What makes an NMD target? Identification and characterization of NMD sensitive endogenous transcripts in human cells

Abstract

Nonsense-mediated mRNA decay (NMD) was initially described as a quality control mechanism clearing transcripts harboring a premature termination codon (PTC) from the cell. Because a PTC can be caused by a mutation in the gene sequence or by aberrant pre-mRNA splicing, NMD was always associated with abnormal or pathological conditions. However, transcriptome profiling revealed that many mRNAs with no PTC are downregulated, often only moderately, by the NMD pathway. NMD therefore emerges as a post-transcriptional mechanism contributing to the fine-tuning of gene expression. The molecular mechanism of NMD is only partially understood and depends on the concerted activity of many factors, but accumulating evidence indicates that aberrant translation termination is the general underlying condition triggering NMD. Previous genome-wide studies showed little agreement regarding the endogenous transcripts subject to NMD, and depletion of different NMD factors affected different sets of transcripts. To address these inconsistencies, we performed RNA-seq experiments following the latest best practices of the field. We carried out knockdowns of three well-characterized NMD factors (UPF1, SMG6 and SMG7) and also operated the respective rescues, which allowed us to increase the stringency and accuracy of the analysis. Furthermore, a differential transcript usage (DTU) analysis was applied, attempting to discern mRNAs directly affected by NMD from indirect effects. Our results show that despite the existing individual differences, UPF1, SMG6 and SMG7 similarly affect the abundance of most mRNAs. Among the NMD-targeted genes, we found a significant enrichment of miRNA host genes, in addition to the already reported snoRNA host genes (Lykke-Andersen et al., Genes Dev 2014). Many non-coding RNAs also appear to be targeted by NMD, depending on the presence of open reading frames (ORFs) in their sequence, consistent with recent reports showing ribosome association of many supposedly non-coding transcripts (Ingolia et al., Cell Reports 2014). Furthermore, we also obtained evidence of transcription upstream of canonical start sites, which appears to be partially cleared by the NMD pathway, but to a lesser extent than has been reported for yeast (Malabat et al., eLife 2015).