RNF10 and RIOK3 facilitate 40S ribosomal subunit degradation upon 60S biogenesis disruption or amino acid starvation

Identification of translation components that alter iRQC pathway activation

Previous studies have demonstrated that the addition of several compounds that inhibit translation initiation or activate the ISR results in uS3 and uS5 ubiquitylation.^15,17,18 Overexpression of the critical iRQC-specific E3 ubiquitin ligase RNF10, or genetic ablation of the antagonizing deubiquitylating enzyme USP10, results in degradation of the entire 40S subunit in both normal and stressed conditions.^16–18 However, other factors that contribute to iRQC activation or downstream 40S decay have not been characterized in higher eukaryotes. Because most characterized pharmacological iRQC activators inhibit translation initiation, we hypothesized that genetically perturbing translation initiation or ribosome biogenesis in a manner that impairs scanning or initiation at start codons may result in iRQC activation. To test this hypothesis, we depleted several known initiation factors and probed the extent of uS3 and uS5 ubiquitylation in the presence and absence of harringtonine (HTN), which was previously shown to stimulate uS3 and uS5 ubiquitylation. While knockdown of eIF5B, DDX3X, and DHX29 modestly decreased HTN-stimulated uS5 and uS3 ubiquitylation, eIF4A1 knockdown nearly completely inhibited ribosome ubiquitylation upon HTN treatment (Figure 1A). In stark contrast with what was observed upon eIF4A1 knockdown, knockdown of the 60S biogenesis factor eIF6 induced uS3 and uS5 ubiquitylation both with and without HTN treatment (Figures 1A and S1A). The opposing effects of eIF6 or eIF4A1 knockdown on ribosomal ubiquitylation were also observed upon treatment with the ISR activator DTT or two inhibitors that block eIF4A1 activity during scanning (Figures 1B and S1B). These results suggest that reducing eIF4A1 levels, which impairs 43S initiation complex scanning of 5′ untranslated regions (UTRs), diminishes the ribosomal population that is ubiquitylated by RNF10.

Disruption of 60S biogenesis broadly stimulates iRQC

The ribosome biogenesis factor eIF6 associates with the pre-60S ribosomal subunit during early stages of 60S biogenesis within the nucleolus and remains associated until eIF6 is released during the final cytoplasmic steps of 60S maturation.³⁹ To assess whether eIF6 depletion uniquely stimulates uS3 and uS5 ubiquitylation, hallmarks of iRQC activation, we depleted several characterized 60S biogenesis factors that function throughout the 60S biogenesis pathway (Figure 1C). Polysome profiling confirmed that depletion of individual 60S biogenesis factors leads to the expected reduction in free 60S subunits and 80S ribosomes without altering free 40S subunit levels (Figure 1D). Consistent with this observation, knockdown of 60S biogenesis factors resulted in a reduction in overall protein synthesis as well as reduced cell proliferation (Figures S1C and S1D). Reducing the abundance of nearly every 60S biogenesis factor tested resulted in enhanced uS3 and uS5 ubiquitylation in otherwise unperturbed cells (Figures 1E and S1E). Further analysis using mass spectrometry confirmed a significant increase in abundance of the ubiquitylated uS5-K58 peptide upon 60S biogenesis disruption both at steady state and with HTN treatment (Figures 1F and S1F). Interestingly, depletion of EFL1 or SBDS, which act at the last step of 60S biogenesis to evict eIF6 from 60S subunits prior to 40S binding during translation initiation,⁴⁰ consistently failed to induce uS3 or uS5 ubiquitylation in multiple cell types (Figures 1E, 1F, and S1G). SBDS knockdown had no impact on overall protein synthesis or cell proliferation (Figures S1C and S1D), which suggests that SBDS plays a minor role in 60S biogenesis or overall translation activity, at least in the cell types tested here. These results indicate that disrupting 60S biogenesis at distinct steps in the pathway induces iRQC activation.

