Main

Parkinson’s disease (PD) affects 1–2% of people older than 60 years.1 In addition to rare mutations that have been linked to familial forms of PD, DNA polymorphisms at several genes could contribute to the risk of developing sporadic PD. The leucine-rich repeat kinase 2 gene (LRRK2, PARK8) has been linked to both sporadic and familial PD.2, 3, 4, 5 The LRRK2 p.G2019S mutation accounts for 30–40% of PD cases among North African Berbers patients and Ashkenazi Jews and has also been found in 5% of Spanish patients.6 The p.R1441G mutation is also common in Spain, with a frequency in PD cases close to 8% in some regions.7 The LRRK2 gene encodes a protein with multiple functional domains, including leucine-rich repeats, a GTPase domain and a kinase domain, which, in addition to the brain, is expressed in multiple tissues.8 The mutations found in PD patients disrupt the enzymatic activities of the protein by affecting the kinase (p.G2019S) or the GTPase (p.R1441G/C) activities.9, 10 Some studies also reported differences in LRRK2 expression between brains from PD cases compared with those from controls.11, 12, 13

Most of the mutational screenings were focused on the LRRK2 coding exons, whereas little is known about the contribution of 3′UTR variants to the risk for PD. However, several studies have provided strong evidence of the role of this gene region on the regulation of LRRK2 expression through microRNA binding, and the control of cellular pathways in which this gene is implicated. Among others, pathogenic LRRK2 antagonizes let-7 and miR-184-3p in Drosophila, leading to the overproduction of proteins implicated in cell cycle regulation and survival.14 It was also reported that miR-205 is able to suppress the expression of LRRK2 by targeting a conserved binding site in the 3′UTR region.15 Our first aim was to characterize the contribution of the LRRK2 3′UTR variation in the development of PD.

The study involved a total of 743 PD patients and 523 healthy controls, all Iberians from the region of Asturias (Northern Spain). The main characteristics of these cohorts are summarized in Table 1 and were previously reported.16 The LRRK2 3′UTR was amplified in six overlapping fragments (Supplementary Table 1), which were subjected to single strand conformation analysis and direct sequencing to identify the nucleotide changes. We found a total of 12 variants (Supplementary Table 2), and only the SNP rs66737902 was linked to the risk of developing PD: allele C-carriers were frequent among the patients (P=0.005; odds ratio=1.47, 95% confidence interval (CI)=1.12–1.92) (Table 2). The sample size was enough for a statistical power >80%. The difference remained significant after correcting for age and gender (P=0.01; odds ratio=1.47, 95% CI=1.09–1.97). To our knowledge, the association between PD and rs66737902 has not been reported by others who also examined the LRRK2 3′UTR.4, 17 This SNP showed a heterogeneous worldwide frequency distribution, but was absent among the Chinese and Japanese (www.ensembl.org). Differences in ethnic background could thus explain these discrepancies.

Table 1 Main characteristics of patients and controls enrolled in the study
Full size table
Table 2 Genotype and allele frequencies of the four 3′UTR SNPs in PD patients and controls, and frequencies in the 1000 Caucasian genomes (www.ensembl.org)
Full size table

We found a significant linkage disequilibrium between rs66737902 and p.R1441G (rs33939927): a total of 15 patients were heterozygous for this mutation, and 14 and 1 were rs66737902 TC and TT, respectively. This mutation was not found among the 105 controls who were rs66737902 C carriers. After excluding the 15 p.R1441G mutation carriers, carriers of rs66737902 C remained significantly increased among the patients. This suggested that the risk effect of this 3′UTR allele could not be fully explained by its linkage disequilibrium with the LRRK2 mutation. In contrast, most of the p.G2019S mutation carriers (13/16) were homozygous for the rs66737902 TT protective genotype.

We measured the LRRK2 cDNA levels in three brain regions (SN, cerebellum, and occipital cortex) of 9 PD donors kindly provided by the London Neurodegenerative Diseases Brain Bank (Supplementary Table 3). Tissue processing and expression assays were performed as reported by Cardo et al.16 Briefly, the level of LRRK2 and GAPDH (constitutively expressed normalization control) cDNA was determined using SYBR Green. The amount of LRRK2 was calculated as the fold difference, 2−ΔΔCt. We found a significant lower mean level of LRRK2 cDNA (P=0.01) in the SN of rs66737902 TC (n=4) vs TT (n=5) donors (Figure 1). This effect was not seen in the cerebellum and cortex samples. Some studies have found significant reductions of LRRK2 mRNA in the brains from PD cases compared with controls.11 In the brains of patients with PD, LRRK2 immunoreactivity was reduced in SN neurons, a fact that reflects the disease-associated loss of dopaminergic neurons in this brain region.12 Another study did not find significant alteration in LRRK2 expression between brain tissue from p.G2019S mutation carriers and those from controls.13 None of the nine brain donors analyzed was a p.R1441G carrier, and thus we could not examine whether this mutation was also linked to differences in LRRK2 expression. The differences in mRNA expression between the genotypes could be attributed to its linkage disequilibrium with p.R1441G, but whether they also exist at the protein level can be confirmed only through additional studies.

Figure 1
figure 1

Mean LRRK2 expression (cDNA) in the SN, cerebellum (CB) and occipital cortex (OC) of the nine PD donors according to the rs66737902 genotype (4 TC, 5 TT). The level of LRRK2 mRNA was determined through real-time PCR using ABI7500 equipment (Life Technologies, Carlsbad, CA, USA) and the GAPDH gene was used as the endogenous control for quantification. Each cDNA was amplified in triplicate using iTaq Universal SYBR Green Supermix (Bio-Rad, Hercules, CA, USA) under the following PCR conditions: 95°C for 30 s followed by 40 cycles at 95°C for 15 s and at 58°C for 34 s, and a final melting curve stage. The amount of LRRK2 was calculated as the fold difference 2−ΔΔCt, in which ΔCt=Ct LRRK2−Ct GAPDH; ΔΔCt=ΔCt sample−ΔCt controls mean. LRRK2 primers: ATA TAA AGG CTC GCG CTT CT and TGG CAT TCA CAA AGT GGT AAT (amplified fragment=162 bp). GADPH primers: GGA AGG ACT CAT GAC CAC AG and TTG GCA GGT TTT TCT AGA CGT (amplified fragment=249 bp).

Full size image

Finally, an online analysis with two miRNA-site prediction software programs (http://www.microrna.org; http://diana.cslab.ece.ntua.gr/) identified miR-138.2 as a candidate that could bind to the rs66737902 T sequence (Supplementary Figure 1). We found that this miRNA was expressed in all the nine SN samples (Supplementary Figure 2). Thus, the PD-protective T allele could result in enhanced miRNA binding and downregulation of the LLRK2 mRNA. However, this would be in opposition to the higher transcript levels among the TT SN samples. Moreover, this binding site was recognized by only one of the programs and with a low score (Supplementary Figure 1). Although we considered unlikely a direct effect of the rs66737902 SNP through different binding of miR-138.2 to the two alleles, in vitro studies with neural cells transfected with reporter vectors coupled to 3′UTRs with the two alleles are required in case other studies confirm the genetic association.

In conclusion, we report the association between the LRRK2 rs66737902 SNP and PD. This effect could be mediated by a significant reduction of gene expression in the SN of individuals with the risk allele, and require validation in other populations with larger samples of brain tissues.