Chaperone-Directed Ribosome Repair after Oxidative Damage

Rps26 is a major target of oxidative damage

While several studies have identified ribosomal proteins (RPs) as targets of oxidation,^19,22-25 the significance of this finding has not been explored. To characterize oxidative damage to RPs, and dissect how cells mitigate this damage, we first re-analyzed previous data and asked if ribosomes are preferentially oxidized. Indeed, compared to all proteins, cysteines in RPs are preferentially oxidized upon exposure of yeast cells to H₂O₂ or during chronic oxidative stress in a yeast mutant defective in mitochondrial protein import, Mia40-4int¹⁹ (Figure S1A & B). Similarly, in Drosophila embryos, which experience oxidative stress, RPs are also preferentially oxidized²⁵ (Figure S1C). Two proteins were among the most oxidized proteins in all cases: Rps26 and Rpl10 (Tables S1&S2) ^19,22-25, indicating that they were most sensitive to oxidative stress. Rps26 binds mRNA directly on the platform of the small (40S) subunit (Figure 1D), thereby helping to establish the mRNA sequence preference during translation-initiation captured in the “Kozak-sequence”.³⁶ In contrast, Rpl10 controls the position of the subunits with respect to each other,³⁹ and forms part of the large subunit’s active site.

What these data do not reveal is the fate of the ribosomes containing oxidized RPs. The damaged ribosomes might stall on mRNAs and then be degraded, as shown for ribosomes that cannot initiate.^5,6 Alternatively, our recent discovery that during high salt or pH stress, Rps26 can be extracted from ribosomes via its chaperone Tsr2,³⁵ to generate Rps26-deficient ribosomes with altered mRNA specificity,³⁶ opened the possibility that the damaged protein could be removed and replaced with a newly-made functional version in a chaperone-dependent manner.

To test this hypothesis, we first confirmed Rps26 oxidation in live yeast cells, using a cell-permeable chemical probe for detecting cysteinyl oxidation in situ, called BTD.^22,40,41 BTD covalently attaches to sulfenic acid, the direct product of thiolate oxidation by H₂O₂. We treated yeast with 1 mM H₂O₂, which activates a defined set of redox pathways,^42,43 followed by addition of a biotinylated BTD-analog (bio-BTD) for facile detection. Lysates were prepared and resolved on sucrose gradients. Western blotting demonstrates BTD-tagged proteins of the size expected for RPs in the 80S ribosome peak (Figure 1A, top), and by blotting for individual RPs we identified Rps26 as a major BTD-reactive band (Figure 1A, middle and bottom), indicating its susceptibility to oxidation. Of note, several proteins, including Rps26, were BTD-labeled in the absence of exogenous peroxide, consistent with constitutive RP oxidation. These findings demonstrate the susceptibility of RPs generally, and Rps26 specifically to cysteinyl oxidation in live cells.

To confirm the preferential oxidation of Rps26, and identify the oxidation sites, we exposed purified ribosomes to H₂O₂, before adding iodoacetamide (IAA). IAA alkylates thiolates and a decrease in alkylation, identified in mass-spectrometry (MS) experiments, is indicative of cysteine oxidation. Among the peptides identified by LC-MS/MS, ~ 9% or 13% of cysteines showed a decrease in IAA-labeling after treatment with 1 mM or 10 mM H₂O₂ treatment, respectively (Figure 1B & Table S3), indicating that most cysteines in RPs are resistant to oxidation by H₂O₂. In contrast, 50% and 90% of cysteine-containing peptides in Rps26 had decreased IAA labeling at 1 and 10 mM H₂O₂, respectively (Figure 1C). C74 and C77 were identified as oxidized (peptides including C23 and C26 were not recovered, as previously seen ¹⁹). Furthermore, Rps26 was the only RP where decreased IAA labeling was accompanied by an increase in peptides containing a hyperoxidized form of cysteine (−SO₃H). This proteomic analysis, together with in situ probe labeling, demonstrate that Rps26 is preferentially oxidized in vivo and in vitro.

