CAT tails drive degradation of stalled polypeptides on and off the ribosome

Abstract

Stalled translation produces incomplete, ribosome-tethered polypeptides that the Ribosome-associated Quality Control (RQC) pathway targets for degradation via the E3 ubiquitin ligase Ltn1. During this process, the protein Rqc2 and the large ribosomal subunit elongate stalled polypeptides with carboxy-terminal alanine and threonine residues (CAT tails). Failure to degrade CAT-tailed proteins disrupts global protein homeostasis, as CAT-tailed proteins can aggregate and sequester chaperones. Why cells employ such a potentially toxic process during RQC is unclear. Here, we developed quantitative techniques to assess how CAT tails affect stalled polypeptide degradation in Saccharomyces cerevisiae. We found that CAT tails enhance Ltn1’s efficiency in targeting structured polypeptides, which are otherwise poor Ltn1 substrates. If Ltn1 fails to ubiquitylate those stalled polypeptides or becomes limiting, CAT tails act as degrons, marking proteins for proteasomal degradation off the ribosome. Thus, CAT tails functionalize the carboxy-termini of stalled polypeptides to drive their degradation on and off the ribosome.

Introduction

Stalled mRNA translation produces incomplete polypeptides that can be deleterious to cells. In eukaryotes, the Ribosome-associated Quality Control (RQC) pathway recognizes these stalled polypeptides while they are attached to ribosomes and targets them for degradation^1–3. RQC targets a variety of stalled polypeptides generated by diverse translation abnormalities, including truncated mRNA^4,5, inefficiently decoded codons⁶, and translation past stop codons into poly(A) tails^1,7.When stalling occurs, the slowdown of translation causes adjacent ribosomes to collide. The interface between the stalled ribosomes forms a signal recognized by the E3 ubiquitin ligase Hel2 (ZNF598 in mammals)^8–10. Hel2 then ubiquitylates 40S ribosomal proteins, triggering RQC^11–13. After this process of stall recognition occurs, RQC factors remodel the ribosome to produce a 60S subunit-stalled polypeptide complex^4,5,11–17.

Studies on RQC mechanism have focused on the 60S subunit-stalled polypeptide complex as the sole opportunity for the cell to target stalled polypeptides (RQC substrates) for degradation. Two proteins, Ltn1 and Rqc2, bind that complex and play central roles in degrading the stalled polypeptide^1–3,5,16. Ltn1, a RING E3 ubiquitin ligase, ubiquitylates the RQC substrate while it is tethered to the 60S ribosome to mark it for proteasome-mediated degradation^1–3,5. Rqc2 facilitates Ltn1 binding to the ribosome and drives the C-terminal addition of alanine and threonine (“CAT tails”, “CATylation”) to the RQC substrate^3,18–20. Unlike conventional translation, CATylation occurs without an mRNA template or the 40S ribosomal subunit^19,20. Failure to degrade CATylated proteins can result in their aggregation and lead to disruption of global protein homeostasis in yeast^21–23. Intriguingly, mutation of Ltn1, which is the only E3 ligase known to ubiquitylate CATylated proteins, leads to progressive neurodegeneration in mice²⁴. Given the inherent risk associated with generating aggregation-prone polypeptides, the evolution and function of CATylation are subjects of intense interest.

As a consequence of CAT tail extension, amino acid residues previously buried in the ribosome become exposed to the environment outside of the ribosome exit tunnel. Kostova and colleagues proposed that the ~12 amino acids of the RQC substrate emerging from the exit tunnel are the only residues that Ltn1 can efficiently ubiquitylate²⁵. If a substrate lacks lysine residues within this window, the authors propose that CAT tail extension extrudes any lysine residues buried in the exit tunnel, exposing them for Ltn1 to ubiquitylate²⁵. This model suggests that Ltn1 activity is limited by RQC substrate primary structure and that CAT tails aid degradation of RQC substrates by expanding the region of Ltn1-accessible residues to include those within the ribosome exit tunnel.

In this work, we assessed the contribution of CAT tails to degradation of model RQC substrates using new, quantitative approaches. We found that CATylation enhances degradation of RQC substrates both on and off the ribosome. On the ribosome, CATylation enhances Ltn1’s ability to ubiquitylate structured substrates, which Ltn1 otherwise targets inefficiently. Contrary to the model proposed by Kostova et al.²⁵, our data indicate that Ltn1 is not restricted to lysine residues proximal (in primary structure) to the exit tunnel and that CAT tails can enable Ltn1 to access residues other than those in the exit tunnel. Moreover, we found that CAT tails provide cells with an additional opportunity to degrade RQC substrates that escape ubiquitylation on the 60S ribosome. Off the ribosome, CAT tails act as degrons to mark those escaped RQC substrates for proteasomal degradation, independently of Ltn1. For the substrates we analyzed, Ltn1-independent degradation required the E3 ubiquitin ligase Hul5.

Results

CAT tails aid decay of RQC substrates

To measure how CAT tails contribute to RQC substrate degradation, we monitored how disruption of CATylation affects steady-state levels of model RQC substrates in Saccharomyces cerevisiae. Each model substrate contained an N-terminal green fluorescent protein (GFP) followed by a flexible linker, tobacco etch virus (TEV) protease cleavage site, and stall-inducing polyarginine sequence (Fig. 1b,c). Ribosomes begin translating these substrates normally but stall within the polyarginine sequence without reaching the stop codon²⁶. This produces a GFP-linker-arginine nascent polypeptide that is a substrate for RQC².

