Editor—Wolfram syndrome is the association of juvenile onset diabetes mellitus and optic atrophy,1 also known as DIDMOAD (diabetes insipidus, diabetes mellitus, optic atrophy, and deafness). This is a progressive, neurodegenerative disorder, with diabetes mellitus and optic atrophy presenting in the first decade,2 cranial diabetes insipidus, and sensorineural deafness in the second, and neuropathic bladder in the third, followed by neurological complications (cerebellar ataxia, myoclonus) and psychiatric illness in the fourth decade. The clinical phenotype is consistent with an ATP supply defect, suggesting a mitochondrial mediated disease.3 Mitochondrial genome deletions and pathogenic point mutations4 5 have been described in Wolfram patients. A nuclear gene WFS1, wolframin,6 7 was recently identified, encoding a polypeptide of 890 amino acids. Wolfram syndrome thus appears to be genetically heterogeneous. Recently, a distinct “mitochondrial haplotype” was described.8 Because recombination does not appear to be characteristic of mtDNA, the accumulation of polymorphisms can be used as a “genetic clock” to estimate diversity within and between populations.9 A cluster of nucleotide exchanges at nucleotide positions 4216 and 11 251 roughly distinguished a series of 6/8 Wolfram patients from controls and patients with Leber hereditary optic neuropathy (LHON). The authors suggested that these mtDNA variants may predispose to Wolfram syndrome.8 We investigated our cohort of 50 Wolfram syndrome patients10 for evidence of a distinct mitochondrial haplotype and mitochondrial DNA rearrangements.

Patients for this study were included from a cohort recruited nationally in the UK.10 Minimal ascertainment criteria were juvenile onset (less than 30 years of age) diabetes mellitus and optic atrophy. These were chosen as the only features consistently present and earliest to develop in 166/168 case reports.11Diabetes mellitus was defined as a fasting plasma glucose of more than 6.0 mmol/l (>3 SD above the mean of the normal population). All affected patients had been examined, with pupils dilated, by an experienced ophthalmologist. All patients were visited at home; blood samples were obtained from all available family members after informed consent was given. DNA was extracted from whole blood using Puregene DNA extraction kits (Gentra Systems), according to the manufacturers' instructions, and diluted to stock solutions of 500 ng/μl.

The mitochondrial haplotype of the Wolfram patients was investigated using PCR and direct sequencing of the region between 16 050 and 16 400 in the first hypervariable region of the large non-coding region as previously described.12 Mitochondrial haplotype analysis was carried out on 32 Wolfram families of European origin. The presence of mtDNA variants at bp 4216, 11 251, and 15 257 were assessed using PCR and restriction digestion.9 13 Mean pairwise differences14 were estimated and compared with a control data set comprising 10015 and 60 UK white subjects.16 An unrooted tree was drawn based on previous studies of Europeans. A PCR ApaI restriction site assay was used to screen for the mitochondrial tRNA Leu (UUR) A to G (3243) mutation17; this has been associated with maternally transmitted diabetes and deafness.18 The final cycle of PCR was labelled with ³³P-dCTP permitting detection of the mutation at levels of heteroplasmy below 1%. DNA from whole blood was also screened for the 11778A:T and 3460A:T mutations associated with Leber's hereditary optic neuropathy (LHON) using allele specific PCR.19 These mutations were chosen as they account for 75% of all cases of LHON. The 14484 mutation was not investigated as it is phenotypically milder and has a better outcome.20 Mitochondrial DNA was also analysed for the presence of major rearrangements by long range PCR21; PCRs were performed in a final volume of 50 μl with 0.25 μl each of primers L1 (nt2695-2720) and H3 (nt16459-16436), 2.5 mmol/l MgCl₂, 0.25 mmol/l of each dNTP, 1 × Bio-Optiform 111 buffer (Bioline), and 1.5 U Bio-X-ACT Taqpolymerase(Bioline). PCRs were hot started at 80°C by the addition of 20-50 ng DNA and denatured at 95°C for 10 seconds and 68°C for 10 minutes plus 30 seconds for each subsequent cycle (25 cycles). PCR products were run on 0.7% Seakem agarose gels. The resolution for long range PCR is about 1 kb (sometimes 0.5 kb depending on the gel).

Needle muscle biopsies under local anaesthetic were undertaken on nine adult Wolfram patients. Informed, written consent was obtained, after a written and verbal explanation. Respiratory chain activity was analysed in fresh muscle biopsy samples from six patients by two separate methodologies. Flux experiments measuring rotenone sensitive cytochrome c reductase with glutamate, 2-oxoglutarate, and pyruvate + malate as substrates to assay complexes I and III, and antimycin A sensitive succinate cytochrome reductase for complexes II and III, were carried out.22 23 Flux experiments on cytochrome oxidase for complex IV were also performed. Specific assays using n-decylubiquinone for complex I and the reduced form for complex III were also used.24 The reference values were calculated from a control population of 100, which included normal muscle obtained from orthopaedic procedures, and muscle biopsy tissue from patients with neuromuscular disease including non-mitochondrial metabolic abnormalities, such as McArdle disease. Histochemistry results were available for three additional patients for cytochrome oxidase, succinate dehydrogenase, and NADH reductase. The study was approved by the ethics committee of South Birmingham district health authority.

