We are grateful to Dr Gattermann and colleagues for raising several interesting issues related to our report (Reddy et al, 2002) of mitochondrial DNA (mt-DNA) mutations in cytochrome c-oxidase (COX) genes I and II in patients with myelodysplastic syndromes (MDS). We wish to address each point in their letter respectively.
We believe that the rate of mt-DNA mutations reported in our paper is higher than that reported previously by Gattermann et al for several reasons. First, we used a more sensitive polymerase chain reaction (PCR)-based technique as opposed to Gattermann et al who used restriction fragment length polymorphism (RFLP) before DNA sequencing. Second, they used unfractionated bone marrow (BM) samples (Gattermann et al, 1997), whereas we used four separate cell fractions comprising low- and high-density BM cells and low- and high-density peripheral blood (PB) cells. We used separated fractions because the percentage of apoptotic cells in density separated PB and BM samples from MDS patients was found to be highest in the high-density cells (Shetty et al, 2000). As we hypothesized an association between mt-DNA mutations and apoptosis, we used fractionated cells to enrich for such cells. In fact, we confirmed that the highest incidence of mt-DNA mutations was in the high-density BM compartment and therefore associated with mainly apoptotic cells. Gattermann et al (1997) may not have detected mt-DNA mutations using RFLP (less sensitive than PCR) in unfractionated BM cells where the cells containing mt-DNA mutations would only constitute a small subpopulation.
The mutations we found were, by definition, heteroplasmic as they were not always present in all the cell fractions. In other words, there were both cells containing mutant and wild-type mt-DNA, with the highest proportion of mutant mt-DNA being present in the high-density BM fraction of cells. As correctly pointed out by Gattermann et al, such heteroplasmy further supports a pathophysiological relevance of mt-DNA mutations being reported.
We reported finding a total of 18 mutations in COX I and 34 mutations in COX II genes. Three COX II mutations reported in eight patients (nucleotide positions 7595, 7594 and 7981) and one COX I (np7582) mutation were found in the primer annealing sites. We clearly recognized this and reported, as well as commented on, this in the paper. Our reasons for suspecting that this was not a technical artefact is that when the same primers were used in matched buccal smears of the same patients, as well as fractionated cells from 10 normal age-matched donors, we did not find the mutations. In addition, we have now examined lymphocytes from some of these patients and found that using the same primers, there were no mutations in these cells either. The specific detection of these mt-DNA mutations only in the haematopoietic cells of MDS patients, and their absence from matched buccal smear cells and lymphocytes from the same patients suggests that the mutations are not likely to be related to a technical error. One possible explanation is that once the PCR primer encounters a mutation, the remaining nucleotides in the primer no longer bind to the target strand, and the loose end is removed by exonuclease activity followed by primer extension. Obviously, this explanation is speculative and to confirm the presence of the mutation, we have designed primers flanking these mutations in individual patients, and these data will be reported shortly. We have used Single Strand Conformational Polymorphism (SSCP) in some of these cases and confirmed the presence of mt-DNA mutation(s) (Fig 1).
In summary, therefore, we believe that the higher incidence of mt-DNA mutations we found in COX genes compared with the previously reported incidence may be due to the more sensitive technique and the fractionated nature of the samples we used. Furthermore, as the mutations were uniquely detected in haematopoietic cells and not in the matched buccal smear samples or lymphocytes, we believe the data were not the results of a technical error. We have also confirmed the presence of mutations in these samples by SSCP now and, as these mutations were found frequently in only selected fractions of cells, they are, by definition, heteroplasmic in nature and therefore likely to be related to pathology. Finally, the three mutations detected in the primer annealing region of COX I and II genes in haematopoietic cells, but not found in matched non-haematopoietic cells, need to be confirmed by independent techniques as pointed out by Gattermann et al, and we are in the process of doing so now. We hope that Dr Gattermann and colleagues will also use PCR-based techniques on fractionated samples in the future so that our data may be similar.