eIF5A and polyamines restrict mRNA levels in response to ribosome stalls

Significance

Cells synthesize a diversity of proteins, and allow for variation in protein synthesis rates while also recognizing and eliminating proteins that take an interminable length of time. Precisely how cells accomplish both tasks is unclear. In this study, [Wohlenberg et al.] describe a role for a core protein synthesis factor (eIF5A) and the small molecule spermidine in helping cells recognize and eliminate prolonged protein synthesis stalls. The work answers a long-standing question and augments the understanding of how cells survey protein quality during synthesis. This work is relevant to biological settings where protein quality declines such as aging, neurodegenerative disorders, and pharmacological inhibition of eIF5A.

Abstract

Obstacles to translation elongation stall ribosomes and allow deleterious proteins to accumulate, which threatens cellular health. Cells recognize and clear stalled ribosomes via several interrelated pathways, although the mechanisms by which cells distinguish stalled from normally elongating ribosomes and mount an appropriate response are incompletely understood. While recent work highlights how ribosome collisions help cells to recognize stalled ribosomes, how other factors contribute to detection remains unclear. Here, we report a requirement for the translational factor eIF5A in the mRNA decay response to ribosomal stalling, i.e., No-Go mRNA Decay (NGD). We identified the Caenorhabditis elegans polyamine transporter, catp-6, via a forward genetic screen as a factor required for NGD. During our mechanistic dissection of the catp-6 phenotype, we uncovered a role for cellular polyamines and the translation elongation factor eIF5A in NGD, and we show this requirement is conserved from C. elegans to Saccharomyces cerevisiae. Our analyses support the idea that cells use eIF5A to identify ribosomal stalls and execute NGD and uncover a molecular function for a core protein synthesis factor in limiting expression from stall-inducing mRNAs. Our work offers insight into how cells identify and remove problematic mRNAs from the translational pool. Our work also raises the possibility that dysregulated mRNA decay is an unrecognized pathophysiology associated with polyaminopathies and eIF5A disorders, of relevance to varied neurodegenerative and aging phenotypes and efforts to pharmacologically inhibit eIF5A.

Translational quality control and associated signaling pathways are critical to manage problems during translation elongation. Loss of translational quality control pathways yields a range of aging- and neurodegenerative diseases (1–6), and modulation of translational stress signaling pathways holds promise to unleash novel therapies (7–10). Thus, insight into the reactions that survey translation elongation is key to develop interventions for human health.

During elongation, ribosomes stall due to the lack of the cognate tRNA, or the presence of ribosome-binding drugs, mRNA damage, and/or mRNA structures or sequences that physically obstruct the ribosome (11–16). One cellular response to prolonged elongation stalls is no-go mRNA decay (NGD), in which cells degrade the offending mRNA (11). More widespread ribosomal stalls are met with global effects such as downregulation of translation initiation, inhibition of growth, and in some cases, programmed cell death (17–21). In addition to repressive stall responses, it is clear that cells also allow transient ribosomal pauses to permit proper nascent chain folding (22), frameshifts (23, 24), protein synthesis at difficult peptides and codon pairs (25–27), and localization of the ribosome to specific subcellular regions (28).

Precisely how cells distinguish long and problematic ribosomal stalls from transient pauses—and how cells calibrate their response—is not well understood. Several studies support a central role for ribosome collisions, which form unique interfaces that are recognized by cellular factors to promote ribosome rescue and mRNA decay (12, 16, 17, 29). A key limiting step in stall management is recognition by ZNF598, a lowly expressed E3 ubiquitin ligase that recognizes stalls and triggers repression (18, 21, 30–32). Precisely how ZNF598 finds its targets is not clear, with some work suggesting a role for ribosome collisions and other work arguing against a strict collision requirement (16, 33, 34). There are many more collisions than ZNF598 binding events (30, 31), which underscores the possibility that additional determinants of stall recognition remain to be described. Whether general translational factors play a role in stall recognition and response is a largely unexplored question.

Among conserved eukaryotic translation elongation factors, eIF5A is the least well-understood. As the “I” in its name implies, eIF5A was originally ascribed a role in translation initiation (35, 36), but later placed in translation elongation (37, 38). eIF5A is the only known eukaryotic protein to contain the amino acid hypusine, which is installed via two enzymatic steps using the polyamine spermidine (39). eIF5A hypusination is essential to eIF5A activity and enhances ribosome binding (37, 40–42). In vitro, eIF5A stimulates peptide bond formation at proline and glycine residues, amino acids with uniquely problematic geometries (43, 44) and in vivo eIF5A functions quite broadly in translation elongation (43–46). eIF5A also stimulates CAT-tailing, a quality control process in which the nascent peptide is elongated with alanines and threonines independent of the small ribosomal subunit (47). eIF5A has also been linked to nonsense-mediated mRNA decay (NMD) (48, 49). Furthermore, eIF5A supports elongation through particularly difficult-to-translate sequences, including lysine-rich polybasic stretches (50). Given the role of eIF5A in aiding the translation of problematic sequences, eIF5A would seem uniquely suited to help cells detect when a translational obstacle is severe enough to induce more drastic measures. However, whether cells employ eIF5A for stall detection or signaling was unknown.

Here, we show that loss of the spermidine transporter CATP-6 (Caenorhabditis elegans’ homolog of human ATP13A2/PARK9) impairs the ribosomal stall response known as NGD. Following this lead, we show that catp-6 animals exhibit hallmarks of decreased eIF5A function, and that eIF5A is also required for NGD. Our identification of eIF5A’s requirement in NGD suggests that eIF5A has specific roles in translation of a subset of mRNAs, in addition to its broad role in protein synthesis. We present data in support of a model in which CATP-6 transports the polyamines necessary for the maturation and hypusination of eIF5A and for NGD. Our work highlights the previously unrecognized role of polyamines and eIF5A hypusination in helping cells repress mRNAs with ribosomal stalls.

Results

CATP-6 Is Required for NGD.

