Introduction

The GBA gene (MIM# 606463) encodes the lysosomal enzyme β-glucocerebrosidase that is mutated in Gaucher's disease (GD [MIM# 230800, 230900, 231000]). A very strong association has been established between heterozygous GBA mutations and Parkinson's disease (PD [MIM# 168600]) in a number of populations [1–10], and a large multi-center analysis further highlighted the importance of these mutations as risk factors for the development of PD worldwide [11]. GBA mutations are particularly common in PD patients of Ashkenazi Jewish origin, and were shown to affect disease course [12–14] and initial presentation [15, 16].

It is now clear that PD has a complex genetic background, which involves multiple affected genes and genetic loci, superimposed by aging and environment factors [17, 18]. The GBA chromosomal region (1q21) is a gene-rich locus that contains nine genes and two pseudo-genes (GBAP and MTX1P) [19]. It is therefore possible that other genetic alterations in the GBA region may affect the risk or the course of PD. One of the genes, MTX1 (MIM# 600605), encodes a mitochondrial protein. MTX1 is located 10 Kb downstream to GBA, and the MTX1 c.1051T>C (p.F351L) missense mutation was previously associated with the common Ashkenazi founder mutation GBA p.N370S [20], suggesting a linkage disequilibrium between these two genetic mutations. Hereby, we examined the possibility that alterations in MTX1 are associated with PD.

Methods

Study population

The study population included 600 PD patients of Ashkenazi Jewish origin with an average age at enrolment of 68.22 (±10.1) years, and the control population included 353 subjects with an average age at enrolment of 65.57 (±10.2) years. Details regarding the cohort recruitment, diagnostic criteria, and interview procedure were previously published [14, 21]. All participants signed informed consent and were enrolled consecutively into the study. The Institutional and National Supreme Helsinki Committees for Genetic Studies approved the study protocols and the informed consent forms.

Mutation detection

Genomic DNA was isolated from peripheral blood using standard protocols, or from saliva according to manufacturer's instructions (Oragene, Ottawa, Canada), and LRRK2 (MIM 609007) and GBA genotypes were determined as previously described [14, 22].

Three isoforms of MTX1 are currently known, encoding for 317 (AAC51819 [20]), 435 or 466 amino acids (NP_942584 and NP_002446, respectively). The nomenclature in our study refers to the long 466 amino acids variant only, and therefore, the previously described MTX1 p.F202L alteration [20] is mentioned here as p.F351L.

The entire MTX1 coding sequence and exon–intron boundaries were amplified by specific primer pairs that were designed to distinguish the MTX1 gene from its pseudogene (Online Resource 1). Exons 1–3 were directly amplified, whereas exons 4–8 were amplified via “nested PCR”: a 2,373-bp amplicon was generated followed by individual exon amplifications. The PCR products of exons 1–3, 5, and 7 were sequenced using BigDye Terminator Chemistry and automated ABI 310 or 3100 Genetic Analyzers (Applied Biosystems, Foster City, CA), and the PCR products of exons 4 and 6–8 were scanned for mutations using the WAVE DHPLC apparatus (Transgenomic Inc., Omaha, NE) [23]. The DHPLC mixing procedure was applied on exon 4 to detect samples homozygous for the c.750A>G variation.

The initial analysis of the entire MTX1 coding regions and exon–intron boundaries included 121 individuals (Table 1). Ninety-six (81 PD patients and 15 controls) were carriers of one of the eight common founder Ashkenazi GBA mutations tested [14], and 25 were PD patients who were non-carriers of these mutations and the LRRK2 G2019S mutation (Table 1). The MTX1 c.184T>A (p.S63T) alteration (rs760077) in exon 1 was further tested in all 600 PD patients and 353 controls using Taqman® assay (StepOne RT-PCR system, Applied Biosystems).

Table 1 MTX1 sequence alterations detected in carriers (PD patients and controls) and patient non-carriers of founder GBA mutations
Full size table

Sequencing of the entire GBA gene, including all coding exons and exon–intron boundaries, was performed as previously described [24], except for the following primers: exons 6 forward 5′-ACAGACATTTTGTCCCCTGC and reverse 5′- CTGATGGAGTGGGCAAGAATT, and exon 7 forward 5′- GGTCTGGTCCACTTTCTTGG.

