UFMylation orchestrates spatiotemporal coordination of RQC at the ER

Abstract

Degradation of arrest peptides from endoplasmic reticulum (ER) translocon-bound 60S ribosomal subunits via the ribosome-associated quality control (ER-RQC) pathway requires covalent modification of RPL26/uL24 on 60S ribosomal subunits with UFM1. However, the underlying mechanism that coordinates the UFMylation and RQC pathways remains elusive. Structural analysis of ER-RQC intermediates revealed concomitant binding and direct interaction of the UFMylation and RQC machineries on the 60S. In the presence of an arrested peptidyl–transfer RNA, the RQC factor NEMF and the UFM1 E3 ligase (E3^UFM1) form a direct interaction via the UFL1 subunit of E3^UFM1, and UFL1 adopts a conformation distinct from that previously observed for posttermination 60S. While this concomitant binding occurs on translocon-bound 60S, LTN1 recruitment and arrest peptide degradation require UFMylation-dependent 60S dissociation from the translocon. These data reveal a mechanism by which the UFMylation cycle orchestrates ER-RQC.

Structures of stalled ER-bound 60S ribosomal subunits identify cooperative interaction between UFMylation and RQC machineries.

INTRODUCTION

Despite the highly processive nature of protein synthesis, ribosomes can stall while translating mRNA, generating ribosomes that are incapacitated by truncated peptidyl-tRNA adducts that obstruct the P-site and clog the exit tunnel. Ribosome-associated quality control (RQC) recognizes these aberrant ribosomes, ensuring efficient extraction and degradation of potentially toxic arrested polypeptides (APs) and recycling of the 60S subunit (1, 2). RQC impairment is linked to neurodevelopmental and neurodegenerative phenotypes, suggesting that this pathway plays a key role in maintaining proteostasis (3–6).

In RQC, splitting of stalled 80S ribosomes yields a free 40S subunit and a 60S subunit that contain an AP covalently bound to a tRNA in the P-site (P-tRNA). The AP-tRNA is recognized by nuclear export mediation factor (NEMF), a component of the core RQC machinery (1, 7–10) that catalyzes template-less translation (TLT), resulting in polymerization of additional amino acids at the C terminus of the P-tRNA (6, 9, 11). These so-called C-terminal alanine and threonine (CAT) tails are composed predominantly of alanine in humans (6) and a mixture of alanine and threonine in yeast (9). NEMF promotes TLT by delivering aminoacylated-tRNAs to the A-site, mimicking the canonical translation elongation cycle and extending the AP in a process called CATylation (6, 9, 11). NEMF also recruits the E3 ubiquitin ligase listerin (LTN1) to P-tRNA-60S, allowing LTN1 to position its catalytic RING domain adjacent to the peptide exit tunnel to ubiquitylate the AP, enabling ubiquitin-dependent AP extraction from the ribosome and degradation by the 26S proteasome (1, 7, 8, 12). CAT tails have been suggested to facilitate AP extraction and degradation by extruding lysine residues that are initially buried in the exit tunnel, allowing them to access cytosolic ubiquitylation machinery including LTN1 (13), or by serving as degrons to ensure efficient destruction of aggregation-prone, cytotoxic APs that escape LTN1-mediated ubiquitylation (14–16).

Ribosome stalls can also occur during cotranslational translocation of secretory and membrane proteins at the endoplasmic reticulum (ER), resulting in P-tRNA-60S docked at SEC61 translocons (17–19). These ER-APs are likely to be partially translocated across the ER membrane, clogging both the SEC61 translocon and the 60S exit tunnel. Like cytosolic APs, ER-APs are also degraded by a proteasome-mediated process that requires CATylation by NEMF and ubiquitylation by LTN1 (17, 18). However, ER-RQC differs from cytosolic RQC by requiring covalent conjugation of the ubiquitin-like protein UFM1 (UFMylation) to uL24 (also known as RPL26), a core 60S protein situated directly adjacent to the translocon docking site (20, 21). UFMylation is highly selective, with the ER membrane–anchored UFM1 E3 ligase (E3^UFM1) predominantly modifying a single lysine residue (K134) on uL24 of translocon-docked 60S (20, 21).

Genetic and biochemical studies (17) suggested a model in which UFMylation facilitates ER-RQC by disrupting the ribosome-translocon junction, thereby enabling cytosolic ubiquitin conjugation machinery like LTN1 to access lysine residues on ER-APs that would otherwise be obscured by the translocon and the ER membrane bilayer. These ubiquitylated ER-APs are extracted from the translocon and the exit tunnel by p97/VCP and degraded by cytosolic proteasomes (17). In the absence of RQC or UFMylation machinery, ER-APs are not properly extracted into the cytosol for degradation but are instead released into the ER lumen (17), where they can form toxic aggregates that could potentially escape the cell via the secretory pathway. These considerations highlight the need for precise spatiotemporal coordination of UFMylation-mediated 60S-translocon dissociation with P-tRNA cleavage and ER-AP ubiquitylation to ensure efficient degradation of ER-APs by cytosolic proteasomes.

