Introduction

Mitochondrial involvement in the pathogenesis of Parkinson's disease (PD) stems from the observation that parkinsonism can be induced by the neurotoxin 1-methyl-4-phenyl-1,2,5,6-tetrahydropyridine (MPTP), which inhibits complex I of the mitochondrial respiratory chain through the metabolite MPP+.1 More recently, a PD-like phenotype has been induced in rodents using rotenone, another potent inhibitor of complex I.2 Partially defective complex I activity has also been documented in tissues3 and cybrids derived from PD patients.4 The latter observation suggests that defects in mitochondrial DNA (mtDNA) may be responsible for biochemical dysfunction. However, although parkinsonian features have occasionally been reported in mtDNA-related encephalomyopathies,5 neither maternal transmission nor specific mtDNA mutations have been associated with idiopathic PD.6 Nevertheless, the analysis of mtDNA haplogroups in a large cohort of idiopathic PD patients from the white US population suggested that European mtDNA haplogroups J and K, and their common single-nucleotide polymorphism (SNP) 10398G in the ND3 gene, exert a protective effect from the risk of PD compared to haplogroup H – the most common mtDNA haplogroup in Europe (40%) – which carries the 10398A SNP.7 On the contrary, Autere and colleagues8 have found that the cluster made up of haplogroups J and T, defined by the mutation 4216C in the ND1 gene, was more frequent among PD patients than the supercluster HVKU. In this study, to further evaluate the role of mtDNA backgrounds in PD, we have genotyped 620 mtDNAs from Italian idiopathic PD patients.

Materials and methods

Samples

All PD patients were Italians. Inclusion criteria included:

  1. 1)

    disease onset after the age of 50 years;

  2. 2)

    disease duration of at least 5 years;

  3. 3)

    asymmetric onset;

  4. 4)

    presence of at least two cardinal signs including:

    1. a)

      tremor at rest,

    2. b)

      nonspastic rigidity,

    3. c)

      hypokinesia;

  5. 5)

    good and sustained response to l-DOPA.

Exclusion criteria included:

  1. 1)

    a positive family history up to second-degree relatives for PD or other movement disorders;

  2. 2)

    the presence of additional neurological symptoms referable to vascular accidents or non-PD neurodegenerative syndromes;

  3. 3)

    age of disease onset >65 years.

Controls included two groups, which are hereafter referred to as CT1 and CT2. CT1 was composed of 1486 random Italian subjects. For this control group, sex and age of individual subjects were not available thus it is not sex- and age-matched with the PD cohort. However, CT1 is by far the largest and most representative data set on the distribution of mtDNA haplogroups in Italians. CT2 was made up of 509 unrelated adult Italian subjects, of these 232 were over the age of 50 years, with no significant signs of cognitive or neurological impairment when enrolled in the study. The age-at-onset (AAO) mean (±SD) in PD individuals was 57.3±5.1 years. AAO was self-reported by the PD patients and defined as the age at which the affected individual first noticed one of the cardinal signs of PD. The age-at-examination (AAE) mean was 67.4±6.3 years in PD patients and 47.9±15.3 years in the CT2 controls. AAE was defined as the age at which the affected or unaffected participant was clinically examined and enrolled in the study. The mean age in the 232 CT2 individuals who were >50 years old was 62.1±7.8 years. Written consent was obtained from all participants in agreement with guidelines from the Ethical Committee of the Istituto Nazionale Neurologico C. Besta. For all subjects enrolled in the study, 10 mtDNA SNPs defining different haplogroups were surveyed by RFLP analysis (Table 1).

Table 1 Diagnostic RFLP markers used for haplogroup classification

Statistics

Statistical analyses were performed using the SSPS statistical package, v. 12.0, and statistical significance was established at α0.05. Binary logistic regression analysis was used to assess the risk of mtDNA SNPs and haplogroups with PD. Taking into account that SNP genotype/haplogroup affiliation is a categorical independent variable with more than two categories, we performed two logistic regressions: one of the SNP genotype categories all at once, and one of the haplogroup categories all at once (to look for any effect). In the first logistic regression, the risk of PD associated with the mutations 4216C and 10398G was considered by using the SNP 7028C, which is diagnostic of haplogroup H (Table 1), as a reference level. In the second logistic regression analysis, we analyzed the haplogroup categories (H, J, K, T, U) in patients and controls using the common haplogroup H as a reference level. In both tests, we evaluated the likelihood of the entire model by the P-value associated with the likelihood ratio test. The odds ratio (OR) values for each SNP/haplogroup, with their associated 95% confidence intervals (CI), were also obtained. The analyses were performed by the use of either CT1+CT2, CT1 alone or CT2 alone as controls. Moreover, to further evaluate the differential distribution of haplogroup K between PD and control subjects (maintaining haplogroup H as a reference group), the Fisher's exact test was performed on a 2 × 2 contingency table.

Results

The results of the RFLP survey of Table 1 were used to determine the frequencies of mtDNAs harboring 4216C and 10398G, as well as the frequencies of haplogroups H, I, J, K, T, U (w/t K) and L–M in PD patients and control groups (Table 2). Table 3 reports the ORs (with relative lower and upper CI values) obtained from the binary logistic regressions in numerous comparisons between patients and controls groups. As the power of the analyses depends on sample size, the most reliable among all comparisons is that employing the combined set (CT1+CT2) of 1995 control subjects. As for the SNPs 4216C and 10398G, we noted that, employing the likelihood ratio test statistic, for each comparison, the general model was always superior to the interaction model in terms of overall fit (P-values >0.05), with only one exception, the comparison PD (males) vs CT2 (males), which suggests a potential protective role of 13098G in males (OR=0.59; 95% CI: 0.39–0.91).