To identify factors that repress mRNAs with ribosomal stalls, we developed and previously described a C. elegans ribosome stalling reporter at the endogenous unc-54 locus (32). Briefly, at the C-terminus of unc-54 we used CRISPR/Cas9 to add a T2A “stop-and-go” peptide, a FLAG tag, a ribosomal stalling sequence, and GFP (Fig. 1 A and B). Since at the time the reporter was made there were no ribosomal stalling reporters in C. elegans, we attempted to stall ribosomes via two mechanisms known from other organisms. First, we encoded a polybasic stretch of amino acids, which can interact with the peptide exit tunnel and stall ribosomes (51, 52). Second, we used rare codons, which can stall ribosomes due to the absence of the corresponding tRNA (11). We satisfied both criteria simultaneously with a stretch of twelve rare arginine codons (CGG, AGG) which are decoded by lowly expressed tRNAs in C. elegans (53, 54). After the stall sequence is GFP, which provides a quantitative, inverse relationship between NGD function and protein production: i.e., more GFP indicates an animal with less NGD. The T2A separates the myosin UNC-54 and FLAG::Arg₁₂::GFP peptides, providing two measures of reporter expression: fluorescent quantification (GFP production) and brood size/body thrashes (UNC-54 production). Together these components make up our primary C. elegans stalling reporter, which we hereafter call the RareArg reporter.

Fig. 1.

CATP-6 loss stabilizes NGD substrates. (A) Simplified representation of NGD genetic screen (32) using the RareArg reporter. The Top depicts the RareArg reporter in which black represents the endogenous unc-54 gene, light gray is the ribosome skipping T2A sequence, dark gray is FLAG, yellow is the stall sequence (12 rare arginines), and green is GFP. Summary of screen hits are shown, with a red box highlighting catp-6, the focus of the current work. (B) Schematic of various NGD reporters. The coloring scheme is the same as shown in A. The stall sequence varies for each reporter, containing 12 rare arginine codons (RareArg reporter), 12 common arginine codons (CommonArg reporter), or 12 lysine codons (PolyLys reporter). (C) Mean relative fluorescence units of the indicated strains in the RareArg reporter background. Bars show mean ± SD and each dot is a single animal. (D) Same as C but in CommonArg reporter background. GFP quantifications have n of 8 to 17 animals/strain. (E) Swimming assay (n = 12 animals/strain) of indicated strains in PolyLys reporter background through quantification of thrashes per minute. unc-54 encodes for a myosin expressed in body wall muscle required for animal movement. (F) RNA-seq fold change (FC) (catp-6(Δ)/WT) of RareArg mRNA level for the indicated strains, for three replicate libraries. *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant. P values determined by the Mann–Whitney U test (C–E) or 1-tailed paired t test (F).

As we reported previously (32), we used the RareArg reporter in a forward genetic screen to identify factors required for NGD in C. elegans (Fig. 1A). A majority of the recovered mutants were in NGD pathway components, but unexpectedly, we also recovered five independent mutants in the polyamine transporter, catp-6 (SI Appendix, Fig. S1). To confirm that catp-6 mutation was causative, we crossed a catp-6 deletion allele into the RareArg reporter and observed derepression of the reporter (Fig. 1C and SI Appendix, Fig. S2A). Thus, CATP-6 is an activator of NGD.

Polybasic Stretches Are NGD Substrates.

To better understand the generality of the catp-6 effect, we first had to understand the stalling requirements for NGD. Thus, we created two additional reporters (Fig. 1B). One of these reporters, hereafter called the CommonArg reporter, contains a sequence of 12 arginines encoded by codons (CGT, CGA, AGA) that are common in C. elegans. The second reporter, hereafter called the PolyLys reporter, contains alternating AAA and AAG codons, both of which encode for lysine and are reasonably abundant codons. In wild-type animals, both the CommonArg and PolyLys reporters exhibited low GFP and an uncoordinated (Unc) phenotype in a movement-based swimming assay, consistent with ribosomal stalling and repression of the unc-54 locus (SI Appendix, Fig. S2 A and B).

We noticed that the arginine-based reporters tended to localize GFP to the nucleus while the PolyLys reporter exhibited a more diffuse localization (SI Appendix, Fig. S2A). Basic-rich peptides localize proteins to the nucleus in eukaryotes (55, 56), and our observation is consistent with the idea that 12Arg but not 12Lys is sufficient for nuclear localization in C. elegans. For the PolyLys reporter, the diffuse GFP proved difficult to accurately quantify, and we instead used unc-54 function as a readout of reporter expression. As unc-54 encodes a myosin important for animal movement and egg laying (57, 58), we measured UNC-54 expression via movement and/or brood size in the PolyLys background.

To test whether the CommonArg and PolyLys reporters were bonafide NGD targets, we measured reporter expression in a known NGD mutant background, znf-598. The ZNF-598 protein is an E3 ubiquitin ligase that ubiquitinates collided ribosomes and enables NGD, and a znf-598 mutant exhibits higher RareArg reporter expression (32). A znf-598 mutant derepressed the CommonArg reporter (Fig. 1D and SI Appendix, Fig. S2A). Since the PolyLys reporter could not be easily quantified via GFP fluorescence, we assessed reporter expression through muscle function, with more muscle function indicative of more UNC-54 protein and less NGD. We measured body wall muscle activity via movement and observed that mutation of znf-598 derepressed the PolyLys reporter (Fig. 1E). We also performed a brood size assay as unc-54 is required in the vulval muscles for egg laying. By brood size analysis, znf-598 increased unc-54 function in the PolyLys reporter background (SI Appendix, Fig. S2C). Thus, we determined both the CommonArg and PolyLys reporters are bonafide NGD targets.

Taken together, these data support the idea that various polybasic peptides stall ribosomes and cause NGD in C. elegans.

Loss of CATP-6 Stabilizes Various NGD Substrates.

We next sought to understand the role of catp-6 in repression of these various NGD substrates. As with the RareArg reporter, a catp-6 mutant derepressed the CommonArg (Fig. 1D and SI Appendix, Fig. S2A) and PolyLys (Fig. 1E and SI Appendix, Fig. S2A) reporters. When we performed a brood size assay on a catp-6 mutant with the PolyLys reporter, we observed only a mild rescue, in stark contrast to the much stronger effect of catp-6 in the body wall muscle-dependent swimming assay (SI Appendix, Fig. S2C, compared to Fig. 1E). However, upon looking at adult animals with the RareArg reporter, we noted that catp-6 animals do not show GFP in the vulva, while znf-598 exhibit GFP in the vulva (SI Appendix, Fig. S2D). We reasoned that the differential requirement for catp-6 for UNC-54 expression across tissues may be rooted in differential expression of catp-6 expression in the tissues.

To test whether catp-6 reporter phenotypes arise from disrupted mRNA decay, we conducted RNA-seq on the various reporter animals. Loss of catp-6 derepressed reporters at the mRNA level, indicating that the observed phenotypes arise from inefficient NGD (Fig. 1F and SI Appendix, Fig. S2 E and F). These data demonstrate that catp-6 loss compromises the cell’s ability to execute mRNA decay in response to ribosomal stalls.