The recently reported structures of E3^UFM1 bound to posttermination (i.e., release factor mediated, nonstall) 60S complexes (22, 23) provide mechanistic insight into how UFMylation promotes ribosome-translocon dissociation. Single-particle cryo–electron microscopy (cryo-EM) analysis of 60S particles affinity purified with E3^UFM1, a heterotrimeric complex composed of UFL1, DDRGK1, and CDK5RAP3, from untreated wild-type (WT) cells identified three distinct conformational states that differed in terms of the E3^UFM1 conformation and the presence or absence of conjugated UFM1 and the SEC61 translocon (22). Two E3^UFM1-bound 60S structures, corresponding to states before and after UFM1 transfer to uL24, exhibited clear densities corresponding to the SEC61 translocon. By contrast, in the third state, the SEC61 density on UFMylated 60S was absent, instead replaced by an α-helical domain of the DDRGK1 subunit of E3^UFM1. These structures, together with biochemical data (22, 23), suggest that, following normal termination, uL24 UFMylation promotes dissociation of 60S from ER translocons. The structures of E3^UFM1 bound to posttermination 60S support our hypothesis that UFMylation promotes ER-RQC by releasing the 60S from the translocon, thus enabling the ubiquitin-proteasome system (UPS) machinery to access ER-APs obscured by the translocon and the ER membrane. However, the previously reported conformation of UFL1 bound to posttermination 60S is incompatible with the known structures of RQC-60S, because the ER-AP and NEMF would obstruct the previously observed binding sites for the UFL1 C-terminal domain (CTD) and PTC loop (22, 23). This clash suggests a fundamental incompatibility of the RQC and UFMylation machineries to operate on the same 60S particle, despite genetic evidence that these machineries must cooperate (17, 22).

In this study, we used single-particle cryo-EM of E3^UFM1-bound P-tRNA-60S particles captured from anisomycin (ANM)–treated cells to resolve this conundrum. We find that during ER-RQC, the UFL1 CTD adopts a rotated conformation on P-tRNA-60S that does not occupy the A- and P-tRNA binding sites. In this conformation, part of the mostly unstructured, 79–amino acid region of UFL1 that encompasses the PTC loop (the “loop domain”) forms a β-augmented interface with NEMF. Structure-directed mutational analysis demonstrates that direct interaction between NEMF and UFL1 is critical for coordinating P-tRNA cleavage, UFMylation-mediated translocon release, and ER-AP ubiquitylation, which must occur in a precise temporal sequence to ensure ER-AP extraction to the cytosol for degradation. We present an ensemble of cryo-EM structures of ER-RQC that support a model in which UFMylation plays a central role in coordinating the temporal and spatial order of events in ER-RQC.

RESULTS

E3^UFM1 adopts an alternate conformation on RQC-60S

We affinity captured ER-RQC-60S complexes via the tagged UFL1 subunit of E3^UFM1 (3xFLAG-UFL1) from cells that were challenged with ANM to induce ribosome collisions and subsequent RQC (Fig. 1A) (24). Mass spectrometric analysis (data S1) of this sample identified E3^UFM1, 60S, SEC61, and the RQC factors, NEMF and LTN1, suggesting that E3^UFM1 and RQC factors can coexist on the same 60S ribosomal subunits at the ER. We then performed a single-particle cryo-EM analysis that revealed four major classes of ER-RQC intermediates that were affinity captured with UFL1 in addition to the previously observed posttermination E3^UFM1 complexes (fig. S1). The ratio of posttermination E3^UFM1 complexes to ER-RQC complexes is roughly 10:1, suggesting that UFMylation acts in ER-RQC in addition to its primary role in posttermination 60S recycling. Among the ER-RQC states, the best-resolved class (fig. S2A) contained E3^UFM1, NEMF, a peptidyl-tRNA in the P-site (fig. S2A), and LTN1 bound to the same 60S subunit (E3^UFM1-RQC-60S; Fig. 1, B to D). A molecular model for this map, which we refer to as “RQC state 3” (fig. S1), was built, initially guided by previous structures (22) and AlphaFold [see Materials and Methods, Fig. 1 (B to D), fig. S3 (A to C), and table S1] (25–27).

Fig. 1. Cryo-EM structure of the E3^UFM1-RQC-60S complex.

(A) Coomassie-stained SDS–polyacrylamide gel electrophoresis (SDS-PAGE) gel of a 3xFLAG-UFL1 pulldown from cells challenged with ANM. (B to D) Molecular model of the E3^UFM1-RQC-60S (RQC state 3) complex, shown as a side view on the intersubunit space (B), a top view (C), and a bottom view on the peptide tunnel exit (D). (E to G) Comparison of the E3^UFM1-RQC-60S complex with the posttermination E3^UFM1-60S complex (22) and RQC-60S complex (RQC state 4) (E) view of the tunnel exit region, (F) comparison of the E3^UFM1-RQC-60S structure (RQC state 3) with the RQC-60S class (RQC state 4, no E3^UFM1 bound) obtained from this study (see also Fig. 4D and figs. S1 and S2), and (G) view focusing on the overall position of the E3^UFM1. Molecular models in (E) to (G) are rendered as surfaces for the 60S and as ribbons for nonribosomal ligands. TE, tunnel exit; EBH, exit binding helix of DDRGK1; CTD, UFL1 C-terminal domain; WH, UFL1 winged helix domain; PT, posttermination.