Figure 1 |. Loss of CAT tails stabilizes specific RQC substrates.

a, Model of the RQC pathway. Stalled ribosomes are recognized by a set of factors that facilitate separation of the ribosomal subunits. Rqc2 binds the 60S ribosome-stalled polypeptide complex and directs extension of the stalled polypeptide (RQC substrate) with a CAT tail. Ltn1 binds and ubiquitylates the stalled polypeptide. b & c, immunoblots (IBs) of lysates containing two model RQC substrates (schematics above) with and without tobacco etch virus (TEV) protease treatment. GFP, green fluorescent protein. d, Schematic of expression-controlled model RQC substrate and definitions of “stability” and “CAT-tail dependence.” RFP, red fluorescent protein. e & f, Stability measurements from expression-controlled model RQC substrates with different linker lengths and mutated lysines (0,1, or 10 C-terminal lysines mutated to arginine in GFP). Stability data are reported as means, normalized to the value from the rqc2-d98a ltn1Δ strain. CAT tail dependence is indicated with a percent value. Error bars indicate s.e.m. from n = 5 independent cultures and p-values from indicated comparisons of CAT tail dependence are indicated above (16 degrees of freedom, d.o.f.). P-values are derived from a two-tailed t-test for particular contrast.

We began by constructing two model RQC substrates that differed in the length of the linker (RQCsub_SHORT, RQCsub_LONG), allowing two variants for how much of the GFP-linker-arginine polypeptide protruded from the ribosome exit tunnel (Fig. 1b,c). ltn1Δ strains accumulated more RQCsub_SHORT and RQCsub_LONG protein products by SDS-PAGE than LTN1-WT strains, confirming that both are substrates for Ltn1 and, thus, RQC (Fig. 1b,c). RQCsub_SHORT and RQCsub_LONG protein products migrated as higher molecular weight smears in the ltn1Δ background when Rqc2 was intact (Fig. 1b,c). These smears collapsed onto discrete bands upon mutation of Rqc2 to the CATylation-incompetent rqc2-d98a mutant¹⁹ or cleavage of the C-terminus by TEV protease treatment (Fig. 1b,c), indicating that the smears contained CAT tails of varying length. Strikingly, loss of CATylation with rqc2-d98a led to an accumulation of RQCsub_LONG but not RQCsub_SHORT (Fig. 1b,c). These qualitative results suggest that CAT tails enable efficient degradation of some RQC substrates but not others.

To explore the differences in degradation between RQCsub_SHORT and RQCsub_LONG, we developed a quantitative assay to measure the extent to which RQC substrate degradation depends on CAT tails. We constructed an internal expression control by adding red fluorescent protein (RFP) followed by tandem viral T2A skipping peptides upstream of GFP-linker arginine (Fig. 1d). During each round of translation, the ribosome skips formation of a peptide bond within the T2A sequence, producing an RFP that detaches from GFP-linker-arginine before stalling occurs^27,28. This detachment ensures that RQC targets GFP-linker-arginine but not RFP. Indeed, ltn1Δ did not increase RFP levels compared to wild-type (Supplementary Fig. 1a), confirming that RFP is not an RQC substrate. Because the ribosome synthesizes the two fluorescent proteins stoichiometrically but RQC only targets GFP, the GFP:RFP ratio reports on RQC substrate stability. By comparing stability between an experimental condition and a condition where CATylation and RQC are inactive (rqc2-d98a ltn1Δ), we controlled for any RQC-independent degradation or change in fluorescence. This strategy eliminated noise inherent in our model RQC substrate expression system (Supplementary Fig. 1b) and allowed us to quantitatively compare different RQC substrates.

Using our quantitative assay, we re-assessed how degradation of RQCsub_SHORT and RQCsub_LONG depends on CAT tails. We defined “CAT tail dependence” as the change in protein stability due to CATylation impairment, i.e. stability_rqc2-d98a – stability_RQC2-WT (Fig. 1d). Consistent with our previous qualitative results (Fig. 1b,c), CAT tail dependences for RQCsub_LONG and RQCsub_SHORT were 24% and 0.8%, respectively (Fig. 1e). Additionally, we observed substantial CAT tail dependence independent of Ltn1 (in the ltn1Δ background) (Supplementary Fig. 1c and below). To ensure that the CAT tail dependence observed for RQCsub_LONG was not an artifact of changes in ribosome stalling, we designed a quantitative stalling reporter similar to one used by the Hegde and Bennett groups^9,11,12: RQC2 or LTN1 mutations did not affect stalling relative to control (Supplementary Fig. 1d). Furthermore, expressing RQCsub_LONG using the strong TDH3 promoter and the moderate MET17 promoter yielded identical stabilities (Supplementary Fig. 1e), suggesting that CAT tail dependence did not require RQCsub_LONG overproduction. These data suggest that CAT tails facilitate degradation of RQCsub_LONG but are dispensable for RQCsub_SHORT.

A previous study proposed that CAT tails aid degradation by extending the RQC substrate so that lysine residues buried in the ribosome exit tunnel emerge from the ribosome and can be ubiquitylated by Ltn1²⁵. RQCsub_LONG has its most C-terminal lysine residue 24 amino acids away from the stall sequence, placing it in the 35–40 amino-acid-long exit tunnel²⁹ at the point of stalling. However, mutation of this single buried lysine to arginine (which cannot be ubiquitylated) preserved the bulk of CAT tail dependence (from 24% to 19%) (Fig. 1f; Supplementary Fig. 1f). Similarly, mutation of ten lysine residues in the C-terminal half of RQCsub_LONG maintained CAT tail dependence (from 24% to 22%) (Fig. 1f). Although these mutations placed the proximal lysine 150 residues from the stall sequence, ltn1Δ still stabilized this substrate (Supplementary Fig. 1f). These data suggest that Ltn1 activity is not restricted to lysine residues close (in primary structure) to the exit tunnel. Therefore, CAT tails can mediate degradation of RQC substrates without displacing lysine from the exit tunnel.