Nonstop Decay and NMD Are Unaffected by CATP-6 Loss.

We next sought to understand whether the translational effect of catp-6 was specific to NGD, and thus we analyzed two other translational quality control pathways. While NGD responds to ribosomal stalls, other pathways respond to different translation-dependent triggers: nonstop mRNA decay (NSD) degrades mRNAs that lack stop codons and is triggered by ribosomes that stall on the 3′ end of an mRNA, and NMD degrades mRNAs with premature stop codons, and is triggered by aberrant translation termination. Since the ribosomal triggers differ for each pathway, we reasoned that knowledge of the scope of the catp-6 effect on each pathway could shed light into its translational mechanism.

To first investigate the effect of catp-6 on NSD, we utilized a published NSD reporter (59) (Fig. 2A). Briefly, the NSD reporter was made at the unc-54 locus, contains a C-terminal GFP fusion with no stop codons, and encodes a single ~7,000 nucleotide open reading frame that translates into the polyA tail. When NSD is functional, little GFP is produced. When NSD is dysfunctional, the mRNA is stabilized and GFP is expressed in the body wall muscle. While known NSD factors nonu-1 and skih-2 derepressed the NSD reporter, a catp-6 mutant had no effect (Fig. 2A), suggesting that catp-6 is not required for NSD.

Fig. 2.

NMD and NSD are unaffected by CATP-6 loss. (A) Schematic of NSD reporter, previously described (59). The bottom graph shows mean relative fluorescence units of the indicated strains (n = 10 to 17 animals/strain) in the NSD reporter background. nonu-1 and skih-2 are positive controls required for NSD loss. (B) Schematic of the r293 reporter, previously described (60, 61). Swimming assay of indicated strains (n = 12 animals/strain) in r293 reporter background through quantification of thrashes per minute. smg-1 is a positive control required for NMD loss. (C) Schematic of R2 reporter, previously described (62), and RFP quantification (n = 6 to 9 animals/strain) of indicated strains in R2 reporter background. Quantifications show mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant. P values determined by Welch’s t test.

We quantified the NMD effect of catp-6 with an NMD-sensitive unc-54 allele, unc-54 (r293) (60, 61). The r293 allele deletes the 3’ polyadenylation signal of unc-54, which causes readthrough transcription into the downstream gene, aex-5 (Fig. 2B). Readthrough transcription fuses the unc-54 mRNA with that of aex-5 and introduces several splice junctions downstream of the normal unc-54 stop codon. The resulting unc-54 mRNA is targeted by NMD (61, 63). Since the r293 reporter encodes a lowly expressed but wild-type UNC-54 protein, we quantified NMD activity via measurements of body wall muscle function with NMD-defective animals exhibiting more muscle (UNC-54) function. The NMD factor smg-1 failed to perform NMD while catp-6 loss had no effect on the r293 reporter (Fig. 2B).

We also examined NMD activity with a second reporter (“R2”), which expresses RFP and GFP from a single locus, with RFP under NMD control (62) (Fig. 2C). NMD-competent animals are RFP-negative, while NMD-defective animals are RFP-positive (62, 64). Unlike the r293 reporter which is only expressed in the body wall muscle, the R2 reporter expresses throughout the animal and thus provides a more animal-wide picture of NMD function. As assessed by the R2 reporter, the positive control smg-1 was deficient in NMD while catp-6 animals were functional for NMD (Fig. 2C). Discussion about these results are described in Supplemental Text 1.

We conclude that catp-6 loss does not perturb NSD nor NMD. These results support the idea that catp-6 animals have defects in translation elongation and/or ribosomal stalling which lead to failures in NGD.

Polyamine Transport Via CATP-6 Is Important for NGD.

ATP13A2, the human homolog of C. elegans CATP-6, is a transmembrane lysosomal P5-type ATPase that transports polyamines, particularly spermidine and spermine (65–67). Since C. elegans lacks spermine, we will focus on spermidine in this study (68, 69). We were curious to relate CATP-6 as a polyamine transporter to the requirement of catp-6 in NGD. We began by investigating two catp-6 missense mutations (A741V, G308R) identified in our genetic screen. To understand how A741V and G308R mutations affect CATP-6 function, we placed the mutations on the structure of the CATP-6 human homolog, ATP13A2 (Fig. 3A).

Fig. 3.

CATP-6 polyamine transport mutants disrupt NGD. (A) Cryo-EM E2P-like structure of ATP13A2 (PDB: 7N70) colored by domain, as described (65). Corresponding residues mutated in the catp-6 missense mutants identified are in red, with key neighboring residues labeled. Below the zoomed-in views of the mutation sites shows the residue conservation of that region between C. elegans (c.e) CATP-6 and Homo sapiens (h.s.) ATP13A2. Conserved residues are in dark gray and nonconserved residues in light gray. (B) Mean relative fluorescence units of the indicated catp-6 strains (n = 15 to 17 animals/strain) in the RareArg reporter background. Bars show mean ± SD. (C) Schematic of the oaz-1::mNG reporter with representative full-body images of WT and catp-6(Δ) animals in the reporter background. (D) Zoomed in representative images of WT and catp-6(Δ) animals in the background of oaz-1::mNG reporter (mNG, Left) and hlh-8p::mCherry marker (mCh, Middle). White arrowheads indicate M-cells. The larger out-of-focus GFP-positive nucleus is an intestinal cell nucleus; the smaller foci are from gut autofluorescence. *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant. P values determined by Welch’s t test.

The enzymatic cycle of the CATP-6 transport mechanism relies on three major steps: (1) ATP binding, (2) autophosphorylation of a key aspartate residue (D476 in C. elegans; D508 in humans), and (3) dephosphorylation of that same aspartate residue. The two catp-6 missense mutations from our screen were located in or near motifs important for CATP-6 transport function (Fig. 3a). G308R is located in the highly conserved TGES sequence, where E309 (E343 in humans) is required for CATP-6 dephosphorylation (65). A741V is near D476 in the structure, the key residue involved in CATP-6 autophosphorylation. Based on our recovery of mutations near CATP-6 catalytic residues, we hypothesized that the polyamine transport function of CATP-6 was required for NGD. To test this hypothesis, we made the catalytically dead mutant E309A (E343A in humans), which disrupts polyamine transport function (66). The catp-6 (E309A) mutation conferred RareArg reporter derepression comparable to that conferred by catp-6 deletion (Fig. 3b). We also attempted to mutate the D476 residue several times but were unable to recover mutants at this site. To test the expression of the catp-6 missense mutants, we tagged wild-type catp-6 at two sites with three different tags via CRISPR/Cas9. While tagged catp-6 alleles were functional as assessed by RareArg repression, none of the tagged CATP-6 proteins were detectable by immunoblot (example blot in SI Appendix, Fig. S3).

