Introduction

Parkinson's disease (PD) is the second most common progressive neurodegenerative brain disorder affecting approximately 2% of the population above the age of 65 years. The pathogenesis of PD is not yet completely understood, however, both genetic and environmental factors contribute to the disease phenotype. To date, six causal genes have been identified1, 2, 3, 4, 5, 6 and more are to be identified in several linked loci. Mutations in the α-synuclein (SNCA), parkin (PARK2), ubiquitin hydrolase L1 (UCH-L1), DJ-1, PTEN-induced kinase 1 (PINK1) and leucine-rich repeat kinase 2 (LRRK2) genes have been implicated in hereditary PD (for recent review see Farrer7). Mutations in LRRK2 were most recently identified and are responsible for PD in at least 5% of familial patients.8, 9 Intriguingly, LRRK2 mutations have also been reported in 1–2% of sporadic PD patients presenting with a clinical phenotype largely indistinguishable from common PD.10

The 51 exons of LRRK2, spanning 144 kb in the chromosome 12q locus, encode an unusually large protein belonging to the ROCO family,11 containing both an Roc-GTPase and a COR domain. In addition, a protein kinase as well as multiple protein interaction domains have been predicted in the C-terminal half of LRRK2.11 Most of these catalytic and protein–protein interaction domains are affected by PD-associated mutations.12 The wide mutational spectrum along LRRK2 is consistent with a gain-of-function disease mechanism, which is supported by three recent functional LRRK2 studies, showing increased levels of kinase activity in vitro for LRRK2 mutations within the kinase domain.13, 14, 15 Further functional analysis of the Roc domain showed that distinct mutations in this domain increase GTP binding.15 Since binding of GTP is essential for kinase activity,16 these data also support a (indirect) gain-of-function mechanism for mutations in the Roc domain.

Thus far, 22 putative pathogenic PD-associated LRRK2 mutations have been identified in genetic studies.12 The most prevalent LRRK2 mutation, p.G2019S, has been observed in more than 250 patients worldwide. In Whites, this mutation explains 0.5–2% of sporadic and 5% of familial PD patients. More recent studies reported that p.G2019S accounted for up to 30% of familial and sporadic PD in Ashkenazi Jews and North African Arabs. Haplotype analyses showed a common Middle Eastern founder haplotype in most p.G2019S carriers, however, at least three European American families carried a distinct disease haplotype suggesting two separate founding events for the p.G2019S mutation in these populations.17, 18, 19, 20 The p.R1441C mutation is the second most frequent LRRK2 mutation and was first identified in a large autosomal dominant kindred from Western Nebraska, family D, presenting with late-onset, levodopa-responsive parkinsonism.6, 21 Since then, p.R1441C mutations were reported in five additional US, three Italian and several single PD families from different nationalities.6, 22, 23, 24, 25, 26, 27, 28, 29 Founder effects have also been reported for the LRRK2 p.R1441 ‘hotspot’ codon. A common disease haplotype containing the p.R1441C mutation was observed in three Italian families,22 while transmission of the p.R1441G mutation from a common ancestor was identified in two neighboring autonomous communities in Northern Spain.30, 31 Although the pathogenicity of many of the reported missense mutations still needs to be proven, it was demonstrated that LRRK2 mutations can also act as potential risk factors. For example, the p.G2385R polymorphism increased the risk for PD in Chinese and Taiwanese populations.32, 33

Materials and methods

Subjects

We studied a large group of 304 Belgian PD patients (Table 1), mainly derived from a retrospective epidemiological study designed to evaluate environmental risk factors in PD (N=181).34 This PD group was supplemented with patients from a prospective Belgian study of neurodegenerative diseases (N=123).35 Patients included in these studies were all residents of Flanders, the Flemish speaking region of Belgium. Overall, the mean onset age in the Belgian PD group was 59.0±11.5 years (range 28–87) and mean age at examination 67.2±10.9 years (range 30–90) with 56.3% male patients. In 18.1% of PD patients, a positive family history of PD with a first-degree relative was reported. The control group consisted of 278 Belgian individuals with a mean age at inclusion of 60.4±12.0 (range 28–89), and included 127 spouses recruited within the epidemiological study34 and 151 community individuals with no clinical evidence or history of PD or another movement disorder.