Disruption of 60S biogenesis leads to reduced 40S abundance

Reducing the abundance of translation-competent 60S subunits without impacting 40S levels results in subunit stoichiometry imbalance. Because 40S subunits engage with mRNA directly and recruit 60S subunits only at the onset of translation elongation, subunit imbalance leads to an increase in scanning preinitiation complexes that arrive at start codons and are unable to pair with 60S subunits. We hypothesized that reducing the abundance of 60S may result in a concomitant degradation of 40S subunits to partially correct the subunit imbalance. To test this hypothesis, we employed quantitative mass spectrometry to precisely measure the abundance of all ribosomal proteins in cells with and without 60S biogenesis factor depletion. The median abundance of all 60S proteins was reduced upon knockdown of five different 60S biogenesis factors, and the ratio of 60S proteins to 40S proteins was reduced upon knockdown of nearly all tested 60S biogenesis factors (Figures 2A and 2B). We confirmed the knockdown of 60S biogenesis factors using mass spectrometry and also observed that RSL24D1, eIF6, and GTPBP4 knockdown resulted in reduced levels of other 60S biogenesis factors known to act at similar steps during biogenesis (Figure S2A). Interestingly, 40S ribosomal protein abundance was also reduced upon knockdown of 60S biogenesis factors (Figure 2A). RSL24D1 and GTPBP4 knockdown resulted in a consistent reduction in all measured 60S proteins and nearly all 40S proteins (Figure 2C). This result suggests that the abundance of the entire 40S or 60S subunit is reduced upon 60S biogenesis disruption rather than a subset of ribosomal proteins. These findings are also consistent with previous observations that 60S ribosomal protein depletion in S. cerevisiae results in reduced 40S ribosomal protein abundance.⁴¹

Ribosome traffic patterns become altered upon 60S biogenesis disruption

Non-stoichiometric reduction in 60S levels relative to 40S is hypothesized to increase the amount of preinitiation complexes engaged with transcripts that fail to transition to elongating 80S ribosomes.²⁴ However, it is possible that unknown pathways prevent 40S engagement with mRNA when the 60S:40S ratio becomes imbalanced to prevent the accumulation of translation preinitiation complexes engaged with mRNA. Polysome profiling upon 60S biogenesis disruption demonstrated a reduction in free 60S, monosome, and polysome abundance without altering the abundance of free 40S (Figure S2B). This result suggests that 40S does not accumulate in its free form when 60S biogenesis is impaired. However, as observed previously, inhibiting 60S resulted in the formation of half-mer ribosome populations that are not observed in control knockdown cells (Figures 2D and 2E and S2B).^{24–26,42,43} These half-mers migrate between canonical monosome, disome, and trisome peaks upon sucrose gradient separation. Previous studies have hypothesized that these half-mers arise from mixtures of 40S-containing preinitiation complexes and 80S ribosomes concurrently engaged on a single transcript.²⁴ To more precisely define the ribosomal protein content within these half-mer populations, we separated extracts from control and GTPBP4-depleted cells on sucrose gradients and collected fractions enriched for half-mers vs. canonical monosome, disome, and trisome populations (Figures 2E and S2B). Subsequent analysis by mass spectrometry revealed that fractions comprising monosomes, disomes, or trisomes (fractions 9, 12, and 14) contained equivalent levels of 40S and 60S proteins as would be expected from 80S enriched ribosomes (Figure 2F). However, half-mer enriched fractions from GTPBP4-depleted cells displayed increased 40S:60S ratios relative to control knockdown (Figure 2F). Summing 40S and 60S protein intensity from fractions encompassing the monosome peak to the end of the gradient (fractions 9–23) revealed an increase in 40S:60S ratio upon GTPBP4 knockdown (Figure 2F). Taken together, these results suggest that lowering 60S levels relative to 40S levels results in enhanced engagement of preinitiation complexes on transcripts with elongating 80S ribosomes. Because 60S biogenesis disruption results in 40S ubiquitylation, these altered ribosome traffic patterns may contain aberrant ribosomal species that are targeted for ubiquitylation and subsequent degradation.

RNF10 is required to reduce 40S abundance upon 60S biogenesis disruption

To identify stress-responsive pathways that may be activated upon 60S biogenesis disruption, we interrogated our proteomic datasets to identify proteins whose abundance reproducibly increases upon 60S biogenesis factor knockdown. GTPBP4 knockdown results in broad changes to cellular protein levels, including the expected decrease in 60S ribosomal proteins (Figure 3A). RNF10 was among the proteins whose abundance increased upon 60S biogenesis factor knockdown (Figures 3A, 3B, and S3A–S3D). GTPBP4 knockdown resulted in increased RNF10 protein levels without significantly altering RNF10 mRNA abundance, implicating a post-transcriptional regulatory mechanism (Figure S3E). Previous studies have shown that constitutive iRQC activation leads to 40S subunit degradation in a manner that requires RNF10-dependent uS3 and uS5 ubiquitylation.^17,18 To directly examine if the reduction in 40S levels upon 60S biogenesis disruption requires RNF10 activity, we depleted GTPBP4 or RSL24D1 in our previously generated and characterized RNF10-knockout (KO) cells. Quantitative mass spectrometry analysis confirmed that both 60S and 40S proteins were reduced upon RSL24D1 and GTPBP4 knockdown (Figure 3C). While loss of RNF10 did not alter the reduction of 60S proteins upon 60S biogenesis disruption, 40S degradation was significantly attenuated upon RSL24D1 or GTPBP4 knockdown in RNF10-KO cells (Figure 3C). We confirmed that the reductions in GTPBP4 or RSL24D1 protein levels were similar in parental and RNF10-KO cells and that RNF10 abundance increased in the parental cells with 60S disruption (Figures S3F and S3G). These results demonstrate that RNF10 is required to reduce 40S abundance when 60S biogenesis is impaired.

