Messenger RNA (mRNA) degradation provides a robust means to regulate gene expression and limit the quantity of protein produced from information transcribed from individual genes. It has long been recognized that the decay rates of mRNAs within a single cell can differ by orders of magnitude and that the stability of a transcript is tightly coupled to its translation (Bicknell and Ricci, 2017). What is only beginning to emerge is an appreciation for the complexity of the relationship between mRNA translation and decay, and how RNA primary sequence and structure interplay with the multitude of RNA binding proteins capable of influencing translation at one or more of the initiation, elongation or termination steps of protein synthesis (Heck and Wilusz, 2018; Schoenberg and Maquat, 2012). One well-documented example of translation-dependent modulation of mRNA stability is the accelerated degradation of transcripts harboring premature translation termination codons (PTCs) by the nonsense-mediated mRNA decay (NMD) pathway. The cell’s ability to discern premature termination events and rapidly eliminate PTC-containing transcripts serves a vital quality control function by preventing the accumulation of carboxy-terminal truncated polypeptides that, depending on the point of truncation, likely lack function or, in some cases, acquire detrimental gain-of-function activity. Despite our appreciation of NMD and its broad role in modulating the levels of ~10% of cellular transcriptomes (Celik et al., 2017; Mendell et al., 2004; Rehwinkel et al., 2005; Smith et al., 2014), our mechanistic understanding of how translation termination is perceived as premature and the subsequent events that transpire to signal rapid decay of the mRNA remain unclear.

The machinery responsible for mediating NMD consists of three core and highly conserved proteins, UPF1, UPF2, and UPF3 (Kervestin and Jacobson, 2012). How the NMD machinery associates with substrates and monitors translation termination has been the focus of significant debate and is envisioned by two prevailing models (He and Jacobson, 2015). The first posits that premature termination, which by default occurs in the context of a lengthened or faux 3’ UTR, is inherently inefficient and results in a delay in ribosome-associated events at the PTC that is sufficient to cause subsequent recruitment and/or activation of UPF proteins on the translation machinery. The second model proposes that NMD components assemble indiscriminately on most or all transcripts, but are displaced from protein coding regions by elongating ribosomes so as to accumulate preferentially on transcripts in a 3’ UTR length-dependent manner, where they are poised to interact with terminating ribosomes and elicit downstream events. This latter model is supported by both the observed enhancement of NMD by exon-junction complexes (of which UPF3 is a peripheral component) and genome-wide binding studies revealing UPF1 binding to both normal and NMD-sensitive mRNA and redistribution of the protein from sites predominantly within 3’ UTRs to those along the entire transcript body upon inhibition of translation (Hurt et al., 2013; Zünd et al., 2013). Extended 3’ UTRs derived from the abbreviated open reading frame of PTC-containing mRNA, therefore, serve as a preferential binding platform for UPF protein interaction and provide a rationale for how a premature translation termination event is preferentially targeted by this pathway.

Independent of the mode of UPF protein association with mRNA, the translation-dependent nature of NMD specifies that UPF protein binding is insufficient to elicit NMD and that a functional interaction between the NMD and translation machinery must occur before initiating degradation of the mRNA. Such an interface between the NMD and translational machineries is supported by biochemical data demonstrating the interaction of one or more UPF proteins with ribosomes, ribosomal proteins, or rRNA (Min et al., 2013; Schuller et al., 2018) and with eukaryotic release factors 1 and 2 (eRF1 and eRF3), proteins involved in stop codon recognition and nascent peptide hydrolysis during translation termination (Ivanov et al., 2008; Kashima et al., 2006; Neu-Yilik et al., 2017; Singh et al., 2008; Wang et al., 2001). Moreover, evidence for UPF1 protein involvement in stop codon readthrough in yeast (Weng et al., 1996a; Weng et al., 1996b) and translation termination efficiency in cell-free extracts (Amrani et al., 2004; Ghosh et al., 2010) provides functional data for NMD components modulating ribosome activity. Despite these observations, recent studies using purified components and reconstituted translation assays have failed to assign a role for UPF1 in influencing either the efficiency of termination or subsequent ribosome subunit recycling (Neu-Yilik et al., 2017; Schuller et al., 2018), leaving our understanding of this critical step in NMD incomplete.

UPF1 is a member of the SF1 family of RNA helicases and exhibits RNA binding and ATP hydrolysis activities, both of which are required for NMD. Mutation of conserved residues within the UPF1 ATP binding pocket that abrogate either nucleotide binding or hydrolysis leads to stabilization of NMD substrate mRNA (Weng et al., 1996a). Structural studies on both yeast and human UPF1 (Chakrabarti et al., 2011) have illuminated how ATP binding and hydrolysis invoke conformational changes to the protein that are thought to underlie the RNA unwinding and translocation activities observed for UPF1 in vitro (Czaplinski et al., 1995; Fiorini et al., 2015) and mRNA target discrimination and ribonucleoprotein (mRNP) remodeling in vivo (Franks et al., 2010; Kurosaki et al., 2014; Lee et al., 2015). Our previous work revealed that in yeast, failure of UPF1 to hydrolyze ATP leads to the accumulation of 3’ RNA decay fragments from nonsense-containing mRNAs that arise due to an impediment in 5’ → 3’ degradation by exoribonuclease XRN1 (Serdar et al., 2016). The dependency on PTC position, translation of the mRNA, and for NMD cofactors UPF2 and UPF3 for fragment accumulation led us to propose that the block to XRN1 was a consequence of a stalled trimeric mRNP complex that forms between the ribosome, mRNA, and the NMD machinery, and that ATP hydrolysis by UPF1 is critical for efficient translation termination at a premature stop codon (Serdar et al., 2016). These data revealed an important functional interaction between the NMD and translation machinery and provided a direct role for ATP hydrolysis by UPF1 in events occurring during premature termination.

To gain mechanistic insight into how UPF1 impinges upon the translation machinery, we further characterized the 3’ RNA fragments that accumulate in ATPase-deficient UPF1 mutants. Herein we provide evidence that the 5’ terminus of decay intermediates originate downstream of the premature termination codon (PTC) and are distinct from fragments that amass when ribosomes are stalled at the PTC due to depletion of translation termination factor eRF1. Consistent with a defect that is distinct from stop codon recognition, we show that decay intermediates co-purify with 80S ribosomes and that occupancy of the translation machinery on the mRNA 3’ UTR is independent of canonical translation events. Further, we demonstrate that accumulation of decay fragments is dependent upon RNA sequence downstream of the PTC predicted to contact the mRNA binding channel of a ribosome residing on the 5’ end of the fragment and that mutations within the active site of UPF1 alter the abundance and position of the 5’ end of the RNA intermediates in an allele-specific manner. Our data reveal that a failure to hydrolyze ATP by UPF1 results in a defect in post-termination ribosome recycling at PTCs and migration of ribosomes into the mRNA 3’ UTR, and provide novel evidence linking UPF1 catalytic activity with ribosome dynamics during premature termination.

We have leveraged the 3’ RNA decay intermediates that accumulate from PTC-containing mRNA in the presence of ATPase-deficient UPF1 to enhance our understanding of the functional relationship between the translation and NMD machinery. Our previous characterization of these fragments showed that they were coincident with the transcript 3’ of the PTC, co-sediment with 80S monosomes, and accrue due to a block in digestion by the 5’ → 3’ exoribonuclease XRN1, leading us to conclude that ATP hydrolysis by UPF1 was required for efficient translation termination at a PTC (Serdar et al., 2016). High-resolution analysis of these decay intermediates presented here confirm that the fragments are ribosome associated, but reveal, unexpectedly, that the 5’ end of the RNAs map downstream of the PTC, signifying that the impediment to decay is beyond the termination codon and that ribosome(s) occupy the 3’ UTR of these transcripts in the UPF1 mutant. These findings alongside the comparative analysis of RNA decay intermediates that amass upon a genuine block in translation termination (i.e. depletion of eRF1) argue against the simple notion that the efficiency of termination by ribosomes at the PTC is impaired in UPF1 ATPase mutants. Indeed, evidence supporting inefficient translation termination including ribosomal pausing at the PTC or stop codon read through is lacking (Figure 1A and Figure 3B), and, in contrast, we observe accumulation of truncated polypeptides in the UPF1 ATPase mutants that by mass spectrophotometric analysis possess a carboxy-terminus expected for accurate peptide hydrolysis at the PTC (Figure 3B and data not shown). Our data thereby establish that UPF1 impinges on the translation machinery in a novel manner such that when hydrolysis of bound ATP is blocked, ribosome recycling is inhibited and post-termination ribosomes migrate into the transcript 3’ UTR and contribute to hindering XRN1 and preventing full degradation of the transcript downstream of the PTC.

