Introduction

Parkinson’s disease (PD) is the most common neurodegenerative movement disorder, and the second most common neurodegenerative disease after Alzheimer’s disease (AD), with a prevalence of more than 1% after the age of 65 years [1]. The incidence of PD increases with age, and the number of patients is expected to double by the year 2030, due to aging of the population, improved diagnosis and prolonged survival, particularly in the developing countries [2].

PD is clinically defined by adult-onset, progressive parkinsonism (the combination of akinesia, resting tremor, and muscular rigidity), which displays a beneficial response to dopamine-replacement therapy [3]. In most patients, this clinical syndrome correlates with neuronal loss and gliosis in the substantia nigra pars compacta and other brain areas, and with formation of cytoplasmic inclusions called Lewy bodies (LB) and Lewy neurites in the surviving neurons.

The molecular mechanisms of PD remain mostly unknown. Several lines of evidence, including biochemical analysis, genomic and proteomic profiling of brain tissue, cell and animal models, implicated mitochondrial defects, oxidative stress, protein misfolding, proteasomal and lysosomal abnormalities in the pathogenesis [4–12]. However, there are many reciprocal interactions between these pathways, making it difficult to disentangle the primary and the secondary events. Moreover, the determinants of the preferential vulnerability of the dopaminergic system observed in PD remain unknown.

In most patients PD appears as a sporadic disorder. In 10–15% of cases the disease runs in families, but a Mendelian inheritance is rarely evident from the pedigree analysis. Yet, the ongoing identification of primary genetic defects in patients with inherited forms of PD is rapidly expanding the possible approaches to unravel the disease pathogenesis [13].

Five genes are today considered as definitely implicated in the etiology of PD. Mutations in the α-synuclein [14, 15] and leucine-rich repeat kinase 2 (LRRK2) [16, 17] gene cause autosomal dominant forms whereas mutations in the parkin [18], DJ-1 [19] and PINK1 (20) gene cause autosomal recessive forms of PD. LBs are found in the brain of patients with α-synuclein mutations and in most of the cases with LRRK2 mutations. On the other hand, LBs are not present in most of the patients with parkin mutations, while their occurrence in cases with DJ-1 or PINK1 mutations remains unknown.

The discovery that duplication and triplication of the whole α-synuclein gene is also a cause of autosomal dominant PD and of the related condition, dementia with LBs, links directly the over-expression of wild-type α-synuclein protein to the disease pathogenesis [15, 21, 22]. Moreover, common allelic variation in the α-synuclein gene might increase the risk for the sporadic form of PD [23]. A central role of α-synuclein in the pathogenesis of PD is further supported by the fact that wild type α-synuclein protein is the major component of the LBs and of other neuronal and glial inclusions found in PD, dementia with LBs, and multiple system atrophy, now collectively termed “synucleinopathies” [24, 25].

LRRK2 mutations as a cause of PD

A genome-wide search for linkage in a large Japanese pedigree with autosomal dominant, late-onset parkinsonism yielded the identification of a novel locus (PARK8) to the peri-centromeric region of chromosome 12 [26]. Interestingly, autopsy study of four affected members of this family revealed no LBs in the brain, a finding considered incompatible with a formal pathological diagnosis of PD. However, linkage to the same chromosomal region was later confirmed in two large families of European ancestry, segregating parkinsonism associated with different brain pathologies with or without LBs in different patients. This suggested PARK8 to be an important locus with a pleomorphic pathology [27].

Using positional cloning strategies, in the year 2004, LRRK2 was identified as the gene defective at the PARK8 locus [16, 17]. Soon thereafter, different groups identified a single LRRK2 mutation (c.G6055A) leading to a Gly2019Ser substitution in the encoded protein, which was present in familial and sporadic PD with unprecedented high frequency. A different mutation affecting the subsequent amino acid, Ile2020Thr, was detected as the cause of disease in the original Japanese PARK8 family [28]. The following two years have seen an explosion of research into the LRRK2 gene in PD and related disorders. Due to the large size of its open reading frame (more than 7.5 kb across 51 exons), a comprehensive screening of the entire LRRK2 coding region has been rarely performed so far, while in most studies large series of patients were screened only for one or few known mutations.

