Association of Sequence Alterations in the Putative Promoter of RAB7L1 With a Reduced Parkinson Disease Risk

Parkinson disease (PD) (OMIM 168600) is a multifactorial, age-related disorder affecting about 2% of the elderly population. More than a decade of study of familial and sporadic PD has revealed PD as a disorder with complex genetic background involving multiple affected genes and genetic loci^{1 - 2} and also multiple environmental risk factors.³ Most of the genes studied were associated with increased risk for PD; however, 2 large-scale genome-wide association studies in Japanese and North American and European populations identified PARK16, a novel locus on chromosome 1q32 (OMIM 613164) that was associated with a reduced PD risk.^{4 - 5} This region encompasses 5 genes: SLC45A3 (OMIM 605097), NUCKS1 (OMIM 611912), RAB7L1 (OMIM 603949), SLC41A1 (OMIM 610801), and PM20D1 (OMIM 613164).^{4 - 5} Recently, the association between PARK16 and PD was also confirmed in ethnic Chinese,^{6 - 7} European,^{8 - 9} and Chilean¹⁰ populations but was not confirmed in other studies in Spanish, European, and North American populations.^{11 - 13}

The high prevalence of founder GBA and LRRK2 mutations in more than a third of the Ashkenazi Jewish patients with PD^{14 - 15} is an additional demonstration for the genetic homogeneity of this population.¹⁶ This phenomenon renders this population useful for detecting other common genetic variations that affect PD risk. Herein, we aimed to examine whether such alterations exist in the PARK16 locus in Ashkenazim.

The PD patient cohort included 720 consecutively recruited individuals (62.6% men) (Table 1), with a mean (SD) age at enrollment of 67.5 (10.4) years. The 642 control samples included 267 sex-matched elderly controls with a mean (SD) age at enrollment of 69.1 (10.9) years (62.9% men) and 375 young controls (mostly women, 20-45 years old). All patients and controls were of full Ashkenazi Jewish origin. Details regarding the cohort recruitment, diagnostic criteria, and interview procedures were previously published.^{14 - 15} All participants signed informed consent before entering the study. The Institutional and National Supreme Helsinki Committees for Genetic Studies approved the study protocols and the informed consents.

Genomic DNA was isolated from peripheral blood using standard protocols or from saliva according to the manufacturer's instructions (Oragene; DNA Genotek). Table 1 details some demographic and clinical characteristics and the frequencies of the GBA and LRRK2 G2019S mutations in our cohort of 720 patients with PD. Mutations in the LRRK2 and GBA genes were identified as previously described,^{14 - 15} and the LRRK2 G2019S mutation (rs34637584) was also detected using TaqMan assay ID C_63498123_10 in the StepOnePlus Real-Time PCR System (Applied Biosystems). The 5 tested single-nucleotide polymorphisms (SNPs) within the PARK16 locus (eTable 1) were analyzed using TaqMan assays.

A 1588–base pair (bp) fragment that contains the putative promoter region of RAB7L1, exon 1 and part of exon 2, and the coding sequences of exons 2 through 6 and their exon-intron boundaries were amplified (Biometra GmbH) (the specific primers are detailed in eTable 2) and sequenced using the BigDye Terminator cycle sequencing chemistry in the automated ABI 3100 Genetic Analyzer (Applied Biosystems).

The following tracks, implemented in the University of California, Santa Cruz, Genome Browser, were used to predict the putative promoter region of RAB7L1: CpG Islands, ENCODE Transcription Factor ChIP-seq, and ENCODE Promoter-Associated Histone Mark (University of California, Santa Cruz). To analyze the possible effects of the rs823144 and rs1572931 SNPs located within this promoter on the putative transcription factor binding sites, 15-bp sequences from each side of the SNP were analyzed using TFSEARCH (http://www.cbrc.jp/research/db/TFSEARCH.html). The parameters used for this comparison were vertebrate matrices only, with a threshold score of 75.0 points (of a maximum 100.0), as was previously described.^{17 - 18}

Differences in continuous variables were tested using analysis of variance or a t test. A χ² or Fisher exact test was used for comparison of categorical variables. Bonferroni correction for multiple comparisons was applied and a cutoff P value of .01 was set for the analysis of the 5 SNPs (Table 2). Since the allele and genotype frequencies of all tested SNPs did not differ between the young and elderly controls (Table 2) or between men and women (data not shown), these variables were not included as covariates in the analysis. To test for any deviation from Hardy-Weinberg equilibrium among patients with PD and controls, a goodness of fit test with 1 df was applied. Calculations of odds ratios (ORs) for minor allele carriers were done using the online Hutchon calculator.¹⁹ For haplotype analysis, HPlus software version 3.2 was used.²⁰ Analysis of the Affymetrix SNP 6.0 array results was done using the Golden Helix SNP & Variation Suite software (Golden Helix, Inc). To calculate linkage equilibrium (D′ and r²), we used the CubeX online calculator.²¹ SPSS software version 17 (IBM SPSS Inc) was used for all other data analyses.

