Genetic screens identify connections between ribosome recycling and nonsense mediated decay

The decay of messenger RNA with a premature termination codon (PTC) by nonsense mediated decay (NMD) is an important regulatory pathway for eukaryotes and an essential pathway in mammals. NMD is typically triggered by the ribosome terminating at a stop codon that is aberrantly distant from the poly-A tail. Here, we use a fluorescence screen to identify factors involved in NMD in S. cerevisiae. In addition to the known NMD factors, including the entire UPF family (UPF1, UPF2 and UPF3), as well as NMD4 and EBS1, we identify factors known to function in post-termination recycling and characterize their contribution to NMD. We then use a series of modified reporter constructs that block both elongating and scanning ribosomes downstream of stop codons and demonstrate that a deficiency in recycling of 80S ribosomes or 40S subunits stabilizes NMD substrates. These observations in S. cerevisiae expand on recently reported data in mammals indicating that the 60S recycling factor ABCE1 is important for NMD (1, 2) by showing that increased activities of both elongating and scanning ribosomes (80S or 40S) in the 3’UTR correlate with a loss of NMD. Author Summary In this work, we aim to understand the mechanism of targeting mRNAs for decay via the long-studied nonsense mediated decay (NMD) pathway. We demonstrate that efficient large and small subunit ribosome recycling are necessary components of NMD. We go on to provide evidence that either scanning or actively translating ribosomes in the 3’UTR disrupt the decay of NMD targets. Our work highlights the importance of the composition of the 3’UTR in NMD signaling and emphasizes the need for this region to indeed be untranslated for NMD to occur. Exon junction complexes (EJCs) in the 3’UTR are known to induce NMD, however, in the budding yeast system used here, the NMD targets are EJC-free. Therefore, our data support a model in which factors other than EJCs may accumulate in the 3’UTR and provide a signal for NMD.

6 99 reasoned that important players in the pathway might remain undiscovered and 100 could shed light on molecular mechanism.

101
Here, using a bi-directional fluorescent reporter in S. cerevisiae, we screen for 102 factors that contribute specifically to decay of the mRNA or its translation 103 repression during NMD. Along with the known NMD regulators in yeast, we 104 identify a group of genes involved in translation termination and recycling. We

134
To confirm that the decreases in RNA levels for the reporter reflect NMD, we 135 deleted the core NMD genes, UPF1, UPF2 and UPF3, and saw a restoration of 136 RNA levels for the NMD reporter by both flow cytometry and northern blots 137 ( Figure 1B and 1C). As expected, loss of these factors have no effect on the OPT 138 reporter. Interestingly, in the NMD-reporter strains we saw that GFP RNA levels 139 were fully restored by deletion of UPF1/2/3 but GFP protein levels were not 8 140 (compare Figure 1B and  reporter activity to be scored in each deletion mutant. We performed independent 9 163 calculated. In this setting Z-scores represent the deviation of the GFP/RFP ratio 164 for a given strain from the mean GFP/RFP ratio of the entire array; these data for 165 the OPT and NMD reporter strains are plotted against one another in Figure 2A 166 (raw data is given in Suppl. Table 1 and 2). From the NMD screen, there were 76 167 hits with a Z-score above 2.0 and 94 hits with a Z-score below -2.0 ( Figure 2B,

172
However, GO enrichment data show that the hits in the NMD screen are strongly 173 enriched in mRNA metabolic processing and nuclear-transcribed mRNA 174 catabolic process nonsense-mediated decay, while there were no GO terms for 175 the hits in the OPT screen with a similar level of enrichment ( Figure 2B).

176
Furthermore, UPF1/2/3 deletion strains exhibited some of the largest deviations 177 from the mean in the NMD reporter strains, immediately validating the potential of 178 the screen ( Figure 2A). Interestingly, NMD4 and EBS1 deletions are also strong 179 outliers that cause an increase in GFP signal for the NMD reporter ( Figure 2A).