Table 1 General characteristics of the Belgian PD study populations
Full size table

The local medical ethical committees of the Middelheim General Hospital Antwerp, the University Hospital Antwerp and the University of Antwerp approved the Belgian PD studies and all participating individuals gave written informed consent. Patients were diagnosed using strict diagnostic criteria for PD, requiring three out of four features of bradykinesia, rigidity, tremor and asymmetrical onset with a positive response to levodopa.

Gene sequencing

The coding sequence and exon–intron boundaries of LRRK2 exons 29–31 and 38–44 encoding the Roc and kinase domains, and exons 5 and 34 harboring informative single-nucleotide polymorphisms (SNPs) implemented in the haplotype analysis, were amplified by standard PCR reactions. Primer pairs were designed using Primer3.36 Amplification products were purified and sequenced using the BigDye Terminator Cycle Sequencing v3.1, and analyzed on an ABI3730 DNA analyzer. Novel variants identified in patients were confirmed and tested in a panel of 278 Belgian control individuals by pyrosequencing on a PSQTM pyrosequencer, genescan analysis on an automated ABI3730 DNA analyzer or by SEQUENOM MassARRAY(R) technology (Sequenom Inc.) (primer sequences available upon request). In silico analysis of evolutionary amino-acid conservation between different types of organisms was performed using the UCSC Genome Browser. The effect of mutations at the protein level was predicted using the SIFT program.37

Haplotype analysis

We genotyped 28 LRRK2 intragenic and flanking markers, both microsatellite markers and SNPs that spanned a region of 16 Mb at chromosome 12q in all six Belgian R1441C carriers and in 20 of their relatives for haplotype phase determination. LRRK2 markers were also analyzed in 178 Belgian control individuals to estimate the population frequency of the shared haplotype. Microsatellite markers17 were amplified and fluorescently labeled during PCR amplification. Amplicons were separated on an automated ABI3730 DNA analyzer. Alleles were scored using the ‘Genotyper’ and ‘Pattern automatic allele caller’ programs developed in house. Genotypes of exonic and intronic LRRK2 SNPs were extracted from the sequence trace files.

Results

Mutations in the Roc and kinase domains

Direct sequencing of PCR amplicons of the 10 LRRK2 exons coding for the Roc and kinase domains identified 10 heterozygous LRRK2 mutation carriers in the Belgian PD patients (10/304=3.29%) (Table 1). Six patients carried the same p.R1441C mutation, while each of the other four patients carried a distinct novel missense mutation (Table 2, Figure 1a). We also observed five known SNPs at the same frequencies as previously reported in other European study populations25 (Supplementary Table 1).

Table 2 Missense mutations identified in the Belgian PD population
Full size table
Figure 1
figure 1

(a) Reported and novel coding mutations in the LRRK2 Roc and kinase domains. Symbols: red stars, putative pathogenic missense mutations detected in the Flanders-Belgian PD population; green stars, novel rare variants, putatively not pathogenic; yellow stars, known polymorphisms; cDNA numbering according to AY792511.1 starting at nt 1. Domain boundaries as depicted in Bosgraaf and Van Haastert.11 Protein numbering according to AAV63975.1 starting at nt 1. (b) Data on protein conservation of all coding mutations in the LRRK2 Roc and kinase domains.

Full size image

p.R1441C founder mutation

p.R1441C (c.4321C>T in exon 31) was observed in six independently ascertained PD patients (6/304=1.97%; Table 2). Five of the p.R1441C carriers presented with a classical PD phenotype with tremor as the predominant initial symptom, while the sixth patient suffered primarily from rigidity (Table 3). The onset age varied largely among carriers from 39 to 73 years with a mean onset of 56.5 years. A positive family history of PD was reported for all p.R1441C carriers in one or more first-degree (DR75, DR133, DR134, DR135 and DR237) or second-degree (DR99) relatives (Figure 2).

Table 3 Clinical characteristics of the probands
Full size table
Figure 2
figure 2

Pedigrees of probands carrying p.R1441C. Symbols: ▪=PD; =dementia; =unclear; *=blood drawn; all at-risk individuals are collected in a diamond symbol (carrying wild-type or mutant alleles) at the right of each generation regardless of their age.

