LRRK2 Mutations and Risk Variants in Japanese Patients with Parkinson’s Disease

SUBJECTS AND METHODS

RESULTS

Mutational analysis of the LRRK2 coding region and intron/exon boundaries in Tier 1-patients revealed 67 variants divided as follows: 60 single nucleotide substitutions, six deletion/insertion polymorphisms, and one short tandem repeat. Fourteen of these variants were novel in that they were not contained in dbSNP. Six of the 15 exonic single nucleotide substitutions found were nonsynonymous, but all have been reported in multiple population controls and are thus likely to be SNPs rather than pathogenic missense mutations.

Sequence analysis of exons 31 and 41 in Tier 2-patients revealed two novel nonsynonymous variants (L1446P and V1450I) in close proximity to codon 1441 which is known to harbor three distinct pathogenic mutations (Table 2). The amino acid residues at positions 1446 and 1450 are highly conserved among vertebrate species and in Drosophila melanogaster, indicating that these two variants might alter protein function. However, both variants were also found in multiple individuals in Control Groups 1 and 2, suggesting that neither are pathogenic mutations of high penetrance. A much larger sample size would be required to adequately assess whether either of these low frequency (≤2%) variants alter risk for PD.

Of the five pathogenic or likely pathogenic mutations previously reported in Asian populations, only one (G2019S) was detected among patients with PD in this study (Table 2). G2019S was found in two patients with sporadic PD; the clinical characteristics of these individuals have been reported elsewhere.⁶ Of the two recurrent mutations previously proposed as potentially pathogenic among Asians, one (R1067Q) was found in our PD sample (Table 2). The two patients heterozygous for R1067Q were both female, had no family history of PD or atypical features, and reported ages at onset (AAO) of 46 and 59 years, respectively. No family members of either individual were available to assess segregation of the variant with disease. R1067Q was also detected in a single male Japanese-American control who was followed biannually in the Kame Project from 1992–2000. At his last assessment at 83 years of age, this individual displayed no signs or symptoms of parkinsonism and reported no family history of PD.

Of the two reported susceptibility variants, only G2385R was observed among patients with PD in this study (Table 2). For G2385R there were no significant deviations from HWE in the PD group (Tier1 + Tier 2; two-tailed P = 1.00), Control Group 1 (P = 0.70), or Control Group 2 (P = 0.58). In the full dataset, we observed a significantly greater frequency of the 2385R allele in patients with PD compared to control subjects under a dominant model (Table 3; G/R + R/R versus G/G, OR, 1.96; 95% CI, 1.42–2.70; P = 4.1 × 10⁻⁵). The association remained highly significant after adjustment for age and sex (OR, 1.83; 95% CI, 1.31–2.54; P = 3.3 × 10⁻⁴). The frequency of 2385R was higher in Control Group 1 (8.4%) than in Group 2 (5.6%), but the difference did not reach significance (χ² = 3.4; P = 0.06).

A meta-analysis of eight previous PD case-control studies on G2385R in Asian populations, together with data from the present study, is presented in Figure 1. There was no significant evidence of heterogeneity between studies (χ² for Q statistic = 11.42; degrees of freedom = 8; P = 0.18). This was further reflected by a relatively low I² value of 30% (95% CI, 0–68%). The summary OR for all nine studies was similar assuming either fixed (OR, 2.55; 95% CI, 2.10–3.10) or random (OR, 2.48; 95% CI, 1.90–3.24) effects. A funnel plot of these studies is presented in Figure 2. The points are essentially evenly divided on either side of the line denoting the pooled effect and all fall within the pseudo 95% confidence limits arguing against substantial publication bias.

All 2385R carriers included in the haplotype analysis were found to share at least one allele at six of the seven markers genotyped (Table 4). We then used PHASE to reconstruct haplotypes for these six markers. All carriers shared a six-locus haplotype (254-C-A-G-A-132) which spanned a distance of 128 kb from marker D12S2516 to D12S2519 (Table 4). The frequency of this haplotype among noncarrier-chromosomes was only 20.7%.

DISCUSSION

Our findings, together with those from several recent studies, suggest two major differences in the spectrum and frequency of PD-related LRRK2 variants in Asians in comparison to individuals from other populations. First, pathogenic LRRK2 mutations appear to be rare in Asians. In our cohort of Japanese patients with PD, we observed mutations in none of the familial cases and less than 1% of the sporadic subgroup. A study of ethnic Chinese patients with PD in which exons 23–51 were sequenced revealed similar results.¹⁶ In contrast, comprehensive studies of the LRRK2 coding region in patients of European ancestry with familial PD have discovered pathogenic mutations at a frequency of 5%–13%,^17–22 and a single mutation (G2019S) has been reported in 26%–30% of Ashkenazi Jews and 37% of North African Arabs with familial PD.^23–25 Second, there is now convincing evidence that a LRRK2 polymorphism (G2385R) modifies risk for PD in Asians, whereas in patients of European origin, no such common risk variants have been identified thus far.^26,27

In our case-control analysis, the largest such study undertaken to date, we observed a strong association between G2385R and PD, though the effect size (unadjusted OR, 1.96) was slightly smaller than that seen in most previous studies (Fig. 1). Overall, the consistency of the effects seen across all studies, which were carried out in samples from three Asian subpopulations (Chinese, Japanese, and Malay), was striking. All but one study¹³ observed a significant over-representation of the 2385R allele in cases, and the magnitude of the effect was similar in all but the two smallest data-sets.^10,11 Repeatedly robust effects in case-control samples of small to modest size raise concern for publication bias. However, we did not observe marked asymmetry in a funnel plot of these data to support the presence of such bias (Fig. 2).

