Mitochondrial sequencing identifies long noncoding RNA features that promote binding to PNPase

The spatial distribution of genetic regulators within eukaryotic cells dictates functional outcomes and metabolic efficiency. Prior research has shown that long noncoding RNAs (lncRNAs) primarily function within the nucleus to modulate gene expression through epigenetic and transcriptional mechanisms. Recent evidence suggests these transcripts also localize to extranuclear compartments like the mitochondrion to influence organellar physiology. The specific mechanisms facilitating the translocation of nuclear-encoded transcripts across double mitochondrial membranes remain largely elusive to researchers. Scientists suspect that specialized proteins recognize specific motifs to guide this transport process across the intermembrane space. Understanding these interactions is vital for developing targeted mitochondrial therapies for complex metabolic diseases. This gap motivated the current investigation into how structural motifs influence transcript trafficking into the mitochondrial matrix.

This investigation evaluates the interaction between nuclear-encoded transcripts and polynucleotide phosphorylase (PNPase) to define mitochondrial entry requirements. The researchers sought to identify specific primary sequences and secondary structures that govern RNA-protein binding within cardiac environments. By utilizing machine learning, the team aimed to predict which noncoding elements possess high affinity for the transport protein. The study addresses how specific protein domains, namely KH and S1, contribute to the recognition of these genetic elements. Determining the abundance of mitochondrial-localized transcripts in different species provided a comparative framework for the entire analysis. The work focuses on establishing a predictive criterion for RNA import based on identified structural features and sequence motifs. Validating these computational predictions through experimental assays ensures the biological relevance of the proposed mitochondrial import pathway.

The research team utilized FVB/NJ mouse and human cardiac tissues to isolate high-purity RNA from both the cytoplasmic and mitochondrial subcellular compartments. Cross-linked immunoprecipitate (CLIP) sequencing of PNPase within the mitochondrion was performed using the Illumina MiSeq platform for high-resolution mapping. Subcellular RNA fractions underwent high-throughput sequencing on the Illumina HiSeq system to quantify transcript distribution across different cellular regions. Supervised machine learning algorithms, specifically classification and regression trees (CART) and support vector machines (SVM), analyzed sequence and structural data. In HL-1 cells, the researchers executed quantitative polymerase chain reaction (qPCR) on PNPase CLIP knockout mutants lacking KH and S1 domains. In vitro fluorescence assays measured the binding capacity of synthesized RNA fragments to the target protein under controlled conditions. The researchers compared bound fragments against randomly generated sequences to refine the predictive accuracy of the computational models.

Mitochondrial sequencing identified 112 mouse and 1,548 human lncRNAs within the organelle, with Malat1 emerging as the most prevalent transcript in both species. The majority of noncoding transcripts associated with PNPase were categorized as lncRNAs, confirming a selective binding preference for this transcript class. Support vector machines (SVM) and classification and regression trees (CART) successfully stratified bound fragments from random sequences based on structural features. Experimental validation demonstrated that RNA sequences designed using the machine learning criterion exhibited significantly higher binding affinity than control sequences. Knockout of the KH and S1 RNA binding domains resulted in a measurable decrease in lncRNA association with the transport protein. The data suggest that PNPase utilizes specific domains to recognize both primary sequence motifs and complex secondary structures during import. These findings establish a clear link between transcript architecture and the efficiency of the mitochondrial import pathway in cardiac cells.

The identification of specific structural features provides a roadmap for understanding how nuclear-encoded transcripts reach the mitochondrial matrix. These results suggest that PNPase acts as a gatekeeper, selecting transcripts based on defined sequence and structural signatures. The ability to predict and design RNA molecules with high mitochondrial affinity holds significant promise for RNA-based therapeutics. Future research may leverage these machine learning models to target specific genetic treatments to dysfunctional mitochondria in human patients. The study underscores the importance of the KH and S1 domains in maintaining mitochondrial transcriptomic integrity and functional capacity. This research clarifies the role of lncRNAs in regulating mitochondrial function across different mammalian species and tissue types. Refining the criteria for mitochondrial import will likely enhance our capacity to manipulate organellar gene expression for clinical benefit.