FTD [8]. CSF Aβ_1 - 42, total tau, phosphorylated tau, and the Aβ_1 - 42: phosphorylated tau ratio all correlated with disease severity in AD, and low Aβ_1 - 42: phosphorylated tau ratio also correlated with low MMSE performance in FTD [9]. Recent findings have support for an active role of chronic neuroinflammation in AD pathophysiology [10]. Meanwhile, further studies also proposed a link between neuroinflammation and specific forms of FTD [11]. In addition, the main mendelian genes shared between AD and FTD include genes coding for microtubule-associated protein tau (MAPT) and Chromosome 9 Open Reading Frame 72 (C9ORF72) gene [12]. Moreover, much effort to date has been made to research other potential genetic risk factors that might contribute to both diseases. Recent researches provided evidence that heterozygous rare variants in TREM2 gene (encoding the triggering receptor expressed on myeloid cells 2 on chromosome 6p21.1-q15), is implicated with a significant increase in the risk of AD [13]. Similarly, using exome sequencing a study has revealed mutations in TREM2 in patients presenting as an FTD-like syndrome [14]. All these connections suggest that AD and FTD might have some shared risk factors and/or common pathogenic mechanisms.

Recently, a large genome-wide association study (GWAS) reported several novel susceptibility genetic loci for FTD, including RAB38 (rs302668), RAB38/CTSC (rs16913634), HLA-DRA/HLA-DRB5 (rs9268877 and rs9268856), and BTNL2 (rs1980493). The expression levels and functional features of these genes might affect expression and methylation in cis and indicate that immune system processes, and possibly lysosomal and autophagy pathways, are potentially involved in the pathogenesis of FTD [15]. Meanwhile, all these genes might be involved in immune system processes though lysosomal and autophagy pathways, which are referred as a secondary pathological mechanism for AD [16]. Therefore, we explored the association of late-onset AD (LOAD) with selected single-nucleotide polymorphisms (SNPs) from the FTD GWAS (rs302668 for RAB38, rs16913634 for RAB38/CTSC, rs9268877 and rs9268856 for HLA-DRA/HLA-DRB5, and rs1980493 for BTNL2) in a large LOAD case-control study.

METHODS

RESULTS

Demographic and clinical characteristics of AD and control subjects are shown in Table 1. The 984 LOAD cases were well matched with the 1,354 control subjects in terms of gender (p = 0.068) and age (p = 0.186). As expected, APOE ɛ4-carriers had higher risk for LOAD (odds ratio [OR], 2.422; 95% confidence interval [CI], 1.970–2.977; p <  0.001).

All SNPs genotype distributions were consistent with the Hardy– Weinberg equilibrium in both AD patients and controls (p >  0.05). In the total sample (Table 2), we observed significant differences in genotype distributions of rs302668 (RAB38, p_c = 0.025), rs9268877 (HLA-DRA/HLA-DRB5, p_c = 0.025), rs9268856 (HLA-DRA/HLA-DRB5, p <  0.001), and rs1980493 (BTNL2, p_c = 0.045) between cases and controls. The SNPs rs16913634 for RAB38/CTSC was unrelated to LOAD risk (p = 0.088).

Furthermore, we performed a multivariate logistic regression analysis after adjusting for age, gender, and the APOE ɛ4-carrier status to assess the effect of each SNP on LOAD risk (Table 3). For rs9268856, A allele homozygosity decreased the risk of LOAD under a recessive model (OR = 0.569; 95% CI = 0.393–0.823; p_c = 0.045). For rs9268877, dominant model revealed that A allele raised the risk of LOAD (dominant model: OR = 1.153; 95% CI = 1.021–1.303; p_c = 0.015).

Next, we stratified the subjects in the AD cases into MMSE ≥21, MMSE between 10 to 20, and MMSE ≤9 subgroups according to MMSE score to investigate whether the SNPs analyzed in this study are associated with altered cognitive function. Minor allele frequencies and genotype distribution of SNPs in each subgroup were summarized in Table 4. Only the allele frequencies of rs302668 (RAB38) differed significantly among the subgroups (p <  0.001). For other SNPs, neither genotype distribution nor minor allele frequency presented significant differences among the subgroups (p >  0.05).

In addition, we conducted LD and haplotype analysis in total sample. Two SNPs (HLA-DRA/HLA-DRB5 (rs9268877 and rs9268856)) formed a haplotype block in total sample (D’ = 1.0, r² = 0.207) (Fig. 1). Furthermore, we found three haplotype (CA, CG, AG) derived from these two SNPs, but they showed no significant association with LOAD in any data set (p >  0.05) (Table 5).

DISCUSSION

RAB38 (rs302668), RAB38/CTSC (rs16913634), HLA-DRA/HLA-DRB5 (rs9268877 and rs9268856), and BTNL2 (rs1980493) were all implicated in FTD pathology and might be involved in immune system processes though lysosomal and autophagy pathways [15]. To explore whether these genes cause susceptibility to LOAD, we analyzed RAB38 (rs302668), RAB38/CTSC (rs16913634), HLA-DRA/HLA-DRB5 (rs9268877 and rs9268856), and BTNL2 (rs1980493) loci in a Han Chinese data set of 2,338 subjects. In this large case-control study, we have identified four novel potential loci for LOAD in a northern Han Chinese population: rs302668 (p_c = 0.025) at the RAB38 locus, rs9268877 (p_c = 0.025) and rs9268856 (p_c <  0.001) at the HLA-DRA/HLA-DRB5 locus, and rs1980493 (p_c = 0.045) at the BTNL2 locus. For rs16913634 at the RAB38/CTSC locus, we did not detect any association with LOAD in this population.

