Association of bipolar disorder with the 5178 polymorphism in mitochondrial DNA

INTRODUCTION

It is well established that bipolar disorder is caused mainly by genetic factors. Recent molecular genetic studies of bipolar disorder have made a great progress. A number of candidate locus such as 4p16, 11p15, 18p11, 18q22, 21q21, and Xq26 have confirmed by independent linkage studies [Berrettini and Pekkarinen, 1996]. However, mode of transmission of bipolar disorder is still a matter of debate. Although it was suggested that single major gene and multi-factorial inheritance contribute to transmission of this disorder [Rice et al., 1987], transmission of bipolar disorder has two characteristics that do not follow Mendel's laws: anticipation [Mclnnis et al., 1993] and parent-of-origin effect [McMahon et al., 1995]. This evidence gives a clue to search for pathogenetic mutations causing bipolar disorder. Molecular basis of anticipation has already been examined, and a triplet repeat expansion at 18q21.1 was found to be associated with bipolar disorder [Lindblad et al., 1998].

With regard to parent-of-origin effect, five lines of evidence have been reported: 1) affected mothers were more frequent than affected fathers [McMahon et al., 1995; Gershon et al., 1996; Kato et al., 1996]; 2) more maternal relatives had the illness than did paternal relatives [McMahon et al., 1995]; 3) bipolar patients with an affected father had lower ages at onset than did those with an affected mother [Grigoroiu-Serbanescu et al., 1995; Kato et al., 1996]; 4) linkage with chromosome 18 was limited to paternally transmitted pedigrees [Gershon et al., 1996; Stine et al., 1995; Nothen et al., 1999]; and 5) segregation analysis suggested that the disease is transmitted by a single major gene in paternally transmitted pedigrees while it may be transmitted by multi-factorial inheritance in maternally transmitted pedigrees [Grigoroiu-Serbanescu et al., 1998].

Mitochondrial DNA (mtDNA) is genetic material, which is localized not in nuclei but in mitochondria, a cytoplasmic organelle. Although it is compact in size, 16,569 bp, its quantity in human body is assumed to reach approximately 10% of nuclear genome. Recent studies clarified that mtDNA mutations may cause wide variety of diseases, such as diabetes mellitus, deafness, cardiomyopathy, Alzheimer disease, and Parkinson disease [Beal et al., 1997]. It should be noted that it is transmitted exclusively from a mother, and all offsprings are at risk. These characteristics of mitochondrial transmission are similar to those observed in transmission of bipolar disorder noted above. In the light of parent-of-origin effect, mitochondrial DNA (mtDNA) is one of candidate locus of bipolar disorder.

We have been studying brain energy metabolism using phosphorus-31 magnetic resonance spectroscopy (³¹P-MRS) and found that brain energy metabolism in bipolar disorder was abnormal. Intracellular pH was significantly decreased in the frontal lobes in euthymic patients with bipolar disorder [Kato et al., 1994a]. This is not an effect of psychotropic drugs because this finding was confirmed in drug-free euthymic patients with bipolar disorder [Kato et al., 1998]. We also found that phosphocreatine (PCr) was decreased in the frontal lobes in depressive patients with bipolar disorder [Kato et al., 1994b, 1995]. Using photic-stimulation paradigm, we found that response of PCr in the occipital cortex to photic stimulation was abnormal in the euthymic patients with bipolar disorder [Murashita et al., 2000]. These findings suggest that brain energy metabolism is trait-dependently impaired in patients with bipolar disorder. Decrease of PCr in the brain at rest and delayed recovery of PCr in skeletal muscles after exercise were reported in mitochondrial diseases. Based on these findings, we hypothesized that abnormal brain energy metabolism in bipolar disorder may be due to mitochondrial dysfunction. Case reports of patients with mitochondrial encephalopathies or mtDNA mutations who have comorbid affective disorder also support that bipolar disorder may be associated with mtDNA abnormality [Stewart and Naylor, 1990; Suomalainen et al., 1992; Onishi et al., 1997; Miyaoka et al., 1997; Kato and Takahashi, 1996]. According to this hypothesis, we examined the levels of the 4,977-bp deletion in mtDNA in the postmortem brains [Kato et al., 1997] and found that it was significantly increased in bipolar disorder. This finding may also support our hypothesis that mtDNA mutation is one of the risk factors of bipolar disorder.

Recently, Tanaka et al. [1998] sequenced total mitochondrial genome in 11 healthy subjects age more than 100 years old in an attempt to find mtDNA polymorphisms associated with longevity. They reported that a C-to-A base substitution at the position of 5178 within NADH dehydrogenase subunit 2 gene in mtDNA was significantly associated with longevity. They also reported that the 5178C genotype increases a risk of Alzheimer disease and Parkinson disease [Tanaka, 1998]. Patients with these disorders having the 5178C genotype were found to have onset of illness approximately 10 years earlier than those with 5178A genotype. We hypothesized that this polymorphism may also be a risk factor of bipolar disorder.