In principle, a number of RNA elements could interfere with XRN1 progression leading to incomplete degradation of the nonsense-containing mRNAs in UPF1 ATPase mutants. RNA structure including thermostable stem-loops (Garcia and Parker, 2015), stretches of poly-guanine nucleotides (Decker and Parker, 1993), and three-dimensional XRN1-resistant RNA folds (recently identified in the flaviviral RNA genome) (Chapman et al., 2014) all obstruct XRN1 and protect downstream RNA from being digested. Inspection and computational modeling of the first 50 nucleotides of the 5’ end of the decay fragments that accumulate in our various reporter mRNAs failed, however, to identify common nucleotides/motifs or predict RNA secondary structure that might contribute to impeding XRN1. Moreover, altering the terminal nucleotide of the fragments that accrue from GFP^PTC125 mRNA did not influence the 5’ end of the decay intermediates, as would be expected if they were engaged in base pairing necessary for preventing XRN1 progression (Figure 4C). While these data do not strictly rule out the possibility that a thermodynamically stable RNA structural element contributes in blocking XRN1 in our PTC-containing mRNA, we think it unlikely as it would require that such elements exist downstream and proximal of each nonsense mutation we arbitrarily introduced into our various reporter genes in order to lead to the 3’ decay intermediates we consistently observe (Serdar et al., 2016). Alternatively, an mRNP tightly bound to the mRNA downstream of the PTC could, theoretically, cause inhibition of XRN1 activity. One plausible component of such an mRNP would be the ATPase-deficient UPF1 protein itself, which displays increased association with both NMD target and non-target RNAs compared to wild-type UPF1 (Kurosaki et al., 2014; Lee et al., 2015). Although we anticipate that UPF1 associates within the 3’ UTR of our PTC-containing transcripts (Hurt et al., 2013; Kurosaki and Maquat, 2013; Zünd et al., 2013), the expected footprint for RNA-bound UPF1 is ~11 nucleotides (Chakrabarti et al., 2011), making it difficult to explain how changes within GFP^PTC125 12–18 nucleotides downstream of the position of the 5’ end of the decay intermediate could impact UPF1 binding so as to alter fragment accumulation (Figure 4C). Moreover, any model in which RNA-bound UPF1, alone or in complex with other proteins, serves as the block to XRN1 activity would have to account for how a ribosome can traverse this block to establish the demonstrated association with the 3’ RNA fragment (Figure 2B).

In contrast, evidence supports a view in which the impediment to complete digestion of the PTC-containing mRNA by XRN1 is the 3’ UTR-bound ribosome itself. The ability of a ribosome to interfere with XRN1 progression is well documented from co-translational decay studies employing reporters harboring rare codons designed to slow translation elongation (Hu et al., 2009) and in global characterization of XRN1-dependent mRNA decay intermediates (Pelechano et al., 2015). We demonstrate that the 3’ RNA decay fragments in UPF1 ATPase mutants physically associate with ribosomes and that mutation of residues expected to be accommodated within the mRNA binding channel of a ribosome protecting the 5’ terminus of the decay fragment strongly influence the accumulation of the intermediates (Figure 4C). Attempts to gather direct evidence for ribosome occupancy at the 5’ end of the decay fragment employing ribosome profiling have, however, been unsuccessful. Although Ribo-Seq libraries from UPF1 ATPase mutants assigned ribosomes to the coding region of GFP^PTC125 mRNA at high density and with periodicity, ribosome protected fragments (RPFs) mapping to the 3’ UTR of the transcript were absent from our datasets. As discussed further below, we speculate that the 3’ UTR-bound ribosome(s) are inherently instable and/or insensitive to cycloheximide, which would contribute to omission of RPFs mapping to the 3’ UTR from our preparations; although it is also possible that the 3’ UTR-associated ribosomes assume a confirmation that protects RNA greater or less than ~28 nucleotides, and were unintentionally precluded from our libraries.

Together, our observations indicate that 80S ribosomes were not released from mRNA after peptide hydrolysis at the PTC and suggest that post-termination ribosome recycling is inhibited in UPF1 mutants defective in ATPase activity. Recent ribosome profiling in yeast suggests that movement of post-termination ribosomes into mRNA 3’ UTRs is rare but markedly enhanced in cells depleted of ribosome recycling factor RLI1 (Young et al., 2015). RLI1 (ABCE1 in mammals) is an Fe-S cluster protein that has been implicated in at least two steps of translation termination, including promoting peptide hydrolysis by eRF1 in an ATPase-independent manner and driving ribosomal subunit separation in a process that requires ATP hydrolysis (Dever and Green, 2012). Consistent with this, in RLI-deficient yeast cells, where both steps would be prevented, ribosomes are found to both queue at stop codons and transit downstream into the mRNA 3’ UTR where they appear competent to reinitiate translation without an apparent reading frame or start codon preference (Young et al., 2015). Although we provide evidence for 3’ UTR-bound ribosomes downstream of the PTC in the UPF1 DE572AA mutant, our data is not consistent with ATPase-deficient UPF1 simply preventing RLI1 from interacting with the terminating ribosome (akin to depleting it from the cell). Indeed, we showed previously that 3’ decay fragments from GFP^PTC125 mRNA in cells with reduced cellular RLI1 levels display greater size heterogeneity on northern blots than in the UPF1 mutant (Serdar et al., 2016) and evidence for 3’ UTR translation products or a queuing of ribosomes at the stop codon is lacking in UPF1 ATPase mutants (Figure 3B and Figure 1A). Then how could a defect in ATP hydrolysis by UPF1 lead to 3’ UTR ribosomes? One model that reconciles our data posits that mutant UPF1 inhibits RLI1 function not during recruitment to the terminating ribosome or in promoting peptide hydrolysis, but subsequently in its ability to catalyze ATP-dependent ribosome subunit dissociation, leading to ribosome migration into the mRNA 3’ UTR (Figure 6). In such a scenario, ribosomes blocked during the second step would be expected to retain eRF1 and catalytically-stalled RLI1 in the A site and thus be incompetent for canonical translation elongation into the mRNA 3’ UTR and any reinitiation of protein synthesis. Consistent with this model and a block in RLI1 function after its association with ribosomes terminating at the PTC, overexpression of RLI1 in UPF1 ATPase mutants does not abrogate accumulation of 3’ decay fragments, and deletion of DOM34, which binds within the ribosome A site to promote rescue of stalled ribosomes (Guydosh and Green, 2014), does not lead to an increase in the accumulation of 3’ decay intermediates (data not shown). Moreover, because RLI1-bound 80S ribosomes would be incapable of re-engaging in translation, they would be insensitive to stabilization by cycloheximide, and likely lost during the sucrose gradient centrifugation step of our ribosome profiling protocol.