Statistical analysis

Data are presented as mean (±standard deviation, SD) for continuous variables. Clinical and demographic categorical variables are presented in percentages, whereas allele frequencies are presented on a range of 0–1. Any differences in continuous variables among groups were tested using the analysis of variance (ANOVA) or Student's t test. The chi-square or Fisher's exact tests were used for comparison of categorical variables. To calculate linkage equilibrium (D′ and r 2) we used the CubeX online calculator [25]. The goodness of fit test with df = 1 was applied to test for any deviation from Hardy–Weinberg equilibrium among PD patients and controls. SPSS software V. 17 (SPSS Inc., Chicago, IL) was used for all other data analysis.

To analyze the possible association between the MTX1 c.184>A alteration and PD, and to exclude any possible confounding effect of GBA or LRRK2 G2019S mutations on this analysis, we stratified our cohort, and analyzed separately the patients (n = 401) and controls (n = 330) that did not carry either LRRK2 p.G2019S or GBA mutations.

For the analysis of the effects of the different MTX1 c.184T>A genotypes on the age at motor symptom onset (AAO) of PD, six genotypic groups of patients were compared using ANOVA: three groups of carriers of a founder GBA mutation with either MTX1 c.184A/A (GBA-A/A), c.184T/A (GBA-T/A), or c.184T/T (GBA-T/T) genotype, and three groups of non-carriers (NC) of founder GBA mutations (NC-A/A, NC-T/A and NC-T/T, Table 3). Since the LRRK2 p.G2019S mutation and homozygous and compound heterozygous GBA mutations significantly decrease the AAO in PD [14], we excluded them from the analysis.

Results

Detection of MTX1 gene variations

Table 1 presents all sequence variations detected in the MTX1 gene in 121 patients and controls, including four exonic alterations (c.184T>A, c.750A>G, c.1040C>T, and c.1051T>C) and one intronic change (IVS4-12C>T). The association between MTX1 c.1051T>C (p.F351L) and the GBA p.N370S mutation [20] was confirmed here, but this MTX1 alteration was not associated with any of the other Ashkenazi founder GBA mutations tested. In contrast, the majority of all the patients and healthy controls that carry GBA mutations also carried the MTX1 c.184T>A alteration (93%, 89/96), compared to 64% of the non-carriers (n = 25, p = 0.0008, Fisher's exact test, Table 1).

Since c.184T>A was the only change that was associated with multiple GBA mutations, we examined its association with GBA mutations in all our PD patients and controls (Table 2). When analyzing all PD patients and healthy controls as one group, the c.184A allele frequency was significantly higher in all GBA mutation carriers (0.67, 185/276) compared to all non-carriers (0.45, 739/1630, χ2 = 44.47, df = 1, p < 0.0001). Significant associations were also shown when analyzing the genotypes of patients and controls separately. GBA mutations were present in 22.1% and 31.9% of the patients with c.184T/A and A/A genotypes, respectively, compared to 5% in patients with c.184T/T genotype (χ 2 = 32.46, df = 2, p < 0.0001). In control individuals, only 1% with c.184T/T genotype carried GBA mutations, compared to 2.8% and 12.7% in those with T/A and A/A genotypes (χ 2 = 16.08, df = 2, p = 0.0003, Fig. 1).

Table 2 MTX1 c.184T>A (p.S63T) genotypes among Ashkenazi PD patients and controls, carriers and non-carriers of GBA mutations
Full size table
Fig. 1
figure 1

The significant majority of founder Ashkenazi GBA mutations were detected in individuals that carry homozygous or heterozygous MTX1 c.184A allele. The Y-axis depicts the percentages of individuals with GBA mutations among the patients and controls with different MTX1 c.184T/A genotypes (X-axis)

Full size image

Linkage disequilibrium analysis was done between all GBA mutations and the MTX1 c.184T>A alteration. There were no evidence for recombination events (D′ = 1) between the c.184T>A alteration and the N370S, 84GG, IVS2 + 1, V394L, and RecTL mutations in GBA. Table 2 details the distribution of the specific GBA mutations among all patients and controls with different MTX1 c.184T/A genotypes. Taken together, our results strongly suggest a preferential linkage of GBA mutations to the MTX1 c.184A allele. Of note, one of the founder GBA mutation, p.R496H, which was sixfold more prevalent among PD patients (11/600, 1.83%) than controls (1/353, 0.28%, p = 0.039), was not found in individuals with homozygous c.184A/A genotype, but only in those with MTX1 c.184T allele (Table 2).