Substrate structure affects degradation

We next sought to find the properties of RQCsub_SHORT and RQCsub_LONG that drive their differences in CAT tail dependence. These substrates differ in their capacity to cotranslationally fold (Fig. 2a). After stalling occurs, the linker in RQCsub_LONG is long enough to displace ten of the eleven GFP beta strands out of the exit tunnel, which enables the nascent GFP to adopt a stable conformation³⁰. By contrast, the linker in RQCsub_SHORT can only displace nine beta strands, preventing the formation of this stable conformation³⁰. Thus, a difference in folding states between RQCsub_SHORT and RQCsub_LONG may account for their different CAT tail dependence.

Figure 2 |. Conditions that favor RQC substrate folding increase CAT tail dependence.

a, Cartoon of folding states for RQCsub_SHORT and RQCsub_LONG, emerging from the ribosome exit tunnel. b, CAT tail dependence measurements at different incubation temperatures for RQCsub_LONG with two different GFP variants (S65T and superfolder). c, Normalized stability and CAT tail dependence for a model RQC substrate that can be measured using a split GFP, which features a spectrin R16 domain with or without a fold-disrupting F11D mutation. d-e, Normalized stability and CAT tail dependence of variants of RQCsub_LONG with added N-terminal domains (folded or unfolded spectrin variants and 6xFLAG and 6xnoK-FLAG) appended to GFP. For all plots, data is normalized as in Fig. 1f and error bars indicate s.e.m. from n = 3 independent cultures. P-values from two-tailed t-tests for particular contrast are displayed above the bars (8 d.o.f.).

To evaluate the hypothesis that CAT tails promote degradation of structured RQC substrates, we tested how modulating folding of the substrate changes CAT tail dependence. To modulate folding of RQCsub_LONG, we took advantage of the temperature-sensitive folding of the GFP variant (GFP-S65T) used in the substrate³¹. As incubation temperature increased (decreasing GFP-S65T folding capacity³¹), CAT tail dependence for RQCsub_LONG decreased (Fig. 2b). When we replaced GFP-S65T with the less temperature-sensitive superfolder-GFP^31,32, RQCsub_{LONG-superfolder} had high CAT tail dependence even at high temperatures (Fig. 2b). These data support a model where CAT tails enhance degradation of structured substrates but are dispensable for unstructured substrates.

To generalize our results beyond GFP, we next tested this hypothesis on spectrin, a protein whose co-translational folding is well-understood and easily modulated^33,34. We designed RQCsub_SPECTRIN using the same RFP-T2A module followed by a spectrin domain, a C-terminal GFP beta strand (β11), and lastly a polyarginine stall sequence. We quantified RQCsub_SPECTRIN levels using a “split-GFP” strategy, co-expressing the N-terminal GFP fragment from another transcript³⁵. We observed that RQCsub_SPECTRIN was 12% CAT tail dependent, while the folding-disrupting F11D mutation^33,34 eliminated CAT tail dependence (Fig. 2c; Supplementary Fig. 2a). These data suggest that the role of CAT tails in promoting degradation of structured RQC substrates is general, and not unique to GFP.

If CAT tails mediate degradation of structured RQC substrates, we would expect that addition of unfolded domains to a structured substrate would relax its CAT tail dependence for degradation. To test this hypothesis, we appended spectrin variants to the N-terminus of RQCsub_LONG. Folding-disrupted spectrin eliminated CAT tail dependence (from 24% to 1.7%) while folding-competent spectrin did not (from 24% to 28%) (Fig. 2d; Supplementary Fig. 2b). This finding prompted us to inquire whether the presence of unfolded domains alone or, instead, flexible lysine residues within the domains abrogate CAT tail dependence. To distinguish between these possibilities, we added unstructured FLAG-tag variants (with or without lysine residues) to the N-terminus of RQCsub_LONG. The lysine-containing FLAG-tag decreased CAT tail dependence (from 24% to 4.9%), 3.8-fold more than did the lysine-free FLAG-tag (19%) (Fig. 2e; Supplementary Fig. 2c). We thus propose that CAT tails preferentially enhance degradation of RQC substrates lacking lysine in unfolded regions.

Ltn1-independent degradation

While impairing CATylation affected the stability of some RQC substrates differently when Ltn1 was intact, impairing CATylation in the ltn1Δ background dramatically increased the stability of every substrate we measured (Supplementary Figs. 1,2). This indicated that CAT tails destabilize proteins independently of Ltn1. To evaluate whether this destabilization was due to CAT tail-induced proteolysis, we perturbed two mechanisms of proteolysis. Treatment with the proteasome inhibitor bortezomib increased the stability of RQCsub_LONG in ltn1Δ, but only when CATylation was intact (Fig. 3a). By contrast, disruption of vacuolar proteolysis with pep4Δ had no effect on the stability of RQCsub_LONG in ltn1Δ (Supplementary Fig. 3a). These data suggest that CAT tails target proteins for proteasomal degradation by Ltn1-dependent and -independent mechanisms.

Figure 3 |. CATylated RQC substrates are degraded independently of Ltn1.

a-c, Mean stability of RQCsub_LONG in indicated strains after perturbation with the proteasome inhibitor bortezomib, TEV protease co-expression, or HUL5 deletion (details in panel legend; results from one-tailed, one-way ANOVA tests with 4 d.o.f. indicated above bars). d, Analysis of RQCsub_LONG and associated ubiquitin by immunoprecipitation (IP) from indicated cell lysates and IB. e, Densitometry analysis of two bands seen in RQCsub_LONG IBs from whole cell extract, with results from a one-way ANOVA test (one-tailed, 4 d.o.f.) shown above the bars. Raw images are shown in Supplementary Fig. 4a. For all plots, error bars represent s.e.m. from n = 3 independent cultures.