Figure 6

Download asset Open asset

Our analysis of several PTC-containing reporters reveals that 3’ RNA decay intermediates differ in their heterogeneity and in the position of the 5’ ends relative to the PTC. We found that site selection for ribosome stalling was dependent upon the sequence downstream of the PTC (Figure 4B) but we were unable to ascertain common RNA features that contribute to establishing the RNA 5’ end. Specifically, we did not observe the consistent presence of an AUG codon in the predicted P site of a ribosome stalled at the 5’ end of the decay intermediates, as has been observed for ribosomes detected upstream of a PTC following premature termination in in vitro translation extracts (Amrani et al., 2004). Notably, for GFP^PTC125 mRNA, which does encode an AUG triplet proximal to the predicted P site, mutation of nucleotides within the AUG did alter the distribution and/or intensity of 5’ ends of the decay intermediates but as did substitution of several non-AUG residues flanking this triplet (Figure 4C). In addition, our finding that the identity of the 5’ ends of decay fragments is independent of the nature of the penultimate codon upstream of the PTC (Figure 4—figure supplement 1B) rules out a role for base complementarily between the mRNA and a P site tRNA that might be retained post termination. One intriguing observation is that the 5’ termini of the decay fragments in UPF1 ATPase mutants are sensitive to allele-specific amino acid substitutions predicted to influence ATP hydrolysis or retention of either the substrate or product within the enzyme active site (Figure 5B and D). Although these data closely link the catalytic function of UPF1 with ribosome dynamics post termination, the precise determinants dictating ribosome stalling remain unclear, and likely are determined by a combination of the kinetics of ATP hydrolysis with sequence elements, mRNP composition, and/or RNA structural landscape that is unique to each mRNA 3’ UTR.

We provide evidence supporting a model in which ATPase-deficient UPF1 inhibits prematurely terminating ribosome release resulting in migration of 80S ribosomes downstream of the PTC where they stall and block XRN1-mediated degradation of the mRNA. In the context of NMD in wild-type cells, our data are supportive of a function for UPF1 in which conformational changes to the protein induced by ATP hydrolysis promotes RLI1 ATP-dependent ribosome subunit dissociation at a PTC. It is tempting to speculate that modulation of this latter step in the translation termination process could serve as the molecular indicator that is sensed and transmitted to the cellular mRNA decay machinery leading to the accelerated degradation characteristic of NMD substrates. This proposed coupling of ribosome dynamics with changes in transcript stability for NMD substrates aligns with an emerging theme exemplified by modulation of mRNA decay by codon identity and translation elongation rate (Radhakrishnan and Green, 2016), as well as in the detection and rescue of stalled ribosomes (Ikeuchi et al., 2018). Future research will be needed to define the precise nature of the perturbation in RLI1 function by the NMD machinery during premature translation termination. Such studies will provide an exciting opportunity to biochemically differentiate translation termination events specific to nonsense codons and for the development of therapeutic agents capable of modulating premature termination with high specificity.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Amrani N
   2. Ganesan R
   3. Kervestin S
   4. Mangus DA
   5. Ghosh S
   6. Jacobson A

   (2004) A faux 3'-UTR promotes aberrant termination and triggers nonsense-mediated mRNA decay

   Nature 432:112–118.

   https://doi.org/10.1038/nature03060

   • PubMed
   • Google Scholar
2. 1. Bicknell AA
   2. Ricci EP

   (2017) When mRNA translation meets decay

   Biochemical Society Transactions 45:339–351.

   https://doi.org/10.1042/BST20160243

   • PubMed
   • Google Scholar
3. 1. Brown A
   2. Shao S
   3. Murray J
   4. Hegde RS
   5. Ramakrishnan V

   (2015) Structural basis for stop Codon recognition in eukaryotes

   Nature 524:493–496.

   https://doi.org/10.1038/nature14896

   • PubMed
   • Google Scholar
4. 1. Celik A
   2. Baker R
   3. He F
   4. Jacobson A

   (2017) High-resolution profiling of NMD targets in yeast reveals translational fidelity as a basis for substrate selection

   RNA 23:735–748.

   https://doi.org/10.1261/rna.060541.116

   • PubMed
   • Google Scholar
5. 1. Chakrabarti S
   2. Jayachandran U
   3. Bonneau F
   4. Fiorini F
   5. Basquin C
   6. Domcke S
   7. Le Hir H
   8. Conti E

   (2011) Molecular mechanisms for the RNA-dependent ATPase activity of Upf1 and its regulation by Upf2

   Molecular Cell 41:693–703.

   https://doi.org/10.1016/j.molcel.2011.02.010

   • PubMed
   • Google Scholar
6. 1. Chapman EG
   2. Moon SL
   3. Wilusz J
   4. Kieft JS

   (2014) RNA structures that resist degradation by Xrn1 produce a pathogenic dengue virus RNA

   eLife 3:e01892.

   https://doi.org/10.7554/eLife.01892

   • PubMed
   • Google Scholar
7. 1. Cheng Z
   2. Muhlrad D
   3. Lim MK
   4. Parker R
   5. Song H

   (2007) Structural and functional insights into the human Upf1 helicase core

   The EMBO Journal 26:253–264.

   https://doi.org/10.1038/sj.emboj.7601464

   • PubMed
   • Google Scholar
8. 1. Czaplinski K
   2. Weng Y
   3. Hagan KW
   4. Peltz SW

   (1995)

   Purification and characterization of the Upf1 protein: a factor involved in translation and mRNA degradation

   RNA 1:610–623.

   • PubMed
   • Google Scholar
9. 1. Decker CJ
   2. Parker R

   (1993) A turnover pathway for both stable and unstable mRNAs in yeast: evidence for a requirement for deadenylation

   Genes & Development 7:1632–1643.

   https://doi.org/10.1101/gad.7.8.1632

   • PubMed
   • Google Scholar
10. 1. Dever TE
    2. Green R

    (2012) The elongation, termination, and recycling phases of translation in eukaryotes

    Cold Spring Harbor Perspectives in Biology 4:a013706.

    https://doi.org/10.1101/cshperspect.a013706

    • PubMed
    • Google Scholar
11. 1. Fiorini F
    2. Bagchi D
    3. Le Hir H
    4. Croquette V

    (2015) Human Upf1 is a highly processive RNA helicase and translocase with RNP remodelling activities

    Nature Communications 6:7581.

    https://doi.org/10.1038/ncomms8581

    • PubMed
    • Google Scholar
12. 1. Franks TM
    2. Singh G
    3. Lykke-Andersen J

    (2010) Upf1 ATPase-dependent mRNP disassembly is required for completion of nonsense- mediated mRNA decay

    Cell 143:938–950.

    https://doi.org/10.1016/j.cell.2010.11.043

    • PubMed
    • Google Scholar
13. 1. Garcia JF
    2. Parker R

    (2015) MS2 coat proteins bound to yeast mRNAs block 5' to 3' degradation and trap mRNA decay products: implications for the localization of mRNAs by MS2-MCP system

    RNA 21:1393–1395.

    https://doi.org/10.1261/rna.051797.115

    • PubMed
    • Google Scholar
14. 1. Ghosh S
    2. Ganesan R
    3. Amrani N
    4. Jacobson A

    (2010) Translational competence of ribosomes released from a premature termination Codon is modulated by NMD factors

    RNA 16:1832–1847.

    https://doi.org/10.1261/rna.1987710

    • PubMed
    • Google Scholar
15. 1. Guydosh NR
    2. Green R

    (2014) Dom34 rescues ribosomes in 3' untranslated regions

    Cell 156:950–962.

    https://doi.org/10.1016/j.cell.2014.02.006

    • PubMed
    • Google Scholar
16. 1. He F
    2. Jacobson A

    (2015) Nonsense-Mediated mRNA decay: degradation of defective transcripts is only part of the story

    Annual Review of Genetics 49:339–366.

    https://doi.org/10.1146/annurev-genet-112414-054639

    • PubMed
    • Google Scholar
17. 1. Heck AM
    2. Wilusz J

    (2018) The interplay between the RNA decay and translation machinery in eukaryotes

    Cold Spring Harbor Perspectives in Biology 10:a032839.

    https://doi.org/10.1101/cshperspect.a032839

    • PubMed
    • Google Scholar
18. 1. Hu W
    2. Sweet TJ
    3. Chamnongpol S
    4. Baker KE
    5. Coller J

    (2009) Co-translational mRNA decay in Saccharomyces cerevisiae

    Nature 461:225–229.