RNF10 translation is induced upon iRQC activation in response to 60S biogenesis impairment

ISR activation results in an overall reduction in total protein synthesis while also stimulating translation of specific stress-responsive transcripts.⁴⁴ The typical mechanism governing this translation regulation involves alternative utilization of upstream open reading frames (uORFs) within the 5′ UTRs of stress-responsive transcripts.⁴⁵ To investigate possible RNF10 post-transcriptional regulatory mechanisms that are utilized to increase RNF10 protein levels upon 60S biogenesis disruption, we examined the RNF10 transcript for the presence of uORFs. The 5′ UTR of RNF10 contains three AUG start codons, which display conservation from humans to frogs (Figure S3H). Interrogating available RiboSeq datasets revealed the presence of at least four translated uORFs within the annotated RNF10 5′ UTR (Figure 3D).^46–49 We also observed a relative paucity of ribosome protected fragments within the RNF10 coding sequence (CDS) compared to the uORFs, which suggests that the uORF sequences within the RNF10 5′ UTR may repress translation of the RNF10 CDS. The two more N-terminal RiboSeq peaks within the RNF10 5′ UTR contain non-canonical initiation codons, whereas putative uORF4 and uORF5 (Figure 3D) are AUG initiated in a reading frame distinct from the RNF10 CDS start codon (Figures 3D and S3H). To test the effect of these 5′ UTR elements on RNF10 translation, we generated a translation reporter placing the RNF10 5′ UTR with the RNF10 CDS initiation sequence upstream of a dual luciferase reporter (Figure 3E). Mutation of all five putative uORF initiation codons increased Renilla luciferase (Rluc) expression relative to the IRES-driven firefly luciferase (Fluc) by more than 9-fold compared to the wild-type (WT) RNF10 5′ UTR, demonstrating that the uORFs repress RNF10 translation (Figure 3E). Systematic elimination of all initiation codons in isolation and all possible combinations revealed a complex regulatory structure for individual uORF sequences (Figures 3E and S3I). To test if 60S biogenesis disruption impacts RNF10 uORF-mediated translational repression, we depleted GTPBP4 and examined Rluc:Fluc expression from the WT and uORF initiation codon mutant reporters. Reducing GTPBP4 protein levels increased the Rluc:Fluc expression of the WT 5′ UTR reporter to an extent similar to the increased endogenous RNF10 protein levels (Figure 3F). Collectively, our results indicate that RNF10 translation is constitutively repressed but translationally upregulated in response to 60S biogenesis disruption by a complex and conserved uORF-dependent regulatory mechanism.

40S subunit reduction resulting from impaired 60S biogenesis requires eIF4A1

Our demonstration that HTN-induced uS3 and uS5 ubiquitylation is abrogated upon eIF4A1 knockdown indicated that active scanning of 5′ UTR sequences within transcripts may be required for 40S ubiquitylation and decay. Consistent with this hypothesis, co-knockdown of GTPBP4 and eIF4A1 prevented the reduction in 40S levels observed upon GTPBP4 knockdown alone (Figures 4A and S4A). As expected, eIF4A1 knockdown did not alter the reduction in 60S levels observed upon GTPBP4 knockdown alone (Figure 4A). Further, the increase in uS3 and uS5 ubiquitylation upon 60S biogenesis disruption was reduced with eIF4A1 knockdown as measured by immunoblotting and mass spectrometry (Figures 4B and S4B). These results suggest that mRNA engagement with the scanning preinitiation complex is required for RNF10-mediated 40S ubiquitylation when 60S abundance falls below 40S abundance.

Interestingly, RNF10 protein and mRNA abundance was decreased upon eIF4A1 knockdown in otherwise unperturbed cells (Figures 4C and S4C). Reduction of eIF4A1 also attenuated the increase in RNF10 protein levels observed upon GTPBP4 knockdown, indicating that eIF4A1-dependent scanning is required for RNF10 translational regulation (Figures 4C and S4C). Because RNF10 protein and mRNA levels are suppressed by eIF4A1 knockdown, it is possible that the observed reduction in RNF10-mediated 40S ubiquitylation and subsequent 40S degradation upon co-knockdown of eIF4A1 and GTPBP4 may be due to reduced RNF10 levels rather than reduced scanning activity. To discern between these two possibilities, we utilized 293 Flp-In cells with inducible RNF10 expression. We previously demonstrated that RNF10 overexpression alone results in reduced 40S abundance, a result that was reproduced here (Figures 4D and S4D). Knockdown of eIF4A1 reduced 40S protein decay upon RNF10 overexpression (Figures 4D and S4D), indicating that eIF4A1-dependent scanning of 5′ UTRs is required for RNF10-mediated 40S ubiquitylation and decay independent of its effect on RNF10 expression.