Either the process of CATylation or CAT tails themselves could serve as an Ltn1-independent degradation signal. To distinguish between these possibilities, we employed a strategy to remove a substrate’s CAT tail in vivo without disrupting the process of CATylation. We co-expressed RQCsub_LONG and TEV protease in vivo to cleave RQCsub_LONG’s C-terminus and remove its CAT tail. TEV co-expression increased RQCsub_LONG’s mobility on SDS-PAGE (Supplementary Fig. 3b), confirming TEV activity in vivo. RQCsub_LONG was stabilized by TEV co-expression in cells with Rqc2 intact but not in cells incapable of CATylation (rqc2-d98a) (Fig. 3b). Taken together, these results indicate that CATylation destabilizes proteins, but removal of the CAT tail blocks this destabilization. We conclude that CAT tails themselves target RQC substrates for Ltn1-independent degradation.

We next searched a set of candidate genes from the ubiquitin-proteasome system to identify an E3 ligase that ubiquitylates CATylated proteins in the absence of Ltn1. Deletion of the proteasome-associated E3 ligase HUL5³⁶ stabilized RQCsub_LONG in the ltn1Δ background as much as removing CATylation (rqc2-d98a) (Supplementary Fig. 3c). Furthermore, during a cycloheximide chase, hul5Δ slowed decay of RQCsub_LONG in the ltn1Δ background as much as impaired CATylation (Supplementary Fig. 3d). This indicates that CATylated proteins are continuously degraded when Ltn1 is limiting. For CATylated RQCsub_LONG, this Ltn1-independent degradation required Hul5. The stabilization we observed after loss of Hul5 was not due to perturbed CATylation, as RQCsub_LONG had identical amino acid composition in ltn1Δ and hul5Δ ltn1Δ (Supplementary Fig. 3e). When Ltn1 was intact, hul5Δ significantly stabilized RQCsub_LONG with RQC2-WT but not rqc2-d98a (Fig. 3c). These modest effects observed in rqc2-d98a were likely non-specific, as hul5Δ also weakly stabilized a non-stalling degradation sequence (“degron”)³⁷ (Supplementary Fig. 3f). These results support a role for the E3 ligase Hul5 in Ltn1-independent degradation of a model CATylated protein.

Hul5 has E4 ligase activity, which extends existing ubiquitin conjugates to create poly-ubiquitin chains that boost proteasome processivity^36,38,39. It is thus possible that we identified Hul5 because degradation of CATylated proteins requires extension of a mono-ubiquitin mark left by another E3 ligase. A hallmark of E4 ligase activity is stabilization of the mono-ubiquitylated substrate after loss of the ligase^40,41, resulting in an 11kDa shift (His-Myc-Ubiquitin) by SDS-PAGE. However, purification of RQCsub_LONG in the ltn1Δ background revealed that hul5Δ, like rqc2-d98a, diminished detection of ubiquitylated conjugates without stabilizing an apparent mono-ubiquitylated species (Fig. 3d). Thus, Hul5 is required for an E3 ligase activity that ubiquitylates our model CATylated protein.

While hul5Δ did not stabilize a mono-ubiquitylated RQCsub_LONG species in the ltn1Δ background, hul5Δ intensified a crisp band within the CATylated smear, ~1kDa above the lowest band (Fig. 3d; Supplementary Fig. 4a). This band disappeared after disruption of CATylation (rqc2-d98a) (Fig. 3d), suggesting that the corresponding protein contains short CAT tails of relatively uniform size (~10–14 residues). To test whether short CAT tails support Hul5-dependent degradation, we monitored RQCsub_LONG stability in the presence of rqc2-d9a, an RQC2 mutant that produces short CAT tails (Supplementary Fig. 4b). In the ltn1Δ background, rqc2-d9a preserved the majority of RQCsub_LONG stabilization after hul5Δ that we observed for RQC2-WT (Supplementary Fig. 4c). We posit that short CAT tails mark proteins for destruction.

CAT tails can decrease the solubility of RQC substrates and drive the formation of aggregates^21–23. We asked whether blocking Ltn1-independent degradation of CATylated RQCsub_LONG potentiates its aggregation. As expected, blocking Ltn1-independent degradation of RQCsub_LONG with hul5Δ led to greater accumulation of the substrate, as seen by SDS-PAGE in hul5Δ ltn1Δ lysates relative to ltn1Δ lysates (Supplementary Fig. 4d, “input” fractions). Fractionation of ltn1Δ lysates containing RQCsub_LONG revealed that disruption of Ltn1-independent degradation with hul5Δ slightly increased RQCsub_LONG deposition into the insoluble pellet fraction compared to when HUL5 was intact (Supplementary Fig. 4d, “supernatant” and “pellet” fractions). However, the bulk of RQCsub_LONG in all ltn1Δ strains remained in the soluble fraction similarly to the soluble protein hexokinase (Supplementary Fig. 4d, “supernatant” and “pellet” fractions). Therefore, the majority of CATylated RQCsub_LONG was soluble whether or not Ltn1-independent degradation was functional.