Is the MTX1 c.184A allele an independent risk factor for PD?

We examined the possibility that the c.184A allele is an independent risk factor for PD, by stratifying our cohort according to mutation carriage. This analysis was performed among all PD patients (n = 401) and controls (n = 330) that did not carry LRRK2 p.G2019S or GBA mutations. No differences were found between the distributions of the MTX1 c.184 genotypes among patients and controls: 50.9% and 50.6% with MTX1 c.184A/A, and 21.9% and 18.5% with c.184T/A, respectively (χ 2 = 1.937, df = 2, p = 0.38). Furthermore, there were no differences in the AAO, and there were no associations between the prevalence of familial PD and any of the MTX1 c.184T/A genotypes or alleles among these 401 PD patients (data not shown). Together, these results suggest that the c.184A allele is not an independent risk factor for PD.

An earlier age at onset in GBA mutation carriers with the homozygous MTX1 c.184A/A genotype

The average ages at of motor symptoms onset (AAO) in patient carriers and non-carriers of GBA mutations are presented in Table 3. A significantly younger AAO was observed in the GBA-A/A group of patients, compared to the other five genotypic groups tested (3.3–5.9 years earlier, one-way ANOVA, F (5,500) = 2.29, p = 0.045). When the group of seven patients with GBA-T/T was excluded from the analysis due to its small size, similar results were obtained (F (4,494) = 2.86, p = 0.023).

Table 3 Age at Parkinson's disease motor symptoms onset (AAO) according to genotypes of MTX1 c.184T>A and carrier status of GBA mutations
Full size table

Post-hoc analysis revealed that the sources of significance were the comparisons between the GBA-A/A group to the three genotypic groups of non-carriers (5.1–5.9 years earlier, p = 0.002–0.013). If mutations in GBA were the only factor responsible for this effect, we would expect significant differences between all three genotypic groups of GBA carriers (A/A, T/A, and T/T) when compared to the three genotypic groups of non-carriers. Therefore, we concluded that the combination of GBA mutation carriage and the MTX1 c.184A/A genotype is responsible for the earlier age of motor symptoms onset.

Early-onset PD (EOPD) may be defined as AAO <50 years. A significantly greater proportion of EOPD was observed in patients with GBA-A/A genotype (35.9%, Table 3), compared to 13.6–17.5% in all other five genotypic groups (χ 2 = 12.54, df = 5, p = 0.028). Of interest, we also examined the AAO in the six GBA-homozygous and GBA-compound heterozygous patients, who were not included in Table 3. The AAO among those with homozygous c.184A allele were 41, 44, 47, and 53 years, earlier than the AAO (54 and 68 years) in the two c.184T/A heterozygotes.

Severe and mild GBA mutations were already shown to differentially affect the AAO of PD [14, 26]. It was therefore necessary to exclude the possibility that the severity of the GBA mutation confounds the AAO analysis. First, we examined the prevalence of severe GBA mutation carriers among the GBA-A/A group and the GBA-T/A and GBA-T/T group (as detailed in Table 2). These two groups included a similar percentage of severe GBA mutation carriers (26% and 24%, p = 0.79). Second, additional ANOVA was conducted with the type of GBA mutation (severe or mild) as a covariate, and resulted in a similarly significant effect of MTX1 genotypes on AAO (p = 0.035, the genotypic groups are detailed in Table 3). Together, these analyses demonstrate that the severity of GBA mutations was not a confounder that affected the earlier AAO observed among the GBA-A/A group.