Ribosome splitting and canonical RQC factors are not required to reduce 40S abundance

The RNF10 ortholog in S. cerevisiae, Mag2, is required for the 18S NRD pathway, which degrades mutant 18S rRNA that fails to support translation elongation. This 18S NRD pathway also requires the activity of Dom34 and Slh1, which facilitate stalled 80S ribosome splitting in a manner distinct from canonical translation termination.³⁸ Further, a second ubiquitin ligase, Fap1, was identified as an 18S NRD factor that binds to slowly translating 80S ribosomes and extends polyubiquitin chains on Mag2 monoubiquitylated uS3.⁵⁰ We knocked down NFX1 or NFXL1, the two human proteins with the highest degree of sequence similarity to Fap1, to directly evaluate if these putative Fap1 homologs are required for 40S decay upon 60S biogenesis disruption. Unlike what was observed upon GTPBP4 knockdown in RNF10-KO cells, combined knockdown of NFX1 or NFXL1 with GTPBP4 did not reduce the 60S:40S ratio compared to GTPBP4 knockdown in isolation (Figures 5A and S5A). Further, the increase in RNF10 abundance observed upon GTPBP4 knockdown is unaffected by co-knockdown of NFX1 or NFXL1, which suggests a limited role for NFX1 or NFXL1 in the iRQC pathway (Figure S5B). To examine if known RQC pathway components function downstream of RNF10 ribosome ubiquitylation to reduce 40S abundance, we co-depleted the canonical RQC E3 ligase, ZNF598, or the ribosome splitting and rescue factors, ASCC2, ASCC3, PELO, HBS1L, and GTPBP2, with GTPBP4 and measured 60S and 40S ribosomal protein abundance by mass spectrometry. Knockdown of all tested RQC factors did not elevate 60S or 40S protein abundance relative to GTPBP4 knockdown alone (Figures 5B, S5C, and S5D). Interestingly, ZNF598, PELO, ASCC3, or GTPBP2 knockdown further reduced 40S protein levels when combined with GTPBP4 knockdown. Loss of RQC factors did not significantly alter the increased uS3 ubiquitylation observed upon GTPBP4 loss of function (Figure S5E). These observations suggest that canonical RQC factors are not required for 40S decay upon 60S biogenesis disruption and may antagonize the iRQC pathway by inhibiting 40S decay upon iRQC activation.

Reducing 40S levels attenuates iRQC activation upon 60S biogenesis disruption

Loss of 60S ribosomal proteins themselves also impairs overall 60S ribosome biogenesis and decreases 60S ribosomal abundance.^25,26,51,52 As such, we would expect to observe a similar reduction in 40S ribosomal protein abundance upon 60S protein knockdown. Consistent with previous results in yeast,⁴¹ uL18 (RPL5) and uL5 (RPL11) knockdown not only results in the expected decrease in 60S subunit abundance but also results in reduced 40S ribosomal protein abundance and increased RNF10 expression (Figures 5C and S5F). Interestingly, 40S ribosomal protein eS19 (RPS19) knockdown resulted in specific 40S subunit loss with no significant corresponding reduction in 60S protein levels (Figure 5C). This result suggests that RNF10-mediated 40S ubiquitylation is specifically activated when the 60S:40S ratio is reduced. We hypothesized that restoring 60S:40S stoichiometry by concomitant reduction of 40S and 60S biogenesis factors would reduce unpaired 40S levels and therefore reduce 40S ubiquitylation. Indeed, co-depletion of 40S biogenesis factors or ribosomal proteins with GTPBP4 either reduced or completely blocked uS3 and uS5 ubiquitylation compared to GTPBP4 knockdown alone (Figure 5D). This result indicates that the iRQC pathway is activated in response to an overabundance of 40S relative to 60S levels, and correcting this imbalance by depleting 40S reduces iRQC activation.