CAT tails are degrons

We hypothesized that short, mRNA-encoded sequences of alanine and threonine residues are sufficient to confer the same degradation we observed for CATylated proteins. To test this hypothesis, we replaced our model RQC substrates’ stalling sequence with three non-stalling arginine residues (preserving the stalling sequence charge) and appended defined alanine and threonine sequences followed by a stop codon (Fig. 4a). These “hard-coded” CAT tails simulated natural CAT tails but had manipulable sequences and were not RQC substrates. If a hard-coded CAT tail suffices for degradation as observed for a naturally CATylated RQC substrate (Fig. 3c), its stability would be higher in hul5Δ cells than wild-type. We define this Hul5-dependence as stability_hul5Δ – stability_wt (Fig. 4a). While the arginine C-terminus control and alanine/threonine two-mers were not Hul5-dependent, “RRRATA” yielded weak Hul5-dependence (13%) (Fig. 4b). Doubling this motif to form “RRR(ATA)₂” increased Hul5 dependence to 80%, but the “RRR(ATA)₄” motif (54%) was weaker than “RRR(ATA)₂” (Fig. 4b). These results suggest that, similarly to nontemplated CATylation, short hard-coded CAT tails destabilize a model protein in the presence of Hul5. Thus, short CAT tails function as degrons.

Figure 4 |. CAT tails are degrons.

a, Schematic of the hard-coded CAT tail construct scaffold, terminating in “RRR” encoded by non-stalling codons, and definition of Hul5 dependence which is used as a measure for CAT tail degron strength. b-c, Normalized stability measurements of hard-coded CAT tail constructs with indicated C-termini in wild-type and hul5Δ cells. Data are presented as mean, normalized to the “RRR” alone construct in wild-type. d, Mean stability measurements from wild-type yeast expressing the indicated hard-coded CAT tails, treated with bortezomib. e, Analysis of “(ATA)₂” hard-coded CAT tail construct by IB after IP from lysate. For all plots, error bars represent s.e.m. from n = 3 independent cultures.

We then performed mutagenesis experiments to identify additional CAT tail properties that modulated their degron strength. We assessed the strength of these CAT tail degrons by measuring their Hul5-dependence. After making modifications to “RRRATATA,” we found that Hul5-dependence decreased after mutating alanine and threonine to glycine and serine (especially alanine adjacent to arginine), replacing basic residues (common sites of ribosome slowdown^42–44) adjacent to the CAT tail with non-basic residues, or capping the C-terminus with two leucine residues (a relatively stable C-terminal amino acid⁴⁵) (Fig. 4b,4c). This mutagenesis series suggests that CAT tails are effective degrons when: 1) adjacent to one or more basic amino acids, especially when alanine is directly adjacent, and 2) C-terminal.

We next investigated whether hard-coded CAT tails target proteins for ubiquitylation similarly to natural CAT tails. Stability increased upon bortezomib treatment for “RRRATATA” but not the arginine C-terminus control (Fig. 4d). Pull-down of the strongest CAT tail degron we tested, “RRR(ATA)₂”, recovered ubiquitin conjugates whose detection was abolished upon hul5Δ (Fig. 4e). As for the naturally CATylated RQCsub_LONG, hul5Δ diminished ubiquitin conjugation and did not stabilize a mono-ubiquitylated species (Fig. 4e). Thus, hard-coded CAT tails are sufficient to mark proteins that are not RQC substrates for ubiquitylation. This sufficiency suggests that Ltn1-independent degradation of RQC substrates can occur off the ribosome, unlike Ltn1-dependent degradation⁴⁶.

Dissection of CAT tail function

Our work supports the following model: Ltn1 efficiently ubiquitylates substrates that contain lysine in unstructured regions regardless of whether CATylation takes place (Fig. 5a). However, CATylation enhances Ltn1’s ability to ubiquitylate structured substrates that lack mobile lysine residues. If Ltn1 fails to ubiquitylate the substrate or is limiting, short (~1kDa) CAT tails mark that substrate for degradation off the ribosome. To support this model, we sought to test its key predictions and quantify the degree to which CAT tails contribute to Ltn1 function compared to acting as off-ribosome degrons.

Figure 5 |. Decomposition of the contribution of CAT tails to RQC substrate degradation.

a, Model for how CAT tails enable degradation of RQC substrates by Ltn1-enhancing and degron-forming activities. For unstructured substrates, ubiquitylation by Ltn1 occurs efficiently without CAT tails. CAT tails facilitate ubiquitylation of structured substrates. If Ltn1 fails, substrates released from the ribosome can be targeted for degradation, which depended on Hul5 for the substrates we measured. b, Mean stability of RQCsub_Rz substrate whose mRNA self-cleaves and leaves thus stalls ribosomes without a polybasic tract (see schematic above). Error bars indicate s.e.m. from n = 3 independent cultures. c, Scheme to decompose the contribution of CAT tails to Ltn1 and Hul5 (degron) function through combined perturbations to delete HUL5, remove long CAT tails with in vivo TEV cleavage, then mutate RQC2. d, Estimation of the contribution of Ltn1 (on-ribosome) and Hul5 (required for off-ribosome degron function) to CAT tail-mediated degradation of RQCsub_LONG and RQCsub_LONG with the C-terminal GFP lysine residue mutated (RQCsub_LONG-KlastR, as in Fig. 1f). Data are presented as mean, and error bars indicate s.e.m. from n = 3 independent cultures. Raw data are presented in Supplementary Fig. 5a. e, Spot assay for indicated strains grown with or without the translation inhibitor cycloheximide (CHX).

Our model predicts that CAT tails are dispensable for degradation of substrates that: 1) are unstructured, and 2) terminate in a non-basic residue (preventing Hul5-dependent degron activity). We constructed such a substrate, RQCsub_Rz, encoded by an mRNA containing a hammerhead ribozyme that self-cleaves before the stop codon to produce a GFP transcript with a truncated 3’ end that stalls ribosomes (Fig. 5b)⁴⁷. Before CATylation, the C-terminal residue of RQCsub_Rz is neutrally charged valine and there are too few residues between GFP and the C-terminus to enable formation of the stable GFP conformation³⁰. As predicted, neither disruption of CATylation (rqc2-d98a) nor loss of Hul5 stabilized RQCsub_Rz (Fig. 5b). Thus, unstructured substrates terminating in non-basic amino acids are not CAT tail dependent.

