Cytoplasmic Rat1 can function co-translationally in mRNA 5’→3’ exonuclease, albeit inefficiently
A quintessential function of Xrn1 is to degrade decapped mRNA during its last round of translation. This helps terminating translation and contributes to half-lives of many mRNAs (see Introduction). We next investigated whether cRat1 could perform ribosome-associated co-translational decay. For this end, we measured the presence of 5’-Phosphate-containing mRNAs genome wide by 5’P degradome RNA sequencing (HT-5Pseq) on both sets of four strains: a) the set of strains transformed with a plasmid containing the Rat1-∆NLS version (cRat1) without FLAG epitopes or with the empty vector (Figure 2); and b) the set of strains (wild-type and xrn1∆ ) transformed, or not, with cRat1-3xFLAG (Supplementary Figure S3). We decided to use the FLAG tagged cRat1 as we could be sure about its equal level in all the studied strains (Fig. S3), whereas using the untagged strain was made to make sure that Rat1 function was not compromised.
The wild-type HT-5Pseq metagene profile was characterized by a 3-nucleotide periodicity pattern and a prominent peak of HT-5Pseq reads at -17 nt from the STOP codon (Pelechano et al, 2015). HT-5Pseq measures the distance between the 5′→3′exonuclease and trailing ribosome during co-translational mRNA decay. We have previously shown that the distance between the 5’P of the mRNA undergoing degradation and the first base of the termination codon is 17 nt at the A site of the trailing ribosome (Pelechano et al, 2015; Zhang & Pelechano, 2021). This distance reflects the steric protection offered by the ribosome to the in vivo trimming action of Xrn1p. In contrast to wild-type the -17 nt peak was completely lost in xrn1∆ and, in the 5’UTR, reads accumulated in the zone immediately upstream of the AUG codon (Figure 2A). cRat1 partially restored the wild-type metagene profile to show intermediate levels for the 5’-upstream and -17 nt peaks (Fig. 2 left and right panels, respectively). An approach to evaluate genuine ribosome occupancy was to determine how well the coding frame was protected from 5’ to 3’ decay in relation to the other two frames. The ratio between the protected frame and the other frames was designated as the “frame protection index” (FPI) (Pelechano et al, 2015). This index measures the effectiveness of single-nucleotide coupling between nuclease activity and ribosome position. As expected, the FPI was compromised inxrn1∆ cells (Figure 2B). The expression of cRat1 in thexrn1Δ cells only partially recovered the FPI (from 0.79 inxrn1Δ increased to 0.82 in cRat1, compared to value 1 for the wild-type FPI; see Figure 2B). This finding suggests that Rat1 is inefficient, closely following the ribosome position during co-translational mRNA decay when placed in the cytoplasm. This might reflect either differences in the enzyme localization, configuration or processivity.
Given that HT-5Pseq reads accumulate at the 3’ (close to the STOP codon) in wild-type cells and shifted to 5’ (close to the AUG codon) inxrn1∆ , we reasoned that the proportion of reads in these two regions could serve as a proxy for the processivity of the 5’→3’ decay level once decay has started, taken as either a genome-wide average or a per-gene index. To calculate this 3’ vs. 5’ proportion for the wild-type, xrn1∆ and cRat1 strains, we first made metagene profiles by spanning the entire gene body (Figure 2C), and then compared the number of reads in the last 20% vs . the first 20% of the average gene body region. These 3’/5’ indices showed that expressing cRat1 in xrn1∆ was half as efficient as the wild-type in 5’→3’ decay (Figure 2D). We next used the per-gene 3’/5’ indices to investigate whether sets of functionally related genes were skewed towards any end of the range. A functional classification analysis of the wild-type revealed that the 250 genes with the highest values in the 3’/5’ index (high 5’→3’ degradation: Figure, 2E upper panel) were enriched in Gene Ontology (GO, Biological Process) terms related to cell cycle and stress response (not shown), whereas the 250 genes with the lowest values (low 5’→3’ degradation: Figure 2E, lower panel) were enriched in GOs related to mitochondria and glucose metabolism, although with low significance (not shown). Next, we generated metagene plots for specific gene sets and observed that while the overall shapes of profiles slightly differed for the various groups, the proportions of reads between samples remained approximately the same in all four yeast strains (Supplementary Figure S3B). This result indicates that the lack of Xrn1 (xrn1∆ ) and its substitution by cRat1 have transcriptome-wide effects. The heatmaps of the individual genes ordered by 5’UTR length (Figure 2F and Supplementary Figure S4) show that the average metagene plots (Figure 2A/C) represent truly uniform behavior for most yeast mRNAs, with two stronger regions at the 5’: being most reads associated with the TSS (see Supplementary Figure S4) and ordered according to 5’UTR length, and a minor one located around the AUG codon as we have previously described for multiple species (Huch et al., 2023). This distribution is clearly observed in the summary average count metagene plots (top panels in Figures 2F and S4). The TSS peak detected by HT-5Pseq was particularly clear for xrn1∆, and slightly less so for cRat1. The position of this peak corresponded to the canonical capped TSS full-length molecules (as measured by 5CapSeq, the rightmost lane in Figures 2F and S4). Altogether, these results suggest that in xrn1∆ , and to a lesser extent in cRat1, a large fraction of the 5’P-detected molecules corresponds to decapped full-length mRNAs.
FPI reflects single-nucleotide coupling between co-translational decay and the ribosome position along the coding region, while the 3’/5’ ratio reflects the processivity of the 5’→3’ exonuclease activity and ribosome protection along mRNA. The 3’/5’ index is mainly influenced by translation initiation and termination, which are the two regions where ribosomes tend to pause the most. Because in the wild-type pausing at the stop codon lasts longer than pausing at the start codon, the general 3’/5’ ratio is expected to be higher than 1. A slow 5’→3’ decay results in a 3’/5’ index lower than 1 (negative Log2 ratio in Figure 2D). Nevertheless, the translation machinery can also protect mRNA from 5’→3’ decay, perhaps because the cap is protected (e.g., by eIF4F). Indeed, highly translated mRNAs are less prone to global co-translational decay (Pelechano et al, 2015) and 5’→3’ decay. In fact, the 3’/5’ index shows a statistically significant inverse correlation with mRNA stability and individual translation rate in wild-type cells (TLRi data from Forés-Martos et al., 2021) (Figure 2G and Supplemental Figure S5). These correlations support the notion that the higher the ribosome density (related to TLRi), the greater the protection against 5’→3’ decay, and the higher the mRNA stability. This effect depends on Xrn1 because the negative correlation disappears in xrn1∆ , which is consistent with Xrn1 being the major 5’→3’ exonuclease (Haimovich et al., 2013; Parker, 2012). The substitution of Xrn1 for cRat1, however, cannot restore ribosome density dependence (Figure S5), consistent with the poor capacity of Rat1 to complement the function of Xrn1 in co-translational decay.