Previous studies in S. cerevisiae suggest that 60S binding to immature 40S intermediates in the cytoplasm assists in the final stages of 40S biogenesis.^29,30 As such, lowering 60S levels may impact late 40S biogenesis steps, resulting in the accumulation of defective 40S subunits that become incorporated into 80S ribosomes. It is possible that these improperly assembled 40S subunits represent the ribosomal species targeted by RNF10. If this is the case, then impairing the final steps of 40S maturation would stimulate RNF10-dependent 40S ubiquitylation. To test this hypothesis, we targeted the late 40S biogenesis factors NOB1, PNO1, and RIOK1, which are required for the final 18S rRNA maturation step and eS26 (RPS26) incorporation into the mature 40S subunit.³² PNO1 but not NOB1 or RIOK1 knockdown reduced 40S protein levels (Figures S5G and S5H). Knockdown of these late 40S biogenesis factors alone failed to induce uS3 or uS5 ubiquitylation, and combining PNO1, NOB1, or RIOK1 knockdown with GTPBP4 knockdown reduced uS3 or uS5 ubiquitylation (Figures 5E and S5G). As observed for eS19, depleting eS26 reduced 40S abundance and blocked the 60S-biogenesis-disruption-mediated uS3 and uS5 ubiquitylation (Figures 5E and S5G). Further, PNO1 and eS26 knockdown reduced RNF10 protein levels in untreated and GTPBP4-depleted cells (Figure 5F). Collectively these results demonstrate that inhibiting the late steps in 40S biogenesis does not induce iRQC activation but instead impairs activation. These results are consistent with our hypothesis that RNF10 acts on scanning-competent 40S subunits rather than improperly assembled subunits.

RIOK3 is a previously uncharacterized crucial iRQC factor

To identify possible iRQC factors that act downstream of RNF10 ribosome ubiquitylation to facilitate 40S degradation, we interrogated our proteomics datasets for proteins that, like RNF10, increase in abundance upon 60S biogenesis disruption. The RIO kinase family member RIOK3 was among those proteins that reproducibly increased in abundance upon 60S biogenesis factor knockdown (Figures 6A and S6A). RIOK3 protein abundance is also elevated upon overexpression of WT, but not catalytically inactive, RNF10 (Figure S6B). Further, steady-state RIOK3 protein levels are reduced in RNF10-KO cells, and the increase in RIOK3 protein abundance observed upon knockdown of 60S biogenesis factors is significantly repressed in RNF10-KO cells (Figures S6A and S6B). The other RIO kinase family members, RIOK1 and RIOK2, have established roles in 40S biogenesis in both yeast and mammals.^53–59 RIOK3 evolved more recently and is found only in higher eukaryotes. While RIOK3 has been demonstrated to associate with 40S subunits, RIOK3 loss-of-function does not impair 40S biogenesis.⁵⁷ In addition, an N-terminal fragment of RIOK3 was identified in a screen for ubiquitin-binding proteins.⁶⁰ These observations together led us to hypothesize that RIOK3 may act as a ubiquitin reader downstream of RNF10-mediated uS3 and uS5 ubiquitylation.

To directly test if RIOK3 contributes to 40S degradation upon iRQC pathway activation, we depleted RIOK3 alone and in combination with GTPBP4 and measured 40S abundance. Similar to RNF10 loss, RIOK3 knockdown blocked the reduction in 40S but not 60S protein levels upon GTPBP4 depletion (Figure 6B). This result was replicated in RIOK3-KO cells, indicating that RIOK3 participates in 40S decay upon 60S biogenesis disruption (Figures S6C). The increase in RNF10 protein abundance upon 60S biogenesis impairment was not inhibited by RIOK3 loss of function (Figures 6D and S6D). Interestingly, uS3 and uS5 ubiquitylation was substantially increased by combining RIOK3 and GTPBP4 knockdown compared to GTPBP4 knockdown alone (Figure 6C). This suggests that loss of RIOK3 leads to elevated levels of ubiquitylated 40S ribosomes that cannot be degraded. These results demonstrate that ribosome ubiquitylation alone is not sufficient to facilitate 40S degradation and that RIOK3 acts downstream of RNF10-mediated ribosome ubiquitylation to facilitate 40S decay.

RIOK3 ubiquitin binding is necessary for 40S decay

Examination of the RIOK3 amino acid sequence, sequence conservation, and predicted AlphaFold structure reveals three general domains: a predicted ubiquitin-binding N-terminal domain, a conserved helical linker domain, and a RIO kinase domain (Figures 7A and S7A). The putative N-terminal ubiquitin-binding domain contains two inverted ubiquitin interaction motifs (MIU1/2).^61–63 AlphaFold3 modeling of the RIOK3 N-terminal domain with ubiquitin highlighted a binding surface comprising hydrophobic MIU residues known to interact with ubiquitin and the well-characterized hydrophobic patch on ubiquitin centered on I44 (Figures 7A and S7B–S7D). Consistent with structural modeling and previous data, WT RIOK3, but not a mutant version of RIOK3 that lacks the ubiquitin-binding domains (Δ1–153; ΔN hereafter), associated with polyubiquitin as demonstrated by co-immunoprecipitation experiments (Figures 7B and 7C). Mutation of two catalytic residues within the kinase domain (K290A/D406A, KM hereafter) did not impair ubiquitin binding (Figures 7B and 7C).