Table 2 Haplogroup affiliation of mtDNAs from Italian PD patients and controls
Table 3 ORsa of mtDNA SNPs and haplogroups

The role of 10398G and 4216C in PD was further evaluated by taking into account that 10398G characterizes haplogroups J and K (haplogroup I is too rare for reliable comparisons), while 4216C marks haplogroups J and T. Haplogroup U was used as a control. When age and sex were not taken into account, the logistic regressions showed a significant likelihood only in the comparison of PD vs CT1+CT2 (P-value=0.048), thus highlighting the importance of the sample size. Overall, the haplogroup comparisons of Table 3, which did not consider age and sex, support a potential protective role of 10398G in the context of haplogroup K, but never on a haplogroup J background. Indeed, significant differences for K were observed in PD vs CT1+CT2 (OR=0.54; 95% CI: 0.35–0.83), PD vs CT1 alone (OR=0.53; 95% CI: 0.34–0.83) and PD vs CT2 alone (OR=0.56; 95% CI: 0.32–0.97). Stratification by age and sex (in other words, comparisons involving only the CT2 cohort) appears to further support the protective effect of haplogroup K (PD vs CT2 >50 years; OR=0.44; 95% CI: 0.24–0.84), particularly in males as shown by PD (males) vs CT2 (males) (OR=0.42; 95% CI: 0.21–0.87), and PD (males) vs CT2 (males) >50 years (OR=0.37; 95% CI: 0.16–0.89). However, in these comparisons, a significant likelihood is observed only in two comparisons – PD vs CT2 >50 years (P-value=0.026), and PD (females) vs CT2 (females) >50 years (P-value=0.011). Again, this could be due to the small sample size of CT2.

The potential effect of haplogroup K on PD was further evaluated by using the Fisher's exact test (Table 4). All comparisons involving haplogroup K, with the exception of those in females, provided significant probability values thus supporting a role of haplogroup K in reducing the risk of PD.

Table 4 Probability values of Fisher's exact test

As for the 4216C mutation, the absence in all comparisons of significant differences for haplogroup J does not support an involvement of this SNP in modulating PD, while some significant differences are observed for haplogroup T (Table 3), the other haplogroup harboring this mutation (Table 1). These significant differences always involve the CT2 cohort: PD vs CT2 (OR=0.62; 95% CI: 0.41–0.93), PD vs CT2 >50 years (OR=0.58; 95% CI: 0.35–0.97) and PD (females) vs CT2 (females) >50 years (OR=0.39; 95% CI: 0.19–0.81). However, we performed an additional binary logistic regression in which the two control groups were compared (CT2 vs CT1) in terms of haplogroup distributions. A significant difference was observed only for haplogroup T (OR=1.46; 95% CI: 1.04–2.04) indicating an excess of mtDNAs belonging to haplogroup T in CT2, thus suggesting that the differences observed between PD and CT2 are due to the smaller sample size of this control sample.

Discussion

The contribution of the 4216C and 10398G SNPs in modulating the risk of PD is controversial. Both higher8 and lower7 frequencies of the cluster JT in PD vs control cohorts have been reported in different European populations. In our study, the 4216C mutation was equally distributed in PD and non-PD subjects. We conclude that neither haplogroup J alone nor the supercluster JT, or the polymorphism 10398G, significantly influence the risk of PD in the general Italian population. A significant effect in modulating the risk of PD was obtained only for haplogroup K. This haplogroup was less frequent in PD patients than in controls, and even less so in >50 year old males. Our findings differ from those obtained by van der Walt and colleagues,7 who observed a reduced risk for PD in association with 10398G, a polymorphism which among Europeans characterizes only haplogroups J, K and I. However, these haplogroups do not belong to the same clade. Haplogroup K is a clade within haplogroup U, J is a sister group of haplogroup T and haplogroup I is a subset of the cluster N1.9 Therefore, the 10398G in J, K and I has not been transmitted by descent, rather it is the result of three mutational events that have occurred independently on different mtDNA backgrounds. In addition, the frequency of 10398G-positive mtDNAs, which is only 20–25% in Europeans, is much higher in other human populations. 10398G characterizes super-haplogroup M – the most common haplogroup in Asia – and virtually all sub-Saharan African mtDNAs (haplogroups L0–L3). Since there is no evidence of major ethnic biases in the prevalence of PD, the very high frequency of the 10398G in non-European populations makes it unlikely that this polymorphism can confer per se a significant protection from PD. Concordant with our own findings, the reduced risk for PD in the supercluster HVKU compared to the supercluster JTIWX found in Finns was exclusively contributed to by the KU cluster.8 The accumulation of nonsynonymous changes in some mtDNA haplogroups can modulate either the efficiency in coupling respiration rate to ATP synthesis or the production of reactive oxygen species during respiration, or both. These effects could play a role in determining neuronal damage or, instead, neuronal protection. Interestingly, the KU cluster carries the lowest average number of amino-acid replacements in the ND genes of complex I, which is a key enzyme for energy production and a site of ROS production. It is possible that the relative paucity of nonsynonymous mutations in the ND subunits characterizing haplogroup K may contribute to the decreased risk for PD. Incidentally, haplogroup K is extremely common in the Ashkenazi Jews (more than 30%).10 Therefore, if haplogroup K reduces the risk of PD, a verifiable consequence would be the observation of a lower frequency of PD in Ashkenazi Jews, relative to other European populations – a possibility that has not yet been investigated.

The modifying role of an mtDNA haplogroup on disease risk is most probably due to the synergistic action of a set of different polymorphisms rather than to the effect of a single polymorphism. Further analysis of haplogroup K within the KU cluster will allow us to recognize founder events and specific mtDNA subclades relevant for PD.