Statistics from Altmetric.com
Editor—Selectively neutral polymorphisms in mitochondrial DNA (mtDNA) have accumulated during human evolution along mtDNA lineages that correlate with ethnic and geographical origin. By contrast, deleterious mutations have arisen repeatedly and occur against various genomic backgrounds. Since they reduce the fitness of carriers, however, the affected maternal lineages eventually become extinct. Mutations of intermediate severity also occur along mtDNA lineages and have become fixed in the population.1Although they do not substantially reduce fitness, they may interact with nuclear or environmental factors, predisposing people to an increased risk of developing neurodegenerative diseases later in life.1
1555A>G in the 12S rRNA gene may be considered a mutation of intermediate severity. It has been shown to predispose carriers to maternally transmitted sensorineural hearing impairment, the expression of which requires additional environmental or genetic factors.2 Similarly, 14484T>C and 11778G>A have been shown to cause Leber's hereditary optic neuropathy, but expression of the disease phenotype is significantly higher when 14484T>C occurs in mtDNA belonging to haplogroup J3 and a preferential association with this haplogroup has also been observed for the 11778G>A mutation.4 The homoplasmic transition A to G at nt 4336 (4336A>G) in the mtDNA tRNAGln gene has been found at low frequency in populations of European origin. This nucleotide connects the amino acid acceptor stem with the TψC stem of tRNAGln and is moderately conserved between species.5 It is thought to entail an increased risk of Alzheimer's disease (AD) and Parkinson's disease (PD),1 5-7 but its possible role in diseases manifesting in middle life has not been evaluated. We therefore set out to study the frequency of this mutation among middle aged patients with various clinical phenotypes and in healthy, age matched controls (group 1), and also in patients with late onset neurodegenerative diseases and elderly, cognitively normal controls (group 2). We determined the entire mtDNA sequence in 10 patients and three controls with 4336A>G by conformation sensitive gel electrophoresis (CSGE) and subsequent sequencing.
Material and methods
Group 1 contained 575 patients and 480 controls (table 1). Patients with diabetes mellitus, epilepsy, sensorineural hearing loss, occipital stroke, ophthalmoplegia, intracranial calcification, white matter disease, ataxia, and migraine were ascertained as described previously.8-11 The patients with diabetes mellitus, epilepsy, or sensorineural hearing loss had a family history of similar diseases in maternal relatives. Furthermore, we obtained blood samples from 20 consecutive patients with hypertrophic cardiomyopathy. To be included as a control in group 1 it was required that the subjects and their mothers should be free of diabetes mellitus, sensorineural hearing impairment, and neurological ailments and that their mothers should have been born in central or northern Finland. After obtaining this information, the samples were anonymised.
Group 2 included 497 patients with AD, non-Alzheimer dementia, or PD. The inclusion criteria for patients with AD were those of NINCDS-ADRDA.12 All the patients with PD satisfied the PD Brain Bank (London) criteria for idiopathic PD, that is, akinetic rigid syndrome with asymmetrical onset, resting tremor, and a good response to L-dopa.13 The non-Alzheimer dementia group included patients with Parkinson plus syndrome or frontotemporal dementia.14 The control samples in group 2 were from 107 elderly subjects recruited from the same population as the patients with AD and assessed for cognitive performance to exclude dementia. The research protocol was approved by the local Ethics Committees and the Finnish Red Cross. Permission for the chart review was obtained from the Finnish Ministry of Social Affairs and Health.
Blood samples were obtained from the patients after written informed consent. Total DNA was isolated from the blood cells using a QIAamp Blood Kit (Qiagen, Hilden, Germany), and restriction fragment length polymorphisms (RFLPs) were used to identify the most informative polymorphic sites. The subsequent definitions of the various mtDNA haplogroups conformed to the published criteria.15
The tRNAGln 4336A>G mutation was detected by restriction fragment analysis using NlaIII. The mtDNA region was amplified in PCR using a forward primer spanning nucleotides 3951 to 397016 and a reverse primer spanning nucleotides 4508 to 4489. The template DNA was amplified in a total volume of 50 μl by PCR in 30 cycles of denaturation at 94°C for one minute, annealing at 55°C for one minute, and extension at 72°C for one minute, with a final extension at 72°C for 10 minutes. The amplified DNA fragment (557 bp) was then digested overnight at 37°C with 10 U of NlaIII (New England Biolabs, Beverly, MA, USA) and electrophoresed through a 1.5% agarose gel. The PCR fragment remained undigested when the wild type nucleotide was present, but was cleaved into fragments of 388 and 169 bp when the 4336A>G mutation was present. The PCR fragment also encompassed a nucleotide variant 4216T>C, which is a common variant in haplogroups T and J15 that creates a novel NlaIII site gain. The PCR fragment was cleaved into fragments of 289 and 268 bp when 4216T>C was present.