To test if RIOK3 ubiquitin binding or kinase activity was necessary for RIOK3 to facilitate 40S decay, we performed knockdown rescue experiments using RIOK3 transgenes that are resistant to small interfering RNAs (siRNAs) used to deplete endogenous RIOK3. Consistent with previous results, RIOK3 knockdown prevented the reduction in 40S protein levels observed upon GTPBP4 depletion. The reduction in 40S protein levels was restored upon expression of WT but not ΔN or KM RIOK3, indicating that both kinase activity and ubiquitin binding are necessary for RIOK3 to facilitate 40S decay (Figures 7D and S7G). Similarly, expression of WT but not ubiquitin-binding-deficient RIOK3 reduced uS3 and uS5 ubiquitylation observed upon GTPBP4 knockdown (Figure 7E). Interestingly, expression of the kinase mutant RIOK3 resulted in elevated uS3 and uS5 ubiquitylation, suggesting that kinase-inactive RIOK3 may block deubiquitylating enzyme access to ubiquitylated 40S (Figure 7E). The results establish that RIOK3 ubiquitin binding and kinase activities are required to facilitate 40S decay.

Depleting amino acids induces 40S decay

Our results, combined with results from others,^41,51,64 demonstrate that lowering 60S protein abundance results in 40S ubiquitylation and degradation. We previously demonstrated that uS3 and uS5 ubiquitylation is stimulated upon conditions that activate the ISR, which suggests that chronic ISR activation may also result in iRQC-dependent 40S decay.^15,17 To investigate this hypothesis, we cultured cells in medium without lysine and arginine, as this amino acid starvation paradigm is known to induce the ISR via ribosome p-stalk-mediated GCN2 activation.⁶⁵ We measured ribosomal protein abundance via mass spectrometry in cells grown in lysine/arginine-free medium over a 96-h time course. Imbalanced ribosome subunit stoichiometry due to reduced 40S relative to 60S (high 60S:40S ratio) was observed after 24 h of amino acid starvation (Figure 7F). This stoichiometry imbalance worsened after 24 h, with a more than 20% increase in 60S relative to 40S observed after 96 h of lysine and arginine depletion. The subunit stoichiometry imbalance observed upon amino acid starvation was blocked in either RNF10- or RIOK3-KO cells, indicating that 40S decay in this context is also mediated by iRQC factors (Figures 7F, S7E, and S7F). Despite the large difference in ribosome stoichiometry, the ISR is similarly activated in RNF10-KO and RIOK3-KO cells as judged by ATF4 protein expression (Figure S7H). Indicative of iRQC activation, RIOK3 levels increased over time when culturing cells in lysine/arginine-free medium (Figure S7I). These results establish that chronic ISR stimulation results in iRQC activation and subsequent RIOK3-dependent 40S decay. Collectively, our findings establish that diverse conditions that impair translation initiation result in RNF10-mediated 40S ubiquitylation, and RIOK3 acts as a ubiquitin receptor for these 40S subunits to facilitate their degradation (Figure 7G).

Biologically relevant contexts for iRQC pathway activation

A variety of defects within ribosomes or mRNA can impair ribosome progression that, left unresolved, would reduce the pool of available ribosomes and lower cellular protein biogenesis capacity.^2–4 Ribosome ubiquitylation has emerged as a key step in regulating ribosome-associated quality control (RQC) pathways.⁶⁶ A growing number of ubiquitin ligases facilitate site-specific ubiquitylation of individual ribosomal proteins in response to distinct perturbations that impede ribosome progression.^15,17,67,68 While ISR agonists, translation initiation inhibitors, and high-dose translation elongation inhibitors all induce RNF10-dependent uS3 and uS5 ubiquitylation, how this pathway is activated in the absence of chemical agonists was unclear. USP10 depletion results in chronic uS3 and uS5 ubiquitylation and reduced 40S abundance in the absence of stress, which suggests that RNF10-dependent ubiquitylation may occur as part of the normal translation cycle.^16,17 We demonstrate that either direct disruption of the 60S:40S stoichiometry by limiting 60S biogenesis activity or chronic ISR induction by limiting amino acid availability elevates uS3 and uS5 ubiquitylation and stimulates 40S decay. These findings illuminate biologically relevant cellular stress conditions that utilize this surprising 40S decay pathway to abrogate translational stress.