Rps26 oxidation leads to Zn release and protein inactivation

The four cysteines in Rps26 chelate the Zn²⁺ ion of a structurally important Zn-finger motif (Figure 1D & Figure S1D). We therefore hypothesized that cysteinyl oxidation would lead to Zn²⁺ release, and Rps26 inactivation. To test this hypothesis, we purified recombinant Rps26 (Figure S1E), and quantified bound Zn²⁺ using the PAR-assay.⁴⁴ Indeed, addition of H₂O₂ increased the rate and extent of Zn²⁺ release (Figure 1E & Figure S1F), demonstrating redox-sensitive Zn²⁺ binding to ~85% of the protein.

To test if the Zn-free protein is functional, we mutated each cysteine to serine, and tested for complementation of a Rps26-deficient yeast strain. This showed that C23 and C74 are essential for cell viability (Figure 1F). Moreover, purified recombinant Rps26_C23S and Rps26_C74S were essentially Zn-free (Figure 1E & Figure S1G-H). C26 was non-essential, and mutation of C77 led to a strong growth defect. Together, these data strongly suggest that Zn-binding to Rps26 is essential for its function.

Tsr2 can selectively release Rps26 from ribosomes when its binding is weakened by loss of a Mg²⁺ ion at the RNA-Rps26 interface³⁵. Because the Zn-finger forms part of the 40S interface (Figure 1D), we hypothesized that Tsr2 could similarly release oxidized Rps26 (Rps26_ox) lacking Zn²⁺. To examine this idea, we incubated untreated or 40S exposed to H₂O₂ with recombinant, purified Tsr2, and then used co-sedimentation to assess Rps26 release.³⁵ Rps26 was released from 40S subunits in a Tsr2-dependent manner, and the extent of release increased with H₂O₂ concentration (Figure 2A). Release was specific for Rps26 and not observed for Rps10, demonstrating the overall integrity of the subunits, consistent with MS results. Additionally, release required the ability of Tsr2 to bind both Rps26 (Tsr2_DWI^28,35) and the 40S subunit (Tsr2_K/E³⁵), demonstrating active release, rather than ‘catching’ of released Rps26 (Figure 2B). Moreover, 40S oxidation, and not Tsr2 oxidation was required for this process (Figure S2A). Furthermore, H₂O₂- and chaperone-dependent release was specific to Tsr2/Rps26, as neither Rps2 nor Rps3 were released by their chaperones, Tsr4 and Yar1, respectively (Figure S2B-C). Thus, these data demonstrate specific Tsr2-dependent release of Rps26_ox from 40S ribosomes in vitro.

We next investigated whether oxidized 40S could be repaired by replacing Rps26. For this, we incubated ribosomes from which Rps26^ox was released with recombinant Rps26•Tsr2. While Tsr2 delivered Rps26 to 40S, Zn-less Rps26_C23S and Rps26_C74S cannot be incorporated into Rps26-deficient ribosomes (Figure 2C). To test if Rps26_C23S and Rps26_C74S are incorporated into ribosomes in vivo we purified ribosomes from yeast strains expressing HA-tagged wt or mutant Rps26. While wt Rps26-HA is readily incorporated into ribosomes, Rps26_C23S and Rps26_C74S are not (Figure 2D). We did not test whether Rps26_C23S and Rps26_C74S were present outside of ribosomes (and therefore produced), as RPs outside of ribosomes are unstable and rapidly degraded²⁶. However, the mutant proteins can be readily produced in E.coli (Figures 1E and 2C), which is generally not possible for unstable proteins. Together, these data demonstrate that Tsr2 extracts non-functional Zn-less Rps26 and replaces it with intact Zn-bound Rps26.