The screening of the complete coding region of LRRK2 revealed mutations in ∼10% of the PD cases with a family history compatible with autosomal dominant inheritance [29–32]; this figure by itself nominates LRRK2 as the most common known genetic cause of the disease. However, a note of caution is due here. For several mutations, co-segregation studies have been very limited or lacking; large series of ethnically matched controls have not always been tested; and assays are not available yet to study the effects of a given mutation on the function of the LRRK2 protein. Due to all these reasons, the pathogenic nature remains unclear for several of the LRRK2 mutations detected in PD cases, and at least some of these (for example: Arg1514Gln) might represent rare, benign variants [33, 34]. The uncertainties regarding the pathogenic nature of the mutations found in the LRRK2 (as well as in any other genes) are an important issue, which complicates the transfer of the results of the genetic screening into the clinics for diagnostic and genetic counseling purposes. To date, only five LRRK2 mutations (Arg1441Cys, Arg1441Gly, Tyr1699Cys, Gly2019Ser, and Ile2020Thr) can be considered as definitely disease causing, on the basis of a clear co-segregation with PD in large pedigrees and absence in large series of controls (Fig. 1).

Fig. 1
figure 1

Schematic representation of the human LRRK2 protein. The known functional domains are shown, the catalytic domains in black and the protein-protein interaction domains in grey. Five definitely PD-causing mutations and one PD-associated polymorphism are also displayed

Full size image

Another very important aspect is the wide pathological spectrum associated with LRRK2 mutations (∼30 cases came to autopsy so far) [35]. Dopaminergic neuronal cell loss and gliosis in the substantia nigra are the common features in the patients carrying LRRK2 mutations [17]. In addition, classical LBs were found in the majority of cases, but in few, there was absence of inclusions, or only tau-positive or ubiquitin-positive inclusions were detected [31, 36–42]. These observations are based on a limited number of brains, and further studies are warranted. However, the pathologic pleomorphism seems a common theme for the different LRRK2 mutations, at least for Arg1441Cys, Tyr1699Cys, and Gly2019Ser, suggesting that our pathological definition of PD and related diseases has to be revised.

The Gly2019Ser story

Gly2019Ser is particularly important among the PD-causing mutations in LRRK2. This mutation was identified by several groups as a common cause of the disease, being detected initially in ∼5–6% of large cohorts of familial PD in Europe and US, and in ∼1–2% of sporadic PD from UK [43–47]. It is now clear that the frequency of Gly2019Ser in PD varies greatly across populations [48]. The results of the different studies are not easily comparable because of the different sample size, different methods for patient ascertainment, different definitions of “familial” versus “sporadic” disease, and different genotyping techniques, and much more work remains therefore ahead. The Gly2019Ser mutation has not been identified in three large series of Chinese patients [49–51], though it was rarely found in Indian [52] and Japanese patients [53, 54]. Studies in large referral series from the US population estimated a mutation frequency of up to ∼3% in familial and ∼0.7% in sporadic cases, respectively [55, 56]. This mutation seems present at lower frequency in patients from Northern Europe [57–61], than in those from Southern Europe such as Italy (∼5% of familial and ∼1% of sporadic cases) [62, 63], Spain and Portugal (up to ∼6–18% of familial and ∼3–6% of sporadic cases) [64–67](Ferreira et al. unpublished data). However, an extremely high prevalence is found among Arab patients from North Africa (∼37% of familial and ∼41% of sporadic cases) and among Ashkenazi Jewish patients (∼29% of familial and ∼13% of sporadic cases) [68, 69]. The prevalence of this mutation remains to be investigated in other large populations, such as those from Brazil, and other countries of Latin America.

Gly2019Ser represents clearly the first common pathogenic mutation identified in PD, establishing for the first time the proof-of-principle for a genetic determinant frequently involved in the classical, late-onset, sporadic forms of this disease.

Most of the patients carrying this mutation and living in disparate countries in Europe and America, share a common, very old founder haplotype [46, 62, 70], which likely originated from North Africa or Middle East ∼2,000 years ago or earlier [71]. A second haplotype has been detected in a few patients of European ancestry [71], while a third haplotype was found in Japanese patients [54]. The occurrence of Gly2019Ser in at least three different haplotypes suggests either an extremely old founder, or a mutational hot spot. Another hot spot might be represented by LRRK2 codon 1441, where three different mutations are known to occur (Arg1441Cys, Arg1441Gly, Arg1441His) [17, 29, 32, 72–74].

Mapping and cloning of genes for dominantly inherited diseases often relies on families with an exceptionally high number of affected individuals. This leads to an inherent ascertainment bias, and an overestimation of the mutation penetrance. However, after a causative mutation is identified, more accurate values of penetrance can be estimated in unselected, consecutive series of patients, ideally from population-based studies. This approach might yield considerably lower figures of penetrance. In the case of PARK8, a reduced penetrance of the underlying mutation was already suggested in the initial linkage study [26], and confirmed after the identification of the LRRK2 gene. Recent estimates of the lifetime penetrance of the Gly2019Ser mutation in large, hospital-based but otherwise unselected series of PD patients (US Jewish, US non-Jewish, and Italians) yielded values of ∼24–33% [56, 69, 75]. Yet, the penetrance might be different in other populations and additional studies are therefore warranted before Gly2019Ser testing is used for genetic counseling. Such a low penetrance explains the high Gly2019Ser prevalence among patients with sporadic PD, and its rare occurrence in controls (∼1%), particularly among the populations with the highest mutation frequencies such as Arabs and Ashkenazi Jews [68, 69].