The purpose of this study is to clarify 1) whether the 5178 polymorphism in mtDNA is associated with bipolar disorder, 2) whether the 5178C genotype is associated with earlier age at onset of bipolar disorder, and 3) whether this polymorphism is associated with previously reported neuroimaging findings by ³¹P-MRS and magnetic resonance imaging (MRI) in bipolar disorder.

MATERIALS AND METHODS

Subjects were 145 unrelated patients diagnosed as having bipolar disorders according to the DSM-IV criteria (90 women and 55 men, 103 with bipolar I and 42 with bipolar II, 49.2 ± 13.6 years old (mean ± SD)) who were hospitalized in the psychiatric wards of Tokyo University Hospital, Shiga University of Medical Science Hospital, and University of Teikyo Hospital. Consensus diagnosis was made for each patient according to the DSM-IV criteria by at least two trained psychiatrists. Their ages at onset defined as the ages at the first major affective episode was 34.5 ± 12.9 years old. Family history was obtained from the patient and available relatives. Of the 145 patients, 70 had family history of mood disorders or psychotic disorders within second-degree relatives. The mode of transmission in these families was classified into four groups: 1) paternal only (n = 17); 2) both sides (n = 3); 3) maternal only (n = 28); and 4) only siblings are affected (n = 22).

One hundred, eighty-four control subjects (100 women and 84 men, 29.4 ± 7.6 years old) were selected from hospital staffs and students. They were not interviewed. None of controls were removed because of their psychiatric history. All subjects were Japanese. Written informed consent was obtained from all subjects. This study has been approved by the ethical committees of Shiga University of Medical Science, Teikyo University, and the University of Tokyo.

Total leukocyte DNA was extracted by the salting-out extraction method. PCR fragments of 300 bp were amplified by standard protocols using primer pairs L5042 (corresponding to base number 5042–5061) [Anderson et al., 1981] and H5342 (corresponding to base number 5342–5323) purchased from Life Technologies Oriental Inc. (Tokyo, Japan). PCR was performed using Ex-Taq (0.2 unit/12.5 μl of reaction volume) and Takara PCR Thermal Cycler MP (Takara Co. Ltd., Otsu, Japan) in the reaction solution recommended by the manufacturer. The parameters of the PCR reaction were as follows; 94°C for 20 sec and 60°C for 20 sec, 30 cycles. Before the first cycle, heat-denaturation was performed at 94°C for 3 min and final extension was done at 72°C for 3 min. The PCR product was digested by 2 units of AluI for 2 hr at 37°C. This enzyme digests the 300-bp PCR product into two fragments of 136-bp and 164-bp when the base at 5178 position is C. These samples were electrophoresed in 2% SeaKem Agarose (Takara Co. Ltd., Otsu, Japan) gels in 0.5× TBE at 100 V for 1.5 hr, stained with ethidium bromide, and visualized by ultraviolet trans-illuminator.

The method of ³¹P-MRS data acquisition and its reliability were described in our previous report [Kato et al., 1994a, 1994b]. In brief, subjects were examined on a 1.5T SIGNA MR system and a surface coil for ³¹P. The volume of interest was the 30-mm slice in the frontal region. ³¹P-MR spectra were obtained using depth-resolved surface coil spectroscopy (DRESS) [Bottomley et al., 1984]. The repetition time (TR) was set at 3 sec, and 128 scans were averaged. Metabolite data were shown as peak area percent to the total phosphorus signal. Intracellular pH was calculated from the difference of chemical shifts between inorganic phosphate and phosphocreatine. Inter-assay intra-individual coefficients of variation (CV) of peak area percentages were less than 10% except for inorganic phosphate peak, and CV of pHi was 0.80% [Kato et al., 1994b].

The method of MRI was described in our previous report [Kato et al., 1998]. In brief, MRI were also obtained using the same MR system. T₂-weighted and proton density-weighted double-spin echo axial images were taken with, slice thickness of 5 mm, inter-scan gap of 2.5 mm, repetition time (TR) of 2,000 msec, and echo times (TE) of 80 msec and 20 msec, respectively. These MRI films were scored by two neuropsychiatrists. Subcortical hyperintensity lesion was assessed using the 4-point scale [Coffey et al., 1989]. Three scores, deep white matter hyperintensity, periventricular hyperintensity, and hyperintensity lesions in the basal ganglia were summed to yield a total subcortical hyperintensity scale.

For statistical analysis, χ² test for independence, logistic regression analysis, three-way analysis of variance (ANOVA), one-way ANOVA with multiple comparison, Student's t-test, and Mann-Whitney U-test were applied using SPSS software (SPSS Co. Ltd., Tokyo, Japan).

