Background

Alpha-synuclein is an essential component of Lewy bodies, protein inclusion found post-mortem in affected subjects with synucleinopathies like PD[1], dementia with Lewy bodies (DLB)[2], and multiple system atrophy (MSA)[3, 4]. Synphilin-1 was identified as an alpha-synuclein interacting protein in vitro in a yeast two-hybrid screen [5], and antibodies of Synphilin-1 have been found to stain Lewy bodies in brain cells in PD and DLB affected [6, 7]. In a study performed by Wakabayashi and colleagues[6], Synphilin-1 was identified in the core of Lewy bodies with alpha-synuclein situated peripherally, suggesting a possible role of Synphilin-1 protein-protein interaction in Lewy body formation. Synphilin-1 has also been implicated in parkinsonism as a protein ubiquinated by Parkin for which loss of function results in juvenile or early-onset disease [8, 9].

Synphilin-1 has several protein-protein interacting motifs, including Ankyrin repeats, a coiled coil domain and an ATP and GTP binding domain [10] (figure 1). By comparing the human and mouse synphilin-1 cDNA sequence it was confirmed that the regions containing these protein interacting motifs were among the most conserved together with a relatively long internal conserved region [11]. Amino acid substitutions in any of these domains may impair the protein's function [12].

Figure 1
figure 1

Organisation of the synphilin-1 gene with motifs as described by Engelender et al 2000 [10] and O'Farrell et al 2002 [11]. The SNPs genotyped are located on the figure relative to exons and motifs. Ankyrin-like motifs are highlighted in yellow. The coiled-coil motif with ATP and GTP binding domain are marked in green. Motifs involved in interactions with other proteins such as alpha-synuclein [29] and Parkin [9, 30] are not known.

Full size image

The region of chromosome 5 in which the synphilin-1 locus resides has been implicated in PD is several independent genome wide screens [13–15] although a recent combined, pooled analysis was not supportive [16]. In this study we genotyped 11 SNPs in the synphilin-1 gene using rare variants in a direct association approach and a group of more informative heterozygous SNPs to assess by indirect means whether there was any evidence for genetic association with idiopathic PD. We also screened an internal microsatellite marker and five additional microsatellites downstream of the synphilin-1 5p23.1-23.3 locus [13–15].

Methods

PD affected and controls

A total of 305 affected have been clinically examined and followed longitudinally by one neurologist (JOA) at the outpatient clinics of three hospitals in Central Norway. Their diagnosis of PD was based on the presence of two or more of the cardinal symptoms including tremor, rigidity, bradykinesia and postural instability, with a positive response to levodopa, according to Gelb et al. 1999[17]. Five affected subjects did not fulfil the criteria and were omitted from the association study. They were diagnosed as PSP (n:2), MSA (n:1), CBD (n:1) and Pick's disease (n:1). All affected and controls were from Central Norway. The mean age at onset for the PD series was 58.9 ± 11.3 years; the mean age at study was 70.2 ± 10.7 (mean ± SD). Among the PD samples there were 113 females (37.7%) and 187 males (62.3%). Control subjects included 412 healthy individuals with age at examination 62.8 ± 10.6 years; (mean ± SD), 171 females (41.5%) and 241 males (58.5%). Four affected were found to have an additional relative included in the samples. No other family relationships were known within the affected samples. The Ethics Committees (IRB) of Central Norway and Mayo Clinic Jacksonville approved the study and informed consent was obtained from all participants.

Marker selection

The strategy for selection of SNPs was bipartite. First, rare SNPs were selected in a direct approach to identify variants associated with PD. These included known variants of low frequency, of which five out of 6 result in non-synonymous or synonymous amino acid substitutions [10, 12]. The low frequency variants were selected through scrutiny of previously published works by Engelender et al 2000 [10] and Marx et al 2003 [12] that identified some variants of interest to the study of idiopathic PD in Synphilin-1. Second, using an indirect approach for testing association within haplotype blocks, a group of informative SNPs (minor allele frequencies > 0.05) were selected from those available in the National Centre for Biotechnology Information database [18] (figure 1). In addition, a microsatellite, D5S2950, within synphilin-1 intron 6 characterized by Engelender and colleagues [10] was genotyped for association analysis. The selection of additional microsatellites, including D5S2115, D5S816, D5S414, D5S2116 and D5S2010, was based on the heterozygosity and location downstream of the synphilin-1 locus, spanning a region of chromosome 5 previously highlighted in genome-wide linkage analyses [13–15].

Isolation of DNA and genotype analysis

DNA was isolated from 200 μl whole blood using QIAmp DNA Blood Mini Kit from Qiagen. All affected and control subjects were genotyped using SNaPshot® multiplex kit and a subset confirmed by sequencing using Applied Biosystems Bigdye® Terminator V3.1 cycle sequencing kit. DNA amplification was performed on an Applied Biosystems 2700 PCR machine, according to conditions recommended by the Applied Biosystems SNaPshot® multiplex kit. Fluorescent SNPs and microsatellite products were subsequently sized and genotyped on an Applied Biosystems 3100 automated capillary machine. PCR conditions and primer sequences are available on request.

