Identification of a novel trigger complex that facilitates ribosome-associated quality control in mammalian cells

Ribosome stalling triggers the ribosome-associated quality control (RQC) pathway, which targets collided ribosomes and leads to subunit dissociation, followed by proteasomal degradation of the nascent peptide. In yeast, RQC is triggered by Hel2-dependent ubiquitination of uS10, followed by subunit dissociation mediated by the RQC-trigger (RQT) complex. In mammals, ZNF598-dependent ubiquitination of collided ribosomes is required for RQC, and activating signal cointegrator 3 (ASCC3), a component of the ASCC complex, facilitates RQC. However, the roles of other components and associated factors of the ASCC complex remain unknown. Here, we demonstrate that the human RQC-trigger (hRQT) complex, an ortholog of the yeast RQT complex, plays crucial roles in RQC. The hRQT complex is composed of ASCC3, ASCC2, and TRIP4, which are orthologs of the RNA helicase Slh1(Rqt2), ubiquitin-binding protein Cue3(Rqt3), and zinc-finger type protein yKR023W(Rqt4), respectively. The ATPase activity of ASCC3 and the ubiquitin-binding activity of ASCC2 are crucial for triggering RQC. Given the proposed function of the RQT complex in yeast, we propose that the hRQT complex recognizes the ubiquitinated stalled ribosome and induces subunit dissociation to facilitate RQC.

mammalian homolog of yeast Rqc1, accelerates this preferential linkage formation by restricting the elongation of K63-linkages 18 . Ubiquitinated peptides are released from 60S-RNCs by the tRNA endonuclease ANKZF1 18,19 and extracted by p97 for degradation by the proteasome 15 (Fig. 1).
Despite the identification of various RQC factors in mammals, the factors that recognize ubiquitinated stalled ribosomes and promote their dissociation into subunits remain mostly unknown. We previously reported that ASCC3 is required for RQC induction 8 . A recent study showed that ASCC2 and ASCC3 bind to the ribosome and mitigate the toxic effects of stalling induced by PF8503 20 . The ATP-dependent helicase ASCC3 is an ortholog of Slh1(Rqt2) 8 , and the ubiquitin-binding protein ASCC2 has significant sequence homology to yeast Cue3(Rqt3), implying the existence of a human RQT (hRQT) complex. Here, we show that the hRQT complex is composed of ASCC3/2-TRIP4 and facilitates RQC in a manner dependent upon the ATPase activity of ASCC3 and the ubiquitin-binding activity of ASCC2. We found that ASCC1 is not an essential component of the hRQT complex, indicating that the hRQT complex is distinct from both the ASCC complex involved in DNA alkylation repairing and the ASC-1 complex that serves as a transcriptional coactivator. Based on these results, we propose that the newly identified hRQT complex facilitates RQC by dissociation of the ubiquitinated ribosomes into subunits in mammalian cells.

Results
Analysis of the domains of ZNF598 required for RQC. To determine which part > region of ZNF598 is required for the induction of RQC ( Fig. 2A,B), we constructed a series of mutants, focusing on the characteristic domains of ZNF598. ZNF598 contains an N-terminal GC-rich region, followed by a RING domain, three C2H2-type zinc-finger domains, and a proline-rich motif at the C-terminus (Fig. 2B). We co-transfected these mutants along with the V5-GFP-K(AAA)24-FLAG-HIS3 reporter into ZNF598 knockdown (KD) cells constitutively expressing shRNA against ZNF598, and then monitored RQC induction by western blotting with anti-V5 antibody. We evaluated RQC by assessing the levels of the full-length and arrest products: ZNF598 is functional, the level of the full-length product will decrease, and the levels of the arrest products will increase. Given that RQC excludes arrest products, the arrest products should not be observed when functional full-length ZNF598 is expressed. Because we tested the function of ZNF598 in cells overexpressing a poly(A)-coding reporter, we suspected that arrest products were produced in excess and could not be completely cleared by RQC (Fig. 2C, lane 2). We observed the frameshift products (Fig. 2C, asterisk at lanes 1, 3-8 and 10), which were in accordance with previous reports 11, 16 , and the size of the frameshift products was also as expected (Fig. 2D). The RING domain deletion mutant (ΔRING) and its conserved cysteine residues mutant (C29S/C32S) did not induce RQC (Fig. 2C, lanes 3 and 4), whereas RQC was partially induced by the Pro-rich region trimmed mutant (1-634) but not the Pro-rich region deletion mutant (1-278) (Fig. 2C, lanes 5 and 6). Deletion mutants lacking the C2H2type zinc-finger domain (1-246, 1-186) did not induce RQC (Fig. 2C, lanes 7 and 8). Moreover, the deletion of the N-terminal GC-rich region (21-904 and 21-278) had no effect on the induction of RQC (Fig. 2C, lanes 9 and 10). Finally, we confirmed that these phenotypes were not dependent on the expression levels of ZNF598 mutants (Fig. S1A). Based on these results, we conclude that the cysteine residues (C29, C32) within the RING domain and C-terminal regions containing the zinc-finger and Pro-rich region are both essential for induction of RQC.
