Neurofibrillary tau Pathology Modulated by Genetic Variation of α-Synuclein

Abstract

We analyzed whether genetic variation of α-synuclein modulates the extent of neuropathological changes in a population-based autopsied sample of 272 elderly Finns. None of the 11 markers was associated with the extent of neocortical β-amyloid pathology. The intron 4 marker rs2572324 was associated with the extent of neurofibrillary pathology (p = 0.0006, permuted p = 0.004; Braak stages IV-VI vs 0-II). The same variant also showed a trend for association with neocortical Lewy-related pathology. These results suggest for the first time that variation of α-synuclein modulates neurofibrillary tau pathology and support the recent observations of an interaction of α-synuclein and tau in neurodegeneration.

Abnormal protein accumulation characterizes common age-related neurodegenerative diseases, such as Alzheimer’s disease (AD), Parkinson’s disease (PD), and dementia with Lewy bodies (DLB). In AD, deposition of β-amyloid leads to the formation of neuritic plaques, and accumulation of hyperphosphorylated microtubule-associated protein tau (MAPT) leads to the formation of intraneuronal neurofibrillary tangles (NFTs) that are distributed in the brain in a predictable manner.¹ In DLB and PD, aggregates of α-synuclein form Lewy bodies and accumulate within Lewy neurites, which together are called Lewy-related pathology. In DLB, these aggregates are localized in the brainstem, limbic areas, and neocortex, whereas in PD, they are mostly confined to the brainstem.² Conventionally, the AD-type and Lewy-related pathologies have been thought to be separate neurodegenerative processes. However, these pathologies are commonly found simultaneously in the brains of elderly people, and it has been recognized that the underlying neurodegenerative processes most likely share certain pathogenetic mechanisms.

Genetic variation in α-synuclein (SNCA) has been shown to have a key role in the pathogenesis of PD. Rare SNCA mutations and multiplication have been implicated in familial forms of PD and DLB phenotypes,^3–5 and common genetic variation in the SNCA gene has been reported to be associated with sporadic PD in multiple studies.^6–8 In addition, MAPT H1 haplotype homozygosity has been identified as a risk factor for sporadic PD or PD dementia in several studies,^9,10 and two recent reports suggest that SNCA alleles and MAPT H1/H1 haplotype may have an interactive effect on PD risk.^10,11

Here, we have examined how genetic variation in SNCA contributes to the extent of common age-associated neurodegenerative changes, that is, β-amyloid, NFT (tau), and Lewy-related (α-synuclein) pathologies in a population-based autopsied sample of very elderly subjects. Our results suggest that allelic variation in SNCA modulates the extent of both the NFT (tau) and Lewy-related (α-synuclein) pathologies, and provide further evidence for the interaction of tau with α-synuclein in neurodegeneration.

Subjects and Methods

Subjects

The Vantaa85+ study population includes all persons aged 85 years or older who were living in the city of Vantaa on April 1, 1991. Of the 601 eligible subjects, peripheral blood (and DNA) samples have been obtained from 515 study subjects, and consented autopsy with neuropathological examination was conducted in 272 cases (Table 1). The study was approved by the ethical review committee of the City of Vantaa.

Table 1.

Characteristics of the Whole Study Population and the Neuropathologically Examined Subpopulation

Characteristics	Whole Vantaa 85 + Study Population	Neuropathologically Examined Subpopulation
N	515	272
Women (%)	79.8	83.5
Age in 1991 (yr ± SD)	88.63 ± 2.9	88.92 ± 3.1
Age at death (yr ± SD)	92.38 ± 3.8	92.55 ± 3.7
CERAD score M + F (n)	—	178
CERAD score 0 (n)	—	66
Braak stage IV-VI (n)	—	126
Braak stage 0-II (n)	—	79
Diffuse cortical Lewy-related pathology-positive (n)	—	41
Lewy-related pathology-negative (n)	—	187

No statistically significant differences were found in sex, age in 1991, or age at death between the whole study population and the neuropathologically examined subpopulation. CERAD = Consortium to Establish a Registry for Alzheimer’s Disease; M = moderate neuritic plaques; F = frequent neuritic plaques. SD = standard deviation.

Neuropathological Assessment

Density of neuritic plaques (as an estimate for the β-amyloid pathology) was scored according to the Consortium to Establish a Registry for Alzheimer’s Disease criteria,¹² and the NFT pathology was assessed according to the Braak staging protocol,¹ as described previously.¹³ For the analysis of Lewy bodies and Lewy neurites, a two-step analysis was used. First, samples from the midbrain and hippocampus were stained with hematoxylin and eosin method and with immunohisto-chemical method for α-synuclein (antibody Clone 42; Transduction Laboratories, Lexington, KY). If no evidence of Lewy-related pathology was present in these sections, the subject was categorized in the group of “no Lewy-related pathology.” Second, if an abnormal α-synuclein immunopositive structure or Lewy bodies (hematoxylin and eosin) were found, samples from the temporal, parietal, and frontal neocortex and the cingulate gyrus were stained for α-synuclein. These sections were assessed according to the DLB consortium protocol.¹⁴ Subjects were defined to have extensive Lewy-related pathology if they had diffuse cortical Lewy-related pathology, according to the DLB consortium protocol.¹⁴