The PARK16 locus was defined by Satake et al⁴ on chromosome 1q32 between 203 910 000 and 204 070 000 (genome assembly GCRh37). To select candidate SNPs for this study, we used unpublished results from a genome-wide analysis of 426 DNA samples done in our laboratory (A.O., unpublished data, June 2009). The samples were from 128 elderly controls and 298 patients with PD, all of Ashkenazi origin. Since the patients samples were greatly enriched for LRRK2 G2019S carriers (25.2%) and for GBA mutations carriers (31.2%), they did not represent the true carrier rate in the Ashkenazi PD population. These 426 samples were processed on the Human SNP array 6.0, according to the manufacturer's protocols (Affymetrix Inc). In this array, 40 SNPs are included in the PARK16 region. Two of them, rs1772153 and rs1775151, with uncorrected P values <.007 (red circles in the eFigure) were chosen for the initial analysis of the entire unselected consecutively recruited cohort of 720 patients with PD.

The third SNP for analysis was rs947211, the most significantly PARK16 -associated SNP in Satake et al⁴ (black circle in the eFigure). Table 2 details the genotypes and minor allele frequencies in all patients and controls. Since the minor allele frequencies were similar in both subgroups of controls, young and elderly, the 2 groups were analyzed together (total controls in Table 2). In addition, although most young controls were women, there were no differences in the allele and genotype frequencies between men and women (data not shown), allowing us to analyze both sexes together. Table 2 (top 3 SNPs) demonstrates that all the minor alleles of these 3 SNPs were significantly associated with a reduced risk for PD.

Since rs1772153 and rs1775151 are closely located to RAB7L1, 1698 and 1600 bp from its translation start site (Figure), we searched for additional variations in this gene. First, the RAB7L1 putative promoter region was defined based on data available from the University of California, Santa Cruz, Genome Browser (Figure). Second, a 1588-bp region that includes the putative promoter was sequenced in 8 individuals: 4 homozygous for both rs1772153 and rs1775151 (C/C and T/T, respectively) and 4 heterozygous (C/T and T/C). All 4 heterozygous individuals carried 2 additional alterations within the putative RAB7L1 promoter, 461 and 133 bp from its translation start site, an A>C alteration in position 205744546 (rs823144) and a C>T alteration in position 205744218 (rs1572931). Linkage disequilibrium (LD) analysis was performed to determine the linkage between these 5 SNPs. Linkage disequilibrium was demonstrated between all these SNPs as detailed in eTable 3 (D′ = 0.89-1.0; r² = 0.37-1.0). The LD statistics were similar in patients with PD and controls.

The online software TFSEARCH was used to predict whether rs823144 and rs1572931 may modify the binding capacity of known transcription factors. The change at rs1572931 adds a putative binding site for the CdxA transcription factor (TFSEARCH score, 75.0) (Figure, B), and the substitution at rs823144 potentially eliminates the binding site for 1 transcription factor (c-Ets-1; TFSEARCH score, 78.4) while adding binding sites for 3 others (p300, GATA-1, and Sp1; TFSEARCH scores: 76.2, 75.5, and 75.3, respectively) (Figure, B).

Table 2 (bottom 2 SNPs) details the genotypes and minor allele frequencies of rs823144 and rs1572931 in all patients with PD and controls. Of the 5 tested SNPs, individuals with the rs1572931 C>T alteration had the lowest risk for PD (OR = 0.64; P = .0002). Among patients with PD, only 0.4% were homozygous for rs1572931 T/T, compared with 3.0% in controls (P = 5 × 10⁻⁵, Fisher exact test).

Since rs823144 sequence changes were fully linked to those in rs1775151 (eTable 3), and since all SNPs tested herein demonstrated a reduced PD risk, we hypothesized that a specific protective haplotype may exist in Ashkenazim. Five haplotypes with a frequency of more than 0.001 were defined for the tested SNPs (HPlus version 3.2). These haplotypes, designated as A to E in Table 3, were tagged by 3 SNPs (rs947211, rs1772153, and rs1572931). As expected, carriers of haplotypes B and D who harbor the rs1572931 C>T change had a significantly reduced risk for PD. Notably, haplotype D demonstrated a particularly high protective effect, a 10-fold lower risk (P = .002) (Table 3).