    https://doi.org/10.1038/nature08265

    • PubMed
    • Google Scholar
19. 1. Hurt JA
    2. Robertson AD
    3. Burge CB

    (2013) Global analyses of UPF1 binding and function reveal expanded scope of nonsense-mediated mRNA decay

    Genome Research 23:1636–1650.

    https://doi.org/10.1101/gr.157354.113

    • PubMed
    • Google Scholar
20. 1. Ikeuchi K
    2. Izawa T
    3. Inada T

    (2018) Recent progress on the molecular mechanism of quality controls induced by ribosome stalling

    Frontiers in Genetics 9:743.

    https://doi.org/10.3389/fgene.2018.00743

    • PubMed
    • Google Scholar
21. 1. Ingolia NT
    2. Ghaemmaghami S
    3. Newman JR
    4. Weissman JS

    (2009) Genome-wide analysis in vivo of translation with nucleotide resolution using ribosome profiling

    Science 324:218–223.

    https://doi.org/10.1126/science.1168978

    • PubMed
    • Google Scholar
22. 1. Ivanov PV
    2. Gehring NH
    3. Kunz JB
    4. Hentze MW
    5. Kulozik AE

    (2008) Interactions between UPF1, eRFs, PABP and the exon junction complex suggest an integrated model for mammalian NMD pathways

    The EMBO Journal 27:736–747.

    https://doi.org/10.1038/emboj.2008.17

    • PubMed
    • Google Scholar
23. 1. Kashima I
    2. Yamashita A
    3. Izumi N
    4. Kataoka N
    5. Morishita R
    6. Hoshino S
    7. Ohno M
    8. Dreyfuss G
    9. Ohno S

    (2006) Binding of a novel SMG-1-Upf1-eRF1-eRF3 complex (SURF) to the exon junction complex triggers Upf1 phosphorylation and nonsense-mediated mRNA decay

    Genes & Development 20:355–367.

    https://doi.org/10.1101/gad.1389006

    • PubMed
    • Google Scholar
24. 1. Kervestin S
    2. Jacobson A

    (2012) NMD: a multifaceted response to premature translational termination

    Nature Reviews Molecular Cell Biology 13:700–712.

    https://doi.org/10.1038/nrm3454

    • PubMed
    • Google Scholar
25. 1. Kurosaki T
    2. Li W
    3. Hoque M
    4. Popp MW
    5. Ermolenko DN
    6. Tian B
    7. Maquat LE

    (2014) A post-translational regulatory switch on UPF1 controls targeted mRNA degradation

    Genes & Development 28:1900–1916.

    https://doi.org/10.1101/gad.245506.114

    • PubMed
    • Google Scholar
26. 1. Kurosaki T
    2. Maquat LE

    (2013) Rules that govern UPF1 binding to mRNA 3' UTRs

    PNAS 110:3357–3362.

    https://doi.org/10.1073/pnas.1219908110

    • PubMed
    • Google Scholar
27. 1. Lee SR
    2. Pratt GA
    3. Martinez FJ
    4. Yeo GW
    5. Lykke-Andersen J

    (2015) Target discrimination in Nonsense-Mediated mRNA decay requires Upf1 ATPase activity

    Molecular Cell 59:413–425.

    https://doi.org/10.1016/j.molcel.2015.06.036

    • PubMed
    • Google Scholar
28. 1. Longtine MS
    2. McKenzie A
    3. Demarini DJ
    4. Shah NG
    5. Wach A
    6. Brachat A
    7. Philippsen P
    8. Pringle JR

    (1998) Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae

    Yeast 14:953–961.

    https://doi.org/10.1002/(SICI)1097-0061(199807)14:10<953::AID-YEA293>3.0.CO;2-U

    • PubMed
    • Google Scholar
29. 1. Mendell JT
    2. Sharifi NA
    3. Meyers JL
    4. Martinez-Murillo F
    5. Dietz HC

    (2004) Nonsense surveillance regulates expression of diverse classes of mammalian transcripts and mutes genomic noise

    Nature Genetics 36:1073–1078.

    https://doi.org/10.1038/ng1429

    • PubMed
    • Google Scholar
30. 1. Min EE
    2. Roy B
    3. Amrani N
    4. He F
    5. Jacobson A

    (2013) Yeast Upf1 CH domain interacts with Rps26 of the 40S ribosomal subunit

    RNA 19:1105–1115.

    https://doi.org/10.1261/rna.039396.113

    • PubMed
    • Google Scholar
31. 1. Neu-Yilik G
    2. Raimondeau E
    3. Eliseev B
    4. Yeramala L
    5. Amthor B
    6. Deniaud A
    7. Huard K
    8. Kerschgens K
    9. Hentze MW
    10. Schaffitzel C
    11. Kulozik AE

    (2017) Dual function of UPF3B in early and late translation termination

    The EMBO Journal 36:2968–2986.

    https://doi.org/10.15252/embj.201797079

    • PubMed
    • Google Scholar
32. 1. Pelechano V
    2. Wei W
    3. Steinmetz LM

    (2015) Widespread Co-translational RNA decay reveals ribosome dynamics

    Cell 161:1400–1412.

    https://doi.org/10.1016/j.cell.2015.05.008

    • PubMed
    • Google Scholar
33. 1. Radhakrishnan A
    2. Green R

    (2016) Connections underlying translation and mRNA stability

    Journal of Molecular Biology 428:3558–3564.

    https://doi.org/10.1016/j.jmb.2016.05.025

    • PubMed
    • Google Scholar
34. 1. Rehwinkel J
    2. Letunic I
    3. Raes J
    4. Bork P
    5. Izaurralde E

    (2005) Nonsense-mediated mRNA decay factors act in concert to regulate common mRNA targets

    RNA 11:1530–1544.

    https://doi.org/10.1261/rna.2160905

    • PubMed
    • Google Scholar
35. 1. Schoenberg DR
    2. Maquat LE

    (2012) Regulation of cytoplasmic mRNA decay

    Nature Reviews Genetics 13:246–259.

    https://doi.org/10.1038/nrg3160

    • PubMed
    • Google Scholar
36. 1. Schuller AP
    2. Zinshteyn B
    3. Enam SU
    4. Green R

    (2018) Directed hydroxyl radical probing reveals Upf1 binding to the 80S ribosomal E site rRNA at the L1 stalk

    Nucleic Acids Research 46:2060–2073.

    https://doi.org/10.1093/nar/gkx1263

    • PubMed
    • Google Scholar
37. 1. Serdar LD
    2. Whiteside DL
    3. Baker KE

    (2016) ATP hydrolysis by UPF1 is required for efficient translation termination at premature stop codons

    Nature Communications 7:14021.

    https://doi.org/10.1038/ncomms14021

    • PubMed
    • Google Scholar
38. 1. Singh G
    2. Rebbapragada I
    3. Lykke-Andersen J

    (2008) A competition between stimulators and antagonists of upf complex recruitment governs human nonsense-mediated mRNA decay

    PLOS Biology 6:e111.

    https://doi.org/10.1371/journal.pbio.0060111

    • PubMed
    • Google Scholar
39. 1. Skabkin MA
    2. Skabkina OV
    3. Hellen CU
    4. Pestova TV

    (2013) Reinitiation and other unconventional posttermination events during eukaryotic translation

    Molecular Cell 51:249–264.

    https://doi.org/10.1016/j.molcel.2013.05.026

    • PubMed
    • Google Scholar
40. 1. Smith JE
    2. Alvarez-Dominguez JR
    3. Kline N
    4. Huynh NJ
    5. Geisler S
    6. Hu W
    7. Coller J
    8. Baker KE

    (2014) Translation of small open reading frames within unannotated RNA transcripts in Saccharomyces cerevisiae

    Cell Reports 7:1858–1866.

    https://doi.org/10.1016/j.celrep.2014.05.023

    • PubMed
    • Google Scholar
41. 1. Wang W
    2. Czaplinski K
    3. Rao Y
    4. Peltz SW

    (2001) The role of upf proteins in modulating the translation read-through of nonsense-containing transcripts

    The EMBO Journal 20:880–890.