Genotyping and Data Analysis

Genotyping of 11 single nucleotide polymorphisms (SNPs) within SNCA, the APOE ε2/ε3/ε4 locus, and an insertion deletion polymorphism in intron 9 of MAPT that discriminates the common H1/H2 haplotype groups was performed as described previously.^11,15,16 Genotyping of the SNCA markers was performed using whole-genome–amplified DNA (Qiagen REPLI-g; Qiagen, Valencia, CA; www.qiagen.com). Six of the SNCA SNPs were selected as tagging SNPs from the LD structure of the SNCA gene, and five were selected from their association with PD. Based on previous evidence,⁷ intron 4 and the 3′ end of the gene were covered most thoroughly. The genotype distributions did not deviate from Hardy–Weinberg equilibrium.

Results

The demographic and neuropathological characteristics of the population-based unselected Vantaa 85+ sample are shown in Table 1. Associations between genetic variation in SNCA and β-amyloid pathology (Consortium to Establish a Registry for Alzheimer’s Disease score), NFT pathology (Braak stage), or Lewy-related pathology were analyzed by comparing genotype distributions in subjects with pronounced pathological changes with the distributions in subjects with mild or no respective pathologies.

Eleven SNCA polymorphisms were genotyped: two SNPs in the intron 2, eight SNPs in intron 4, and one SNP in the 3′ untranslated region. In addition, we attempted to genotype the Rep1 marker in the promoter region, but we were unable to obtain reliable genotype data and this marker was excluded. None of the SNCA markers was associated with the extent of β-amyloid pathology (Consortium to Establish a Registry for Alzheimer’s Disease moderate or frequent versus no neuritic plaques; Table 2). Five of the 11 polymorphisms showed nominally significant associations with the extent of NFT pathology (Braak stages IV-VI vs stages 0-II; see Table 2). The marker rs2572324 showed the strongest association with the extent of NFT pathology (nominal p = 0.0006), and this association remained significant after correction for multiple testing (permuted p = 0.004; Bonferroni corrected p_c33 = 0.021). The marker rs2572324 A/A genotype was overrepresented and the A/G and G/G genotypes underrepresented in subjects with Braak stages IV-VI. The odds ratio for high NFT pathology was 2.81 (95% confidence interval, 1.56–5.07) in subjects with A/A genotype versus subjects with A/G or G/G genotypes. Similar genotypic association was found with diffuse cortical Lewy-related pathology with odds ratio of 2.57 (95% confidence interval, 1.18–5.59) (p = 0.023), but it did not remain statistically significant on permutation (permuted p = 0.127; see Table 2). Genotype data of the significantly associated markers are available in Supplementary Table 1.

Table 2.

Association of α-Synuclein Variants with Different Types of Brain Pathology (Nominal/Permuted^a p Values Are Shown)

SNP ID	α-Amyloidb	NFTc	Lewy-Relatedd
rs2583985	NS	0.017/0.069	NS
rs2583978	NS	0.047/0.260	NS
rs2737020	NS	NS	NS
rs1812923	NS	NS	NS
rs3775439	NS	NS	NS
rs356186	NS	0.098	NS
rs356164	NS	NS	NS
rs356198	NS	NS	NS
rs2572324	NS	0.0006e/0.004	0.023/0.127
rs356168	NS	0.045/0.194	NS
rs356165	NS	0.009/0.062	NS

Number of tests was 11 × 3 = 33 (11 markers, 3 pathological variables). χ² or Fisher’s exact tests or χ² or exact tests for linear trend were used in the statistical analysis by SPSS v12.0 (SPSS, Chicago, IL).

^aPermuted p values were determined by 10,000 permutations by Haploview (http://www.broad.mit.edu/mpg/haploview/index.php).

^bSubjects with high amyloid load (Consortium to Establish a Registry for Alzheimer’s Disease [CERAD] moderate or frequent; n = 178) were compared with subjects with minimal amyloid load (CERAD no neuritic plaques; n = 66).

^cSubjects were categorized in two groups: the low-stage group (stages 0-II; n = 79) and the high-stage group (stages IV-VI; n = 126). p values < 0.05 are in boldface.

^dSubjects with diffuse cortical Lewy-related pathology (diffuse dementia with Lewy bodies; n = 41) were compared with subjects with no Lewy-related pathology in the brainstem (CERAD score of 0; n = 187).

^eAfter Bonferroni correction, p_c33 = 0.021. NFT = neurofibrillary tangles; NS = not significant (p > 0.1).