Tsr2 is recruited to 80S ribosomes in vivo

We next tested whether Tsr2 was recruited to ribosomes upon H₂O₂ treatment in vivo. Indeed, sucrose gradient fractionation demonstrates that exposure of yeast to H₂O₂ leads to increased Tsr2 binding to 80S ribosomes (Figure 3A). This finding is specific to Tsr2, and not observed for the Rps3 chaperone Yar1, and complemented by in vitro binding studies showing increased affinity of Tsr2, but not Yar1, to oxidized ribosomes (Figure S2D-F).

Rps26 can be released from 80S and 40S ribosomes

Because in vivo most Tsr2 binds to 80S ribosomes rather than 40S ribosomes (Figure 3A), we tested whether 80S ribosomes are substrates for Tsr2-mediated Rps26 release. Indeed, Tsr2 can release Rps26_ox from 40S and 80S ribosomes, although more Rps26 is released from 40S ribosomes (Figure S2G). Nonetheless, the H₂O₂ concentration dependence shows that the maximal release differs (Figure S2H), suggesting that less Rps26 is oxidized in 80S ribosomes, consistent with the location of Rps26 at the subunit interface. To test if the Tsr2 affinity for oxidized 40S and 80S ribosomes differs, we compared the Tsr2 concentration dependence for Rps26 release from 40S and 80S ribosomes. Because in this experiment, oxidized ribosomes are substoichiometric and binding saturated by Tsr2, any difference in Rps26 oxidation does not matter. Rps26 release from 40S requires ~ 2-fold less Tsr2 than release from 80S (Figure 2E), indicating that its binding to 40S ribosomes is ~ 2-fold tighter. Nonetheless, because 80S ribosomes are much more abundant they are the relevant substrates in vivo.

Oxidized Rps26 is released from non-translating ribosomes

To test if Rps26_ox is released from ribosomes in vivo, we utilized bio-BTD⁴⁰ to label the oxidized protein and determine its half-life. Relative to total Rps26, Rps26_ox decreased over time and in a Tsr2-dependent manner (Figure 3B-C & Figure S3A), strongly suggesting that the decay of Rps26_ox reflects degradation of released Rps26 and not the entire ribosome.

To identify the substrate for Tsr2-mediated release of Rps26_ox we measured the kinetics of Rps26_ox decay from 40S, 80S, and polysomes. After exposure to exogenous peroxide, cells were incubated with bio-BTD, collected by centrifugation and resuspended in rich media without H₂O₂ and bio-BTD for 10, 30 or 90 minutes. Cells were harvested, lysates generated and fractionated on sucrose gradients (Figure 3D). Quantifying the distribution of Rps26_ox over time demonstrates that its disappearance from 40S and 80S ribosomes is faster than from polysomes (Figure 3E). This observation, together with the findings that Tsr2 binds to and releases Rps26 from 80S ribosomes, suggests that upon Rps26 oxidation, the damaged ribosomes first form idle 80S ribosomes. After this rate-limiting step, Tsr2 binds these 80S ribosomes to release Rps26_ox.

To further test this model, we probed the distribution of Rps26_ox over time when its release is impaired with the Tsr2_K/E mutation (Figure 3F). Indeed, as predicted from this model, Rps26_ox accumulates in 80S ribosomes when Rps26 release is impaired, demonstrating that the removal of the damaged ribosomes from the polysomes is chaperone-independent (Figure 3G).

The data also indicate that Rps26_ox in 40S, although a minor species, is even less stable than Rps26_ox in 80S (Figure 3E), consistent with the stronger binding of Tsr2 to 40S ribosomes observed in our in vitro experiments (Figure 2E).

