How frequent is mutation in the mitochondrial cytochrome c oxidase gene in patients with myelodysplastic syndromes?

Reddy et al (2002) recently reported a high frequency of mitochondrial DNA (mtDNA) mutations in cytochrome c oxidase genes in patients with myelodysplastic syndromes (MDS). In a series of 20 MDS patients, they examined two fragments of mtDNA (375 and 400 bp), which together represent about 4·6% of the mitochondrial genome. The first segment includes the terminal 12% of subunit I of cytochrome c oxidase (COX), and the second segment represents about 50% of COX subunit II. It is not clear why only a small segment of COX I and the first half of COX II were selected for analysis. On the basis of looking at less than 5% of the mitochondrial genome, mtDNA mutations were allegedly found in 16 out of 20 MDS patients. As these results were not in accordance with our own experience, we looked for possible explanations and found methodological problems that may account for the unexpectedly high rate of mtDNA mutations.

Reddy et al (2002) stated that certain frequently involved nucleotide positions ‘can definitely be considered as ‘hot spots’ for mtDNA mutations'. These were nucleotide positions 7264 (25%) and 7289 (15%) in the COX I gene, and nucleotide positions 7595 (40%) and 7594 (30%) in the COX II gene. However, the COX II mutation hot spots, and another COX II mutation (nt7981), are within primer annealing sites. One of the COX I mutations (nt7582) is also within a primer annealing site. Reddy et al (2002) claimed that ‘although these mutations were within primer binding sites, they could not be artefactual because both DNA strands were sequenced to detect these mutations’. We wish to take issue with the authors over this statement. Both ends of a PCR product consist of the sequences introduced by the PCR primers. In the primer annealing sites, automated sequencing of PCR products can therefore not detect any mutations in the target DNA. Sequencing of both strands is thus not appropriate for verification of suspected mutations at these sites. The alleged mutations at 7595 (40%) and 7594 (30%) represent a major proportion of the mutations in COX II that were identified by Reddy et al (2002).

Furthermore, Reddy et al (2002) reported deletional and insertional mtDNA mutations in close proximity to PCR primers, where sequencing fluorograms are often distorted and unreliable (‘hot spot’ 7264 close to Cyt.c.OXI F in four cases, 7572 close to Cyt.cOX IR in one case, and 7622/23 close to Cyt.c.OXII F in three cases). Such deletional or insertional mutations in protein genes encoded from mitochondrial DNA are exceedingly rare, probably because the resulting frameshift would render the gene products non-functional. There was no indication as to whether the mutations were reproducible in a repeat experiment. Reddy et al (2002) did not use an independent method (e.g. restriction fragment length polymorphsim, single-stranded conformation polmorphism, denaturing gradient gel electrophoresis or temperature gradient gel electrophoresis) to confirm the mtDNA mutations suggested by automated sequencing.

Unfortunately, Reddy et al (2002) did not state the proportion of mutant mtDNA (the level of heteroplasmy) in the blood or bone marrow cells of their patients. Heteroplasmy, i.e. the coexistence of mutant and wild-type mtDNA, is important to researchers in the field of mitochondrial DNA. It is typical of disorders of the mitochondrial genome and is widely accepted as a criterion supporting the pathophysiological relevance of mtDNA mutations. MtDNA mutations producing a mild phenotype tend to reach much higher proportions of mutant mtDNA than those producing a more severe phenotype. In hereditary mitochondrial encephalomyopathies, the level of heteroplasmy can be quite high in non-dividing cells (e.g. muscle cells), but is much lower in blood cells, most likely because mitochondrial defects in proliferating cells cause selection against cells carrying a high proportion of mutant mtDNA.