    https://doi.org/10.1093/emboj/20.4.880

    • PubMed
    • Google Scholar
42. 1. Weng Y
    2. Czaplinski K
    3. Peltz SW

    (1996a) Genetic and biochemical characterization of mutations in the ATPase and helicase regions of the Upf1 protein

    Molecular and Cellular Biology 16:5477–5490.

    https://doi.org/10.1128/MCB.16.10.5477

    • PubMed
    • Google Scholar
43. 1. Weng Y
    2. Czaplinski K
    3. Peltz SW

    (1996b) Identification and characterization of mutations in the UPF1 gene that affect nonsense suppression and the formation of the upf protein complex but not mRNA turnover

    Molecular and Cellular Biology 16:5491–5506.

    https://doi.org/10.1128/MCB.16.10.5491

    • PubMed
    • Google Scholar
44. 1. Young DJ
    2. Guydosh NR
    3. Zhang F
    4. Hinnebusch AG
    5. Green R

    (2015) Rli1/ABCE1 recycles terminating ribosomes and controls translation reinitiation in 3'UTRs In Vivo

    Cell 162:872–884.

    https://doi.org/10.1016/j.cell.2015.07.041

    • PubMed
    • Google Scholar
45. 1. Zünd D
    2. Gruber AR
    3. Zavolan M
    4. Mühlemann O

    (2013) Translation-dependent displacement of UPF1 from coding sequences causes its enrichment in 3' UTRs

    Nature Structural & Molecular Biology 20:936–943.

    https://doi.org/10.1038/nsmb.2635

    • PubMed
    • Google Scholar

Acceptance summary:

This work is significant in providing evidence that genetically impairing the ATPase activity of UPF1, a factor central to the process of nonsense-mediated mRNA decay (NMD), interferes with the process of ribosome recycling by ABCE1/RLI1 following translation termination at a premature termination codon, allowing the unrecycled 80S ribosomes to migrate into the 3' UTR and stall at particular sequences. As such, it represents a valuable contribution to our understanding of the influence of NMD proteins on the process of translation termination at premature stop codons.

Decision letter after peer review:

Thank you for submitting your article "Inhibition of post-termination ribosome recycling at premature termination codons in UPF1 ATPase mutants" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Alan Hinnebusch as the Reviewing Editor and James Manley as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Oliver Mühlemann (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data. If you choose to pursue this option and re-submit the manuscript without performing the requested additional experiments, and you cannot convince the reviewers that they are not required to justify your main claims, then you will be committing yourself to eventually carrying out the additional experiments and reporting on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

In this report, the authors study decay intermediates from a reporter mRNA harboring a PTC that accumulate in a UPF1 ATPase mutant, but not in the absence of UPF1 protein, and which were shown previously to originate from inhibition of XRN1 5' to 3' exonucleolytic decay by ribosomes on the mRNA. Here they show that the 5' ends of these intermediates map downstream of the PTC rather than coinciding with the PTC itself as had been proposed previously. Thus, rather than providing evidence of a ribosome stalled at the PTC owing to inefficient termination; they instead appear to represent an unrecycled 80S ribosome that has migrated from the PTC and become stalled at specific positions downstream in the UPF1 ATPase mutant.

Coimmunoprecipitation analysis of tagged Rps13 shows that these decay fragments are ribosome bound. Other evidence suggests that the protecting ribosomes are not engaged in either translational read-through of the PTC, nor from reinitiation of translation 3' of the PTC. Evidence further indicates that the distance from the PTC is not a critical factor in determining the site of the block to 5'-3' decay presumed to arise from the stalled 80S ribosome; and instead, the nucleotide sequence following the PTC determines the site of presumed stalling and XRN1 inhibition. Other experiments show that the sequence at the site of XRN1 inhibition is not essential for inhibition, but that the accumulation of the decay intermediates is reduced by mutations that would affect mRNA contacts with the predicted stalled ribosome, and can be completely prevented if the distance between the PTC and the stalled site is reduced to the point where the presumed stalled ribosome would collide with the termination complex at the PTC. The latter finding provides some of the best evidence that XRN1 is inhibited by a stalled ribosome rather than by an RNA-binding protein, as the latter would have a much smaller footprint on the mRNA. They further show that the nature of the UPF1 mutation strongly influences the position of the putative stalled ribosome downstream of the PTC, as only mutations that impair ATP hydrolysis and not ATP binding yield the XRN1-resistant decay products, and various active site mutants affecting hydrolysis in different ways produce somewhat different protection patterns. They propose that when hydrolysis of bound ATP is blocked by particular UPF1 mutations, ribosome recycling by ABCE1/RLI1 following termination at the PTC is inhibited and post-termination 80S ribosomes subsequently migrate (without translating) downstream, stall at particular sequences, and thereby block 5'-3- degradation by XRN1. They further propose that in wild-type cells, UPF1 normally promotes RLI1 function in ribosome recycling in a manner dependent on conformational changes to the protein induced by ATP hydrolysis.

Essential revisions:

1) Perform additional experiments to bolster the conclusion that "upon collision with a terminating ribosome, the stalled ribosome is displaced and does not re-establish a stable interaction with the downstream RNA", for example, by constructing deletions with increasing length as in Figure 4D for two other PTCs.

2) Provide control experiments using primer extension to confirm that the protected fragments are not detected downstream of a normal termination codon and are uniquely found downstream of PTCs.

3) Revise the manuscript to address as many as possible of the other major and comments or raised by the reviewers.

Reviewer #1:

In this report, the authors study decay intermediates from a reporter mRNA harboring a PTC that accumulate in a UPF1 ATPase mutant, but not in the absence of UPF1 protein, and which were shown previously to originate from inhibition of XRN1 5' to 3' exonucleolytic decay by ribosomes on the mRNA. Here they show that the 5' ends of these intermediates map downstream of the PTC rather than coinciding with the PTC itself as had been proposed previously. Thus, rather than providing evidence of a ribosome stalled at the PTC owing to inefficient termination; they instead appear to represent an 80S ribosome that has migrated from the PTC and become stalled at specific positions downstream in the UPF1 ATPase mutant. CoIP analysis of tagged Rps13 shows that these decay fragments are ribosome bound. Other evidence suggests that the protecting ribosomes are not engaged in either translational read-through of the PTC, or from reinitiation of translation 3' of the PTC. Their results further indicate that distance from the PTC is not a critical factor in determining the site of the block to 5' -3' decay presumed to arise from the stalled 80S ribosome; and instead, the nucleotide sequence immediately following the PTC determines the site of presumed stalling and XRN1 inhibition. Other experiments show that the sequence at the site of XRN1 inhibition is not essential for inhibition, but that the accumulation of the decay intermediates is reduced by mutations that would affect mRNA contacts with the predicted stalled ribosome, and can be completely prevented if the distance between the PTC and the stalled site is reduced to the point where the presumed stalled ribosome would collide with the termination complex at the PTC. The latter finding provides some of the best evidence that XRN1 is inhibited by a stalled ribosome rather than by an RNA-binding protein, as the latter would have a much smaller footprint on the mRNAs. They further show that the nature of the UPF1 mutation strongly influences the position of the putative stalled ribosome downstream of the PTC, as only mutations that impair ATP hydrolysis and not ATP binding yield the XRN1-resistant decay products, and various active site mutants affecting hydrolysis in different ways produce somewhat different protection patterns. They propose that when hydrolysis of bound ATP is blocked by particular UPF1 mutations, ribosome recycling by ABCE1/RLI1 following termination at the PTC is inhibited and post-termination 80S ribosomes subsequently migrate (without translating) downstream, stall at particular sequences, and thereby block 5'-3- degradation by XRN1. They further propose that in wild-type cells, UPF1 normally promotes RLI1 function in ribosome recycling in a manner dependent on conformational changes to the protein induced by ATP hydrolysis.