RESULTS

DISCUSSION

In this study, the frequency of the 5178C genotype was slightly and significantly higher in patients with bipolar disorder compared with controls when patients with family history in paternal side were excluded. Although this is a very weak risk factor for bipolar disorder with relative risk of 1.3, it may be a somewhat stronger risk factor for bipolar II disorder (relative risk of 2.3). This seems to be compatible with our previous results that alteration of brain energy metabolism detected by ³¹P-MRS was more prominent in bipolar II disorder compared with bipolar I disorder [Kato et al., 1994b]. Other findings that the 5178 polymorphism is related to age at onset of bipolar disorder and brain intracellular pH also support that this may have certain pathophysiological significance in bipolar disorder.

There is an ongoing debate regarding the execution of association studies. The Japanese population may be viewed as a homogenous sample of individuals to be examined by association study. However, it should be noted that control groups collected in a hospital environment may not be representative of the population. Although the controls were not interviewed and their psychiatric histories were not known, the use of non-screened population controls is thought to be valid [Malhotra and Goldman, 1999].

How this mitochondrial DNA polymorphism may regulate the vulnerability to bipolar disorder in an individual is not known. This C-to-A polymorphism at the position of 5178 in mtDNA changes leucine in NADH dehydrogenase subunit 2 (ND2) into methionine [Anderson et al., 1981]. Because functional alteration caused by this substitution has never been examined yet, we can only speculate the mechanism by which this polymorphism regulates vulnerability to bipolar disorder. There are several possible mechanisms that this polymorphism might play a certain part in pathophysiology of bipolar disorder.

First, this polymorphism may alter brain energy metabolism through alteration of activity of NADH dehydrogenase. We previously reported that intracellular pH in the frontal lobes was significantly lower (7.01 ± 0.04) in patients with bipolar disorder using ³¹P-MRS compared with controls (7.05 ± 0.04) [Kato et al., 1994a]. The 5178A polymorphism may decrease a risk of bipolar disorder by increasing intracellular pH. Second, this polymorphism may affect free radical generation when exposed to a certain environmental factor. This may be supported by our previous report that partial deletion of mtDNA was increased in the postmortem brains in bipolar disorder [Kato et al., 1997]. Third, this polymorphism may be linked to some other mtDNA mutation that causes bipolar disorder. Fourth, this polymorphism may be a risk factor for silent cerebral infarction, which increases vulnerability to bipolar disorder [Steffens and Krishnan, 1998]. However, the present result that 5178 polymorphism was not related to subcortical hyperintensity does not support this hypothesis.

Another possible explanation, although very speculative, is an effect of this mtDNA polymorphism on intracellular calcium signaling system, because recent studies have clarified that mitochondria play an important role in buffering intracellular calcium increased by agonist-stimulation [Simpson and Russell, 1997].

Anyway, these hypotheses noted above are still immature and need to be validated by further studies before making a definite conclusion.

The frequency of the 5178A genotype in controls in this study (46.7%) is comparable with that in Japanese blood donors described in a previous report [Tanaka et al., 1998] (45%), which validates the genotyping method used in this study. It was reported that this polymorphism was found in only five Asians and one European among 147 samples from various areas of the world [Tanaka et al., 1998]. Therefore, the 5178A polymorphism may be relatively rare among the global population, although the frequency of this polymorphism has never been systematically examined in any other ethnicity. This may explain not only high life expectancy of Japanese people (83.5 years old for females and 77.0 years old for males) but also higher age at onset of bipolar disorder in Japan (34.5 ± 12.9 years old in this study) compared with that in Caucasians (28.1 years old) [Goodwin and Jamison, 1999].

It should be noted that we should be cautious to generalize these results. Observed significant relationship between the 5178 polymorphism and bipolar disorder in this study is marginal and it cannot be ruled out that this is only a sampling bias. This finding needs to be replicated in other ethnicities before conclusion. In this study, many statistical tests were performed, which might have affected the levels of significance. Therefore, apparently significant results might be due to multiple testing. The age of the cases and controls were not matched. Although logistic regression analysis did not reveal a significant effect of age, it cannot be ruled out that difference of age of these groups might have affected the results.

The observed difference in intracellular pH is also very small. However, intracellular pH is the most reliable parameter in ³¹P-MRS and a slight difference of intracellular pH (0.04) is not negligible because such a small difference of intracellular pH should cause alteration of enzyme activities of many kinds.

Despite these limitations, this is the first report that suggests association of affective disorder with a mitochondrial DNA polymorphism and possible relationship between a neuroimaging finding and a genetic risk factor in bipolar disorder. If these findings were confirmed in other ethnicities, it may help us to pursuit a new treatment strategy of bipolar disorder.