Introduction

Schizophrenia is a disorder marked by hallucinations and abnormalities of thinking, mood and behaviour often accompanied by marked social withdrawal, that affects about 1% of the population [1]. A genetic predisposition to the disease is evident from the high concordance among monozygotic as compared to dizygotic twins and the familial clustering of the disease [2]. Schizophrenia has been reported to be linked to chromosomes 6 [3–5], 3 and 8 [6], 15 [7] and 22 [8, 9]. Evidence for linkage is not conclusive at any of these loci and has not been detected in some studies [10, 11], emphasizing the probable genetic heterogeneity of the disease [12].

In recent years, a large number of human diseases have been attributed to defects in the mtDNA. Most of these diseases belong to the group of neurological diseases called mitochondrial myopathies and encephalomyopa-thies [13]. However, there is increasing evidence that other more common disorders might involve mtDNA mutations, such as certain types of diabetes [14, 15], deafness [16–18] and late-onset Alzheimer’s disease [19, 20]. Mitochondrial mutations have also been proposed in the aetiology of maternally transmitted bipolar affective disorder [21] and Parkinson’s disease [22].

Several lines of evidence indicate that defects in mitochondrial energy production could be involved in the pathogenesis of schizophrenia. Metabolic changes, such as a decrease in creatine kinase levels, have been found in the brains of schizophrenic patients, suggesting that local concentrations of ATP might be altered [23]. Schizophrenia has also been associated with neuromuscular abnormalities that cannot be attributed to medication or drug abuse [24] and retinitis pigmentosa and sensorineural deafness, both possible symptoms of mitochondrial disease [25]. We have previously reported a 50% reduction in mitochondrial COX activity in the nucleus caudatus and cortex gyrus frontalis of schizophrenia patients as compared to controls [26]. A decrease in COX activity might reflect a defect in any of the COX subunits. On the other hand, any gene that affects mitochondrial target-ting, transport or metabolism might affect COX activity. Therefore, a decreased COX activity might be coupled with malfunction of any other mitochondrial or nuclear genes involved in mitochondrial function.

In this report, we have searched for mtDNA substitutions that might have a deleterious effect on oxidative phosphorylation in schizophrenic patients. To this end, the entire mtDNA genome was sequenced from (1) two unrelated Swedish schizophrenic patients with extremely low COX activities [26], and (2) two Scottish patients belonging to a family that shows a possible maternal inheritance and multiple affected offspring, a pattern that might be consistent with mtDNA mutations.

Experimental Procedures

Patient Materials

Brain samples were obtained at autopsy from a total of 12 schizophrenic individuals with a significant reduction in COX activity compared to healthy controls. These patients have been described previously [26]. Brain tissue specimens of the nucleus caudatus were dissected, freeze-dried, crushed into a coarse powder and stored at −70°C.

Diagnoses were according to DSM-III-R (American Psychiatric Association, 1987). All schizophrenic patients had suffered from a chronic form of the disease and had been treated to varying degrees with neuroleptic drugs. The mean post-mortem delay before autopsy was 42.1 ± 18.5 h for schizophrenic patients and 90.4 ± 36.3 h for the controls.

Blood samples were obtained from individuals 1:2 and II: 1 in the pedigree in figure 1. These patients were diagnosed according to the criteria above. There was no history of schizophrenia in previous generations in this family. The onset of the disease was at age 35 for the mother and during their teens for generation 2. The two patients 1:2 and 11:3 are considered as one patient in table 3 due to the close relationship. An additional set of 68 blood samples from Swedish patients were extracted to be used for the screening of variants.

Fig. 1.
figure 1

Pedigree from a family with suspected maternal inheritance of schizophrenia. The mtDNA of individuals 1:2 and 11:1 were sequenced.