Taken together, our EMS and CRISPR/Cas9 mutants support the idea that polyamine transport by CATP-6 is important for NGD.

catp-6 Animals Exhibit Global and Muscle-Specific Changes in Polyamines.

Polyamines affect myriad cellular processes, including translation. Thus to understand the nature of the polyamine alterations in catp-6 animals, we performed a series of translational and genetic analyses.

First, we analyzed translation genome- and animal-wide using ribosome footprint profiling (Ribo-seq). Cells regulate polyamine levels via several polyamine biosynthesis genes the translation of which is polyamine-responsive (70). Based on polyamine metabolism regulation from other experimental systems, and under a simple model in which catp-6 animals have decreased cellular polyamines, we determined the expected translational changes in four polyamine synthesis genes: odc-1, smd-1, oaz-1, and F53F10.2 (C. elegans homolog of AZIN1) (SI Appendix, Fig. S4A). Two genes’ mRNAs (oaz-1 and F53F10.2) exhibited translational changes consistent with a lower polyamine environment in catp-6 animals, smd-1 exhibited changes consistent with a higher polyamine environment in catp-6 animals, and odc-1 exhibited inconsistent changes across replicates (SI Appendix, Fig. S4 B and C). These translational changes are consistent with polyamine dysregulation, but point to a more complicated picture than global lower polyamines in catp-6 animals. We suspect that this is due to tissue-specific increases and decreases in polyamines which are obscured by the animal-wide Ribo-seq assay. This idea is also substantiated by the observed effect of catp-6 in the body wall muscles, but not the vulval muscles (Fig. 1E and SI Appendix, Fig. S2C).

There are two paralogs of catp-6 in C. elegans: catp-5 and catp-7. catp-5/6/7 have overlapping substrate specificity, but different localization and expression patterns across tissues (71). Thus to test if the other polyamine transporters affect NGD, we crossed mutants of each paralog with the RareArg reporter. Only catp-6 loss derepressed the RareArg reporter; we saw no effect of catp-5 or catp-7 alone (SI Appendix, Fig. S5A). Since all three paralogs have at least partial redundancy (71), tissue-specific differences in paralog expression patterns could explain the specific role of catp-6 on the RareArg reporter. Thus we analyzed published C. elegans single-cell RNA-seq data (72), which showed that catp-6 is the predominant paralog in the body wall muscle where the RareArg reporter is expressed (SI Appendix, Fig. S5B).

With the differences in expression and function of catp-5/6/7 in mind, we created a tissue-specific reporter of polyamine-dependent translational regulation. The reporter is conceptually similar to that described previously (73, 74), and takes advantage of a polyamine-dependent ribosomal frameshift (75, 76) in the oaz-1 gene. When polyamine levels are low, translation terminates at an in-frame stop codon in oaz-1. When polyamine levels are high, ribosomes instead frameshift and translate to a downstream stop codon. To detect the frameshift cell-specifically, we integrated a fluorescent protein (mNeonGreen) downstream of the genomic oaz-1 frameshift site (Fig. 3C). In a wild-type background, oaz-1::mNG animals exhibited robust expression of mNeonGreen across many tissues, consistent with a high rate of oaz-1 frameshifting and an abundance of cellular polyamines. catp-6 animals exhibited tissue-specific differences in oaz-1::mNG expression, most notably more mNeonGreen in intestinal cells and lower mNeonGreen in neurons and muscle (Fig. 3C). We validated low expression of oaz-1::mNG in the body wall muscle where unc-54 is expressed by coexpression with an hlh-8p::mCherry M-cell lineage marker which labels a subset of body wall muscle cells (77) (Fig. 3D). These data are consistent with the idea that catp-6(Δ) animals have an aberrant distribution of polyamines across tissues, including decreased polyamines in the body wall muscle.

These data show that catp-6 animals exhibit tissue-specific alterations in translation and polyamines and that the effect of catp-6 on the RareArg reporter is rooted in its muscle-specific expression pattern. We next sought to mechanistically understand how polyamines impact NGD.

CATP-6 Loss Reduces eIF5A Activity.

One critical role of polyamines, in particular spermidine, is to install the rare hypusine modification on eIF5A at a highly conserved lysine, K50 in humans, K51 in Saccharomyces cerevisiae, and K54 in C. elegans. Hypusine is critical for eIF5A function and cell viability (37, 78–82), and loss of eIF5A causes ribosomes to accumulate at proline-rich regions in S. cerevisiae (43–45). Thus we hypothesized that polyamine reduction in catp-6 animals decreases eIF5A hypusination and eIF5A function. Our Ribo-seq analyses showed that some polyamine-dependent effects were detectable at the level of whole-animal mRNA translation; we therefore examined our Ribo-seq data for translational signatures associated with eIF5A dysfunction.

To examine eIF5A function, we analyzed ribosomal accumulations at proline-rich motifs.

Proline-rich regions are eIF5A-sensitive in S. cerevisiae (43–45), and thus we looked for changes in ribosomal occupancy at PPX/XPX/XPP motifs in catp-6 animals. We observed that catp-6 loss led to overall increased pausing at these PP motifs (Fig. 4A). We then wanted to expand upon this observation by examining translation of other Pro-rich regions. To do so, we aligned our C. elegans Ribo-seq data about windows of ten amino acids with 3, 4, 5, or 6 prolines. We noted increased ribosomal occupancy in catp-6 animals compared to wild type as the number of prolines in the window increased, both globally and on specific gene examples (Fig. 4B and SI Appendix, Fig. S6 A and B). We also noted a catp-6-dependent increase in pausing at R-rich (but not K-rich) sequences (SI Appendix, Fig. S6 C and D), though the rarity of such endogenous motifs increased noise and hampered interpretation. We computed a pause score by calculating the relative ratio of ribosomes within and outside the proline-rich window, with a higher pause score indicating slower relative translation within the window. We computed pause scores for all 20 amino acids and observed that the catp-6-dependent accumulation of ribosomes was specific for prolines and arginines and reproducible across replicates (Fig. 4C and SI Appendix, Fig. S6E). Ribo-seq data from znf-598 animals did not show proline-dependent ribosomal pauses (SI Appendix, Fig. S6F). These data support the idea that catp-6(Δ) animals display increased ribosomal pausing at proline-rich regions, a sign of eIF5A dysfunction.

