Effect of MDR1 Haplotype on Risk of Parkinson Disease

Background  MDR1, a multidrug transporter, encodes a P-glycoprotein that regulates the bioavailability of xenobiotics and is highly expressed at the blood-brain-barrier. Two single nucleotide polymorphisms (SNPs) (e21/2677[G/T/A] and e26/3435[C/T]) in the MDR1 gene can lead to differences in MDR1 expression and function. Specific MDR1 alleles of the 2 SNPs are positively selected among ethnic Chinese but not in the white population.

Objective  To determine whether specific haplotypes formed by SNPs e21/2677 and e26/3435 may protect against Parkinson disease (PD) among ethnic Chinese in Hong Kong.

Design  Case-control study.

Setting  Tertiary referral centers in Hong Kong.

Subjects  One hundred eighty-five patients with PD and 206 control subjects.

Interventions  The two SNPs were amplified in a single multiplex polymerase chain reaction. Five other SNPs that span 100 kilobases of the gene were also analyzed.

Main Outcome Measures  Haplotypes frequencies, degree of haplotype association with the disease status, and estimated odds ratio for each haplotype with associated 95% confidence intervals.

Results  In addition to 2677 G→T/A (exon 21) and 3435 C→T (exon 26), the other SNPs that were analyzed were –41 A→G (intron –1), –145 C→G (exon 1), –129 T→C (exon 1), 1236 T→C (exon 12), and 4036 A→G (exon 28). Haplotypes containing SNPs e21/2677 and e26/3435 were found to be significantly associated with risk of PD. In particular, the 2677T-3435T haplotype was strongly associated with a reduced risk of PD (P<.001; χ² = 14.521; odds ratio, 0.33; 95% confidence interval, 0.19-0.59).

Conclusions  An MDR1 haplotype containing SNPs e21/2677T and e26/3435T protects against PD in ethnic Chinese, compatible with the observation of a recent positive selection of the T alleles of these 2 SNPs in this ethnic population.

Parkinson disease (PD) is a progressive neurodegenerative disease characterized by loss of dopaminergic cells in the substantia nigra pars compacta and presence of Lewy bodies. The cardinal clinical symptoms and signs of PD are bradykinesia, rigidity, tremor, and postural instability.¹ Pathogenic mutations in several genes, including α-synuclein, parkin, UCHL1 (ubiquitin–C-terminal hydrolase-L1), and DJ1 have been described in some familial forms and/or early-onset PD.² Studies of genetic polymorphisms of candidate genes have thus far been inconclusive.^3-6

The relative role of genetic and environmental factors in the pathogenesis of PD has not been clarified.^7,8 The discovery of parkinsonism induced by methyl-4-phenyl-1,2,3,6-tetrahydropyridine in intravenous drug users suggests other potential putative environmental causative agents.⁷ Exposure to pesticides and rotenone recapitulates some parkinsonian features in animal models.²

Genetic susceptibility to sporadic PD may be modulated by genes involved in xenobiotic management. A recent meta-analysis³ of genetic association studies demonstrates that polymorphisms in 4 genes are significantly associated with PD. These genes either are responsible for xenobiotic metabolism, such as NAT2 and GSTT1, or may potentially interact with environmental agents, such as monoamine oxidase. Poor metabolizer alleles of the cytochrome P450 xenobiotic metabolism enzyme, CYP2D6, can also be associated with increased risk of PD.⁹

MDR1 is a multidrug transporter and a member of the adenosine triphosphate binding cassette superfamily of transporter genes that regulate the bioavailability of xenobiotics. MDR1 is highly expressed at the interface of major organs.^10-13 Two single nucleotide polymorphisms (SNPs; e21/2677[G/T/A] and e26/3435[C/T]) in the MDR1 gene can result in differences in MDR1 expression and function.

A recent study¹⁴ in an ethnic Chinese population in Singapore found haplotypes comprising SNPs e21/2677 (G/T/A) and e26/3435(C/T) significantly associated with risk of PD. Two smaller studies^15,16 on limited MDR1 SNPs in an Italian and Polish PD population did not demonstrate significant association with PD. Haplotype mapping is increasingly recognized to be a more effective method in genetic association studies, where study of haplotype blocks (regions of low recombination) of a candidate gene of interest may shed light on ancestral conservation and ethnic differences, factors that can confound population genetic studies. The identification of haplotype-tagging SNPs reduces redundant association studies and improves power for a given sample.¹⁷ To test the hypothesis that specific haplotypes formed by SNPs e21/2677T and e26/3435T may protect against PD among ethnic Chinese, we conducted an independent study of MDR1 haplotypes in an ethnic Chinese Hong Kong population.

All study subjects were ethnic Chinese and recruited from 2 major hospitals in Hong Kong (Prince of Wales and United Christian). The diagnosis of PD was made by neurologists or geriatricians according to the United Kingdom PD Brain Bank Criteria.¹⁸ Exclusion criteria were extensor plantar reflexes, ophthalmoplegia, early dementia, or early autonomic failure. Control subjects were recruited from the same sources and examined by us with no evidence of neurodegenerative diseases. In the selection of controls, 10-year age group, sex, and locality (hospital catchment area) were taken into account to make the control group as comparable with the PD group as possible.