Conformation sensitive gel electrophoresis (CSGE) was carried out as described previously.17 In short, 63 pairs of primers were designed for amplification of the mtDNA coding region (nt 574-16023). The template DNA was amplified in a total volume of 50 μl by PCR in 30 cycles of denaturation at 94°C for one minute, annealing at a primer specific temperature for one minute, and extension at 72°C for one minute, with a final extension at 72°C for 10 minutes. The quality of the amplified fragment was estimated visually on a 1.5% agarose gel, and a suitable amount of the PCR product, usually 3-10 μl, was then taken for heteroduplex formation. Each amplified fragment of mtDNA from the patients with 4336A>G was mixed with the corresponding fragment amplified on two control templates, respectively, the complete sequences of which were known.17 The amplified fragments were denatured at 95°C for five minutes and the heteroduplexes were subsequently allowed to anneal at 68°C for 30 minutes.
The polyacrylamide gel was prepared as described previously,17 pre-electrophoresed for 30 minutes, and the heteroduplex samples were electrophoresed through it at a constant voltage of 400 V overnight at room temperature. After electrophoresis, the gel was stained on a glass plate in 150 μg/l of ethidium bromide for five minutes followed by destaining in water. It was then transferred to an UV transluminator and photographed (Grab-IT Annotating Grabber 2.04.7, UVP Inc, Upland, CA).
PCR fragments within the coding region and with differential mobility in CSGE, together with the hypervariable segment I (HVS I) in the D loop of the patients and the population samples, were analysed by automated sequencing (ABI PRISMTM 377 Sequencer using Dye Terminator Cycle Sequencing Ready Kit, Perkin Elmer, Foster City, CA) after treatment with exonuclease I and shrimp alkaline phosphatase.18 The primers used for sequencing the coding region were the same as those used in the amplification reactions for CSGE. The HVS I was amplified in a fragment spanning nts 15714 and 16555 and the sequence was determined between nts 16024 and 16400.
Eight middle aged patients in group 1 (1.4%) with various clinical phenotypes and three of the group 1 controls (0.63%) carried the 4336A>G mutation (table 1), this being most frequent among the patients with sensorineural hearing impairment and migraine. In addition, the mutation was found in occasional patients with diabetes mellitus or hypertrophic cardiomyopathy. The odds ratios (OR) were higher than unity for many patient groups (table 1), but calculation of the 95% confidence intervals suggested a significant increase only in the case of migraine and sensorineural hearing loss.
Five patients in group 2 (1.0%) but none of the controls harboured 4336A>G. The mutation was found in three patients with Alzheimer's disease, one with non-Alzheimer dementia, and one with Parkinson's disease, but it was not found among the 107 cognitively normal elderly subjects.
The mtDNAs harbouring the 4336A>G mutation belonged to haplogroup H. Furthermore, we determined the entire sequence in the mtDNA coding region and HVS I in 10 out of the 13 patients with 4336A>G, the three AD patients being excluded because of an insufficient amount of sample. The data were used to construct a phylogenetic network19(fig 1). There were eight mtDNA substitutions in the coding sequence that differed from the revised Cambridge reference sequence.20 Three of the variants were silent mutations and three were common polymorphisms (709G>A in 12S rRNA, a 9 bp deletion in a non-coding region between nts 8272 and 8280, and 14766T>C in the cytochrome b gene). The remaining two substitutions, 4336A>G and 9128T>C, were considered rare variants. The 9128T>C mutation, found in a patient with frontal lobe dementia, leads to an isoleucine to threonine amino acid replacement in ATPase subunit 6, and this amino acid was found to be moderately conserved in evolution according to information from the Entrez-Protein data bank (http://www.ncbi.nlm.nih.gov). This site is almost invariably occupied by a hydrophobic aliphatic amino acid, the only exception being inGorilla gorilla, where it contains methionine (table 2).
A 5 bp insertion between nts 956 and 965 in the 12S rRNA gene and 3397A>G in the ND1 gene have been reported previously in patients with AD and 4336A>G.5 Neither of these variants were found in our 10 patients harbouring 4336A>G.
The 13 patients harbouring 4336A>G were not clinically distinct, although six of them presented with more than one phenotype (table 3). Syndromic features were seen only in patient 4, who presented with hypertrophic cardiomyopathy, epilepsy, non-migrainous headaches, psychiatric symptoms, and cognitive decline. Age at the onset of the leading phenotype among the 13 patients did not differ from that among the patients without 4336A>G, but, interestingly, the age at onset for the patient with Parkinson's disease (patient 9) was 40 years.
The 4336A>G mutation has previously been found at an increased frequency among patients with neuropathologically defined AD,5-7 but other studies have failed to detect any difference in the frequency of this mutation between AD patients and controls21 or have shown a decreased frequency.22 23 We found 4336A>G in three out of 175 patients with AD (1.7%), whereas none of the 107 cognitively normal elderly subjects harboured the mutation. A total of 748 patients with AD have been screened for 4336A>G in the six previous studies and the present study combined, and 17 (2.3%, 95% confidence interval 1.20-3.34%) have been found to harbour the mutation, suggesting that the frequency of 4336A>G may be increased among patients with AD. Our results, moreover, show a similar finding among middle aged patients with phenotypes commonly associated with mitochondrial diseases, as patients with migraine harboured 4336A>G at a frequency of 4.8% and patients with matrilineal sensorineural hearing impairment at a frequency of 3.6%.