Fig. 4.

Translation in catp-6 animals shows hallmarks of decreased eIF5A function. (A) Relative ribosome occupancy from catp-6(Δ)/WT Ribo-seq libraries at PPX/XPP/PXP motifs, with the 3 amino acids aligning with the ribosome E-P-A sites, compared to all 3 amino acid motifs. Boxes for “all 3mers” show median and upper/lower quartiles, with individual motifs omitted for visualization purposes. For PPX/XPP/PXP, each individual motif is represented by a single dot, with median in a solid black line. (B) Metagene analysis showing log₂ (catp-6(Δ)/WT) Ribo-seq read density for two Ribo-seq replicates. Black box outlines a 10 amino acid window. Each graph has a different number of prolines within the 10 amino acid window, as written in the top right of each row. (C) Relative pause scores from catp-6(Δ)/WT Ribo-seq libraries (replicate 2), for each amino acid frequency (3–6) within a given 10 amino acid window. *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant. P values determined by Mann–Whitney U.

Recent work also showed that the gene TIM50 in S. cerevisiae is regulated by eIF5A (83). We identified TIM50’s C. elegans homolog (scpl-4) and analyzed its ribosome distribution. catp-6 animals showed an irregular distribution of ribosomes across scpl-4 compared to wild-type animals, especially in the vicinity of two proline-rich regions (SI Appendix, Fig. S7A). This further supports the idea that catp-6 animals experience eIF5A dysfunction.

Prior work noted that decreased eIF5A function causes a global shift of ribosomes toward the 5′ end of coding sequences in S. cerevisiae (44). We did not observe a concerted shift of ribosomes toward either the 5′ or 3′ end of all mRNAs (SI Appendix, Fig. S7B). Instead, some mRNAs exhibited a shift of ribosomes toward the 5′ end of their CDS (SI Appendix, Fig. S7C), while other mRNAs exhibited a shift of ribosomes toward their 3′ end (SI Appendix, Fig. S7D). These results point to a global shift in the positions of ribosomes within coding sequences in catp-6 animals. As noted previously, since the Ribo-seq data were obtained from whole animals, the differences in ribosome shifts between mRNAs may reflect the mRNAs’ cellular environment and cell-specific differences in polyamines and/or eIF5A hypusination.

eIF5A also acts in termination, though we saw no consistent effect of catp-6 on ribosomes adjacent to stop codons (SI Appendix, Fig. S7E). This result is challenging to interpret as the conditions used to prepare the Ribo-seq libraries (cycloheximide, 28 to 30 nt fragments) are not well-suited to capture terminating ribosomes (84).

Our Ribo-seq analyses revealed hallmarks of eIF5A dysfunction in catp-6 animals. These results are consistent with a model in which CATP-6 transports spermidine that supports eIF5A hypusination and function.

eIF5A Hypusination Status Affects NGD.

We next tested whether less eIF5A hypusination could explain the loss of NGD seen in catp-6 animals. To do so, we crossed hypusine biosynthesis mutants into the RareArg reporter. To install hypusine on eIF5A, two enzymes (DHPS-1 and DOHH-1) sequentially act on eIF5A to make deoxyhypusinated and hypusinated eIF5A, respectively (Fig. 5A). Loss of either of dhps-1 or dohh-1 compromised NGD (Fig. 5 B and C). We noted a stronger phenotype (more GFP) with dhps-1 compared to dohh-1; such a difference may reflect residual activity of deoxyhypusinated eIF5A in dohh-1 animals (85). We also analyzed the effect of iff-2 (the somatic C. elegans eIF5A homolog), and noted derepression of the RareArg reporter in iff-2 mutant animals (Fig. 5D). These results show that eIF5A and its hypusination are required for NGD in C. elegans.

Fig. 5.

eIF5A hypusination is required for NGD in C. elegans. (A) Schematic of eIF5A/IFF-2 hypusination pathway. (B–E) Mean relative fluorescence units of the indicated strains (n = 13 to 18) in the RareArg reporter background. Bars show mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant. P values determined by Welch’s t test.

While analyzing iff-2, dohh-1, and dhps-1 mutants, we noticed variable levels of the RareArg-dependent GFP phenotype, with more GFP in posterior regions of the animal (SI Appendix, Fig. S8A). The difference was less pronounced over time: As animals developed, the GFP phenotype intensified throughout the body wall muscle (SI Appendix, Fig. S8B). This result is consistent with NGD rescue via maternal eIF5A protein. Maternal rescue results when functional wild-type mRNA or protein from the mother is passed into the developing egg, yielding homozygous mutant offspring with residual wild-type protein. As cells divide, the cytoplasmically inherited wild-type protein will be diluted. Consistent with this idea, we noticed a delay in the essentiality of all three of iff-2, dohh-1, and dhps-1; all three alleles confer viable but sterile homozygous mutant adults. Also consistent with the maternal rescue model, the cells that first appear GFP-positive in the mutants are the M cell lineage (SI Appendix, Fig. S8C), a body wall muscle cell lineage that divides more than other body wall muscle cells and would be expected to dilute out maternal eIF5A more quickly (86). These observations temporally align loss of mature eIF5A with onset of the NGD phenotype, consistent with a role for eIF5A in NGD.

To contextualize the effect of the essential protein eIF5A on NGD, we analyzed the effect of loss of other essential cellular proteins. We selected mutants using the criteria that: 1) mutants were reported to exhibit a delay in their essentiality, yielding homozygous but sterile animals similar to eIF5A, and 2) mutants would be expected to disrupt core gene expression processes. These four mutants were rpt-5 (protein degradation), rbp-9 (mRNA transcription), eEF1A.1 (translation elongation), and eIF3.H (translation initiation). All four mutants failed to produce a RareArg phenotype comparable to that of catp-6 and the eIF5A mutants (Fig. 5E and SI Appendix, Fig. S8 D and E). Animals of the last mutant (eIF3.H) were sick even as heterozygotes, which made strain maintenance, animal survival, and imaging difficult. Among the mutants, eEF1A.1 is notable as it is involved in translation elongation, though it carries out a distinct molecular role than that of eIF5A. These data support the idea that loss of NGD is not simply activated due to general insults to cellular health, but specifically occurs upon loss of eIF5A.

The Role of eIF5A in NGD Is Conserved.

