To the Editor:
In a recent article in Blood, Shirafuji et al1report the isolation of a cDNA encoding an antisense RNA for subunit I of mitochondrial cytochrome c oxidase (MARCO). Expression of this cDNA in hematopoietic cell lines caused morphological changes and cell death. Although Shirafuji et al1 convincingly demonstrate that the antisense RNA induced these changes, we would take issue with the investigators on the suggested mechanism, namely interference with the corresponding mitochondrial gene transcript (COX I).
Animal mitochondria represent a distinct cellular compartment that has its own mitochondrial DNA, encoding 13 proteins, including 3 subunits of cytochrome c oxidase. The mRNAs of the mitochondrial genes are translated on mitochondrial ribosomes. Whereas the protein components required for mitochondrial gene expression (polymerases, ribosomal proteins) are imported into mitochondria, all the necessary RNA components (transfer-RNAs, ribosomal RNAs) are encoded by the mitochondrial genome and are therefore not taken up from the cytoplasm.2 There is no indication of significant DNA or RNA uptake into mitochondria. Even for the nucleus-encoded small RNA components of two mitochondrial riboproteins (MRP-RNAse3and RNAse P4) the import mechanism is unclear and may involve cotransport with the associated proteins. Nonspecific uptake of polynucleotides is unlikely, because the inner mitochondrial membrane is a nonpolar lipid bilayer with only a very limited permeability towards polar molecules. To support an antisense mechanism, the investigators would have to demonstrate that the MARCO-RNA actually crosses the mitochondrial membrane to interfere with the corresponding mitochondrial COX I gene transcript. Second, the investigators should have shown that other mitochondrial transcripts are not affected, because, otherwise, the reported reduction of the COX I mRNA could be interpreted as an unspecific feature associated with early stages of cell death.
Finally, we would like to speculate on an alternative explanation for the findings of Shirafuji et al.1 Because all the mitochondrial genes encode very hydrophobic membrane proteins, they contain nucleotide sequences that are similar to a variety of other membrane spanning proteins. It is therefore conceivable that MARCO could affect the cytoplasmic translation of an important membrane protein that is located in the cytoplasm rather than in the mitochondria. Decreased expression of this protein may trigger cell death, with an associated early change in mitochondrial gene transcription.
Nevertheless, even the demonstration that mitochondrial (antisense) mRNAs can induce cell death when present in the cytoplasm might open new perspectives regarding the role of mitochondria in apoptosis.
I thank Dr Hofhaus for the interest in my study of expression cloning and characterization of MARCO (mitochondrial antisense RNA for cyto-c oxidase).1-1 He points out that it is doubtful that antisense RNA can cross mitochondrial membrane. It has been reported that tRNAs in the cytoplasm can cross mitochondrial membrane using their specific receptors, which can recognize specific nucleotide sequences and integrate tRNAs into mitochondria in accordance with ATP and TAS factor (sequence-specific RNA binding factor) or other cytoplasmic proteins.1-2 As I referred to in my report, small RNAs, including antisense RNAs, can also cross mitochondrial membrane with this mechanism.1-3,1-4 These receptors have been reported to recognize nucleotide sequence GAAA A/G G in Leishmania mitochondria system,1-4 GAAGGG, CGAGAGG, and CGAAGGG in trypanosomes and crithidia in 180 nt srRNA system,1-5 CGAGAG in 140 nt srRNA system,1-5 CGAATG in trypanosome 5S RNA system,1-6and GGCAGAG and GGUAGAG in L tarentolae tRNA system.1-7 These observations have not been reported in the human mitochondria system; however, it is possible that also in the human system similar mechanisms work to integrate tRNAs and small RNAs into mitochondrial organella across mitochondrial membrane because of very high homology of mitochondrial nucleotide sequences beyond species specificity. The nucleotide sequence of MARCO contains three GAAAGG and one GGTAGAG.1-1 These sequences may be recognized by mitochondrial membrane receptor, and MARCO may be able to cross membrane. Also, it is not clear how large nucleotides can cross the mitochondrial membrane. It may be possible that, after MARCO is transcribed in the cytoplasm, partial digestion of RNA occurred, and small fragments of MARCO with receptor-recognized sequence can cross the mitochondrial membrane, as reported in the Kinetoplastid srRNA system in which cytoplasmic 28S rRNA precursor is processed posttranscriptionally.1-5 Otherwise, it may be possible that the entire MARCO RNA can cross mitochondrial membrane after binding to the receptor.
Furthermore, as Dr Hofhaus points out, other mechanisms should be possible to induce cell death when MARCO is expressed in the cytoplasm. Thus, I agree with Dr Hofhaus that precise experiments are required to make clear the mechanism of the action of MARCO.