The human RQC-trigger (hRQT) complex consists of ASCC3, ASCC2, and TRIP4. We previously reported that an ortholog of yeast Slh1, ASCC3, is required for RQC 8 . In addition, a recent study suggested the involvement of ASCC3 and ASCC2 in co-translational quality control 20 . ASCC3 was originally identified as a component of the activating signal cointegrator 1 (ASC-1) complex, which consists of ASCC3, ASCC2, ASCC1, and TRIP4/ASC-1 (Fig. 3A) 21,22 . The ASC-1 complex promotes transactivation by serum response factor (SRF), activating protein 1 (AP-1), and nuclear factor κB (NF-κB) through direct binding to SRF, c-Jun, p50, and p65 22 . ASCC3 is also a component of the ASCC complex, which is composed of ASCC3, ASCC2, and ASCC1 (Fig. 3A) 22 . ASCC3 binds to the demethylation enzyme ALKBH3 and repairs alkylated DNA 23 . Proper recruitment of the ASCC repair complex requires recognition of K63-linked poly-ubiquitin chains by the CUE (coupling of ubiquitin conjugation to ER degradation) domain of ASCC2 24 . ASCC1 binds to ASCC3 and mediates the proper recruitment of the ASCC complex during alkylation damage 25 . TRIP4 (TR-interacting proteins) is a transcription coactivator in the nucleus and is also involved in trans-repression between nuclear receptors and Proposed model for the Ribosome-associated quality control (RQC) pathway in mammals. A stalled ribosome collides with the following ribosome, and ZNF598 ubiquitinates the collided ribosomes. In yeast, it was proposed that the RQT complex dissociates the ubiquitinated ribosome(s) into subunits but the complex that recognizes and splits ubiquitinated ribosome(s) is unknown in mammals. Nascent peptide on the 60S subunit is ubiquitinated by Listerin, and released from the ribosome by ANKZF1 and p97. Then, the ubiquitinated polypeptides are degraded by the proteasome.
To identify the components of the hRQT complex, we first examined the interaction between ASCC3, ASCC2, ASCC1, and TRIP4 in HEK293T cells. In these experiments, a FLAG-tagged bait protein (i.e., one of the aforementioned proteins) was overexpressed with HA-tagged prey (the rest of the proteins listed above) and immunoprecipitated with anti-FLAG antibody ( Fig. 3B-E). In all cases, immunoprecipitation of FLAG-tagged proteins resulted in co-purification of the remaining HA-tagged proteins. These results indicate that ASCC3/2/1 and TRIP4 form a complex, consistent with previous reports 21,22 .
After confirming the physical interactions between ASCC3/2/1 and TRIP4, we investigated whether these factors are required for RQC. To this end, we transfected the V5-GFP-K(AAA)24-FLAG-HIS3 reporter into the corresponding KD cells and monitored RQC induction by western blotting (Fig. 4A). shRNA-mediated KD efficiency is shown in Fig. S1B. In accordance with previous results 8 , ASCC3 KD abolished the induction of RQC. ASCC2 KD and TRIP4 KD partially disrupted the induction of RQC, whereas ASCC1 KD had no effect (Fig. 4A). In ASCC2 or TRIP4 KD cells, the levels of the arrest products were slightly reduced, whereas the levels of the full-length and frameshift products were significantly higher than those in control cells (Fig. 4A). The changes in the full-length products suggest that ribosome stalling was reduced in ASCC2 KD or TRIP4 KD cells, as well as in ZNF598 KD or ASCC3 KD cells. These results indicate that only three members of the ASC-1 complex, ASCC3, ASCC2, and TRIP4, are required for RQC (Fig. 4E,F). We also confirmed that the decrease in the full-length products depends on ribosome stalling induced by the poly(A) sequence (Fig. S1C). Considering that ASCC3, ASCC2, and TRIP4 are orthologs of the yeast RQT complex (Slh1, Cue3, and Rqt4, respectively), we suspected that the hRQT complex was composed of these factors, and that inhibition of hRQT complex-mediated ribosome recognition and/or dissociation may have abolished ribosome stalling and RQC induction. To explore the possibility that the hRQT complex might include novel components, we investigated the interaction of ASCC3/2-TRIP4 in ASCC1 KD cells. As expected, purification of one of these factors resulted in the co-purification of the other two factors, indicating that ASCC3/2-TRIP4 formed a complex even in the absence ASCC1 ( Fig. 4B-D). We named the components of this novel hRQT complex hRqt2 (ASCC3), hRqt3 (ASCC2), and hRqt4 (TRIP4).