We next tested the conservation of the eIF5A–NGD relationship in S. cerevisiae. We began by utilizing a published stalling reporter (87). The reporter expresses GFP and RFP from a single mRNA, with an intervening stall of 12 lysine codons (AAG) (Fig. 6A). To test a requirement for eIF5A, we treated cells with a deoxyhypusine synthase inhibitor, GC7 (88), and measured reporter expression. Upon treatment with GC7, GFP, and RFP levels increased from the (AAG)₁₂ reporter, but not from a nonstalling control reporter (Fig. 6 B and C and SI Appendix, Fig. S9A). These data are consistent with increased protein production upon eIF5A inhibition. That both GFP and RFP increased to a similar extent upon GC7 treatment (Fig. 6D and SI Appendix, Fig. S9A) is consistent with a loss of NGD upon eIF5A inhibition. We also extensively sought to measure GC7’s effect on eIF5A hypusination directly via immunoblot but could not confidently determine the identity of the hypusine bands.

Fig. 6.

eIF5A hypusination is required for NGD in S. cerevisiae. (A) Schematic of single transcript (AAG)₁₂ reporter as DNA, RNA, and expected protein output depending on NGD functionality. (B–D) FC of the GFP (B), RFP (C), and GFP/RFP (D) between 1 mM GC7 and non-GC7 treated cells, between three replicates. The dashed line represents the mean. (E) Schematic of dual transcript (CGA)₁₂ reporter as DNA, RNA, and expected protein output depending on NGD functionality. (F) Distribution of dual transcript (CGA)₁₂ reporter expression (log₂ (GFP/RFP)) from yeast cells indicated (replicate 1). *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant. P values determined by paired t test.

eIF5A has a published role in ribosome quality control (RQC) (47), which could confound the interpretation of the single transcript (AAG)₁₂ reporter. Therefore, we designed a second reporter in which the reporter signal is derived from the protein downstream of the stall sequence and would be expected to be independent of RQC. The reporter normalizes GFP expression to a second transcript that expresses RFP, produced from the same promoter (Fig. 6E). As expected, elongation through a (CGA)₁₂ stall was inefficient and the reporter expressed low GFP relative to RFP. GFP expression was increased upon mutation of hel2, the S. cerevisiae homolog of znf-598. GC7 treatment stabilized the reporter comparable to the extent observed upon hel2 mutation (Fig. 6F and SI Appendix, Fig. S9B). This result is consistent with a function for eIF5A in mRNA decay upon ribosome stalling, and demonstrates the reproducibility of the effect across multiple NGD reporters. We also performed RNA-seq on cells with the (CGA)₁₂ reporter and confirmed the reporter is stabilized at the mRNA level upon GC7 treatment (SI Appendix, Fig. S9C). To deconvolute potential effects from transcriptional changes, we also measured mRNA half-life after transcription shutoff, and determined that the (CGA)₁₂ reporter mRNA is stabilized by NGD inhibition via hel2 mutation, validating the dual transcript reporter (SI Appendix, Fig. S10 A–C). While GC7 treatment stabilized the (CGA)₁₂ reporter, GC7 treatment also stabilized RFP mRNA half-life, complicating interpretation of the experiment (SI Appendix, Fig. S10 A–C). Taking all these data together, the data support that hypusinated eIF5A is a conserved cellular factor required for NGD across eukaryotic systems.

The Integrated Stress Response Is Not Strongly Activated by catp-6 Loss.

We next sought to test potential models by which eIF5A could be required for NGD. One model could be that loss of eIF5A activates a translational response that decreases the number of ribosomes per mRNA. Widespread ribosomal pauses feedback to translation initiation via the general control nonderepressable (GCN), integrated stress response (ISR), and other pathways so as to lower rates of translation initiation (17, 18, 89, 90). We reasoned that widespread pauses in catp-6 animals (Fig. 4 A–C) may signal to translation initiation and decrease the number of ribosomes per mRNA. Less ribosomes on the reporter would be expected to lead to less ribosomal collisions, which would allow ribosomes more time to elongate through the ribosomal stall and thus decrease NGD. This model suggests that eIF5A regulates NGD via the balance of the number of ribosomes per mRNA. We set out to test whether feedback via the ISR could explain the NGD phenotype of catp-6 animals.

Upon ISR activation, the kinase GCN2 phosphorylates eIF2α, which broadly reduces mRNA translation except for certain stress-response mRNAs (91). To test whether the ISR is activated in catp-6(Δ) animals, we analyzed the effect of key ISR mutants: gcn-2 and eif-2alpha(S49A), which ablates the GCN-2 phosphorylation site on eIF2α. Mutation of either gcn-2 or eif-2alpha(S49) alone had no effect on the RareArg reporter (SI Appendix, Fig. S11 A and B). Mutation of both catp-6 and gcn-2 led to a slight–but reproducible–restoration of RareArg repression compared to catp-6(Δ) alone (SI Appendix, Fig. S11A). We also attempted to measure eIF2α phosphorylation via immunoblots, but these experiments failed for technical reasons. The genetic results suggest that the ISR contributes to the observed RareArg phenotype, but the bulk of the derepression observed in catp-6 animals is independent of the ISR.

Taken together, our results failed to support ISR activation as an explanation for the catp-6 NGD phenotypes. An alternative model that is consistent with all of our data is that eIF5A differentiates paused from stalled ribosomes to facilitate NGD (Fig. 7). As ribosomes elongate, some ribosomes encounter difficult motifs, and eIF5A would be expected to bind and assist elongation through such a transient pause. However, if a ribosome persists at a site and fails to elongate, eIF5A may remain bound, and could serve as a physical readout of failed elongation. In this model, NGD factors such as ZNF598 would recognize eIF5A in the E-site of a ribosome and execute downstream repression.

Fig. 7.

Model of eIF5A in NGD. A ribosome stalls during translation of a problematic sequence (yellow triangle). eIF5A binds to the ribosome and attempts to resolve the stall. If eIF5A is unable to resolve the stall, eIF5A will remain in the ribosome for an extended period of time. Eventually, a trailing ribosome will collide with the leading ribosome, and together the collision and eIF5A occupancy will enable ZNF598 to recognize the substrate. ZNF598 ubiquitinates the ribosome, triggering subsequent mRNA decay.

Discussion

Here, we describe how a mutant (catp-6) from a genetic screen revealed a central and conserved role for the small molecule spermidine and the translation elongation factor eIF5A in the cellular response to ribosomal stalls. The discovery was enabled through the genetic tractability of polyamine levels in C. elegans (SI Appendix, Supplemental Text 2). Among translation elongation factors, eIF5A would be uniquely well-suited to help cells sense stalled ribosomes as eIF5A is thought to bind more frequently and spend more time at problematic sequences. That eIF5A would be expected to dissociate from resolved pauses–but persist at prolonged stalls–makes eIF5A well-suited to serve the molecular function of a pause-stall determinant.