To test for possible gene interactions, we analyzed the SNCA marker rs2572324 (A/A genotype) in combination with the APOE ε4, a known risk factor for β-amyloid deposition and NFT pathology.^15,17 When assessing the extent of β-amyloid pathology, a prominent effect of the APOE ε4 was observed but no effect of the SNCA alleles (Table 3), whereas both independent and joint effects of SNCA and APOE ε4 were found on the NFT pathology (see Table 3).

Table 3.

Combined Analysis of SNCA and APOE ε4 on β-Amyloid (Consortium to Establish a Registry for Alzheimer’s Disease) and Neurofibrillary (Braak Stage) Pathologies

				Braak Stage
SNCA A/A a	APOE ε4	CERAD M+F	CERAD 0	IV-VI	0-II	OR (95% CI)
+	+	44	3			11.8 (3.36–41.46)
−	+	33	1			26.6 (3.45–204.60)
+	−	54	28			1.55 (0.81–2.96)
−	−	41	33			Reference
+	+			35	3	23.3 (6.42–84.8)
−	+			26	6	8.67 (3.09–24.3)
+	−			42	25	3.36 (1.63–6.91)
−	−			21	42	Reference

^ars2572324. OR = odds ratio; CI = confidence interval; CERAD = Consortium to Establish a Registry for Alzheimer’s Disease; M = moderate neuritic plaques; F = frequent neuritic plaques.

Discussion

The most striking novel finding in this population-based study was the association between common variation in SNCA and the extent of the Alzheimer-type NFT pathology, suggesting that SNCA variation may have an effect on the intraneuronal aggregation of hyperphosphorylated tau. Thus, our results add a novel population-based perspective to the interplay of tau and α-synuclein in neurodegeneration. Previously, tau and α-synuclein have been shown to interact at the protein level, and they can seed each other’s aggregation at low protein concentrations.¹⁸ Two recent genetic studies have suggested that SNCA alleles and MAPT haplotype H1 homozygosity may have an interactive effect on PD risk.^10,11 Furthermore, there is increasing clinical and pathological evidence suggesting that synucleinopathies and tauopathies are linked.¹⁹

The marker rs2572324 showed the strongest association with the extent of NFT pathology. We analyzed the genotype frequencies of this marker in a sample of 136 Finnish middle-aged control subjects²⁰ (Eerola and colleagues, unpublished data). The genotype frequencies in subjects with high Braak stage (IV–VI) were similar to those in the middle-aged control subjects (p = 0.31), whereas the genotype frequencies in subjects with low Braak stage (0–II) markedly differed from the Finnish middle-aged control subjects (p = 0.005) (data not shown). This suggests that the association found in this study is largely due to a protective genetic factor (rs2572324 allele G) enriched in the subjects with low Braak stage at the age of 85 years or older.

Interestingly, the same SNCA variant also showed a nominally significant association with diffuse cortical Lewy-related pathology, although this association was lost on permutation. This may be because of the lack of statistical power because the number of cases with diffuse cortical Lewy-related pathology was lower (n = 41 vs 187 control subjects) than those with Braak stage IV-VI NFT pathology (n = 126 vs 79 control subjects). Previously, familial SNCA mutations and multiplication have been shown to underlie DLB with extensive cortical Lewy-related pathology.^4,5 Our results provide preliminary evidence that genetic variation of SNCA may also modulate nonfamilial cortical Lewy-related pathology.

Our results show the first genetic factor that dissects the two principal pathological features of AD, NFT and β-amyloid, because SNCA variation associated only with the NFT pathology but not with the β-amyloid pathology. We have previously found in the Vantaa 85+ series that APOE ε4 is associated with both β-amyloid and NFT pathologies.^15,17 The two-locus analysis suggested that the ε4 allele and the SNCA A/A genotype have both independent and joint effects on the formation of NFT pathology. An interactive effect of SNCA and MAPT H1 haplotype homozygosity has been suggested in PD.^10,11 In our material, only 11% of the subjects were other than MAPT H1 homozygous. Because of the small number of non-H1/H1 subjects, this interaction could not be adequately addressed here (data shown in Supplementary Table 2).

Genetic association studies typically compare genotype distributions in clinically defined cases and control subjects. These studies are often prone to selection bias and, particularly among the elderly subjects, diagnostic ambiguity. Here the genotype distributions were compared between subjects with discordant neuropathologically defined phenotypes drawn from the same population-based Vantaa 85+ sample. Even though the number of subjects in our study is relatively low compared with many studies composed of clinically defined patient and control groups, our sample is one of the largest among population-based studies in which neuropathological data are available.²¹ Although population-based studies have been until recently underemphasized, it is being recognized that they may provide unique insights in neurodegenerative disorders by giving unbiased information of the occurrence of diseases and pathology, as well as genetics, at the population level.²¹ Our data provide evidence that a functional interaction exists between SNCA gene variation and NFT formation in the population at large. It will be important to verify these findings in other elderly neuropathologically examined materials and further explore functional effects linked with rs2572324.