Triplex DNA:RNA,3′-to-5′ Inverted RNA and Protein Coding in Mitochondrial Genomes

Abstract

Abstract

Triple-stranded DNA:RNA helices of unknown function in vertebrate mitochondria associate with replication and transcription. Antiparallel Hoogsteen pairings form triplexes at physiological conditions. Intermolecular antiparallel triplexes require inverted 3′-to-5′ RNA polymerization, which was never observed. Three rare, long natural 3′-to-5′ inverted GenBank RNAs from mice mitochondria suggest occasional inverted transcription, putatively coding for proteins. BLAST aligns 18 GenBank-stored proteins with hypothetical proteins translated from the 3′-to-5′ inverted Mus musculus mitochondrial genome. Three are DNA-binding, five are membrane proteins. 25% of main frame codons contribute to their 3′-to-5′ overlap coding. Properties of these codons match those of overlap coding protein genes, as compared to codons not expected involved in inverted coding: a) nucleotide contents at synonymous codon positions in mitochondrial genomes fit replicational deamination gradients (A->G and C->T), but digress from gradients when functioning as nonsynonymous positions in putative 3′-to-5′ overlapping genes; b) bias against ‘circular code’ codons (codon groups creating unambiguity between frames), and favouring homogenous codons (AAA, CCC, GGG, TTT) characterize overlapping genes, including putative 3′-to-5′ overlapping genes, as compared to nonoverlapping coding sequences from the same main frame gene. This signature correlates with digression from deamination gradients. Deamination and circular code tests confirm independently alignment-based predictions of overlapping 3′-to-5′ protein coding genes. Results indicate varying expression for different 3′-to-5′ overlapping genes. Inverted 3′-to-5′ RNA is produced, perhaps by an unknown RNA polymerase (invertase) putatively coded by 3′-to-5′ inverted RNA.