Full size image
Table 1. Primers used for PCR amplification of mtDNA
Full size table
Table 2. mtDNA mutations in four schizophrenia patients
Full size table
Table 3. Evolutionary conservation and frequencies of novel mtDNA changes found in the schizophrenia patients
Full size table

DNA Isolation

Genomic DNA was prepared from brain tissue and blood by proteinase K digestion followed by phenol extraction and ethanol precipitation. DNA amounts were estimated by DNA fluorescence (Hoechst 33258).

mtDNA Sequencing Strategy

The complete mitochondrial genome was amplified by PCR in 13 overlapping fragments (table 1) as shown in figure 2. Amplified mtDNA fragments were purified using a QIAEX Gel extraction Kit (QIAGEN) and directly sequenced using the Taq Dye Deoxy Terminator Cycle Sequencing Kit (Perkin-Elmer) employing fluorescent nucleotide terminators. A total of 120 sequence reactions was performed for each patient to determine the complete mtDNA sequence from both strands. The sequence of both strands was necessary for unambiguous sequence determination. The sequences were assembled in a contig using the program STADEN and the resulting contig was aligned to the Cambridge sequence [27].

Fig. 2.
figure 2

Linear map of the mtDNA indicating the thirteen fragments amplified for sequencing. The sizes of the fragments and the oligonucleotides used for amplification are indicated in table 1.

Full size image

Screening for Substitutions by Restriction Analysis and Oligo-Hybridization

The substitution at nucleotide position 2780 (C-T) was detected by digestion of PCR products with AvaII and the substitution at position 15758 (A–G) by digestion of PCR products with DdeI. The T–C change at position 3197, the A–G at position 14793, and the A–G at position 15218 were detected by hybridization with specific oligonucleotides as described [28]. The biotinylated oligonucleotides used for hybridization were:

(1) Position 3197 T–C: wild type 5′-GGTATAAT(A)CTAAGTT-G-3′ and mutant 5′-CAACTTAG(C)ATTATACC-3′. (2) Position 14793 A–G: wild-type 5′-TAACC(A)CTCATTCATCG-3′ and mutant 5′-CGATGAATGAG(C)GGTTA-3′. (3) Position 15218 A–G: wild-type 5′-ACATTGGG(A)CAGACCTA-3′ and mutant 5′-TAGGTCTG(C)CCCAATGT-3′. Hybridization was performed with 10 pmol probe in 25 ml 2 × SSPE (0.34 M NaCl, 20 mM NaH2PO4, 2 mM EDTA pH 7.7) at 42 °C for 20 min. Washes were performed with 0.1 × SSPE/0.1 % SDS at 44–52 ° C, according to the Tm of the oligonucleotides. Hybridizing probe was detected using chemilumi-nescence (ECL, Amersham).

Results

The mtDNA sequence was determined from brain samples of two unrelated Swedish schizophrenic patients (A and B in table 2) and blood samples of two maternally related Scottish patients (C and D in table 2, I:2 and II:1 in fig. 1). The mtDNA sequences of the patients differed from the Cambridge sequence [27] by a number of substitutions, all homoplasmic (table 2). Most of these have previously been reported as polymorphisms or errors in the Cambridge sequence [29]. Among the changes not previously reported, three of the substitutions do not alter the amino acid sequence and four are located in the displacement loop (D-loop), a region known to show the highest variability between individuals [30, 31]. For example, a six-base insertion was found at nucleotide position 524 in one of the patients, representing the largest insertion reported in the D-loop in humans. In the Swedish patients, four substitutions were found in coding regions that have not been previously described as polymorphisms; two substitutions in the 16S RNA gene (C–T at position 2780 and T–C at position 3197) and two mis-sense mutations in the cytochrome b gene (A–G at position 14793, and A–G at position 15218). One of the two Swedish patients carried all four substitutions, while the other patient had the changes at positions 3197 and 14793. In the Scottish patients, we found a missense mutation in the cytochrome b gene (A–G at nucleotide position 15758). In addition, the Scottish individuals also had one substitution in the last base of the anticodon loop of tRNAthr (A–G substitution at position 15924) (table 2).