Damaged ribosomes are repaired

Next, we tested if release of Rps26_ox allowed for repair of the ribosomes with newly made Rps26 (Figure 3H & Figure S3B). Addition of H₂O₂ was coupled to the induction of HA-tagged Rps26, which is fully functional³⁵ and can be readily distinguished by Western blotting. Ribosomes made prior to H₂O₂ addition were traced with Rps3-TAP, which allows for affinity purification.³⁶ Thus, by purifying preexisting ribosomes using the TAP-tag, and probing for Rps26-HA, we asked if new Rps26 was incorporated into old ribosomes when cells were exposed to oxidative stress. Indeed, >6-fold more newly-made Rps26-HA incorporated into pre-made ribosomes after H₂O₂ treatment (Figure 3H, I-top). Importantly, the incorporation of the newly-made Rps26-HA requires Tsr2 (Figure S3C), demonstrating that damaged ribosomes are repaired with newly made, functional Rps26 in a Tsr2-dependent process in vivo. The reduction in untagged Rps26 provides an estimate of 15-29% for the fraction of ribosomes that are targeted by this mechanism (Figure 3H&I-bottom,). Finally, constitutive Rps26-HA incorporation occurs even without H₂O₂ treatment, indicating the importance of ribosome repair for maintenance of functional 40S ribosomes.

Ribosome repair is needed for oxidative stress resistance

To examine the importance of ribosome repair, we next tested whether Tsr2 inactivation affects H₂O₂ resistance. Indeed, yeast expressing inactive Tsr2 mutants (DWI or K/E) are more sensitive to H₂O₂, even when overexpressing Rps26 to maintain full Rps26 occupancy (Figure 3J & Figure S3D-E). Moreover, Rps26-deficient yeast are sensitive to oxidative stress (Figure S3F-G), indicating that repair, not just release are required for oxidative stress resistance. Finally, in the absence of H₂O₂, Tsr2 mutant cells grow more slowly, even when Rps26 is overexpressed to compensate for effects on Rps26 incorporation (Figure S3H, ^28,35), consistent with constitutive ribosome repair.

Release and repair of oxidized Rpl10

Our analysis of previous data indicated that in addition to Rps26, Rpl10 is also oxidized in every model system analyzed to date (Tables S1&S2). Intriguingly, Rpl10, which is part of the P-site of the large subunit (Figure 4A) also has a dedicated chaperone, Sqt1.^38,45 To test the generality of ribosome repair, we therefore determined if Rpl10 could be released from mature ribosomes in an Sqt1-dependent manner.

To validate our in vitro system, we re-analyzed our MS data for Rpl10 oxidation. While C49 and C71 were not significantly oxidized and peptides including C8 were not recovered, C105 was significantly oxidized (Figure 4B & Table S3) and conserved to humans (Figure S4A). Next, we incubated purified Sqt1 and Puf6/Loc1, the chaperones for Rpl10^38,45 and Rpl43,⁴⁶ respectively with 60S subunits in the presence of H₂O₂. While oxidized Rpl10 (Rpl10_ox) can be released by Sqt1 (Figure 4C), Puf6/Loc1 did not release Rpl43 (Figure S4B) in vitro. Moreover, sucrose gradient analysis demonstrates that upon H₂O₂ addition Sqt1 was recruited to ribosomes in vivo (Figure 4D).

To determine whether Rpl10_ox degradation depended on Sqt1 we produced a mutant unable to bind Rpl10²⁹, Sqt1_E315A. This mutant displays a small slow growth defect, which is rescued by addition of excess Rpl10 (Figure S4C), consistent with a role of Sqt1 in delivery of Rpl10 to assembling ribosomes.^38,45 Next, we utilized the BTD probe in a pulse-chase experiment as described above for Rps26 to measure the rate of disappearance of Rpl10_ox in yeast strains containing wild type or mutant Sqt1. Importantly, while in wt yeast more than half of Rpl10_ox is degraded after 30 minutes, with Sqt1_E315A, Rpl10_ox is stable for over 90 minutes (Figure 4E-F). Thus, Rpl10_ox decay requires the activity of its chaperone Sqt1. Finally, quantitative growth assays demonstrated that yeast encoding Sqt1_E315A are sensitized to oxidative stress (Figure 4G), even when supplemented with excess Rpl10.