184
As a validation, we re-transformed the NMD reporter into deletion strains from 185 the deletion mutant collection corresponding to the NMD screen hits with a Z-score less than -2.0 or greater than 2.0. In this list, we excluded 29 genes that 187 were also hits in the OPT screen, yielding 113 genes. We individually analyzed 188 these 113 newly constructed strains by flow cytometry assays and plotted the 189 fold change of each candidate gene deletion relative to a HIS3-deletion control 190 ( Figure 2C and Suppl.  Figure 2A), which was shown to affect NMD in 198 mammalian cells (32). As constructed, the YOR309C deletion results in a 199 truncation of NOP58 and uncharacterized effects on its 3' UTR (due to insertion 200 of a kanamycin cassette) (Suppl. Figure 2B). We therefore asked if truncation or 201 loss of NOP58 resulting from deletion of the YOR309C was responsible for 202 blocking NMD that we observed in our screen (rather than from deletion of 203 YOR309C). To test this, we complemented the yor309c∆ (NOP58 truncated) 204 strain with either full length Nop58 (Nop58-full) or the truncated version (Nop58-205 trunc) and evaluated the level of NMD by flow cytometry (Suppl. Figure 2C).

206
Overexpression of either full length or truncated Nop58 restored NMD for the 207 reporter (Suppl. Figure 2C). These data show that Nop58 truncation is not 208 responsible for the NMD phenotype, consistent with studies demonstrating the 11 209 functionality of Nop58 truncation proteins (33), and instead indicating that 210 integration of KanMX at the YOR309C locus affects NOP58 expression or Nop58 211 mRNA or protein stability.

212
We next decided to focus on genes that were thought to be involved in 213 ribosome termination and/or recycling. We reconstructed the TMA20-and

246
Because TMA20, TMA22, and TMA64 deletions have been shown to function 247 in 40S ribosome recycling, we next asked if NMD was also affected by disrupting 248 60S ribosome recycling by deleting the Rli1-accessory factor gene HCR1. While 249 the hcr1Δ strain did not survive the initial screen due to a general growth defect, 250 this deletion indeed strongly stabilizes the NMD reporter ( Figure 3A and 3B).

251
These data together lead to the hypothesis that deficiencies in both 40S and 60S 252 ribosome recycling can disrupt NMD.

253
We next assessed the effects of these same genes on a well characterized 254 endogenous target of NMD, the rare intron-retaining transcript of CYH2 (35  Figure 4A).

286
The first reporter contained a second PTC <20 nucleotides downstream of   Figure 4A). In WT cells, addition of 303 this sequence to the 3'UTR had no impact on reporter levels ( Figure 4C).

304
However, we also saw no effect of this sequence in the 3'UTR for the hcr1Δ 305 strains and only a modest decrease in the tma20Δ tma64Δ strain, in which NMD

306
was not completely restored ( Figure 4C). These data suggested that either an 307 unimpeded ribosome population could still be accessing the 3' UTR and 308 diminishing NMD, or that the ribosome recycling factors are affecting NMD 309 through another mechanism.

310
In a final attempt to block ribosome access to the 3' UTR, we added a  (Figure 4). These data argue that efficient ribosome 364 recycling is required to maintain a ribosome-free 3'UTR for robust NMD

518
Based on this LOESS normalized value, Z-scores were calculated. Strains for the 519 NMD reporter with a Z-score greater than 2.0 or less than -2.0 were considered a 520 hit if their Z-score in the OPT reporter was not also an outlier.

522
The Venn diagram was created using BioVenn and the input was deletion strains 523 with a Z-score greater than 2.0 or less than -2.0.

525
Go analysis was performed with the data from the screen using a list of hits that 526 were in the cutoff of Z > 2.0 or Z < -2.0 and the input gene list is the genes that 527 GFP/RFP data was successfully acquired for in the deletion array screen. The P 528 value cutoff was set to < 10^-7 using GOrilla (56,57). Duplicate strains were 529 removed during analysis.

771
The GFP/RFP signal for each strain, normalized to the WT strain with the OPT reporter is plotted.