Mutations in a variety of genes encoding ribosomal proteins result in a diverse collection of human ribosomopathies.⁶⁹ Diamond-Blackfan anemia (DBA) is among these ribosomopathies where heterozygous loss-of-function mutations in many different ribosomal protein genes result in impaired erythropoiesis.⁷⁰ Studies examining the effect of DBA-associated ribosomal protein loss in human cells revealed that reducing 60S proteins results in loss of both large and small subunit ribosomal proteins, whereas depleting 40S ribosomal proteins had a minor effect on 60S ribosomal protein abundance.⁵¹ These observations mirror our results, suggesting that RNF10-dependent 40S decay occurs in a subset of DBA patients with mutations that selectively compromise 60S ribosome assembly. These observations are consistent with studies in yeast showing that the expression of non-functional 60S ribosomal protein mutants results in a general loss in both ribosomal subunits, whereas expression of 40S mutants has no corresponding effect on 60S abundance.⁴¹ Proteasomal gene expression is also induced solely with the expression of 60S mutants in these yeast studies. These data suggest the presence of conserved cellular mechanisms that specifically target excess 40S ribosomal proteins for degradation when 40S abundance exceeds 60S.

Acute ISR pathway activation stimulates RNF10-dependent uS3 and uS5 ubiquitylation in a manner that requires eIF2α phosphorylation.¹⁵ Here, we show that chronic ISR activation via amino acid starvation also results in 60S:40S imbalance through selective 40S decay (Figure 7F). Similar to what we observe when 60S biogenesis is blunted, this 40S decay pathway also requires RNF10 and RIOK3. We also observe a general decrease in both 40S and 60S abundance in RNF10- or RIOK3-KO cells, indicative of a global increase in ribosome flux through the autophagy pathway that is activated upon amino acid starvation (Figures S7E and S7F).⁷¹ The iRQC-dependent 40S decay appears to operate separate from this general autophagic ribosomal degradation pathway to further limit 40S abundance when protein biosynthesis building blocks become limiting. It remains unclear whether this conserved 40S decay pathway provides a selective advantage during starvation conditions. Because several distinct initiation stress conditions that lead to diverse ribosomal outcomes all lead to iRQC activation, it remains unclear what precise ribosomal species RNF10 recognizes for ubiquitylation. Structural studies under various initiation inhibition conditions will be necessary to determine how RNF10 recognizes 40S subunits in these distinct contexts (Figure 7G).

Similarities and distinctions between mammalian iRQC and yeast NRD pathways

The mammalian iRQC pathway bears some similarities to the established 18S NRD pathway in yeast. Mutations in 18S rRNA lead to uS3 ubiquitylation by the RNF10 ortholog, Mag2, at the same conserved residue of yeast uS3 that RNF10 utilizes for mammalian 40S decay.³⁸ While the initial ubiquitylation event and ubiquitin ligase are conserved, there are notable differences between the mammalian iRQC and the yeast 18S NRD pathways. First, knockdown of the proposed human homologs of Fap1, NFX1, and NFX1L fails to rescue the reduction in 40S levels observed upon 60S biogenesis disruption. Furthermore, the resolution of the 18S NRD pathway requires the ribosome disassociation factors Dom34 and Slh1, both of which play integral roles in the canonical RQC pathway. Knockdown of Pelota or ASCC3, the Dom34 and Slh1 homologs, did not impair iRQC-mediated 40S decay, indicating that stalled ribosome disassociation is not required to reduce 40S levels when 60S biogenesis is inhibited. This finding is consistent with in vitro studies demonstrating that the ASCC2/3 complex did not appreciably dissociate 80S ribosomes that were ubiquitylated by RNF10.⁷² Further, the yeast genome does not contain a clear RIOK3 ortholog. These observations collectively suggest that, despite similarities in the initiating ubiquitylation event, the downstream mechanisms leading to 40S degradation may have diverged during evolution.

It remains possible that other, NRD-like, pathways exist to target defective or improperly assembled 40S ribosomes for degradation. In support of this idea, previous studies identified the deubiquitylating enzyme USP16 as a late-acting 40S biogenesis factor. Depleting USP16 increased 40S ubiquitylation on a specific lysine residue of eS31 but failed to alter uS3 or uS5 ubiquitylation, implicating other ubiquitylation enzymes in the targeting of improperly assembled 40S.⁷³ Consistent with these findings, we show that loss of late 40S biogenesis factors does not induce iRQC activation and instead reduces uS3 and uS5 ubiquitylation upon 60S biogenesis loss. Further, recent studies documenting a role for the yeast Rio1 kinase in a 40S assembly quality control pathway show that this pathway co-opts Hel2, but not Mag2-dependent ribosome ubiquitylation.⁷⁴