Full size image

On the basis of the segregation data of 28 STR and SNP markers in families DR75, DR99 and DR135, we deduced a disease haplotype that was shared among all three families (Table 4). Further, examination of the individual genotype data from the other three probands DR133.1, DR134.1 and DR237.1 indicated that all three shared alleles for several markers from within the disease haplotype. Together, the allele and haplotype data identified a shared segment of 438 kb between centromeric D12S2194 and telomeric D12S1301 (Table 4). The shared haplotype was absent in 178 Belgian control individuals.

Table 4 Allele and haplotype sharing analyses
Full size table

Novel missense mutations

The four novel missense mutations were identified in exons 29, 31 and 44 (Table 2, Figure 1a) and were absent from 278 Belgian control individuals (MAF <0.2%). In silico conservation analyses demonstrated that the mutations in exons 29 and 31 affected evolutionarily conserved amino-acid residues (Figure 1b). For p.Y2189C in exon 44, the in silico conservation analysis showed that the Y residue itself was not conserved through evolution, though the aromatic nature of the respective amino-acid residues was (UCSC Genome Browser Human March 2006 Assembly) (Figure 1b).

Genealogy showed a positive family history of neurodegenerative disease for the respective probands except for DR136.1 (p.K1468E) (Table 3, Figure 3b). A positive family history of PD was reported for the patients carrying p.R1483Q and p.Y2189C, though we have no access to DNA of additional family members of DR241.1 (p.Y2189C; Figure 3d). Two first-degree relatives of DR103.1 (p.R1483Q) were also diagnosed with PD (Figure 3c; I.2 and II.1). The mutation was not detected in an affected sib (II.1) of the proband (II.2), while an older healthy brother (II.3) carried p.R1483Q. The mother (Figure 3a; I.4) of proband DR108.1 (II.1) was not diagnosed with PD though presented with a mild tremor in her late 70s. She turned out to be a homozygous carrier of p.R1325Q. Loss of heterozygosity due to a large genomic deletion of the LRRK2 locus was ruled out by genotyping 18 STR markers. An older unaffected sibling of the proband also carried the mutation but in the heterozygous state.

Figure 3
figure 3

Pedigrees of probands carrying novel missense mutations. (a) p.R1325Q; (b) p.K1468E; (c) p.R1483Q; (d) p.Y2189C. Symbols: ▪=PD; =dementia; =recent tremor; *=blood drawn. All at-risk individuals are collected in a diamond symbol (carrying wild-type or mutant alleles) at the right of each generation regardless of their age.

Full size image

Discussion

LRRK2 is the most frequent mutated PD gene with a prevalence of 5% in familial and 2% in sporadic patients in European populations8, 9, 10, 17 and the majority of the reported pathogenic mutations are located in two major functional LRRK2 domains that is the Roc and kinase domains. In this study, we examined the prevalence of mutations in these domains in a large PD sample of 304 unrelated patients. Although the patient group consisted mainly of sporadic patients (82%), we identified five distinct mutations in 10 patients suggesting an overall LRRK2 frequency of 3.29%. The majority of these mutations were observed in patients with a positive family history of PD (8/56=14.28%). This frequency is higher than what has been reported in other studies,8, 9 however, we should point out that six of eight familial patients carried p.R1441C, indicating that this mutation was the major genetic cause of PD in Belgian patients. All patients were recruited in Flanders, the Flemish speaking region of Belgium. In previous genetic studies of other neurological diseases, we showed that several pathogenic mutations that were identified in Flemish patients had a strong founder effect most likely due to linguistic and geographic isolation of this region.38, 39, 40, 41 Also, in this study, we were able to show that the Flemish-Belgian p.R1441C was present on one single disease haplotype spanning a minimal region of 438 kb, and absent in 178 control individuals. These data strongly support the notion that p.R1441C was transmitted from a single common founder. Comparison of the p.R1441C disease haplotype with that in Italian PD patients22 revealed that in these two European populations, p.R1441C resulted from two independent mutation events. The latter again supports the observation that the p.R1441 codon represents a mutation hotspot since three different missense mutations at this codon have already been reported (p.R1441C, p.R1441G, p.R1441H).6, 26, 30, 31