The experiments are carefully designed and well-executed and the paper is well written. The results largely justify the model that the ATPase-defective UPF1 mutants somehow interfere with ribosome recycling by RLI1 following termination at the PTC, allowing the unrecycled 80S ribosomes to migrate into the 3'UTR and stall at particular sequences. While it is reasonable to propose that wild-type UPF1 would stimulate recycling by RLI1, it's unclear why UPF1 mutants defective for ATP binding (vs. hydrolysis), nor the absence of UPF1, do not give rise to scanning unrecycled 80S ribosomes that stall at the same sequences 3' of the PTC. It's also not clear whether the proposed activity of UPF1 in stimulating RLI1 recycling function is important for NMD; although it would be surprising if it were not.

1) It seems feasible to test the model that ribosome recycling by ABCE1/RLI1 is impaired by the UPF1 ATPase mutants by determining whether depleting RLI1 from WT UPF1 cells, using a published degron mutant, gives rise to the same reporter decay intermediates.

2) It would enhance the story if they were able to map the 3' ends of the presumed stalled 80S ribosomes downstream of the PTC in order to confirm protection of 3'UTR mRNA sequences by entities with a ribosome-size footprint. They tried conventional ribosome profiling to no avail, but perhaps crosslinking could be attempted in an effort to stabilize the putative scanning ribosome.

Reviewer #2:

This manuscript from the Baker lab describes detailed characterization of RNA decay intermediates arising from expression of ATPase-deficient mutants of UPF1 in S. cerevisiae. Building on their earlier work, they find that expression of UPF1 that can bind but not hydrolyze ATP causes accumulation of 80S ribosome-bound fragments with 5' termini at various sites downstream of NMD-inducing termination codons. The authors use several reporters to map the fragment 5' ends, finding that a set of generally well-defined fragments was generated from each reporter, in a manner dependent on the sequence context downstream of the termination codon. While overall rules governing this process remain unclear, they present compelling data that residues positioned at the E/P/A sites of the ribosome are the major determinants of 5' end position. Further, in contrast to various manipulations that alter translation termination and ribosome recycling, they find no evidence that the 80S ribosomes bound downstream of TCs engage in reinitiation or readthrough. Many questions remain, but this work presents convincing evidence that ATPase-deficient UPF1 can impinge on the process of ribosome recycling downstream of PTCs. For that reason, it is a valuable contribution to the field's understanding of roles of NMD proteins at the terminating ribosome.

Due to the current COVID-19 situation, I am focusing on whether the authors’ claims are substantiated by the existing evidence and am limiting my requests to those which do not require further experimentation.

Essential revisions:

1) Figure 4D: Based on deletion analyses which result in the disappearance of bands corresponding to decay fragments, without corresponding appearance of a band elsewhere, the authors conclude that, "upon collision with a terminating ribosome, the stalled ribosome is displaced and does not re-establish a stable interaction with the downstream RNA." I think the general applicability of this conclusion is unclear without more information about why some sequences do not support robust fragment generation. What if, for example, a reporter was constructed in which the sequence used in Figure 1—figure supplement 1D that generates a +31 fragment (with additional bands at +43 and +51) was inserted downstream of the 15 nt deletion reporter sequence? To be clear, I'm not asking for this experiment to be completed-just offering it as an illustration that alternatives to the stated conclusion are possible.

2) I suggest moving Supplementary Figure 3A to the main text so all assays of protein expression in the experiment in Figure 3B can be evaluated as a whole. Likewise, incorporating the model and RNA abundance measurements from Supplementary Figure 5 to the main Figure 5 would make it easier to interpret the data currently in Figure 5. Also, measures of error and stats should be provided if possible.

3) Figure 4A: I think the statement that "the nucleotide distance from the PTC is not a critical factor in determining how XRN1 progression is impeded" is not fully supported. Based on the overall data presented, position likely plays some role, as the products characterized are in a small range downstream of the TC and it is clear that certain positions are incompatible with cleavage or protection of the resulting fragment. It would be better to state that fragment accumulation is likely determined by a combination of sequence and position.

Reviewer #3:

This manuscript represents a follow-up of the work presented in Serdar et al., 2016, in which the showed that expressing of an ATPase-deficient UPF1 mutant in yeast caused the accumulation of 3´RNA decay fragments of NMD-sensitive RNAs whose size was coincident with a block of XRN1-dependent 5´-to-3´exonucleolytic degradation around the PTC position. Based on their data, they concluded that UPF1-mediated ATP hydrolysis was required for ribosome disassembly at PTCs.

Here they now used primer extension assays to map the 5´end of these RNA fragments originating from a GFP^PTC125 reporter gene and detect two bands 20 and 17 nucleotides downstream of the PTC (Figure 1). The 5´ ends of the decay fragments from GFP reporters with PTCs in two different positions also mapped downstream of the PTC, but with different distances and band patterns: GFP^PTC67 showed bands around +20 and +12, and GFP^PTC135 many bands at different positions between +51 and +9 (Figure 1—figure supplement 1). A control I very much would like to see here: Are such fragments also detected downstream of the normal GFP stop codon in the GFP WT reporter when UPF1 fails to hydrolyze ATP? In other words, is this phenomenon PTC-specific?

Immunoprecipitation of tagged ribosomes co-precipitated mainly full-length reporter RNA but also the 3´decay fragment in cell expressing the UPF1 DE572AA mutant, suggesting that these RNA fragments are associated with ribosomes (Figure 2 and Figure 2—figure supplement 1).

How do the ribosomes get to these positions downstream of the stop codon? The authors could show that this seems not to occur due to readthrough, because (i) no readthrough was detected in cells expressing the UPF1 mutant and (ii) introducing a frameshift did not change the pattern of these bands (Figure 3 and Figure 3—figure supplement 1). The mapped 5´ends of these 3´decay fragments do neither occur at a specific distance from the PTC nor depend on the identity of the stop codon but rather depend on the sequence context downstream of the PTC (Figure 4 and Figure 4—figure supplement 1).

While the experiments shown in Figure 4 and Figure 4—figure supplement 1 aim at gaining some mechanistic insight into how these XRN1-resistant decay fragments might arise, they appear to stop halfway. To corroborate the conclusion that the sequence context is important, the single nucleotide substitution approach (Figure 4C) should be extended to the other two PTCs and critical residues and some sort of a "consensus sequence" might be revealed by comparing the results from all three contexts. Similarly, deletions with increasing length as in Figure 4D should also been tested for the other two PTCs to further corroborate the hypothesis that upon collision with the terminating ribosomes positioned at the PTC, these 3´ fragments disappear.

Figure 5 finally further corroborates a finding already reported in the previous paper: UPF1 mutants impaired in ATP binding do not cause the appearance of the 3´ fragments, while mutants impaired in ATP hydrolysis do. Interestingly, among the many UPF1 mutants tested, including those that promote rather than inhibit ATP hydrolysis, there was no correlation between appearance of the 3´ fragment and impairment of NMD. Thus, we do not really gain more insight with these mutants regarding the mechanism by which the 3´decay fragments accrue.

In summary, the authors convincingly document the accumulation of ribosome-associated 3´ RNA decay fragments in cells expressing ATPase-deficient UPF1 whose 5´ends map downstream of the stop codon, but the fail to provide much mechanistic insight for this unexpected finding. The presented model is plausible but remains quite speculative without further data explaining the proposed connection between UPF1 ATP hydrolysis and RLI1's function in splitting the ribosomal subunits.

Essential revisions:

1) Perform additional experiments to bolster the conclusion that "upon collision with a terminating ribosome, the stalled ribosome is displaced and does not re-establish a stable interaction with the downstream RNA", for example, by constructing deletions with increasing length as in Figure 4D for two other PTCs.

We appreciate this suggestion and admit that we did not perform deletion analysis on the 3’ UTRs of additional PTC-containing mRNA reporters to address whether decay intermediates are also lost when the position of the 5’ end is brought within the footprint of a ribosome positioned at the PTC. We felt that the step-wise loss of the two primer extension (PE) products for GFP^PTC125 mRNA shown in Figure 4D was particularly convincing and clearly showed that the impediment residing at the 5’ end of the decay intermediate that blocks XRN1 progression is eliminated when this position is within 12 nucleotides of the PTC (with no evidence for a new site for XRN1 blockage arising downstream).