A feedforward mechanism of iRQC activation and 40S decay

Our results demonstrate that iRQC capacity is generally regulated under iRQC-activating conditions by increasing the abundance of both RNF10 and RIOK3. RNF10 protein levels are regulated by a translation control mechanism that utilizes conserved uORFs within the 5′ UTR of the RNF10 transcript. Our reporter results indicate that RNF10 uORFs suppress RNF10 translation under normal cellular conditions. However, upon disrupting 60S biogenesis, translation from the RNF10 start codon is increased. This leads to an interesting model where conditions that favor stalled preinitiation complexes, the putative triggering event for iRQC, also inherently favor increased RNF10 expression. RIOK3 protein levels also increase upon conditions that stimulate iRQC pathway activation, including amino acid starvation (Figure S7I). Our findings suggest that RIOK3 binds ubiquitylated 40S ribosomes downstream of RNF10 activity, which positions RIOK3 as a crucial reader that helps interpret the ribosome ubiquitin code. RIOK3 uses tandem MIUs to bind ubiquitin. MIU domains from Rabex-5 and MINDY-1 have been shown to interact with ubiquitin using the canonical hydrophobic patch of ubiquitin.^61–63 AlphaFold structural modeling also predicts that the MIU domains of RIOK3 interact with ubiquitin using similar interaction surfaces (Figure 7A). The tandem MIU domains found in the MINDY-1 deubiquitylating enzyme show a binding preference for K48-linked polyubiquitin, and structural studies reveal that a single MINDY-1 MIU domain can interact with three separate ubiquitin entities with the polyubiquitin chain using distinct MIU binding sites.⁶³ Similarly, AlphaFold structural modeling predicts interaction of the RIOK3 tandem MIUs with multiple ubiquitin molecules (Figure S7B). These observations raise the intriguing possibility that the tandem MIU domains within RIOK3 may be used to engage 40S with multiple ubiquitin moieties that occur from concurrent uS3 and uS5 ubiquitylation events. This mode of ubiquitin engagement may provide the mechanistic rationale for why both uS3 and uS5 ubiquitylation events are required for iRQC-mediated 40S decay.

Limitations of the study

Using quantitative mass spectrometry to measure ribosome abundance, we document a 10%–20% decrease in 40S abundance upon iRQC pathway activation. The source of the variability when measuring 40S abundance could arise from a variety of both known and unknown factors. We do not know the extent of pathway activation in other cell or tissue types or how varying cellular conditions may modify iRQC pathway activity. For example, a 10-fold overexpression of RNF10 results in a 13% decrease in 40S abundance. These results suggest there are other limiting factors, like the specific 40S conformation that allows for RNF10-dependent ubiquitylation, that contribute to 40S decay. This study does not define the ribosomal species targeted by RNF10 and RIOK3. While our results suggest that the iRQC pathway acts on 40S subunits during scanning or initiation, RNF10 may recognize any number of related molecular species, including slowly translating 80S ribosomes or unique collision interfaces. Similarly, it is unclear how chronic ISR induction, which limits 43S recruitment to mRNA, results in iRQC pathway activation. Structural and biochemical studies will be necessary to determine the precise ribosomal species that recruits RNF10. We combine sucrose gradient density centrifugation and proteomic analysis to characterize half-mer-enriched fractions. While sucrose density centrifugation is routinely used to examine distinct ribosomal populations, separation by density alone cannot conclusively identify the molecular constituents of ribosomal populations within individual fractions.

Lead contact

Requests for further information and resources and reagents should be directed to and will be fulfilled by the lead contact, Eric Bennett ([email protected]).

Materials availability

All reagents generated in the study are available from the lead contact with a completed materials transfer agreement.

Data and code availability

• All raw mass spectrometry proteomics data have been deposited with the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers listed in the key resources table.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

METHOD DETAILS

QUANTIFICATION AND STATISTICAL ANALYSIS

All mass spectrometry experiments were performed in triplicate (n=3) as biologically distinct samples unless otherwise indicated in the figure legends. Median ribosomal protein abundance was determined by mass fraction normalization. All comparisons were done to the relevant control or siGTPBP4 conditions as indicated in the figure legends. Significance (p-value) was determined using unpaired two-tailed Student’s t-test in Microsoft Excel. Volcano plots were generated using limma analyses⁷⁸ from maxLFQ normalized data output from DIA-NN. Significance between conditions in the amino acid starvation experiments was determined with two-way analysis of variance (ANOVA) followed by the appropriate multiple comparisons test (Tukey’s HSD or Dunnett’s) as indicated in the figure legends using GraphPad Prism 10.2.3.