In total, among the four patients, we found five changes in coding regions that have not previously been described as polymorphisms. These substitutions occurred at positions showing varying degrees of evolutionary conservation (table 3). The frequencies of these substitutions were estimated in a set of schizophrenic patients from southern Sweden, northern Sweden and Edinburgh, UK, as well as in five groups of controls, derived from three parts of Sweden and Edinburgh, UK, in order to evaluate their association with schizophrenia (table 3). The Fisher’s exact test was calculated for all possible comparisons between frequencies in the patients and the control groups. Among the five substitutions, the variants at positions 14793 and 15218 showed a significantly higher frequency in the combined patients as compared with the combined controls from Sweden (n = 259) (Fisher’s exact test p = 0.016 and p = 0.035, respectively). However, the patients from northern Sweden did not show any significant difference in frequency as compared to the controls from the same area (not shown). Thus, the differences between the combined patients and controls might reflect population stratification, rather than disease association. On the other hand, the substitutions at positions 3197, 14793 and 15218 showed significantly higher frequencies in the patients as compared to the controls from southern Sweden. These results emphasize the need to exert great care in choosing an appropriate control population when evaluating mtDNA mutations. For example, the substitution at 2780 was only found in southern Sweden, while 15218 was more prevalent in northern Sweden.

Discussion

We have examined the mtDNA sequences from two unrelated patients and a pair of maternally related patients with schizophrenia. This study followed earlier findings that defects in mitochondrial energy production could be associated with schizophrenia [23–26]. Our hypothesis was that mtDNA sequence variants might underlie the disease susceptibility in these patients.

Our results revealed the presence in the schizophrenic patients of five mtDNA sequence variants not previously reported as polymorphisms. Three of these are located in the cytochrome b gene. Many polymorphic positions were found in the cytochrome b gene [29] but they usually do not involve missense mutations in conserved amino acid residues [16, 22, 32]. On the other hand, three missense mutations in the cytochrome b gene have been associated with different diseases: substitution 15257 (G–A), 15812 (G–A) [29] and 15615 (G–A) [33]. In our study, the three cytochrome b substitutions were found to affect moderately conserved positions (at nucleotides 14793 and 15218) and a very conserved position (at nucleotide 15758) [34]. The 14793 change alters an amino acid in the N-terminal domain of cytochrome b, the 15218 change modifies a residue in the second intracellular loop and the 15758 change is located in the seventh transmembrane domain. The 15218 substitution has previously been described in patients with idiopathic cardiomyopathy [35] and Leber hereditary optic neuropathy [36]. A mutation was also found in the last base of the anticodon loop of tRNA thr (position 15924). The 15924 mutation was first described in patients with fatal infantile respiratory enzyme deficiency [37], but has recently been detected in about 11 % of control subjects, suggesting that it is a polymorphism [9]. The 3197 substitution, in the 16S RNA, has previously been described in a patient with ischaemic colitis. It has been suggested that this substitution may modify the expression of another mitochondrial mutation [38]. The frequencies of the five substitutions found were determined in 81 patients and five control groups. The combined patients and controls showed significant differences at two positions. However, when the groups of patients and controls from the same area were compared, significant differences were only found between patients and controls from southern Sweden. The frequency differences observed among regions within the same country stress the importance of using a large and well defined control material when assessing the association of mitochondrial sequence variants with disease. Our results indicate that a sample of 100–200 controls, a frequently used sample size in many other studies [22, 39, 40], might not be enough to evaluate the importance of mitochondrial substitutions.

When compared to the distribution of continent-specific mtDNA variants [41], the mtDNAs of the schizophrenia patients were found to resemble lineages of European origin most closely, in agreement with the Swedish and Scottish origin of the patients (data not shown). On the other hand, the Swedish and Scottish mtDNAs are not found on any of the major branches of the European mtDNA clade. Therefore, the possibility remains that the mtDNA of patients with schizophrenia might belong to the same ancestral lineage, as recently suggested for some late onset Alzheimer patients [19].

In summary, our analysis of 81 schizophrenic patients and 400 controls does not lend strong support for the association of a particular mtDNA variant with increased risk for schizophrenia. However, the trend towards a higher frequency of substitutions in the patients, as well as the presence of certain substitutions in higher frequency in the group of patients from southern Sweden, deserves further attention.