There are two simple, nonmutually exclusive molecular interpretations of the eIF5A-detection model. First, eIF5A occupancy may be directly sensed by NGD factors that bind eIF5A or read out its E-site occupancy. Such a model is similar to stall recognition by NOT3, which binds the empty E-site and the P-site tRNA to recognize stalls at particular codons (92). Second, eIF5A may be indirectly sensed, e.g., by NGD factors that bind to an eIF5A-stabilized ribosome conformation.

Either model would be consistent with a number of observations surrounding eIF5A across systems. The eIF5A-detection model would place eIF5A at ribosomal stalls to assist in later RQC (47, 50) of the nascent peptide. The model would also complement known stall recognition mechanisms via ribosome collisions and may provide a feature for cells to distinguish among different kinds of collisions (12, 16, 34, 93) so as to limit erroneous decay. The model is also consistent with recent computational studies of initiation and elongation dynamics upon eIF5A knockout, which suggested a role for eIF5A in coordinating stall resolution in human cells (94). A requirement for eIF5A in the ribosome pause-stall decision would rationalize the ubiquity of eIF5A in recent translational quality control structures, including Ski complex-associated ribosomes (92), RQC complex-associated ribosomes (47), Rbg1/Tma46-bound ribosomes (95), and GCN-associated disomes (96). Furthermore, eIF5A’s association with the ribosome is stimulated by UV stress (97), suggesting a role of eIF5A in response to widespread translational stress. The structural data make it apparent that eIF5A directly binds to sites of problematic translation. Our data raise the possibility that cells leverage the abundance and affinity of eIF5A for ribosomes to triage translational complexes for quality control pathways.

A third explanation for our observations is that polyamines and/or eIF5A could weaken peptide bond formation at stall sequences. We do not favor this explanation as polyamines and eIF5A are widely known to stimulate protein synthesis (37, 44–46), though we also acknowledge that polyamines and eIF5A are notorious for exhibiting feedback mechanisms that would be challenging to parse out (98–100). While we tested one possible mechanism by which translation elongation could feed back to initiation (ISR), it is possible that another ISR-independent mechanism contributes.

Regardless of the precise mechanism, our results reveal an unexpected requirement for eIF5A in determining the fate of mRNAs that experience ribosomal stalling. While it is well established that eIF5A supports translation through problematic sequences in vitro and in vivo across systems, here, we identified that eIF5A is also required to repress protein production in response to ribosome stalls. These findings extend eIF5A’s role beyond simple core protein synthesis functions, and reveal remarkably mRNA-specific effects of eIF5A on translation. Other translation factors such as Rgb1, New1 (eEF3), and TACO1 may also help cells identify stalled ribosomes, and additional work will need to be done to determine how cells coordinate stall responses across such a network (95, 101, 102).

Our identification of eIF5A’s requirement in NGD also raises the possibility that NGD dysfunction could be an underlying consequence of diseases rooted in polyamine and eIF5A dysregulation. Various pathogenetic variants of ATP13A2/PARK9 are associated with neurodegenerative conditions, including a familial, juvenile-onset form of Parkinson’s Disease (103). Based on our work here, we expect that PARK9 mutation would compromise eIF5A biogenesis and NGD function, the translational consequences of which are largely unexplored. eIF5A and its hypusination decrease with age, and the importance of polyamines and eIF5A is known from cellular models of Huntington’s Disease (104, 105). We speculate that NGD dysfunction is an unrecognized facet of eIF5A/polyamine-associated disease pathophysiology. Additionally, our work has implications for pharmacological efforts to target polyamines and eIF5A (106). Based on our work, eIF5A inhibition would be expected to increase expression of endogenous mRNAs that trigger ribosomal stalling, which–by allowing expression of normally repressed proteins–could exacerbate proteostress and/or enable production of neoantigens.

Overall, our work expands knowledge of the early events associated with the ribosome pause-stall decision. The unexpected and important role for polyamines and eIF5A in reducing expression of mRNAs with ribosomal stalls contrasts to the established role for eIF5A in promoting translation of other mRNAs. Our work lays the foundation to understand how dysregulated NGD could contribute to the pathogenesis of eIF5A and polyamine-related disorders.

Methods

C. elegans Strain Construction.

C. elegans were grown at 16 or 20 °C on NGM or MYOB plates with OP50 bacterial food source. Mutant strains created in this work were made via CRISPR/Cas9 as previously described (107) in which multiple isolates were obtained and validated. Other mutant strains were obtained from the Caenorhabditis Genetic Center (CGC). Combinations of mutants were produced via crossing. A complete description of all C. elegans strains as well as associated sequences and oligonucleotides are described in Dataset S1.

Fluorescent Imaging.

Animals were grown at 20 °C and imaged as L4s, unless otherwise noted (i.e., dhps-1(Δ) animals in SI Appendix, Fig. S8). Animals were anesthetized in EN50 with 1 mM levamisole in a microscope slide with a 0.15 mm coverslip. Slides were imaged on the Zeiss AxioZoom microscope with 1.0× objective and 80X zoom. Parameters include exposure time of either 250 ms (GFP) or 800 ms (RFP), and 50% shift. All images shown within the same figure panel were imaged during the same imaging session, using the same exposure time, and processed identically. To quantify fluorescence, the mean pixel intensity was quantified per animal. For strains containing a fluorescent pharynx from a balancer, all strains imaged together had the pharynx region manually excluded from the animals’ total area for quantification. Scripts to quantify RFP are described (108).

For imaging the hlh-8p::mCherry animals, images were acquired using a Zeiss AxioImager M2 microscope with a Plan-Apochromat 100×/1.4 Oil DIC lens. For oaz-1::mNG;hlh-8p::mCherry animals, exposure time was 1 s (mNG) and 8 s (mCh). For dhps-1(Δ);hlh-8p::mCherry animals, exposure time was 250 ms (GFP) and 3 s (mCh). All bar graphs were produced using GraphPad Prism version 10.4.1 for Mac, GraphPad Software, Boston, MA, https://www.graphpad.com.

UNC-54 Muscle Function Assays.

For the swimming assay, L4 animals were individually put into 5 μL 1X M9 on a microscope slide. The number of complete oscillating thrashes of the head of the animal was counted for 1 min. A total of 12 animals were counted per strain.