We next revisited RQCsub_LONG to dissect how CAT tails mediate its degradation via degron-forming and Ltn1-enhancing activities. We first estimated the contribution of CATylation to Hul5 (required for CAT tail degron activity on this substrate) and Ltn1 by measuring the stabilization caused by hul5Δ, assuming that Hul5 and Ltn1 activities are independent (Fig. 5c). Using this analysis, we estimated that Hul5 mediates 40% of CAT tail-dependent degradation and Ltn1 mediates the remaining 60% (Fig. 5d). We were additionally interested in analyzing the size of CAT tails that facilitated Ltn1-mediated degradation. To estimate the contribution of long CAT tails, we co-expressed TEV to cleave RQCsub_LONG from the ribosome (and evade Ltn1) if its CAT tails were long enough to expose the buried TEV-cleavage-site (greater than 21 residues) (Fig. 5c). TEV co-expression further dissected Ltn1-mediated degradation into 19% contributed by long CAT tails (TEV-sensitive) and 41% by short CAT tails (TEV-insensitive) (Fig. 5d). To assess the effect of lysine buried in the exit tunnel on CAT tail-mediated degradation, we repeated this analysis after mutating RQCsub_LONG’s exit tunnel-buried lysine. This mutation eliminated the contribution of short CAT tails to Ltn1 function, but increased the relative contributions of CAT tails to Hul5 and long CAT tails to Ltn1 (Fig. 5d). We conclude that CAT tails facilitate degradation of RQC substrates via Ltn1-dependent (on ribosome) and Ltn1–independent (off-ribosome) pathways. Short CAT tails can enhance Ltn1 function by exposing lysine residues buried in the exit tunnel, as proposed by Kostova et al²⁵. Long CAT tails, some of which may emerge from the exit tunnel, can enable Ltn1 to access lysine residues distal from the exit tunnel.

We asked whether the dual functions of CAT tails in degradation of RQC substrates contribute to cellular fitness when RQC is stressed by elevated substrate levels. We increased the influx of endogenous RQC substrates by growing cells in the presence of the translation inhibitor cycloheximide, which increases translational stalling. All RQC2 and LTN1 mutant strains grew equally well in the absence of cycloheximide, as assessed by a spot assay (Fig. 5e). While cycloheximide inhibited growth of all the strains we measured, simultaneous disruption of CATylation (rqc2-d98a) and Ltn1 resulted in a dramatic synthetic growth defect (Fig. 5e). This synergistic growth defect indicates that CATylation is crucial for cellular fitness in cells with increased stalling when Ltn1 becomes limiting.

Discussion

We propose that CAT tails are bi-functional marks for RQC substrate degradation. On the ribosome, CAT tails enhance Ltn1 activity on structured substrates. Off the ribosome, CAT tails form degrons that mark escaped RQC substrates for degradation when Ltn1 is compromised.

We observed that CAT tails enable Ltn1 to target structured RQC substrates, which are otherwise poor Ltn1 substrates. Structured RQC substrates may arise when translation fails after synthesis of domains competent for co-translational folding; for example, this might occur during non-stop translation when the ribosome synthesizes the entire reading frame but stalls within the poly(A) tail or during abnormally slow translation of inter-domain linkers⁴⁴. The inefficiency of Ltn1 activity on structured RQC substrates is consistent with the mechanism of RING E3 ubiquitin ligases. RING E3s bind to E2 ubiquitin ligases, stabilizing a conformation that primes the E2-ubiquitin bond for ubiquitin transfer^48–50. For ubiquitin transfer to occur, a nucleophile from the substrate (e.g. the ε-amino group from a lysine residue) attacks the E2-ubiquitin thioester bond⁵¹. Unstructured regions of an RQC substrate may have enough conformational flexibility for their lysine residues to reach the E2 active site and efficiently acquire ubiquitin. Conversely, lysine residues confined to rigid regions of the RQC substrate may lack the flexibility to move in three-dimensional space and access the E2 active site. We propose that CAT tails act as flexible linkers that allow RQC substrates to sample more three-dimensional space, enabling lysine residues within rigid regions of the substrate to acquire ubiquitin from the E2 active site.

The experiments herein succinctly test the model proposed by Kostova and colleagues regarding the role of CAT tails in degradation of RQC substrates²⁵. While this model posits that Ltn1 only ubiquitylates lysine residues immediately proximal to the exit tunnel, our results indicate that Ltn1 can access residues distal to the exit tunnel. LTN1 perturbation still stabilized substrates lacking residues close (in primary sequence) to the ribosome exit tunnel (Fig. 1f, Supplementary Fig. 1e). Furthermore, addition of lysine residues to the N-terminus of RQC substrates enhanced their degradation, provided the residues appeared in unstructured regions (Fig. 2d–e). In agreement with Kostova and colleagues, we found that short CAT tails can enhance degradation of an RQC substrate by exposing lysine residues buried in the ribosome exit tunnel (Fig. 5d), although this degradation mechanism represented roughly one-fifth of total CAT tail-mediated degradation of a model substrate (Fig. 1f). Ltn1 may efficiently ubiquitylate these newly-exposed residues because polypeptides typically emerge from the ribosome exit tunnel unfolded^52,53. We propose that RQC substrate tertiary structure, rather than primary structure, determines the efficiency of Ltn1-mediated degradation.