We proposed the most parsimonious explanation of our data and believe that our conclusion is well justified, and point out that reviewer 1 appears to agree – ‘…accumulation of the decay intermediates is reduced by mutations that would affect mRNA contacts with the predicted stalled ribosome, and can be completely prevented if the distance between the PTC and the stalled site is reduced to the point where the presumed stalled ribosome would collide with the termination complex at the PTC. The latter finding provides some of the best evidence that XRN1 is inhibited by a stalled ribosome rather than by an RNA-binding protein, as the latter would have a much smaller footprint on the mRNAs’. Notwithstanding, we have revised the text associated with this result to reflect that additional experiments are needed to confirm the generality of this conclusion.

2) Provide control experiments using primer extension to confirm that the protected fragments are not detected downstream of a normal termination codon and are uniquely found downstream of PTCs.

We agree that this experiment represents a critical control to show ATP-deficient UPF1 impinges exclusively on translation termination events that are ‘premature’. We direct the reviewers to our previous work characterizing this UPF1 mutant (Serdar et al., 2016) where we performed this precise experiment. We evaluated whether 3’ RNA fragments accumulated downstream of a normal translation termination site in UPF1 ATPase mutants using a reporter lacking a PTC, and did not detect such fragments (see Figure 1C in Serdar et al.). Based on that result, we concluded that accumulation of the 3’ RNA fragments (that we later showed were ‘decay intermediates’) are dependent on premature translation termination. Admittedly, in that experiment, RNA from the non-PTC containing reporter were analyzed by Northern blot and not primer extension, but notably, we have never detected cDNA products using primer extension that were not previously revealed by Northern blot. Notwithstanding, those data show that 3’ RNA decay fragments do not accumulate downstream of a normal termination codon.

3) Revise the manuscript to address as many as possible of the other major and comments or raised by the reviewers.

See responses below.

Reviewer #1:

[…] The experiments are carefully designed and well-executed and the paper is well written. The results largely justify the model that the ATPase-defective UPF1 mutants somehow interfere with ribosome recycling by RLI1 following termination at the PTC, allowing the unrecycled 80S ribosomes to migrate into the 3'UTR and stall at particular sequences. While it is reasonable to propose that wild-type UPF1 would stimulate recycling by RLI1, it's unclear why UPF1 mutants defective for ATP binding (vs. hydrolysis), nor the absence of UPF1, do not give rise to scanning unrecycled 80S ribosomes that stall at the same sequences 3' of the PTC. It's also not clear whether the proposed activity of UPF1 in stimulating RLI1 recycling function is important for NMD; although it would be surprising if it were not.

1) It seems feasible to test the model that ribosome recycling by ABCE1/RLI1 is impaired by the UPF1 ATPase mutants by determining whether depleting RLI1 from WT UPF1 cells, using a published degron mutant, gives rise to the same reporter decay intermediates.

We agree with this reviewer that this a critical experiment to evaluate whether ATPase-deficient UPF1 is functioning to simply preclude/prevent RLI1 activity on a prematurely terminating ribosome. We direct the reviewers again to our previously published work where we depleted RLI1 (using the inducible/repressible GAL promoter) and monitored accumulation of 3’ RNA fragments from our PTC-containing mRNA (Serdar et al., 2016; Supplementary Figure 4D). As reproduced there, we did detect decay intermediates from RLI1-depleted cells, however, these fragments were generally shorter and displayed significantly more heterogeneity in size compared to fragments from UPF1-DE572AA cells (as would be predicted based on ribosome profiling data of RLI1-depleted cells by the lab of Rachel Green at JHU; see Young et al., 2015). It is clear from our experiment that depletion of RLI1 from cells does not phenocopy the impact of ATPase-deficient UPF1 on prematurely terminating ribosomes. We discussed these previous observations in our current manuscript and used them to argue that UPF1 ATPase mutants likely work by impinging on RLI1 function (and not preventing RLI1 function; Discussion, fourth paragraph). Consistent with this interpretation, we also mention that over-expression of RLI1 from a high-copy plasmid does not reduce accumulation of (or alter the nature of) the 3’ RNA decay fragments that accrue in UPF1-DE572AA cells – as might be expected if mutant UPF1 was, for example, competing against RLI1 for binding to the terminating ribosome.

Author response image 1

Download asset Open asset

Incidentally, despite these results reported in 2016, we did perform PE on RLI1-depleted cells containing our PGK^PTC125 reporter mRNA and show these data as Author response image 2. As you will see, depletion of RLI1 did not give rise to the +17 and +20 cDNA products that we see in the presence of UPF1-DE572AA. In fact, we fail to see accumulation of strong cDNA products at any specific position despite clearly detecting full length (FL) cDNA, consistent with the low level and heterogenous distribution of 3’ UTR ribosomes in RLI1-depleted cells detected by ribosome profiling (Young et al., 2015). You may note that we observe a reduction in the level of the 3’ decay intermediates found in the UPF1 ATPase mutant during the RLI1 depletion time course (note it takes ~6 hrs to achieve 90% depletion of Rli1p in our cells; Serdar et al., 2016). While it is tempting to suggest that this indicates a requirement for RLI1 in the accumulation of the 3’ decay fragments – which would be completely supportive of our model – we note that in these cells, ribosomes are amassing globally on mRNA 3’ UTRs and being depleted from the cytoplasmic pool. Given that this would be predicted to reduce general translation and that 3’ decay intermediate accumulation requires translation of the PTC-containing mRNA (as specifically shown in Serdar et al., 2016), we feel that this experiment neither adds to our observation reported in 2016 (that 3’ RNA decay fragments are distinct in UPF1-DE572AA versus RLI1-depleted cells) nor provides definitive support for our model, so we prefer to exclude this data from our current submission.

Author response image 2

Download asset Open asset

2) It would enhance the story if they were able to map the 3' ends of the presumed stalled 80S ribosomes downstream of the PTC in order to confirm protection of 3'UTR mRNA sequences by entities with a ribosome-size footprint. They tried conventional ribosome profiling to no avail, but perhaps crosslinking could be attempted in an effort to stabilize the putative scanning ribosome.

We appreciate this suggestion and whole-heartedly agree that this data would provide additional direct evidence in support of our interpretation that a stalled 80S ribosome resides at the 5’ end of the 3’ decay intermediate. We attempted to gather this evidence in two different ways. First, as mentioned, we attempted ribosome profiling which aimed to define the precise positioning of ribosomes on the 3’ decay fragments. Despite extensive experience with this method (see for example Smith et al., 2014), we failed to map any ribosome protected fragments associated with the 3’ decay fragment – this is despite deep library coverage and strong biochemical evidence that one or more is bound to this intermediate (i.e. polyribosome analysis and ribosome IP). We think this our inability to map the ribosome footprints by this method is likely due to dissociation of the ribosomes during the extensive in vitro processing steps associated with our rigorous ribosome profiling protocol that includes recovering ribosome-protected fragments created by RNase digestion from the monosome fraction of sucrose gradients. We did not attempt ribosome profiling using cells that were subjected to crosslinking, and are not aware of any published method or report where this has been successfully performed.

Second, we attempted toeprint analysis which employs PE as a means to map the 3’ most position of an RNA-bound protein. We endeavored to detect PE products for GFP^PTC125 mRNA from UPF1-DE572AA cell lysates with or without chemical crosslinking (formaldehyde at a concentration and for a duration we have previously shown efficiently crosslinks proteins to RNA without causing extensive inter-transcript complex formation; see Smith and Baker, Methods Mol Biol 2017). Unfortunately, data was messy and the results inconclusive owing from what we believe are challenges associated with generating cDNA products from an mRNA template that is bejeweled with covalently bound RNA-binding proteins (and only a single mRNA population in a sea of transcripts in the cell extract). We note that in our assessment, toeprint analysis has been most successfully employed using nuclease-treated cell extracts and in vitro transcribed templates, which we do not confidently feel would recapitulate NMD-mediated decay. We would be delighted, however, if our reviewers can point us to any alternative approaches that we might try to get at this precise question.