Due to a lower frequency, the penetrance of other LRRK2 mutations is more difficult to estimate accurately, but reduced values are also suggested by analysis of pedigrees segregating the second most recurrent LRRK2 mutation, Arg1441Cys [62, 76].

The clinical phenotype of Gly2019Ser-positive patients appears very similar, or indistinguishable from that of the classical form of PD, but a wide range of onset age is evident [45, 55, 56, 62, 63, 77]. Several patients, mostly from Tunisia and Algeria, were identified who carry the Gly2019Ser mutation in homozygous state [78–80]. This is likely due to the high prevalence of the mutation, and the high frequency of consanguineous marriages in those populations. Homozygous carriers of this mutation seem not to develop PD at an earlier age, nor to have a more severe disease, or a more aggressive course, compared to the heterozygous carriers [80]. However, it is difficult to draw firm conclusions, as the clinical spectrum associated in heterozygous mutation carriers is also very broad. Interestingly, the penetrance might be higher in homozygous carriers [79], arguing for the presence of a mutation dosage effect. The low penetrance and variable phenotypic expressivity of the mutation suggest the existence of further important modifiers, which might include other genetic as well as non-genetic factors. Their identification is an important area of the current research.

Gly2385Arg: a common risk allele for PD in Asia

At the beginning of the year 2006, we found that a different LRRK2 variant, Gly2385Arg, is a polymorphism in the Han Chinese population from Taiwan (frequency of heterozygous carriers ∼5% among controls), and it is significantly more frequent (∼10%) among PD cases [81]. We therefore proposed that Gly2385Arg is a common risk factor for PD in the Han population. Interestingly, this variant was initially detected in a single, small PD family from Taiwan [29], but it has not been observed so far among whites [30–32], and it appears therefore specific for the Asian population. The association between Gly2385Arg and PD has now been confirmed in at least four independent replication studies, involving more than 2000 individuals (three targeting the Chinese, and one the Japanese population) [82–85] (Table 1).

Table 1 Allelic association studies of the Gly2385Arg variant as risk factor for PD
Full size table

Using the observed frequency of the Gly2385Arg genotype among controls and the observed value of odds ratio as estimates of the risk genotype frequency in the general population, and of the relative risk, respectively, one can calculate a population attributable risk of ∼4% for the Gly2385Arg heterozygous genotype in the Han Chinese population [82]. Cross-sectional case control studies are prone to several biases, including survival bias. It is therefore crucial to replicate this finding also in large prospective studies. However, the replication of the association in the same direction-of-effect and with similar effect size (odds ratio ∼2.5 in most studies) in four independent, large samples of different geographic and ethnic origin (Table 1), and the potential functional effects of this missense, non-conservative variant, all strongly support the contention that this represents a real, causal association. Gly2385Arg might be the first identified genetic risk factor for the common PD form in the Asian population, and the most frequent genetic determinant of PD worldwide, also considering the large and expanding size of the Chinese population (projected number of ∼5 millions patients by the year 2030) [2].

As observed for the Gly2019Ser carriers, the clinical spectrum in PD patients who carry the Gly2385Arg variant is very broad and indistinguishable from that of the cases who do not carry it.

The Gly2385Arg variant is located at the surface of the C-terminal WD40 domain of the LRRK2 protein, where it introduces an additional, net positive electric charge. WD40 domains are involved in protein–protein interactions, and they might be important for the formation of complexes between LRRK2 and other proteins, or for the LRRK2 dimerization. It is possible that the Gly2385Arg variant increases the risk of PD by affecting these biochemical properties of the LRRK2 protein.

The LRRK2 protein

LRRK2 mutation causes a disease that most closely resembles the common forms of PD. The LRRK2 protein is likely to be a very important player in the pathogenesis of PD in general, and the pharmacological manipulation of the LRRK2 activity might become a future important therapeutic strategy. It is therefore urgent to unravel the biology of the LRRK2 protein, and how its mutation leads to neurodegeneration, but very little is known about these crucial aspects.