For the brood size assay, one L4 animal was isolated onto individual NGM plates and put at 16 °C, and then sequentially transferred to new plates every 2 to 3 d, across a total of 6 d. For each plate, the number of hatched animals was counted, and total brood size calculated by adding progeny counts across plates. The process was repeated for 11 animals per strain.

Ribo-seq.

Animals were grown at 16 °C and harvested as L4s. To lyse pellets, animals were ground in frozen PLB (20 mM Tris pH8.0, 140 mM KCl, 1.5 mM MgCl₂, 1% Triton) with 100 μg/mL cycloheximide and liquid nitrogen with a mortar and pestle.

Samples were broadly prepared as described in ref. 109, except run on 10 to 60% gradients and size-selecting for 28 to 30 nt footprints. Specific details of the Ribo-seq preparation described in SI Appendix, Supplemental Text 3.

Polyamine gene analyses were done by calculating the ribosome occupancy within two different regions of each polyamine gene (labeled as in SI Appendix, Fig. S4) that are expected to differ in ribosome occupancy depending on cellular polyamine levels (70). Using our Ribo-seq data, we then calculated specific ratios based on the ribosome occupancy in Region 1 (uCC, uORF, or ORF1) divided by the ribosome occupancy in Region 2 (main ORF, or ORF2). The gene-specific ratios for each library were then normalized to library depth and compared to the WT library to calculate the relative ribosome occupancy scores per gene.

To calculate the 3mer amino acid pausing scores (Fig. 4A), we first identified all instances of a motif genome-wide. For each motif site, we then counted the number of times that motif was found centered in a ribosome, with the first amino acid in the E-site, the second amino acid in the P-site, and the third amino acid in the A-site. We divided this count by the abundance of ribosomes up- and downstream of the motif, generating a fractional abundance of ribosomes at this site. We then averaged across all instances of that motif genome-wide, generating a ribosome occupancy score for that motif. We divided the ribosome occupancy score of catp-6 animals by wild-type animals to obtain the value shown.

The per-amino acid pause scores in a 10 amino acid stretch (Fig. 4B) were calculated similarly, with the motif being at least n (n = 3, 4, 5, 6) of an amino acid in a 10 amino acid window. A pause score was then calculated for each amino acid and library by dividing the average Ribo-seq signal in the window by that outside the window. To calculate relative pause scores (Fig. 4C and SI Appendix, Fig. S6 E and F), each pause score from the mutant library was normalized to the wild-type library.

For stop codon analyses, Ribo-seq data were normalized as described in (110) filtering for genes with at least 10 Ribo-seq reads. Polarity scores were calculated as previously described (44). CDF plots used custom scripts, available on the Arribere lab github.

S. cerevisiae Strain Construction and Maintenance.

Yeast strains used in this work are described in Dataset S1 and derived from BY4741. Strains were constructed via standard lithium acetate transformations. The dual transcript (CGA)₁₂ reporter was integrated into the ADE2 locus with StuI-digested pJA1037. Recipes for yeast media are described in Dataset S1. For all galactose-inductions, cultures were first grown in 2% raffinose and 2% galactose (YPGR).

GC7 Treatment and Imaging of Single Transcript (AAG)₁₂ Reporter Yeast.

Log phase cultures grown in YPD were treated with 1 mM GC7 (Cayman Chemical) for 12 to 16 h at 30 °C before imaging. Cells were fixed by pelleting 0.5 mL of culture, washing once in ice-cold PBS, and incubating in 4% paraformaldehyde for 15 min at room temperature. After a final PBS wash, cells were resuspended in PBS, immobilized onto polylysine-coated coverslips (Neuvrito) and mounted with ProLong™ Diamond antifade.

Fixed yeast cells were mounted on concanavalin A–coated slides and imaged on a Zeiss LSM 980 confocal microscope using a Plan-Apochromat 63×/1.4 NA oil objective. GFP and RFP were excited with 488 and 561 nm laser lines, and emission was collected using GaAsP detectors. Z-stacks spanning the full cell volume were acquired using identical settings for all conditions. Maximum-intensity projections were generated in Zeiss ZEN software, and fluorescence intensities were extracted from the resulting images.

Due to reagent access changes, replicate 3 of the (AAG)₁₂ reporter was performed as the other replicates described above, except for the following deviations: Fixed cells were immobilized onto polylysine-coated slides without antifade and imaged on Zeiss LSM 880 confocal microscope. 488 nm laser emission was collected using GaAsP detectors, and 561 nm was collected with a standard PMT detector.

Images were analyzed using Fiji to segment individual cells based on the GFP signal, and mean fluorescence intensities for GFP and RFP were quantified for each cell. Background fluorescence was subtracted using cell-free regions of the same field. For each sample, per-cell mean GFP and RFP intensities were extracted. Mean intensities for GC7-treated cells were normalized to non-GC7-treated cells for each reporter strain. Normalized values from independent experiments were visualized.

GC7 Treatment and Flow Cytometry of Dual Transcript (CGA)₁₂ Reporter Yeast.

For the (CGA)₁₂ reporter strains, log phase cultures were grown in YPGR liquid media. Cells were then diluted back in fresh YPGR media, and GC7 (Millipore Sigma or Santa Cruz Biotech) was added to a final concentration of 0, 1, or 10 mM. Cells were then grown at 30 °C for 24 h to early-log phase. Cells were immediately run on a BD Accuri C6 Plus Flow Cytometer for RFP and GFP detection using 640 nm and 488 nm excitation lasers, respectively. Per strain, 20,000 events were collected with no gating. For all flow cytometry data, plots show histograms of the number of cells for log₂ (GFP/RFP) values for individual cells.

Transcription Shut Off.

Yeast were grown in YPGR liquid media to mid log phase. Cells were then swapped into YPD media to inhibit transcription of the Dual Transcript (CGA)₁₂ Reporter. Cells were harvested at timepoints of 0, 2.5, 5, 10, and 20 min post YPD swap. Cells were lysed and RNA-seq was performed as described above.

Immunoblot, RNA-seq, and Plasmid Construction.

Described in SI Appendix, Supplemental Methods.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Data, Materials, and Software Availability

Data are available through SRA via Bioproject No. PRJNA1304949 (111). Summaries of libraries generated in this study are in Dataset S2. The Arribere Lab pipeline for Ribo-seq and RNA-seq analyses is available at GitHub and scripts specific for this study are therein (250811_WohlenbergCatp6Paper) (112). All other data are included in the manuscript and/or supporting information.