Because of high frequencies of mutations in the GBA and LRRK2 genes that increase PD risk among Ashkenazim^{14 - 15} (Table 1), it was possible that the carriage of these mutations affected the risk estimates in Table 2. We therefore reanalyzed the haplotypes after excluding all 241 carrier patients (Table 3). In noncarrier patients, only haplotype D showed a significant protective effect (OR = 0.077; P = .01) (Table 3), suggesting that this allele reduces the risk for PD independently of the existence of founder mutations in the GBA and LRRK2 genes. Since the status of GBA and LRRK2 mutations in most controls was not known, and the expected number of control carriers was low, we did not include these variables as covariates and did not estimate the risks for the GBA and LRRK2 carrier groups separately.

To examine the possibility that the significant SNPs and haplotypes detected herein are merely markers for functional sequence changes in RAB7L1, we completed the analysis of its coding exons and exon-intron boundaries. Nine individuals were sequenced: 3 homozygous carriers of haplotype A, 3 homozygous carriers of haplotype B, and 3 heterozygous carriers of haplotype D (there was not even 1 homozygous carrier of haplotype D in our cohort). No additional sequence variations were found.

To our knowledge, our study demonstrates for the first time the association between the PARK16 locus and PD risk in the Ashkenazi Jewish population. The significance of this locus for PD was recently demonstrated in multiple population-based studies,^{4 - 10} suggesting that it may be involved in PD pathogenesis worldwide. However, other genome-wide association studies in Europeans and North Americans, including a large meta-analysis,^{12 - 13} and a study of a Spanish population¹¹ could not confirm this association. Our results from the genetically homogeneous Ashkenazi population strengthen the importance of PARK16 in PD and add it to the short list of genes (eg, SNCA [PARK1/4 ], LRRK2 [PARK8 ], and GBA²]) that were already shown to affect PD risk in multiple populations.

We identified new genetic elements related to the reduced risk for PD. A highly protective haplotype associated with a 10-fold risk reduction (designated herein as haplotype D) was defined between the SLC41A1 and RAB7L1 genes. This haplotype includes 2 SNPs in the putative promoter region of RAB7L1 that alter predicted transcription factor binding sites. No coding or splice site alterations in RAB7L1 sequence were found that could explain the protective effect, suggesting that other, yet unknown, molecular mechanisms are involved. Although our analysis focused on the minor alleles and haplotypes that are associated with reduced PD risk, it is clear that in this locus there are genetic elements that are associated with increased risk for PD, such as haplotype A (Table 3). Our results do not rule out the possibility that other genes or genetic variations within the PARK16 region are responsible for its effect on PD.

The small LD block detected in our analysis of the PARK16 locus is approximately 10 kilobases long. This is smaller than the sizes of the PARK16 haplotypes expected in the Ashkenazi population. To better define the actual size, sequencing of the entire region is necessary. While the LD block described herein (eFigure) involves the RAB7L1 and SLC21A1 genes, previously published results^{4 - 5} suggested that the LD blocks in the Japanese and North American and European populations encompass a much larger region of about 160 kilobases, which includes 3 additional genes (SLC45A3, NUCKS1, and PM20D1). Differences in LD block sizes could have resulted from the different genotyping platforms used; the differences between populations tested, ie, their sizes and/or genetic homogeneity; or from a combination of any of these.

The function of the protein encoded by RAB7L1 (RAB7, member RAS oncogene familylike 1) is not known. The RAB7L1 gene homology to human RAB7 (OMIM 602298)²² suggests that it is a member of the Rab (Ras-related in brain) family of proteins. More than 60 Rab genes are spread across the human genome, and they function as regulators in different membrane traffic pathways.²³ Mutations in RAB7 cause Charcot-Marie-Tooth type 2B neuropathy²⁴ and knockdown of the Caenorhabditis elegans Rab7 gene caused severe growth and motor abnormalities selectively in α-synuclein transgenic worms.²⁵ Other Rab proteins were also suggested to be involved in neurodegeneration and PD, including Rab1,²⁶ Rab3a and Rab8a,²⁷ and Rab11a.²⁸

To conclude, our data confirmed the association between the PARK16 locus and reduced PD risk and demonstrated that alterations within the RAB7L1 putative promoter are associated with this effect. However, further studies are necessary to determine whether it is RAB7L1 or other genes or genetic elements in the PARK16 locus that are involved in modifying the risk for PD.