An additional rare, non-founder GBA mutation was detected on the MTX1 c.184A allele

Since most of the GBA mutations (94%) were associated with the MTX1 c.184A allele, we hypothesized that other GBA mutations, which were not included in our panel of mutation screening, might be associated with this allele. We therefore sequenced the entire coding region of the GBA gene and exon–intron boundaries in 17 PD patients: 10 GBA-carriers with homozygous MTX1 c.184A/A genotype and early AAO (between 30 and 47 years), and 7 randomly selected non-GBA carriers, with MTX1 c.184A/T or A/A genotypes. No alterations were detected, except for a c.130C>T substitution in one patient previously diagnosed as a GBA mutation carrier. These results ruled out the possibility that an additional, yet unknown rare GBA mutation, is responsible for the earlier AAO in the other nine EOPD patients that carry a founder GBA mutation and a homozygous MTX1 c.184A/A genotype.

The c.130C>T substitution detected here causes an arginine to cysteine change in position 44 (p.R44C, Fig. 2a [27]), altering a polar hydrophilic positively charged amino acid (aa) to a polar hydrophobic neutral aa. To estimate the potential functional role of this mutation, we used the online SIFT software [28]. Seventy-one sequences represented at this position were found, determining a score of 0.01 that this aa change affects protein function (aa with a score of <0.05 is predicted to be deleterious). Furthermore, the GBA p.44R aa was conserved during evolution (Fig. 2b), resulting in a median sequence conservation score of 2.12 (SIFT software). Taken together, these data suggest that the GBA p.R44C missense mutation has a potentially deleterious functional consequence. Of note, the patient carrying this mutation was previously diagnosed as EOPD, with AAO of 36 years. Since he was a known GBA carrier of the p.V394L mutation, he was consequently categorized as a compound heterozygous.

Fig. 2
figure 2

a The GBA p.R44C mutation. Sequence chromatograms of a control non-carrier (wt wild type, arrow in upper panel) and a GBA p.R44C mutation carrier (c.130C>T, arrow in lower panel). b The alignment of amino acids sequences from eight different species demonstrates that the GBA p.R44 (arrow) was conserved during evolution

Full size image

Discussion

The worldwide importance of GBA mutations as risk factors for PD was established [11], although the majority of Ashkenazi GBA mutation carriers will probably never develop PD [14]. The age at motor symptoms onset in PD varies greatly between different GBA mutations carriers, and a large range of AAO was noted among carriers of the same GBA mutation [14, 26]. It is therefore apparent that other genetic, as well as environmental and age-related factors, modify the risk and onset of PD among GBA mutation carriers. Here, we report for the first time, the existence of a modifier marker in Ashkenazi GBA-associated PD, the homozygous c.184TA/A genotype in the MTX1 gene. Whether this effect on AAO is caused by this missense mutation is currently unknown. Another possible explanation is the existence of a yet uncharacterized genetic change on the “trans allele”, the opposite allele to the one with GBA mutation. Our results, which demonstrated that only homozygous MTX1 c.184A/A genotype is related to earlier AAO, suggest the involvement of the trans-GBA allele. Therefore, a comprehensive sequence analysis of the entire GBA region in multiple patients is necessary to detect all the variations potentially involved in PD pathogenesis.

Defects in mitochondrial protein transport and accumulation were implied in PD pathogenesis [29, 30]. Metaxin1, the protein encoded by MTX1, is a member of a mitochondrial outer membrane complex transporting proteins into the mitochondria [31–33]. One of the proteins transported by this complex is VDAC [32] [voltage-dependent anion-selective channel (MIM# 604492)], which is involved in PINK1/Parkin- (MIM# 608309 and MIM# 602544) mediated mitophagy [34]. Taken together, these data raise the possibility of MTX1 involvement in PD, but additional studies in other populations are necessary.

Several studies have already reported that interactions between genes affect the AAO of PD. The interactions between SNCA and APOE (MIM# 107741) [35], between LRRK2 and MAPT (MIM# 157140) [36] and between IL6 (MIM# 147620) and ESR2 (MIM# 601663) [37] were all shown to be associated with earlier disease onset. Interestingly, genetic variations in individual genes were already shown to affect the AAO of PD, but not increase the risk for PD independently, including PGC1-α [38], PITX3 [39], and GLUD2 [40]. However, the biological mechanisms that may explain these findings, as well as the effect of MTX1 c.184T>A alteration on AAO of PD among GBA mutation carriers, but not on its risk, are currently unknown. These reports, together with our results, support the idea that the course of PD, and specifically, the AAO in GBA-associated PD, is affected by more than one gene or genetic factors.