Our work uncovered a second novel function for CATylation: CAT tails form Ltn1-independent degrons. Rather than acting as an inert extension of the RQC substrate, the alanine and threonine residues in CAT tails mark RQC substrates for degradation off the ribosome. Because Ltn1 activity on RQC substrates is restricted to the 60S ribosome^5,46, in the absence of CATylation, cells would be left with a single opportunity to target RQC substrates for degradation. However, by serving as off-ribosome degrons, CAT tails indelibly mark RQC substrates for degradation throughout their lifetime. The CAT tail degrons we tested depended on Hul5 for ubiquitylation; Hul5 was thus a useful tool to genetically tune CAT tail degron activity. However, we stress that this does not indicate that all CATylated proteins rely on Hul5 for ubiquitylation. Proteasomal degradation of CATylated proteins could involve multiple E3 ubiquitin ligases or could occur though ubiquitin-independent means. Further studies will elucidate the precise degradation mechanism of CAT tail degrons.

Online Methods

Yeast strains, growth conditions, and plasmids

All yeast strains and plasmids used are listed in Supplementary Tables 1 and 2, respectively.Yeast cultures were grown at 30°C (unless otherwise noted) in YPD media or synthetic defined media with appropriate nutrient dropouts.Deletion strains were constructed in the BY4741 background via transformation with PCR products bearing antibiotic selection cassettes (NATMX6 or HYGMX6). These PCR products contained 40bp of homology to the genome on their 5’ and 3’ ends. Transformants were verified by genomic PCR.

RQC2 mutants were constructed by first replacing 1.5kb 5’ and 300bp 3’ of the RQC2 start codon with a NATMX6 cassette. This gap was repaired in transformants by transformation with a PCR product containing a LEU2 cassette and pRQC2-RQC2 variant N-terminus, amplified from plasmids containing these RQC2 variants. The resultant colonies were verified to have incorporated the desired RQC2 allele by genomic PCR and sanger sequencing.

Plasmids used in this study were cloned by the Gibson Assembly method⁵⁴ using NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs). All plasmids were derived from either the pRS315 or pRS316 vectors. For experiments where multiple exogenous proteins were co-expressed, all of the proteins were expressed from a single plasmid.

TEV treatment

1-liter log-phase yeast cultures at OD600 = 0.6–0.8 were harvested by vacuum filtration and flash frozen in liquid nitrogen. Frozen yeast pellets were lysed by cryo-grinding in a Spex 6750 Freezer Mill for 3 rounds of 10Hz for 2 minutes. The resulting “grindate” was re-solubilized in TEV buffer (25mM HEPES-KOH pH 7.4, 150mM KOAc, 0.5mM EDTA, 10% glycerol, 0.04% Antifoam-B, and EDTA-Free Pierce Protease Inhibitor Mini Tablets; added 1:1 vol:mass) to produce lysate. The lysate was clarified by centrifugation for 5 min at 5000g, then two rounds of 5 min at 10,000g. 1mM DTT was freshly added to the lysate before treatment with 2ul ProTEV Plus enzyme (Promega) per 23ul of lysate at 30°C for 3 hr in a thermocycler. 4x NuPage LDS Sample Buffer (Thermo Fisher Scientific) with 5% β-mercaptoethanol was added at 1:1 vol:vol before boiling at 95°C for 5 min to denature the samples.

Flow cytometry

Log phase yeast growing in synthetic defined media were measured on a BD Accuri C6 flow cytometer (BD Biosciences) with 3–5 biological replicates (independent cultures from separate clones) per condition. Data were analyzed in Matlab; a detailed, step-wise graphic example of this analysis is presented in Supplementary Fig. 6a–d. Plasmid-expressing yeast were selected by gating based on RFP fluorescence. Background signal bleeding from the RFP channel into the GFP channel was calculated using an RFP-only control strain expressing RFP-(T2A)₂-GST (Supplementary Fig. 1a) and subtracted before additional calculations. In the case where cultures were treated with bortezomib (LC Laboratories), treatment lasted for 4 hrs.

Statistical analysis

A two-tailed t-test for particular contrast was used to determine whether differences in CAT tail dependence values between substrates were significant. The null hypothesis for the t-test was: μ_{rqc2-d98a + substrate 1} – μ_{RQC2-WT + substrate 1} = μ_{rqc2-d98a + substrate 2} – μ_{RQC2-WT+ substrate 2}. The linear model was constructed using the raw fluorescence means from each replicate using the “lm” function in R and the linear hypothesis was tested using the “linearHypothesis” function in R.

A one-way ANOVA test (one-tailed) was used to determine whether mean measurements differed between substrates in two different conditions. The null hypothesis was: μ_{condition 1} = μ_{condition 2}.

Immunoprecipitation and immunoblot

In most cases where results were analyzed by immunoblots or immunoprecipitation, a version of the substrate without the expression-normalizing RFP-(T2A)₂ module was used. This choice was made to improve clarity by reducing the number of products detected on the gel. The two exceptions to this are Supplementary Fig. 3b and Supplementary Fig. 4d, where an RFP-(T2A)₂-containing substrate was shown in an immunoblot.

For whole-cell immunoblots, (0.375/OD600) × ml of log-phase yeast culture between OD600 = 0.4–0.8 were pelleted and resuspended in 15ul 4x NuPage LDS Sample Buffer with 5% β-mercaptoethanol. The sample buffer-resuspended pellets were lysed and denatured by boiling at 95°C for 5 min.