The clinical features of our cohort of 45 patients have been described.10 There were 29 index patients (14 male, 15 female) and 16 secondary patients (all sibs, seven male and nine female). Twenty seven of the 29 families were white UK patients. In addition, we included three affected sibs from Ireland and two affected sibs from New Zealand. The mitochondrial haplotypes of 32 white UK Wolfram patients are shown in the “unrooted tree” (fig 1). There was no evidence of clustering of the Wolfram haplotypes by eye or by calculating the mean number of differences between subjects in the Wolfram and control populations.14 Only 4/32 (12%) fell into haplogroups 2A and 2B (equivalent to J and T of Torroniet al 25 and collectively to 4216+11251 of Hofmann et al 8), which was less than the expected 6 based on prevalence of these lineages in our control population. This proportion was significantly different from the 7/8 (88% p=0.0001, Fisher's exact test) Wolfram patients found in this haplogroup by Hofmann et al.8 Hence there was no evidence of a founder mitochondrial haplotype in our patients. The patients were then divided into those from families showing genetic linkage to the Wolfram locus on chromosome 4p and sporadic cases. Neither group showed evidence of haplotype clustering either on the unrooted tree or by comparison of the mean pairwise differences. Mutations inWFS1 have been identified in 15/16 patients investigated; there is no evidence of haplotype clustering of patients with WFS1 mutations. Lymphocyte derived DNA was available from 40 patients in the cohort from 28 unrelated white families. There was no evidence for the mitochondrial tRNA Leu 3243G:C mutation or the 11778 A:T and 3460 A:T mutations associated with LHON. In addition, there was no evidence for major mitochondrial rearrangements. Nine Wolfram patients had muscle biopsies; of these, six showed normal respiratory chain complex activity and three normal histochemistry (table 1). The respiratory chain complex activity of four of these patients has been reported previously.26Sixteen of the 32 families were included in a mutation analysis ofWFS1.27 Loss of function mutations in WFS1 were found in 15 of these families, including nonsense, missense, in frame deletions, in frame insertions, and frameshift mutations.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Unrooted mitochondrial phylogenetic tree of UK Wolfram patients. Each numbered open circle represents a UK Wolfram patient. Shaded circles represent patients in whom a (nuclear) WFS1 mutation has been identified. Each solid line represents a change in one base relative to the Cambridge reference sequence. Numbers in italics indicate the base change, for example, 298 represents a base change at nucleotide 16 298. Dashed lines represent divisions between Clades (for example, 1-5) and within Clades (for example, 1A, 1B).

We found no evidence supporting a role for mtDNA in Wolfram syndrome. Firstly, our data show no evidence for distinct mitochondrial haplotypes in Wolfram syndrome as previously described.8 A cluster of nucleotide exchanges were reported at nucleotide positions 4216 and 11 251 (designated haplogroup B by Hofmannet al,8 haplogroup 2B by Richards et al,28 and haplogroup T by Torroni et al 25), which was found in 6/8 (75%) of Wolfram patients but in only 13/67 (19%) controls. Only 2/28 (7%) of our patients fell within this haplogroup and there was no evidence of polymorphism clustering in the remainder of our cohort of 40 patients. We conclude that because of the small size of the previous study, the results most probably occurred by chance.8 On the other hand, it is clear that in LHON the 11 778 A:T mutation is associated with sequence changes at base pairs 4216 and 13 708 (which define haplogroup A of Hofmann et al,82A of Richards et al,28 and J of Torroni et al 25). This suggests that mtDNA sequence changes found in this haplogroup (not necessarily the 4216 and 13 708 variants) may play an aetiological role in LHON. Similarly, recent studies suggest that a common variant in the non-coding region of mtDNA at bp 16189 is associated with insulin resistance in adult life.29 It remains possible that an unrecognised mitochondrial DNA variant may predispose to Wolfram syndrome.

Secondly, we found no abnormal mitochondrial function or mtDNA mutations in our series. Our search for mtDNA rearrangements only excluded them from one tissue (lymphocytes); one might expect patients to harbour mutations in other non-dividing tissues such as brain or pancreas. Wolfram syndrome is a neurodegenerative disease; it is known that mtDNA mutations normally accumulate with age in postmitotic tissues, and this may be increased in neurodegenerative disease.30 We believe that the phenotypes of patients in whom mtDNA mutations have been reported are not typical of the majority of our cases and will not prove to have mutations inWFS1. The recent identification of a Wolfram syndrome gene (WFS1, wolframin), allowed us to screen our patients for mutations.27 We found loss of function mutations in WFS1 in 16 of the 17 families investigated. At present, the intracellular location of the WFS1 protein is not known. It is expressed in most organs, but exact localisation awaits the development of specific antibodies. The amino acid sequence does not have significant homology with other proteins in the databases and there is no evidence that the WFS1 protein has a mitochondrial targeting sequence.

Our study has implications for clinical practice; definition of inherited diabetes syndromes at a molecular level will help us to distinguish overlapping clinical phenotypes such as Wolfram syndrome and mitochondrial diabetes and deafness. In addition, exclusion of a role for mitochondrial DNA should simplify genetic counselling for families with Wolfram syndrome.