Oxidatively damaged RPs are extracted from ribosomes for repair with undamaged proteins

Our findings demonstrate the sensitivity of Rps26 and Rpl10 to oxidation, consistent with previous data.¹⁹ What was unanticipated was the finding that the damaged ribosomes are repaired rather than degraded (Figure 4H). Specifically, the data demonstrate that after exposure to oxidative stress, damaged ribosomes containing Rps26_ox or Rpl10_ox are removed from the translating pool to produce idle 80S ribosomes. After this rate-limiting step the Rps26- or Rpl10-specific chaperones Tsr2 and Sqt1 are recruited to ribosomes to release Rps26_ox or Rpl10_ox from ribosomes, allowing for their degradation. The Rps26- or Rpl10-deficient ribosomes are then repaired with undamaged protein. Thus, this pathway allows for specific, targeted repair of oxidatively damaged ribosomes.

Future experiments will address how damaged ribosomes are released from the translating pool and converted to idle 80S ribosomes. Notably, Rps26-deficient ribosomes can translate³⁶ suggesting the possibility that ribosomes with Rps26_ox could finish translation of their bound mRNA. How the damaged ribosomes are then blocked from re-initiation remains unknown, but we note that eIF3 binds the 40S platform, where Rps26 is located. ⁴⁷ Rps26 oxidation may block eIF3 binding, either directly, or because it leads to Tsr2 recruitment. The inability to recruit eIF3, which blocks binding of the 60S subunit,⁴⁸ would also explain why 80S ribosomes and not free 40S and 60S ribosomes accumulate under oxidative stress.

Rpl10 regulates the conformational changes associated with translocation, the movement of mRNA and tRNA through the ribosome.³⁹ Whether its oxidation would block this movement, leading to stalling of the ribosome, or rather increases the likelihood of a frameshift during this process, which would ultimately lead to nonsense-mediated decay, remains unknown. However, Rps26_ox or Rpl10_ox decay with different rates, suggesting that the mechanism by which idle 80S form differs depending on whether Rps26 or Rpl10 are damaged.

Rps26 and Rpl10 are the last proteins to be incorporated into the small and large ribosomal subunit, respectively.^2,38 Thus, extraction of these proteins should be possible without disruption of the entire subunit. Indeed, we have previously shown that Rps26-deficient ribosomes, which have important roles for metal ion homeostasis³⁶ are generated under conditions of high salt and high pH stress in a Tsr2-dependent manner.³⁵ We note that 12 additional RPs have individual chaperones,^{30,32,38,46,49-53} and of those 5 (Rps2, Rps3, Rpl3, Rpl4, Rpl23) are also oxidized in all but one of the model systems for which there is data (Tables S1&S2). Moreover, all but Rpl3 are located on the outside of the subunits. Thus, it is conceivable that ribosome repair is not confined to Rps26 and Rpl10, but that other RPs might be extracted and then repaired under oxidative stress in a chaperone-dependent manner as shown here for Rps26 and Rpl10.

While our data demonstrate that ribosome repair is stimulated upon exposure to exogenous peroxide, the data also demonstrate constitutive repair, likely to deal with the constitutive oxidative stress in actively growing cells¹³. Indeed, Tsr2_DWI, which can bind ribosomes but not release Rps26, accumulates on 40S and 80S ribosomes,³⁵ suggesting that Tsr2_DWI is trapped there as it cannot release Rps26. The concentration of Tsr2 relative to ribosomes suggests that at least 1-2% of all ribosomes are oxidized even without addition of exogenous peroxide.

How are defective ribosomes recognized to release only damaged RPs?

The Zn-finger of Rps26 is located at the interface with rRNA. Thus, its unfolding is expected to weaken Rps26 binding to ribosomes, allowing for its release. Interestingly, K19, which is adjacent to the Zn-finger residues C23/C26, is crosslinked to Tsr2,²⁸ but inaccessible when Rps26 is bound to ribosomes. Moreover, a beta-strand in Rps26, comprising amino acids 66-72, contributes to Tsr2 binding ²⁸ and is adjacent to C74 and C77. Thus, we hypothesize that unfolding of the Zn-finger allows for better access of Tsr2 to Rps26, thereby explaining our observation that Tsr2 binds more strongly to damaged ribosomes than undamaged ribosomes.