To detect ubiquitin conjugates, cells expressing a plasmid with the bidirectional pGAL1,10 promoter driving expression of His-Myc-tagged ubiquitin and the construct of interest were used. An additional plasmid that lacked a construct of interest and only contained His-Myc-tagged ubiquitin was used to assess non-specific ubiquitin detection. A 20ml culture of these cells was grown in SD media overnight containing 1% galactose and 2% raffinose to induce expression of tagged ubiquitin and the construct of interest. The culture was pelleted by centrifugation, weighed and resuspended in 100mM Tris pH 7.4, 10 mM EDTA at 500ul:25mg pellet. The resuspended culture was added dropwise into liquid nitrogen and cryo-ground as described above. The resultant grindate was resolubilized 1:1 mass:vol in buffer to produce lysate that had a final composition of 50mM Tris pH 7.4, 5mM EDTA, 20mM N-ethylmaleimide (added from a fresh 2M stock in ethanol; irreversibly inhibits deubiquitinases), 0.5% NP-40, 0.04% Antifoam-B, and EDTA-Free Pierce Protease Inhibitor Mini Tablets. The lysate was clarified by centrifugation at 5000g for 5 min and twice at 10,000g for 5 min. The clarified lysate incubated with buffer-equilibrated GFP-Trap magnetic agarose resin (Chromotek) (15ul slurry:25mg pellet) for 1 hr 4°C with rotation, washed 5 times in buffer, and eluted by boiling in 20ul 2x NuPage LDS Sample Buffer with 5% β-mercaptoethanol per 15ul resin for 95°C for 5 min.

For SDS-PAGE, all samples were run on Novex Nupage 4–12% Bis-Tris gels (Thermo Fisher Scientific) and transferred onto 0.45um nitrocellulose membranes (Bio-Rad) using a Transblot Turbo (Bio-Rad). For ubiquitin detection, the membrane was cut at the 50kDa marker to separate unmodified bands from potential poly-ubiquitylated conjugates; these two halves of the membrane were stained separately to enhance detection of the ubiquitin conjugates. Membranes were blocked for 1 hr with 5% milk in TBST at room temperature before staining with antibodies, either overnight at 4°C or for 4 hr at room temperature. Membranes were stained with the following primary antibodies: 1:2000 Pierce mouse anti-GFP (Thermo Fisher Scientific), 1:1000 rabbit anti-GFP (Life Technologies), 1:3000 rabbit anti-Hexokinase (US Biological), 1:1000 Pierce mouse anti-6xHis (Thermo Fisher Scientific), 1:1000 Pierce mouse anti-Myc (Thermo Fisher Scientific). The following secondary antibodies were used to stain membranes at 1:5000: IRDye 800CW donkey anti-mouse, IRDye 800CW goat anti-rabbit, IRDye 680RD goat anti-rabbit, or IRDye 680RD goat anti-mouse (LiCor Biosciences). A LiCor Odyssey (LiCor Biosciences) was used to scan immunoblots.

Amino acid analysis

Yeast lysates were produced as in the “TEV treatment” section, except the following buffer was used to resolubilize the grindate: 50mm HEPES pH7.4, 100mM NaCl, 0.5 mM EDTA. 400ul of clarified lysate was immunoprecipitated 20ul magnetic agarose GFP trap resin. After six washes, the resin was eluted by pipetting up and down in 40ul of buffer adjusted to pH 2.5. Eluates were subjected to amino acid analysis at the UC Davis Genome Center as described in Shen et al¹⁹.

Supernatant-pellet fractionation

The protocol for supernatant-pellet fractionation of yeast lysates was derived from Wallace et al⁵⁵. Yeast grindate were produced as described in the “TEV treatment” section. Soluble Protein Buffer (20mM HEPES-KOH pH 7.4, 120mM KCl, 2mM EDTA, 0.2mM DTT, 0.04% Antifoam-B, EDTA-Free Pierce Protease Inhibitor Mini Tablets; added 4:1 vol:mass) was used to resolubilize the grindate at 4°C. The resolubilized grindate was centrifuged at 4°C for 1 min at 3000g for clarification. This clarified lysate was used as the “input” fraction, 20ul of which was mixed with 20ul of 4x NuPage LDS Sample Buffer with 5% β-mercaptoethanol and denatured by boiling at 95°C for 5 min. 300ul of clarified lysate was spun at 100,000g for 20 min in a TLA-110 rotor (Beckman Coulter) at 4°C. The “supernatant” fraction was denatured as described for the input fraction. The pellet was washed with 500ul of Soluble Protein Buffer and vortexed for 2 min, then centrifuged again at 100,000g for 20 min at 4°C. The supernatant was thoroughly removed from the tube and 250ul of room temperature Insoluble Protein Buffer (20mM HEPES-KOH pH 7.4, 150mM NaCl, 2mM EDTA, 2%SDS, 2mM DTT, 0.04% Antifoam-B, 8M urea, EDTA-Free Pierce Protease Inhibitor Mini Tablets) was added to the pellet. The pellet was resuspended by dislodging with a pipette tip and vortexing for 25 min until the solution was clear. The resuspended pellet was centrifuged at 19,000g for 5 min at room temperature, and the supernatant was collected. This supernatant was defined as the “pellet” fraction, and was denatured as described above for the input fraction. 5ul of the denatured fractions were run on SDS-PAGE.

Yeast spot assay

Late log phase yeast grown in YPD were diluted to OD600 = 0.4 and 200ul of culture was placed in a sterile 96-well plate (Greiner Bio-one). 4 1:10 serial dilutions of the cultures into YPD were made in the plate. 10ul of yeast culture from the wells were transferred to YPD agar plates without drugs or YPD agar plates with 50ng/ml or 100ng/ml cycloheximide (Sigma-Aldrich). After drying, the plates were incubated at 30°C. The drug-free and 50ng/ml cycloheximide plates were imaged after two days of growth. The 100ng/ml cycloheximide plate was imaged after two weeks of growth.