LRRK2 encodes a 2,527 amino acids protein of unknown function, belonging to the “ROCO” group within the Ras/GTPase superfamily [86], and characterized by the presence of several conserved domains: a Roc (Ras in complex proteins) and a COR (C-terminal of Roc) domain, together with a leucine-rich repeat region, a WD40 domain, two ankyrin-like motifs, and a protein kinase catalytic domain (Fig. 1). Review of the ROCO family members reveals involvement in diverse cellular processes (regulation of cell polarity, chemotaxis, cytokinesis, cytoskeletal rearrangements, and programmed cell death), making impossible to predict the function of human LRRK2 on the basis of homology (reviewed in [86, 87]).

Initial studies suggest that the LRRK2 mRNA [88–91] and the LRRK2 protein [39, 88, 92, 93] are broadly expressed throughout the brain, including nigral neurons, and that the LRRK2 protein shows a cytosolic localization, perhaps in association with membranous structures [93]. There is also evidence that the LRRK2 protein regulates the length and branching of neurites and this function might be impaired by PD-causing mutations [94].

LRRK2 immunoreactivity has been reported in some LBs from PD brains [92, 95]. However, most of the currently available LRRK2 antibodies lack optimal sensitivity and specificity, and further investigations are definitely warranted.

One of the two predicted catalytic domains (GTPase and kinase) represents likely the output activity of LRRK2. Small GTPases (Rho, Rac, Cdc42) usually act upstream of protein kinases. By analogy, the GTPase domain might regulate the LRRK2 kinase domain via intramolecular signaling. Whether the LRRK2 kinase activity is required for the phosphorylation of target proteins, or whether it plays an auto-regulatory role, is currently unclear. The PD-causing mutations replace highly conserved residues, but, in addition, the Glycine2019 residue is remarkable because it is conserved in all human kinase domains. These mutations could destabilize the kinase domain, resulting in loss-of-function of the kinase activity, and suggesting haploinsufficiency as disease mechanism. Another possibility is that Gly2019Ser and other mutations enhance the kinase activity. Of note, the three known PD-causing mutations in the kinase domain (Ile2012Thr, Gly2019Ser, and Ile2020Thr) all introduce novel potential auto-phosphorylation sites, and similar mutations in the activation segment of other kinases induce hyper-activity [96]. This mechanism would confer a gain of function to the mutant protein, fitting with the dominant pattern of inheritance seen in families with LRRK2 mutations.

Over-expressing the human wild-type LRRK2 protein in different cell systems is associated with formation of cytoplasmic inclusions [92, 97]. Moreover, LRRK2 shows protein kinase activity in vitro toward generic substrates and is capable of auto-phosphorylation [92, 98, 99]. Importantly, some of the PD-causing mutations (particularly those located in the kinase domain) appear to enhance the LRRK2 kinase activity in vitro, as well as the inclusion formation, and they induce cell toxicity and ultimately, cell death [92, 97–99]. LRRK2 also displays GTP-binding properties in vitro, and GTP binding seems required for the kinase domain of LRRK2 to be in an active state [100–102]. However, LRRK2 seems devoid of intrinsic GTPase activity, suggesting the involvement of other interacting proteins, such as GTPase activating proteins (GAPs), and guanine nucleotide exchange factors (GEFs) [100–102]. Here, the caveat is that all these findings need validation using in vivo models, and after the (currently unknown) physiological interactors and substrates of the LRRK2 protein are identified. A study focusing on the homologue LRRK1 protein came to the opposite conclusion that the LRRK1 kinase activity might be decreased by amino acid substitutions corresponding to the PD-causing mutations in LRRK2 [103]. Much more work remains ahead in order to understand the biology and pathology of this complex, fascinating protein.

Conclusions

The discovery of LRRK2 mutations in PD led to a turning point in the field. For the first time, gene mutations, and particularly the low-penetrance Gly2019Ser mutation prove to be a frequent genetic determinant of familial and sporadic forms of this disease in several populations; the Gly2385Arg polymorphic variant is a common risk factor for PD in Asia. In both cases, the associated clinical phenotype is indistinguishable from the classical, late-onset PD, and brain study reveals a broad pathological spectrum, which includes in most cases the typical LB pathology. Importantly, this frequent low-penetrance mutation and the frequent risk allele provide etiological links between the familial and the sporadic forms of PD.

Elucidating the function of the LRRK2 protein and how LRRK2 dysfunction leads to neurodegeneration might provide crucial insights for understanding the molecular mechanisms of PD and yield novel avenues to the development of a cure. The pharmacological modulation of one or both catalytic LRRK2 activities could become innovative, important therapeutic strategy for all patients with PD and related neurodegenerative diseases.