A new tool for quantifying mitochondrial RNA processing intermediates reveals increased levels of immature mRNA species under oxidative stress
Mitochondrial gene expression in mammals requires the processing of polycistronic transcripts into mature RNAs, generating multigene maturation intermediates. Yet these species and their potential association with mitochondrial stress remain poorly characterized. We analyzed intact RNA species in healthy mouse hearts using direct RNA sequencing and identified polyadenylated mitochondrial transcripts containing complete mRNA sequences with uncleaved tRNA extensions, which we termed immature mRNAs (imRNAs). Guided by these findings, we developed a new computational tool to estimate imRNA levels using conventional short-read RNA-seq datasets. Analysis of mouse hearts deficient in mitochondrial superoxide dismutase 2 (SOD2), a model of mitochondrial oxidative stress, revealed selectively elevated levels of imRNAs retaining 5' tRNA extensions, without increases in those with 3' tRNA extensions. Comparative analyses of several mouse models with defects in mitochondrial respiration, transcription, translation, or proteostasis did not reproduce this pattern. In fact, a mouse model of mitochondrial import stress with reduced ROS generation exhibited decreased levels of imRNAs retaining 5' tRNA extensions. Conversely, human induced pluripotent stem cells treated with rotenone, which elevates mitochondrial ROS, showed increased levels of imRNAs retaining 5' but not 3' tRNA extensions, reproducing the pattern observed in SOD2-deficient mouse hearts. These findings identify an imRNA signature potentially associated with mitochondrial oxidative stress. The presence of imRNAs described here suggests that widely used transcriptome analysis pipelines may count these species as mature mRNAs, overestimating mitochondrial mRNA levels and overlooking defects in RNA maturation. By enabling retrospective analysis of processing intermediates, our tool provides access to previously unexamined information in existing RNA-seq datasets.