In 1981, Ojala, Montoya and Attardi published an article that led to a universally accepted model for mitochondrial DNA (mtDNA) transcription. Memorable papers often provide long-awaited answers but, in this case, it is also the questions that this work has raised that make it unforgettable.

The human mitochondrial genome encodes only 13 proteins, 2 ribosomal RNAs (rRNAs) and 22 mitochondrial transfer RNAs (tRNAs). All these genes are encoded almost contiguously, in a peculiar arrangement where most protein-coding genes are separated by a single tRNA gene. In their paper, Attardi and colleagues showed that each of the two mtDNA strands is transcribed as a single polycistronic transcript, which is later processed by RNases that excise out the tRNA sequences. Each transcription cycle should therefore generate one copy of each encoded mRNA, rRNA and tRNA. This is the elegant ‘tRNA punctuation model’, which can be found in all textbooks and review articles covering mitochondrial biology (for example, see D’Souza & Minczuk, 2018).

But in the elegance of the model lies a hidden mystery: how do mitochondria manage to translate their protein-coding genes if only one set of tRNAs is produced in each transcription cycle? Moreover, if all mitochondrial tRNAs are synthesized at exactly the same rate, how do mitochondrial ribosomes adapt to the bias in amino acid composition of mitochondrial proteins?

In the cytosol of eukaryotic cells, tRNAs are extremely abundant and their relative concentrations are optimized to fit the codon composition of highly translated mRNAs. This is partly because protein-coding genes and tRNA genes are transcribed and processed differently, thereby ensuring a high tRNA to mRNA ratio. There is evidence that the relative abundance of mitochondrial rRNA may depend on slower turnover rates compared with mitochondrial mRNAs, but how mitochondria accumulate enough tRNAs for translation remains unknown.

how do mitochondria manage to translate their protein-coding genes if only one set of tRNAs is produced in each transcription cycle?

Is the turnover rate of mitochondrial tRNAs much slower than that of mitochondrial mRNAs? Do mitochondrial genomes undergo ‘futile’ transcription cycles that generate only tRNAs? Is there an unknown mechanism for the selective transcription of mitochondrial tRNA genes? These questions remain open, awaiting someone to follow in the footsteps of Ojala, Montoya and Attardi.