Statistical analysis

Single marker and haplotype block association were estimated using Haploview version 3.2, estimating statistical significance by performing 10 000 permutations [19]. Haplotype blocks are defined according to Gabriel et al [20]. All chromosome marker locations were deduced from Ensembl gene [ID; ENSG00000064692][21]. Hardy-Weinberg equilibrium was estimated using PHARE [22]. Multiple marker association for microsatellites was estimated using CLUMP[23]. Differences in allele frequencies, OR, RR and chi-squared p-values were estimated using the Interactive Statistical Pages [24] by John C. Pezzullo PhD, of Georgetown University, Washington DC. Splice-site predictions of human synphilin-1 exon-intron sequence were performed using BDGP Splice Site Predictor set [25] for analysis of human sequence.

Results

Using a direct approach, genotyping of known variants in the synphilin-1gene confirmed the presence of all the 6 variants in affected and/or control subjects (table 1).

Table 1 Genotype frequencies for single markers used in both direct and indirect association analysis obtained with the Haploview 3.2 software package. P-values and OR are for allelic analysis.
Full size table

The Arg621Cys (R621C)[12] substitution was identified as heterozygous in four affected and 10 control subjects. In addition, one of the two patient diagnosed with progressive supranuclear palsy (PSP) was homozygous for this relatively rare variant. Subsequent screening of an additional patient with PSP from the same geographical region did not reveal the variant. However, all patients with atypical parkinsonism were omitted from subsequent association analyses (table 1). A heterozygous Glu709Gln (E706Q) substitution was detected in one affected patient and four control subjects. A Val44Ala (V44A) non-synonymous substitution [10] in the first base of exon 4 was also detected in only one heterozygous female patient with a diagnosis of 'probable' PD with onset at 52 years. A synonymous substitution, C636>T, Pro212Pro[12] was identified in one heterozygous male patient with a diagnosis of 'atypical' PD, 60 years at onset, and in no control subjects. The base pair change C1134>T, leading to another synonymous substitution Leu378Leu [12] was observed in two patients and one control subject. A low frequency polymorphism g.IVS6-62G>A was found in 8 control subjects and 6 patients of both genders and at equal frequency (table 1).

Highly heterozygous SNPs were used in an indirect association approach and tested for allele and haplotype association with disease. As presented in figure 1, all SNPs are situated in intronic regions in close proximity to exons. Single marker association (table 1) and haplotype analyses (table 2) showed no evidence of association between synphilin-1 and PD. The haplotype block structure observed (figure 2) corresponded with that identified for synphilin-1 in the HapMap project, using HapMap version 3.5.

Table 2 Haplotypes and haplotype frequencies in haploblock 1 as defined by default settings using Haploview version 3.2
Full size table
Figure 2
figure 2

Software Haploview version 3.2 were used to estimate haplotype blocks. Haplotype blocks were defined according to Gabriel et al [20].

Full size image

A downstream external microsatellite marker, D5S2010, showed significant association for PD (p = 0.006, 27 df). The remaining microsatellite markers, D5S2115, D5S816, D5S414 and D5S2116 showed no allelic differences when comparing PD with control subjects or in stratified analyses by diagnostic certainty, age or gender (data not shown). The microsatellite D5S2950 of synphilin-1 showed no overall association with disease (p = 0.48, 11 df) although some marginal allelic association was observed.

Discussion

Despite functional research anticipating a role for synphilin-1 in Lewy body disorders, our results, and those of others performing association studies on family material and sporadic populations, do not provide support for a genetic association between PD and synphilin-1 [26–28]. Our results of rare SNPs were not significant and suggest that these genetic variants are unlikely to contribute to risk of PD although it must be noted that some of the SNP frequencies are very low and can not be conclusively excluded within the limits of the power of these analysis. A heterozygous intron-exon 4 splice acceptor site non-synonymous substitution T131>C (V44A), previously described by Engelender et al 2000[10], was identified in a female patient with a 'probable' diagnosis of PD, with onset at 52 years, but was not found to segregate with PD in her affected brother. The substitution was absent in control subjects. Thus, while the variant may affect mRNA splicing it is unlikely to be a major contributor to disease risk.

Previously, Marx and colleagues [12] identified the C1861>T (R621C) substitution in two German patients with sporadic PD. In our sample from the Norwegian population, this variation was identified in four patients with PD (frequency 0.01) and 10 control subjects (frequency 0.02). Rather surprisingly, one patient with PSP was homozygous for the R621C substitution. However, PSP is a tauopathy in which synphilin-1 is unlikely to play a role, and no other PSP patients were identified with the mutation. Our findings suggest C1861>T (R621C) is only a polymorphism. The case-control distribution of other SNPs was approximately equal between groups. No statistical difference was found for single marker or haplotype block associations, adjusting for age and gender. We examined five external microsatellite markers, including D5S816 marker which has previously been implicated in PD through a genome wide analysis of affected sib-pairs [14]. In our population there were no significant differences between affected, controls and diagnosis for any of the external markers except D5S2010 (p = 0.006) for which neighbouring genetic variability may warrant further analysis. Interesting and surprisingly, the D5S2010 marker is more or less directly linked to CENTD3 an effector of PI3K the gate controller of autophagy, cell survival and cell death that complexes with SH3KBP1 on Xp22.1-p21.3.

Conclusion

We found no evidence for association between genetic variability in synphilin-1 and PD. There were no statistical differences in allele or genotype frequencies between PD or control subjects, for common or rare variants within the synphlin-1 gene. Although synphilin-1 appears to be present within Lewy bodies, may interact with alpha-synuclein, and may be a substrate of Parkin, any genetic evidence for its involvement in disease, either through sequencing and the identification of rare coding substitutions, association or linkage, is undetected in this Norwegian population study and remains to be confirmed.