An alternative UPF1 isoform drives conditional remodeling of nonsense‐mediated mRNA decay

Abstract

The nonsense‐mediated mRNA decay (NMD) pathway monitors translation termination in order to degrade transcripts with premature stop codons and regulate thousands of human genes. Here, we show that an alternative mammalian‐specific isoform of the core NMD factor UPF1, termed UPF1LL, enables condition‐dependent remodeling of NMD specificity. Previous studies indicate that the extension of a conserved regulatory loop in the UPF1LL helicase core confers a decreased propensity to dissociate from RNA upon ATP hydrolysis relative to UPF1SL, the major UPF1 isoform. Using biochemical and transcriptome‐wide approaches, we find that UPF1LL can circumvent the protective RNA binding proteins PTBP1 and hnRNP L to preferentially bind and down‐regulate transcripts with long 3’UTRs normally shielded from NMD. Unexpectedly, UPF1LL supports induction of NMD on new populations of substrate mRNAs in response to activation of the integrated stress response and impaired translation efficiency. Thus, while canonical NMD is abolished by moderate translational repression, UPF1LL activity is enhanced, offering the possibility to rapidly rewire NMD specificity in response to cellular stress.

No-nonsense: insights into the functional interplay of nonsense-mediated mRNA decay factors

“…In the cell, UPF1 isoform 2 (UniProtKB: Q92900-2) is the most abundant isoform and accordingly has been studied extensively. More recently however, interest in the less abundant isoform 1 of UPF1 (UniProtKB: Q92900-1) has increased [ 68 , 81 ]. The two UPF1 isoforms exist in mammals only and result from the usage of an alternative 5′ splice site in exon 7 of the Upf1 gene [ 68 ].…”

“…Consistently, UPF1 LL has higher RNA unwinding and translocation activities than UPF1 SL [ 68 ]. Current findings indicate distinct roles for UPF1 LL and UPF1 SL in NMD [ 81 ]: protective RNA-binding proteins, such as polypyrimidine tract-binding protein-1 (PTBP1) or heterogeneous nuclear ribonucleoprotein L (hnRNP L), were shown to bind to mRNAs with long 3′-UTR and to inhibit translocation of UPF1 SL helicase along the mRNA, thus promoting UPF1 SL dissociation [ 83–85 ]. Because of its higher RNA-binding and unwinding activity, UPF1 LL helicase is suggested to overcome this NMD-inhibition by protective RNA-binding proteins and thus promote (EJC-independent) NMD of mRNAs with very long 3′-UTRs [ 81 ].…”

“…A further example of branched NMD is UPF1 which exists in two splice variants in mammalian cells, UPF1 LL and UPF1 SL , with different functional roles and outcomes for NMD (see above). These variants are differentially expressed in human tissues: while UPF1 LL represents ∼20% of expressed UPF1 in most tissues, it is up-regulated to ∼35% in muscles and down-regulated to ∼15% in the liver [ 81 ]. The specific knockdown of UPF1 LL indicates that it participates in NMD under normal cellular conditions.…”

“…The specific knockdown of UPF1 LL indicates that it participates in NMD under normal cellular conditions. Transcriptome analyses revealed that UPF1 LL NMD targets were enriched in mRNAs encoding proteins preferentially translated at the ER, such as integral membrane proteins [ 81 ]. Interestingly, during ER stress, these UPF1 LL NMD targets remained down-regulated.…”

“…Interestingly, during ER stress, these UPF1 LL NMD targets remained down-regulated. In contrast, mRNAs preferentially targeted by UPF1 SL were up-regulated in agreement with stress-induced NMD inhibition [ 81 ]. This differential regulation of NMD is suggested to result from the down-regulation of translation which is induced by the integrated stress response and mediated by the phosphorylation of eIF2α.…”