Blood samples were obtained from patients with PD and controls. All study patients gave informed consent. The study was approved by the ethics committee of the Chinese University of Hong Kong.

In addition to 2677 G→T/A (exon 21) and 3435 C→T (exon 26), the other SNPs that were analyzed were as follows: –41 A→G (intron –1), –145 C→G (exon 1), –129 T→C (exon 1), 1236 T→C (exon 12), and 4036 A→G (exon 28). The 7 SNPs span 100 kilobases of the gene. The 5 genomic segments containing the 7 SNPs were amplified in a single multiplex polymerase chain reaction. We amplified 20 ng of genomic DNA in a T3 thermal cycler (Biometra GmbH iL, Goettingen, Germany) in a total volume of 10 μL containing 0.15 pmol of each of the 10 primers per microliter (see Gwee et al¹⁹), 5mM magnesium chloride, 200μM of each of the 4 deoxynucleotide triphosphates, and 0.75 U of Taq DNA polymerase (HotStarTaq; Qiagen GmbH, Hilden, Germany) in the polymerase chain reaction buffer that was supplied (Qiagen GmbH). The reaction mixture was subjected to initial denaturation at 94°C for 15 minutes followed by 40-step cycles of denaturation at 94°C for 30 seconds, annealing at 56°C for 30 seconds, and extension at 72°C for 1 minute. This was followed by a final extension at 72°C for 5 minutes. The expected polymerase chain reaction fragments and their sizes were described in detail previously.¹⁹ The polymerase chain reaction products were subjected to a multiplex minisequencing reaction to examine the 7 SNP loci simultaneously. The SNP-specific probing primers (or minisequencing primers) were designed to anneal to template DNA next to each SNP site such that extension by DNA polymerase added a single dideoxyribonucleoside triphosphate complementary to the nucleotide at the polymorphic site.¹⁹

Differences in sex distribution, allele and genotype frequencies between groups, and conformation to Hardy-Weinberg equilibrium were compared by means of Pearson χ² and Fisher exact test. Odds ratios and their 95% confidence intervals (CIs) were calculated to evaluate the effects of different genotypes or alleles. Multivariate logistic regression models were used to evaluate the effects of covariates on phenotypes, and a stepwise analysis of deviance was used to generate the final model. Where appropriate, assumptions of Hardy-Weinberg equilibrium were made, and models were assessed by the use of likelihood ratio tests to determine the effects of genetic covariates.

Haplotypes frequencies were estimated by means of the expectation-maximization algorithm. Lagrange multiplier tests (score tests) were used to assess the degree of haplotype association with the disease status, and estimated odds ratio for each haplotype was calculated with associated 95% CI obtained via the profile likelihood method. Our sample size had sufficient power (90%) to detect an increased odds ratio of 2.5 at minor haplotypic frequencies of 0.05.

The degree of linkage disequilibrium was determined by using 2 standard measures, Lewontin D′ and the correlation coefficient r. Statistical significance of linkage disequilibrium was obtained through Monte Carlo permutation tests. We made a modest correction for multiple testing and defined statistical significance at P<.01. All analysis was performed on the statistical software S-Plus (Insightful Corporation, New York, NY).

There were 185 patients with PD and 206 controls matched for age, sex, and ethnic race (Table 1). Univariate analysis of the 7 MDR1 loci did not show any significant differences between patients with PD and controls (Table 2). Only haplotypes formed by e21/2677T and e26/3435T were significantly associated with risk of PD (Table 3). Further analysis demonstrated that specific haplotypes containing e21/2677T and e26/3435T modulated the risk of PD (Table 4). The incidence of haplotype e21/2677T–e26/3435T was significantly lower in patients with PD than controls (χ² = 14.52; P<.001; odds ratio, 0.33; 95% CI, 0.19-0.59), and haplotype e21/2677/T–e26/3435C increased the risk of PD (χ² = 10.51; P = .001; odds ratio, 1.73; 95% CI, 1.25-2.38) (Table 4). Age and sex did not significantly affect these observations, although a marginal significant difference for haplotype e21/2677T–e26/3435T was found between those with age at onset 60 years or older compared with those who were younger (χ² = 3.48; P = .06; odds ratio, 0.48; 95% CI, 0.24-0.98). All 7 SNPs in our study population were consistent with the Hardy-Weinberg equilibrium (each P>.05).

There is strong evidence of recent positive selection for the MDR1 e21/2677T and e26/3435T alleles among ethnic Chinese compared with whites.²⁰ Furthermore, our group recently demonstrated that haplotypes containing these 2 MDR1 SNPs, e21/2677(G/T/A) and e26/3435(C/T), were significantly associated with a reduced risk of PD in an ethnic Chinese population in Singapore.¹⁴ Because of a number of potential confounding variables, it is important to be able to replicate findings of population genetic studies. Herein we extended the MDR1 haplotype analysis to a similar ethnic race (Chinese) but from a different geographic location (Hong Kong), to test the hypothesis that the T alleles of these SNPs may confer better protection of the brain against xenobiotic insults in the Chinese population.