Reviewer #2:

[…]

Essential revisions:

1) Figure 4D: Based on deletion analyses which result in the disappearance of bands corresponding to decay fragments, without corresponding appearance of a band elsewhere, the authors conclude that, "upon collision with a terminating ribosome, the stalled ribosome is displaced and does not re-establish a stable interaction with the downstream RNA." I think the general applicability of this conclusion is unclear without more information about why some sequences do not support robust fragment generation. What if, for example, a reporter was constructed in which the sequence used in Figure 1—figure supplement 1D that generates a +31 fragment (with additional bands at +43 and +51) was inserted downstream of the 15 nt deletion reporter sequence? To be clear, I'm not asking for this experiment to be completed-just offering it as an illustration that alternatives to the stated conclusion are possible.

See response to this inquiry under Essential revisions (point 1) above.

2) I suggest moving Supplementary Figure 3A to the main text so all assays of protein expression in the experiment in Figure 3B can be evaluated as a whole. Likewise, incorporating the model and RNA abundance measurements from Supplementary Figure 5 to the main Figure 5 would make it easier to interpret the data currently in Figure 5. Also, measures of error and stats should be provided if possible.

We thank the reviewer for these suggestions and have moved Supplementary Figures 3A and 5B to their respective main figures (and added required text to the figure legends and Results section accordingly). We note that we have performed Northern blot analysis of CYH2 transcripts on RNA isolated from the various UPF1 active site mutants only twice; the values provided in the figure are for the image shown, but we have added to the figure legend that this experiment has been done in duplicate and that the abundance of CYH2 pre-mRNA (the NMD substrate) normalized to CYH2 mRNA (NMD insensitive) differs less than 10% between experiments.

3) Figure 4A: I think the statement that "the nucleotide distance from the PTC is not a critical factor in determining how XRN1 progression is impeded" is not fully supported. Based on the overall data presented, position likely plays some role, as the products characterized are in a small range downstream of the TC and it is clear that certain positions are incompatible with cleavage or protection of the resulting fragment. It would be better to state that fragment accumulation is likely determined by a combination of sequence and position.

The cumulative data from our various PTC-containing PGK1 mRNA reporters show that the 5’ termini of the 3’ RNA decay fragments are found at a number of distances downstream of the PTC (e.g. +17, +20, variable distances between +25 and +31, +43, +51) clearly indicating distance from the PTC alone does not determine where unrecycled ribosomes stall and block to XRN1 progression. However, we cannot rule out that there is some influence of distance (as indicated by the reviewer, some positions are incompatible [due to their proximity to the PTC] and the range of observed stall sites is restricted to the limited number of PTC-containing mRNA we analyzed); given this we have modified the text to more clearly reflect this limitation of our data.

Reviewer #3:

This manuscript represents a follow-up of the work presented in Serdar et al., 2016, in which they showed that expressing of an ATPase-deficient UPF1 mutant in yeast caused the accumulation of 3´RNA decay fragments of NMD-sensitive RNAs whose size was coincident with a block of XRN1-dependent 5´-to-3´exonucleolytic degradation around the PTC position. Based on their data, they concluded that UPF1-mediated ATP hydrolysis was required for ribosome disassembly at PTCs.

Here they now used primer extension assays to map the 5´end of these RNA fragments originating from a GFP^PTC125 reporter gene and detect two bands 20 and 17 nucleotides downstream of the PTC (Figure 1). The 5´ ends of the decay fragments from GFP reporters with PTCs in two different positions also mapped downstream of the PTC, but with different distances and band patterns: GFP^PTC67 showed bands around +20 and +12, and GFP^PTC135 many bands at different positions between +51 and +9 (Figure 1—figure supplement 1).

A control I very much would like to see here: Are such fragments also detected downstream of the normal GFP stop codon in the GFP WT reporter when UPF1 fails to hydrolyze ATP? In other words, is this phenomenon PTC-specific?

Immunoprecipitation of tagged ribosomes co-precipitated mainly full-length reporter RNA but also the 3´decay fragment in cell expressing the UPF1 DE572AA mutant, suggesting that these RNA fragments are associated with ribosomes (Figure 2 and Figure 2—figure supplement 1). How do the ribosomes get to these positions downstream of the stop codon? The authors could show that this seems not to occur due to readthrough, because (i) no readthrough was detected in cells expressing the UPF1 mutant and (ii) introducing a frameshift did not change the pattern of these bands (Figure 3 and Figure 3—figure supplement 1). The mapped 5´ends of these 3´decay fragments do neither occur at a specific distance from the PTC nor depend on the identity of the stop codon but rather depend on the sequence context downstream of the PTC (Figure 4 and Figure 4—figure supplement 1). While the experiments shown in Figure 4 and Figure 4—figure supplement 1 aim at gaining some mechanistic insight into how these XRN1-resistant decay fragments might arise, they appear to stop halfway.

See our response to this particular comment in the Essential revisions (point 1) section above.

To corroborate the conclusion that the sequence context is important, the single nucleotide substitution approach (Figure 4C) should be extended to the other two PTCs and critical residues and some sort of a "consensus sequence" might be revealed by comparing the results from all three contexts.

We agree with the reviewer that determination of a consensus sequence for which 3’ UTR ribosomes prefers to stall could provide exciting mechanistic insight into how mutant UPF1 impinges on RLI1 and/or the prematurely terminating ribosome. We had this very idea in mind as we interrogated the sequence within 50 nts of the 5’ termini of the decay fragments that accumulate for our various PTC-containing reporter transcripts – but were unable to identify either a sequence motif or putative RNA secondary structure. While we agree that analysis of a battery of additional point mutations within this region for a number of our reporters might help to identify such a consensus, we have opted to apply a more genomic approach where libraries of transcripts with randomized sequences within the +17 and +20 regions of the PGK1^PTC125 mRNA reporter will be correlated with the accumulation of 3’ RNA decay fragments in vivo and measurement of 5’ ends by RNA-Seq. We think this unbiased and more comprehensive analysis of the sequence landscape around the site of the stalled ribosomes is much more likely to provide insight into the sequence requirements for ribosome stalling within the 3’ UTR. This experiment has been delayed due to COVID-19, but we feel is nevertheless beyond the scope of this current manuscript; we hope that the reviewer agrees. We have altered the text associated with Figure 4C to state that additional analyses are required to establish how sequence context influences the site of 3’ UTR ribosome stalling in our UPF1 ATPase mutants.

Similarly, deletions with increasing length as in Figure 4D should also been tested for the other two PTCs to further corroborate the hypothesis that upon collision with the terminating ribosomes positioned at the PTC, these 3´ fragments disappear.

See our response to this particular comment in the Essential revisions (point 2) section above.

Author details

1. Lucas D Serdar

   1. Department of Genetics and Genome Sciences, Case Western Reserve University School of Medicine, Cleveland, United States
   2. Department of Biochemistry, Case Western Reserve University School of Medicine, Cleveland, United States

   Contribution

   Formal analysis, Validation, Investigation, Writing - original draft

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4574-1668

2. DaJuan L Whiteside

   Department of Genetics and Genome Sciences, Case Western Reserve University School of Medicine, Cleveland, United States

   Contribution

   Formal analysis, Validation, Investigation

   Competing interests

   No competing interests declared

3. Sarah L Nock

   Department of Genetics and Genome Sciences, Case Western Reserve University School of Medicine, Cleveland, United States

   Contribution

   Investigation

   Competing interests

   No competing interests declared

4. David McGrath

   Department of Genetics and Genome Sciences, Case Western Reserve University School of Medicine, Cleveland, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

5. Kristian E Baker

   1. Department of Genetics and Genome Sciences, Case Western Reserve University School of Medicine, Cleveland, United States
   2. Department of Biochemistry, Case Western Reserve University School of Medicine, Cleveland, United States

   Contribution

   Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Project administration, Writing - review and editing

   For correspondence

   kristian.baker@case.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2262-5434

• 1,501

  Page views

• 233

  Downloads

• 7

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.