Despite intensive efforts which have included hundreds of case-control analyses and two modest-sized genome wide association studies,^28,29 the discovery of common genetic risk factors in PD has proved to be a very difficult task. In populations of European origin, arguably only two variants, the MAPT H1 haplotype clade^30,31 and the SNCA REP1 polymorphism,^32,33 have been consistently associated with PD. In other populations, evidence in favor of common PD risk alleles has proved largely inconclusive. We believe that sufficient evidence now exists to consider LRRK2 G2385R a bona fide PD susceptibility factor, which represents a major finding for several reasons. First, it is the first well-validated PD risk variant in a non-European population. Second, the observed effect size for the 2385R allele (Fig. 1, summary OR, 2.55) is substantially greater than that seen for MAPT H1 or SNCA REP1 (ORs, 1.25–1.5),^30–33 which convey risks in a range thought more typical for common variants in complex diseases.³⁴ This could have important implications for future research and potentially for clinical care. For example, G2385R genotype might prove to be useful in selecting high risk groups for clinical trials of neuroprotective agents, and eventually, for treatment decisions as the era of personalized medicine develops. Finally, G2385R occurs on the surface of a LRRK2 domain (WD40) that likely mediates protein-protein coupling,³⁵ and in vitro data suggest that it might have pro-apoptotic activity.¹² Thus, G2385R is itself likely to be a “true risk variant” in that it modifies susceptibility for PD by directly altering protein function. In contrast, it is not entirely clear whether REP1 exerts a biologically relevant effect on SNCA expression or is instead in linkage disequilibrium with the true risk variant. The same is true of the MAPT H1 haplotype clade where the identity of the functional variant(s) that convey disease risk is entirely unknown.³¹ Therefore, of the currently known common genetic risk factors for PD, G2385R provides the most direct route to new translational research.

As is the case in many dominant diseases, discerning rare benign nonsynonymous variants from pathogenic missense mutations in LRRK2-related PD has not been straightforward. Over 20 LRRK2 “mutations” which display evolutionary conservation and were absent in samples of control subjects have been reported.³⁶ However, all but five of these variants have only been observed in sporadic cases or small families in which there was insufficient information to assess cosegregation with disease.³⁷ Thus, the biological relevance of most of these variants remains unresolved. The problem is illustrated by the nonsynonymous substitution R1067Q which occurs in the leucine-rich repeat domain of the protein at a position that is conserved in vertebrates.¹⁶ R1067Q was originally observed in a single, sporadic, early-onset patient in a sample of 630 PD cases of predominantly ethnic Chinese origin.¹⁶ It was absent in 630 matched controls. In our Japanese PD cohort, which was of similar size, we observed R1067Q in two patients with sporadic disease. However, we also detected this variant in a control subject who at 83 years of age displayed no signs of parkinsonism and had no family history of PD. Since pathogenic LRRK2 mutations (e.g., G2019S) do not display complete penetrance, how should R1067Q be classified? We believe that for now it must remain on a long list of variants considered “potentially” pathogenic until sufficient data from extended pedigrees, large case-control samples, and/or functional assays in model systems become available to make a more definitive determination.

Ross et al. recently reported that the LRRK2 R1628P polymorphism, which alters a highly conversed amino residue within the “COR” domain of the protein, was associated with PD in a Han Chinese case-control sample.³ However, we did not observe this variant in PD Tiers 1 and 2, or in Control Group 1, which suggests that R1628P is rare or absent in the Japanese population.

All 2385R carriers analyzed in this study shared a single background haplotype, suggesting that these individuals might have arisen from a common founder. Farrer et al. performed a haplotype analysis of 2385R carriers in an ethnic Chinese PD cohort from Taiwan.⁷ They found that all carriers also shared a common haplotype, and estimated that these individuals originated from a single ancestor who lived approximately 4,800 years ago. Three markers overlapped between our haplotype analysis and theirs: D12S2516, D12S2519, and D12S2521. Comparison of data indicates that the vast majority of carriers in the two studies share alleles at all three markers. Thus, most if not all 2385R carriers of Japanese and Chinese ancestry might have descended from a common ancestor.

Our study had some limitations. The familial PD subgroup screened in Tier 1 was relatively small (n = 36), so it is possible that we might have missed rare pathogenic mutations present in the Japanese population. Control Group 2 was comprised of Japanese-Americans recruited from a restricted geographical region of the northwestern US. This raises the possibility of unrecognized admixture with non-Asians, which might result in lower allele frequencies since G2385R is essentially absent in European-Americans. Indeed, we did observe a lower frequency of 2385R carriers in our Japanese-American control subjects (Group 2; 5.6%) than in our controls from Japan (Group 1; 8.4%), though the difference was not significant. However, we were reassured by the fact that the 2385R carrier frequency reported by Funayama et al.⁹ (4.8%) in a sample of 457 control subjects from Japan was more similar to the frequency seen in Control Group 2 than in Group 1. Furthermore, Control Group 2 represents a cohort of subjects born earlier than 1936 whose families immigrated to the United States largely in the late 19th and early 20th century. During that era, there were substantial cultural barriers to intermarriage between Asians and non-Asians. Also, any intermarriage that did occur within a given family after arrival in the United States would likely have been relatively recent and known to the subject, who would have reported this information during the enrollment interview and consequently been excluded. Thus, we believe that the admixture proportion in Control Group 2 is extremely low.

Studies on the genetics of PD have largely focused on individuals of European origin. However, as demonstrated by the work presented here, a great deal of knowledge stands to be gained from analyses of non-European populations. Such work would best be accomplished by large-scale studies using standardized diagnostic criteria and uniform datasets across Asia, Africa, and Latin America.