Univariate analysis of the 7 MDR1 loci did not demonstrate any significant association with PD (Table 2). However, haplotype constructs composed of e21/2677(G/T/A) and e26/3435(C/T) were different between patients with PD and controls. Specifically, haplotype e21/2677/T–e26/3435T had a strong inverse relationship with risk of PD, with an odds ratio of 0.33, as did haplotype e21/2677/T–e26/3435C, with an odds ratio of 1.73 (Table 4). These findings were compatible with the earlier study in Singaporean Chinese, where haplotypes containing these 2 SNPs significantly modulated the risk of PD, with haplotype e26/3435T–e21/2677/T also a strong protective factor.¹⁴ In the Singaporean cohort, univariate analysis yielded strong association with the 2 SNPs e26/3435T and e21/2677/T. In addition, the protective effect of the haplotype e26/3435T–e21/2677/T was stronger in the older age group (age at onset, ≥60 years). Interestingly, the latter trend was similarly observed in the Hong Kong cohort, where this haplotype conferred greater protection in the older group compared with the younger ones. This may be consistent with the fact that other genetic factors likely play a greater role in early-onset PD, whereas MDR1 variability is more important in the older patients.

Although the ancestors of both Hong Kong and Singaporean Chinese migrated from the southern provinces of mainland China, Hong Kong Chinese originated mainly from the southern Canton province, while there were more Singaporean Chinese from the Hokkien province. This may partly account for the minor differences observed in the 2 studies. It is also arguable that neither SNP e21/2677(G/T/A) nor e26/3435(C/T) is the culprit, but that they are merely in strong linkage disequilibrium with an unknown causal SNP. Strong association of these 2 SNPs with PD may suggest that the linked causal variant could reside within a region defined by strong linkage disequilibrium.

The findings from the 2 studies in Chinese contrasted with 2 recent case-control studies in Italy and Poland that examined MDR1 gene polymorphisms (SNPs e1/−129[T/C], e21/2677[G/T/A], e26/3435[C/T]) but found no significant association.^15,16 The recent finding of only marginal evidence of positive selection in whites compared with Chinese supports the possibility that the MDR1 haplotype and pattern of linkage disequilibrium may not be sufficient to detect a difference in whites. Another reason for the ethnic difference could be that we examined an extended MDR1 haplotype instead of a single SNP and our sample size was twice that of the studies of white patients.^15,16

Is there a biological explanation for a possible causative role of MDR1 (in particular, e26/3435[C/T] and e21/2677[G/T/A]) in PD? The blood-brain barrier is important in regulating environmental xenobiotics, which can play a causative role in PD. The MDR1 multidrug transporter represents an important component of this barrier, since it can regulate uptake of drugs and xenobiotics.^21-23 Knockout MDR1 mouse studies demonstrate that certain drugs can accumulate to toxic levels in the brain as a result of the disruption of this barrier.^21,22

We suggest that functional MDR1 SNPs can alter the exposure to xenobiotics and neurotoxic compounds and modulate the PD risk. The SNP e26/3435(C/T) may lead to differential MDR1 protein expression and plasma drug concentration^13,24,25 and drug response.²⁶ The observed correlation with e26/3435T could be due to differential codon usage of the C or the T allele at the wobble position of the isoleucine codon, RNA splicing, or translation regulation. The nonsynonymous SNP e21/2677(G/T/A) has been demonstrated to be associated with a change of digoxin efflux in vitro.¹¹

In conclusion, we have replicated the finding that a MDR1 haplotype containing functional SNPs e21/2677T and e26/3435T was associated with a significantly reduced risk of PD in an independent ethnic Chinese population in Hong Kong. The lack of association in earlier MDR1 studies in the white PD population may be compatible with the observation of a recent positive selection of the T alleles of these 2 SNPs among ethnic Chinese.

Correspondence: Eng-King Tan, MD, Department of Neurology, Singapore General Hospital, Outram Road, Singapore 169608 (gnrtek@sgh.com.sg).

Accepted for Publication: June 2, 2004.

Author Contributions:Study concept and design: E.-K. Tan, Chan, Chong, Zhao, Lee. Acquisition of data: E.-K. Tan, Chan, Ng, Woo, Teo, Tang, Wong, Shen, Zhao, Lee. Analysis and interpretation of data: E.-K. Tan, Chan, Ng, Woo, Teo, Tang, Wong, Shen, Zhao, Lee. Drafting of the manuscript: E.-K. Tan, Lee. Critical revision of the manuscript for important intellectual content: E.-K. Tan, Chan, Ng, Woo, Teo, Tang, Wong, Chong, C. Tan, Shen, Zhao, Lee. Statistical analysis: Teo. Obtained funding: Lee. Administrative, technical, and material support: E.-K. Tan, Chan, Ng, Woo, Tang, Wong, Chong, C. Tan, Shen, Zhao, Lee. Study supervision: E.-K. Tan, Chan, Chong.

Funding/Support: This study was supported by grant NMRC/0657/2002 from the National Medical Research Council (Drs E.-K. Tan, Chong, and Lee) and SingHealth Research, Singapore.