LOAD is characterized by a broad range of clinical manifestations, differential pathological signatures, and substantial genetic variability, which imply complex disease mechanisms [19]. Beside the two hallmark pathologies, extracellular Aβ peptides and intracellular neurofibrillary tangles, an impressive array of studies clearly implicate the important role of innate and the adaptive immune response in the pathology of AD [20]. Additionally, lysosomal and autophagic dysfunctions are referred as a secondary pathologic mechanism for AD [21]. Moreover, the immune system and autophagy were also proved to be involved within the spectrum of neurodegenerative diseases, including FTD, Parkinson’s disease (PD), and amyotrophic lateral sclerosis [22].

RAB38, the gene encoding the transmembrane protein RAB38, is located at chromosome 11q14. The SNP rs302668 shows a relative association with AD in the entire dataset. Our present results are not consistent with those of from individuals of European ancestry. The meta-analysis of 74,046 individuals on 4 previously published GWAS data revealed the negative results (p = 0.322) [23]. Similarly, the meta-analysis of PD GWAS got the same results (p >  0.05) [24]. It is possible that different genetic backgrounds or environments make a difference between Caucasian and Han Chinese subjects. RAB38 is preferentially expressed in melanocytes, in the thyroid, in elements of the immune system, and in the brain [25]. It plays a crucial role in the docking/fusion of transport vesicles or lysosomal-related organelles [26]. In addition, RAB38 might be involved in autophagic processes [27]. A possible role for autophagy has also been shown in AD [28]. Hence, we hypothesizes that RAB38 might be involved in the AD pathology through the lysosomal and autophagic processes. Further studies need to be done to lend support to this hypothesis. CTSC encodes the cathepsin C protein (dipeptidyl-peptidase I), a lysosomal exo-cysteine proteinase belonging to the peptidase C1 family. Cathepsin C can remove dipeptides from the amino terminus of proteins and participates in the zymogen activation of serine proteinases, which was involved in immune and inflammatory processes, including phagocytosis of pathogens and local activation and deactivation of inflammatory factors [29]. The SNP rs16913634 at the RAB38/CTSC locus shows no association with AD. Similarly, this negative result was proved in the meta-analysis of GWAS data (p = 0.331) [23]. Meanwhile, there is no association of rs16913634 with PD (p >  0.05) [24].

The HLA locus plays a key role in susceptibility to many neurodegenerative diseases, including AD [30], FTD [15], and PD [31]. In this large case-control study, we have identified three SNPs exceeding significance at the HLA locus (6p21.3). Except for these significant SNPs pursued in this study, the most powerful evidence comes from the meta-analysis of GWAS data, which catalogs all genetic association studies in AD [23], currently lists rs9271192 polymorphisms that are significant when tested for allelic association. Interestingly, we find positive results in the meta-analysis of PD GWAS (rs9268877; p <  0.001) [32]. The SNPs rs9268877 and rs9268856 were located on chromosome 6p21, in the intergenic region between HLA-DRA/HLA-DRB5. This region is functionally associated with immunocompetence and histocompatibility. The SNP rs1980493 at the BTNL2 locus codes for the butyrophilin like protein 2. BTNL2 protein reduces proliferation and cytokine production from activated T cells [33]. The role of functional BTNL2 as a negative costimulatory molecule involved in T-cell activation might influence the innate immune response. This inappropriate activation of innate immune response will lead to reinforcement of amyloid load and neurotoxicity [20]. Although common variants at the BTNL2 locus have been previously shown to be associated with FTD, we did not get the positive results of BTNL2 (rs1980493) from the AD and PD GWAS (p >  0.05). Future studies should aim to replicate our findings and elucidate the functional basis of BTNL2.

In conclusion, our study mainly suggests that the rs302668 in RAB38, rs9268877, and rs9268856 in HLA-DRA/HLA-DRB5, and rs1980493 in BTNL2 might play a role in the susceptibility to LOAD in the Han Chinese population. The SNP rs16913634 at the RAB38/CTSC locus might not play a major role in the susceptibility to LOAD in the Han Chinese population. To the best of our knowledge, this is the first attempt to explore the association of FTD associated loci in LOAD pathogenesis in a northern Han Chinese population. Therefore, the present results require confirmation in additional and larger studies in Han Chinese as well as in other ethnic groups. Further studies are needed to investigate how these SNPs may influence the function of the encoded protein in LOAD patients. Additionally, our research lends further support to the viewpoint that common pathways and processes might underlie different forms of neurodegenerative disorders, including AD, PD, and FTD. Furthermore, exploring the possibility of current therapeutic measures targeting general damage responses could be harnessed to open exciting new therapeutic perspectives for aging and neurodegenerative disorders, including AD.