Interestingly, five out of six R1441C carriers presented with tremor as the dominant symptom, a feature that generally occurs in only one out of three PD patients. Also, most previously reported p.R1441C carriers presented with tremor as the initial symptom,6, 24, 25, 27, 28, 41 though detailed clinical data have not been reported in all studies.22, 23, 26 An extensive clinical comparison of as many carriers as possible will be obligatory to determine the relationship between p.R1441C and this specific phenotype. Previous segregation analyses in families carrying the most prevalent pathogenic LRRK2 mutations, p.G2019S and p.R1441C, highlighted a considerable variation in age at onset of PD within6, 17, 42 and between families17, 43 due to a highly variable disease penetrance. The onset age between the six probands carrying p.R1441C ranged from 39 to 74 years. Within family DR75, the onset age ranged from 39 to 72 years. Moreover, we also observed an older sibling carrying the mutation but not showing any symptoms at the age of 63 years, 24 years later than the onset age of the proband, supporting the idea of reduced disease penetrance for p.R1441C. In contrast, another older sibling of the proband showed disease symptoms, but was not a carrier of p.R1441C. Non-carrier patients have also been reported in other LRRK2 mutation families including p.R1441C (Zimprich et al6; Di Fonzo et al, 200622) and p.G2019S (Hernandez et al43; Khan et al44; Nichols et al9) families. These patients likely represent phenocopies that are not uncommon for frequent diseases as PD.

In addition, we identified four novel missense mutations in the PD group (4/304=1.32%), with each mutation present in a single unrelated patient. Although these rare mutations were not present in 278 Belgian healthy control individuals their pathogenic nature remains obscure. For the three novel mutations identified in the Roc domain, our data do not support a role in PD pathogenesis. For p.K1468E observed in patient DR136.1, no segregation data could be obtained due to lack of other patients or relatives in this small family. In family DR103, p.R1483Q did not segregate with disease since it was absent in another first-degree patient. In family DR108, the patient was diagnosed with early onset PD at age 33 years and was heterozygous for p.R1325Q. His mother, however, had only a mild tremor at age 79 years. She unexpectedly carried two mutant alleles though on a different haplotype. She refused further medical examination preventing a more accurate clinical diagnosis of the parkinsonian nature of her rest tremor. We cannot exclude that the lack of cosegregation in these families was due to reduced age penetrance of PD. However, in silico analyses were supporting a role as benign polymorphism rather than causal mutation for all three novel variants in the Roc domain. The occurrence of benign coding polymorphisms has been reported earlier for LRRK2 (overview in Mata et al25), and supports the idea that some missense mutations do not drastically affect protein function. Indeed, SIFT analyses demonstrated that the three new LRRK2 variants in the Roc domain, p.K1468E, p.R1483Q and p.R1325Q, are most likely tolerable as is true for another previously reported coding polymorphism in the Roc domain, p. R1398H (rs7133914).

In the exons encoding the kinase domain, we identified only one novel missense mutation (p.Y2189C) located 46 amino acids downstream of the kinase domain. Despite the fact that p.Y2189C was allocated a dbSNP ID (rs35658131) based on a screening in carcinoma DNAs, there were no reports on its presence in healthy control individuals. Moreover, p.Y2189C was not found in 556 Belgian control chromosomes and was never reported in relation to neurodegenerative disorders before. In silico analyses implied that this position may be important for the structure of the protein. Indeed, SIFT analyses provided evidence that p.Y2189C is deleterious for LRRK2 function. Together, these data suggest a potential role for p.Y2189C in the etiology of PD. Further functional analyses will be necessary to pinpoint its exact role in the disease mechanism.

More than 20 LRRK2 mutations have been reported to date,12 but pathogenicity of only six has been confirmed through segregation analysis (p.I1122V, p.R1441C, p.R1441G, p.Y1699C, p.G2019S and p.I2020T). This together with our findings indicates that it will be extremely important to obtain functional data establishing the true pathogenic nature of these mutations. This will also contribute to a more accurate estimate of the frequency of pathogenic mutations and benign polymorphisms in LRRK2.

In conclusion, we identified five distinct missense mutations in a large PD group, four of which were novel (p.R1325Q, p.K1468E, p.R1483Q and p.Y2189C). We demonstrated that p.R1441C is a frequent cause of PD in Flanders-Belgium due to a founder effect. Of interest is that we did not observe p.G2019S known as the most common mutation in LRRK2 in PD in most countries. In our opinion, screening of unrelated PD patients for LRRK2 mutations is necessary to increase our understanding of their role in the disease mechanism. Also, a straightforward functional assay will be needed to unambiguously determine the pathogenic character of any novel mutation, particularly when conclusive genetic data are lacking.