Similarly, in Rpl10, C105 is part of a loop that extends into the rRNA active site (Figure 1D). As cysteine oxidation to sulfenic acid introduces a negative charge, Rpl10_ox is expected to bind ribosomes more weakly than the undamaged protein, allowing for its preferential extraction. Thus, we hypothesize that the specific extraction of damaged RPs is due to their weakened binding to ribosomes.

Ribosome repair is important in neuronal cells

While the present work demonstrates the importance of this pathway in yeast, several lines of evidence suggest its conservation. First, Tsr2 and Sqt1 are conserved throughout evolution (Figure S4D-E). Moreover, exchange of RPs, including Rps26 and Rpl10, in ribosomes from dendrites has been previously described,^54,55 and is enhanced by exposure to oxidative stress.⁵⁴ While it has been speculated that this might be part of a repair mechanism, ruling out the possibility that the incorporation of new proteins reflected late assembly events is challenging without knowledge of an exchange mechanism. Our data not only validate these findings, but strongly suggest that the chaperone-directed ribosome repair we have uncovered is responsible for this process, which is especially important as dendrites are sensitive to oxidative stress,⁵⁶ especially as they age. Moreover, we note that neurons have a low translational load, with many mRNAs being actively translated by just a single ribosome.^57-62 This could preclude detection of damaged and dysfunctional ribosomes by collisions, ⁶⁵ the alternative to repair.

Why do ribosomes get repaired instead of degraded?

Previous work has shown that oxidatively damaged mRNAs are identified in a translation-dependent mechanism, dubbed RQC, to allow for their degradation.^20,21 Similarly, many oxidized proteins are degraded¹³. We speculate that one advantage to the repair rather than the degradation of ribosomes is that it conserves ribosome numbers. Because ribosome assembly ceases under oxidative stress^18,63 degradation of damaged ribosomes would lead to a rapid reduction in ribosome numbers. Because the detection of oxidized mRNAs by RQC malfunctions when ribosome numbers are reduced,¹¹ this would impair the detection and degradation of oxidatively damaged mRNAs. Thus, ribosome repair during oxidative stress ensures the integrity of both ribosomes and mRNAs.

In addition, many mRNAs in neurons⁵⁷ and developing cells⁶⁴ are translated by a single 80S ribosome, thus precluding collisions as a mechanism to detect partially functional ribosomes.⁶⁵ Thus, ribosome repair might have evolved to address the problem of ribosome damage when translational load is low.

Limitations of the Study

Here we describe the discovery of ribosome repair that allows for the release of oxidatively damaged RPs and repair of the ribosomes with new proteins. To produce enough demand for ribosome repair to enable its study, we incubated the cells with 1 mM H₂O₂, a concentration typically used to study oxidative stress in yeast.^42,43 Under these conditions about 20% of all ribosomes are subject to oxidation and repair. Nonetheless, the data also show that this is a constitutive pathway in actively growing cells without the addition of exogenous stress. However, just how important it is in those cells, what fraction of ribosomes are damaged even without additional oxidative stress, and what the consequences for translation and protein homeostasis are when repair is impaired, remains unclear and will require future study.

RESOURCE AVAILABILITY

EXPERIMENTAL MODEL AND SUBJECT DETAILS

QUANTIFICATIONE AND STATISTICAL ANALYSIS

We used standard statistical analyses (unpaired t-test) as implemented in Prism 9 (Graphpad) to analyze the data in here. Significance was defined as a p≤0.05, and all relevant data were included in the analyses. Additional detail, including the number of replicates, precision measures etc., can be found in the legend for each relevant Figure.

METHOD DETAILS