ATPase-dependent helicase activity of ASCC3(hRQT2) is required for RQC. ASCC3(hRQT2), which belongs to the helicase family, harbors two sets of RecA helicase and Sec. 63 domains (Fig. 5A). In yeast, www.nature.com/scientificreports www.nature.com/scientificreports/ the helicase activity of Cue3(Rqt3) is necessary for RQC before the ribosome dissociation step. To investigate whether the ATPase-dependent helicase activity of ASCC3 is essential for RQC, we mutated the conserved lysine residue in the first RecA domain to arginine residue (K505R), leading to a deficiency in ATPase activity (Fig. 5A). Overexpression of the ASCC3-K505R mutant in ASCC3 KD cells did not complement the disruption of RQC induction (Fig. 5B), although the interaction between the ASCC3-K505R mutant and ASCC2 remained unchanged (Fig. 5C). To see whether ASCC3 WT and ATPase-deficient K505R mutant are associated with ribosome, we performed sucrose density gradient analysis followed by western blotting. In these experiments, we overexpressed ASCC3 WT or K505R mutant in ASCC3 KD cells and analyzed the cell lysates. ASCC3 WT was recruited to the ribosome (Fig. 5D,E), like its yeast ortholog Slh1. The ASCC3 K505R mutant was also recruited to the ribosome, although its distribution was partially shifted to the lighter fraction in comparison with the WT (Fig. 5D,E), indicating that the ribosome binding activity of the ASCC3 K505R mutant was weakened by its deficiency in ATPase activity. These results suggest that as in yeast, the ATPase-dependent helicase activity of ASCC3 is indispensable for RQC in mammalian cells.
Ubiquitin-binding activity of ASCC2(hRQT3) is required for RQC. ASCC2(hRQT3) prefers to bind to K63-linked poly-ubiquitin chains via its ubiquitin-binding CUE domain, and recognition of the K63-linked poly-ubiquitin chain is required for proper localization of the ASCC-ALKBH3 repairing complex 24 . In yeast, the ubiquitin-binding activity of Cue3(Rqt3) is crucial for the induction of RQC 8 . Given that ubiquitination of eS10 by ZNF598 is a key step for RQC 27 , we hypothesized that the ubiquitin-binding activity of ASCC2 is crucial for RQC. To test this idea, we mutated the conserved ubiquitin-binding domain of ASCC2 (Ub-m) (Fig. 6A) and pulled down with GST-tagged ubiquitin (Fig. 6B). The ubiquitin-binding activity of the recombinant ASCC2 Ub-m mutant protein was lower than that of the ASCC2 WT protein (Fig. 6B), although the interaction with ASCC3 remained unchanged (Fig. 6C). Expression of ASCC2 WT in ASCC2 KD cells slightly decreased the amount of full-length product, whereas the levels of the arrest products were elevated. On the other hand, the expression of Ub-m mutant only partially complemented the phenotype (Fig. 6D). This partial complementation of RQC induction is in accordance with the partial ubiquitin-binding activity of the Ub-m mutant. These results imply that a ubiquitinated stalled ribosome can be recognized by the ASCC2 ubiquitin-binding domain.