The frequency of 4336A>G among the controls, comprising middle aged Finnish blood donors, was 0.63%, which is similar to that reported for controls previously,5 6 although this mutation has been found at frequencies as high as 2.0 to 3.8% in control samples from British,23 24 German,25French-Canadian,21 and US populations.22 The variation in frequency found in the three studies with more than 200 subjects is 0.34-0.71%5 6 (present study), whereas that in the six studies with fewer than 200 subjects7 21-25is much larger (0-3.8%), suggesting that a sampling error may be involved, although true differences between the populations cannot be ruled out. In the eight previous studies,5-7 21-25 a total of 2751 subjects had been examined for 4336A>G, yielding 27 cases of the mutation (proportion of carriers 0.98%, 95% confidence interval 0.79-1.17%). The frequency of 4336A>G in the Finnish population may thus be lower than that in other white populations. Interestingly, we have previously found that a haplotype harbouring 5656A>G within haplogroup U is more than 30-fold more common among the Finns than elsewhere in Europe.26
The mtDNAs bearing 4336A>G belonged to haplogroup H, which is the most common European specific haplogroup, being found at an average frequency of 50%.27 Since haplogroup H has been found less than three times among 1175 non-whites,15 the lack of the 4336A>G mutation in Japanese patients28 is not surprising. Sequence analysis of our patients and controls indicated that 4336A>G occurs together with 14766T>C and 16304T>C, suggesting that it occupies a specific branch in the phylogenetic network. Nine patients and 15 controls had also been characterised previously by the 16304T>C variant6 22 or the correspondingRsaI site loss orAvaII site gain at nt 16303,5suggesting that they also belong to the same branch. We determined the mtDNA sequence of the coding region and the HVS I segment in 13 samples with 4336A>G (10 patients and three controls), but although the samples belonged to 10 different haplotypes, the polymorphisms characterising the haplotypes were more peripheral in the network in every case, suggesting that 4336A>G had arisen earlier. Interestingly, one control22 with 4336A>G had been found to harbour not 16304T>C but 16356T>C, a variant which characterises haplogroup U427 but which has also been found in a sample belonging to haplogroup H harbouring 16189T>C and 16223C>T.29Comparison of the latter haplotype with those in the 4336A>G network based on our samples nevertheless suggests that 4336A>G has indeed arisen at least twice in human evolution.
It has been suggested that 4336A>G may not contribute to the pathogenesis of AD in itself but may serve as a marker of a haplotype harbouring a pathogenic mutation. Our results showed, however, that the mtDNAs with 4336A>G harboured nine different substitutions, none of which was held in common. The additional substitutions included three polymorphisms in the HVS I, four in the protein coding sequence, one in the 12S rRNA gene, and one in a non-coding segment between the COX II gene and the tRNALys gene. Only 9128T>C in the ATPase6 gene was considered potentially pathogenic, as it was not found among 480 controls and sequence comparison showed that it changed a moderately conserved isoleucine to threonine. This substitution was found in a patient with frontotemporal dementia (patient 10), the clinical features of which did not conform to the clinical criteria for AD.
The patients with sensorineural hearing impairment or migraine and with 4336A>G were not clinically distinct. Interestingly, one of those with AD (patient 12) had insulin dependent diabetes mellitus and another (patient 13) had sensorineural hearing impairment. Both of these disorders are common phenotypes of mitochondrial disorders.30 Patient 10 was diagnosed with clinically typical frontotemporal dementia at the age of 66 years. Apathy and reduced speech were the main symptoms at onset and single photon emission computed tomography with99mTc-hexamethylpropyleneamine oxime as the tracer isotope showed left temporoparietal hypoperfusion. Patient 9, with PD and harbouring 4336A>G, was clinically unremarkable except for the early onset.
Previous studies have implied a role for 4336A>G in late onset neurodegenerative diseases. We found that the frequency of 4336A>G in the tRNAGln gene was significantly higher in patients with matrilineal sensorineural hearing impairment or migraine than in the controls, suggesting that the mutation may be involved in diseases already manifest in middle life. 4336A>G was the only mtDNA variant that was common to the patients, and it could therefore have a causal role in sensorineural hearing impairment or migraine and is not simply a marker linked to another, more significant mtDNA mutation.
The expert technical assistance of Ms Anja Heikkinen is acknowledged. This work was supported in part by grants from the Medical Research Council of the Academy of Finland, the Sigrid Juselius Foundation, the Finnish Medical Foundation, the Neurology Foundation, and the Maire Taponen Foundation, and by EVO grants from the Kuopio University Hospital.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.