Discussion
RQC is an indispensable quality control system for guaranteeing accurate gene expression, and the detailed process of RQC is highly conserved: ribosome collision caused by translation arrest triggers RQC [8][9][10][11] , and the collided ribosomes are ubiquitinated by ZNF598 10 and dissociated into subunits by an unknown mechanism. In this study, we investigated the dissociation of collided ribosomes and identified a novel hRQT complex consisting of ASCC3(hRQT2), ASCC2(hRQT3), and TRIP4(hRQT4) (Fig. 4F). We showed that the ATPase-dependent helicase activity of ASCC3 was essential for RQC induction (Fig. 5B), and that the ubiquitin-binding activities of ASCC2 and TRIP4 also contributed to RQC, although the requirement was only partial (Figs. 4A and 6D). Hel2, an ortholog of ZNF598 in yeast, poly-ubiquitinates collided ribosomes with K63-linkage 6,8 , and ASCC2 preferentially binds to K63-linked poly-ubiquitin chains 24 . Based on these observations, we propose a model in which ubiquitinated 80S ribosomes are first recognized by ASCC2, and then dissociated into 40S and 60S subunits by ASCC3. We suspect that TRIP4 promotes binding of the hRQT complex to the ribosome via its zinc-finger domain. This model explains the partial RQC phenotypes of ASCC2 KD and TRIP4 KD. In ASCC2 KD, the RNA binding activity of TRIP4 localizes ASCC3 to the ribosome without specificity, resulting in lower efficiency in RQC induction. In TRIP4 KD, ASCC3 is located to ubiquitinated ribosomes by ASCC2, but without TRIP4, it cannot promote dissociation due to unstable association with the ubiquitinated ribosomes, resulting in lower efficiency of RQC induction.
Yeast Slh1(Rqt2) is homologous to the RNA helicase Brr1, a pre-mRNA-splicing factor that plays crucial roles in the regulated remodeling of the spliceosome structure during the splicing reaction 28 . The ATPase activity of Slh1 is crucial for triggering RQC 8 , strongly suggesting that ATPase-dependent RNA helicase activity is required for RQC in mammals. ASCC3 is associated with the alpha-ketoglutarate-dependent dioxygenase AlkB homolog 3 (ALKBH3) and unwinds DNA to generate the single-stranded substrate of the ALKBH3-mediated DNA repair pathway 23 . Our results suggest that the ATPase activity of mammalian ASCC3(hRqt2) is required for www.nature.com/scientificreports www.nature.com/scientificreports/ RQC (Fig. 5). We propose that the RNA helicase activity of ASCC3(hRqt2) is crucial for the dissociation of the ubiquitinated ribosome into subunits in mammals.
ASCC3 activity is inhibited by its own non-coding short form in the later stages of the response to UV irradiation 29 . UV irradiation slows down transcriptional elongation and induces a shift from the expression of long mRNAs to shorter isoforms. The members of the ASCC complex, ASCC3, ASCC2, and ASCC1, globally suppress nascent transcription in the later stages of the DNA damage response. The expression of the short ASCC3 RNA isoform is also induced by UV radiation and is required to recover transcription after UV irradiation. The short ASCC3 RNA isoform functions as a non-coding RNA that counteracts the function of the protein-coding isoform. Based on these, we speculate that RQC is regulated by UV irradiation.
We demonstrated that ASCC1 is not required for RQC induction and that the novel hRQT complex, composed of ASCC3/ASCC2/TRIP4, can form a complex without ASCC1 (Fig. 4B-D). On the other hand, ASCC1 coordinates the proper recruitment of the ASCC3 to ASCC2-positive foci via its RNA ligase-like domain during DNA alkylation damage 25 , and ASCC3/ASCC2/ASCC1 (i.e., the ASCC complex) and ASCC3/ASCC2/ASCC1/ TRIP4 (i.e., the ASC-1 complex) play roles in DNA alkylation damage repair and transcriptional regulation, respectively [21][22][23] . The main difference between these events is localization: RQC is a cytosolic event, whereas the others are nuclear event. According to the Human Protein Atlas 30 , all four factors localize in both the cytosol and nucleus, so we suspect that localization of factors are regulated depending on the cellular environment, such as translation stalling, alkylation damage, and serum depletion. Further studies should seek to reveal the sophisticated complex formation process in response to changes in circumstances.   (Table 3) and corresponding secondary antibodies ( Table 3). The membrane was incubated shTRIP4-4 5′-GATCCTGGAAGAAGAAAATT-3′  Ubiquitin-binding assay. For ubiquitin-binding assays, 20 µg GST and GST-Ub were incubated with 10 µL Glutathione Sepharose 4B in a total volume of 200 µl GB100 buffer at 4 °C for 1 h, followed by three washes with   Table 4. Primers used in this study.