In the structure of RQC state 3 (Fig. 1, B to D), NEMF is observed in a previously identified intermediate state of the CATylation cycle, namely, the postinitiation/predecoding TLT intermediate before accommodation of the next A-site tRNA [state E in yeast cytosolic RQC (11); fig. S4]. In this state, NEMF contacts the P-tRNA, which is in the P^RQC conformation [as observed in yeast complexes (11)] via its NFACT domains, while the connecting coiled-coil-middle (CC-M) domain of NEMF contacts the 60S stalk base at the sarcin-ricin loop, as well as uL11 and LTN1 (7). As described before (7, 9, 11, 28), LTN1’s HEAT repeats arch over the 60S toward the peptide tunnel exit (Fig. 1, B to D), while its RING domain is located in close proximity to the long N-terminal α helix of DDRGK1 [exit binding helix (EBH)] (Fig. 1, D and E).

We also observed a RQC-60S complex, RQC state 4 (fig. S1), in which E3^UFM1 was not visible. This state likely arose from dissociation of the UFMylation machinery before cryo-EM analysis. LTN1 and NEMF are in virtually identical conformations in RQC states 3 and 4 (Fig. 1F), except for the notable difference that NEMF’s NFACT-C domain is more rigid in RQC state 3 (fig. S2C). This conformational stabilization of NEMF’s NFACT-C domain by E3^UFM1, together with the occupancy of the E-site by UFL1, suggests that NEMF’s CATylation activity is likely to be paused upon E3^UFM1 binding.

RQC state 3 (Fig. 2A) differs substantially from the previously reported posttermination E3^UFM1-60S structures (Fig. 2B) (22, 23) in that UFL1’s globular CTD is markedly repositioned, and a part of the PTC loop–containing loop domain (Fig. 2C) establishes a direct interaction with NEMF. Otherwise, the overall architecture of E3^UFM1 in the E3^UFM1-RQC-60S structures is nearly identical to its conformation on posttermination E3^UFM1-60S (22, 23). In both, E3^UFM1 forms a clamp-like structure (Fig. 1G) where the EBH of DDRGK1 spans from the peptide exit tunnel toward UFMylated uL24 (Fig. 1E), and the UFL1-CDK5RAP3 winged helix scaffold reaches toward the L1 stalk (Fig. 1G). Thus, repositioning of UFL1’s CTD and loop domain allows E3^UFM1 and the RQC factors NEMF and LTN1 to coexist on the same ER-AP containing 60S particles.

Fig. 2. Conformational change of the UFL1 C terminus and its interaction with NEMF.

(A and B) Molecular models of the E3^UFM1-RQC-60S complex (RQC state 3) (A) and E3^UFM1-60S complex (posttermination state) (22) (B). View focuses on the tRNA binding sites in the 60S intersubunit space, and thumbnails indicate the orientation. Note that in the E3^UFM1-RQC-60S complex, the UFL1 C terminus flips toward the L1 stalk and uL1 protein. (C) Schematic representation of NEMF and UFL1 domain organization and interactions between them. The vertical dashed line marks the site of the β-augmented residues. (D to F) Close-up views on the interactions between the UFL1 CTD and (D) uL1 (white), (E) 5S rRNA (gray), and (F) the NEMF C-terminal NFACT domain (violet; two views). CTD, UFL1 C-terminal domain; CIM, CTD-interacting motif of NEMF; WH, winged helix; CC, coiled-coil (NEMF and UFL1); M, middle domain of NEMF; NFACT, domain found in NEMF, FbpA, Caliban, and Tae2.

In the posttermination E3^UFM1-60S complex (22, 23), the UFL1 CTD occupies the A- and P-sites, being locked between ribosomal RNA (rRNA) helices H38 (A-site finger) and H69 (Fig. 2B) (22, 23). However, in RQC state 3, this space is occupied by the P-tRNA and NEMF, and UFL1’s CTD is rotated backward by approximately 150° into a position between ribosomal protein uL1 and 5S rRNA (Fig. 2, A and C). Two basic patches in the UFL1 CTD (R769/K770 and K658/K659/R662) are likely to stabilize this conformation by interacting with complementary acidic patches in either uL1 (Q71/Q72/D75) or the backbone of 5S rRNA (Fig. 2, D and E). Notably, a region of UFL1 encompassing amino acids N390 to E399 interacts with a portion of NEMF’s NFACT-C domain (Fig. 2, A, C, and F, and fig. S3, D to F). Specifically, the UFL1-NEMF interface displays a β-augmentation of the residues S634-M636 of NEMF with the residues H394-I396 of UFL1, which are part of the PTC loop domain (Fig. 2C) (22, 23). In the posttermination 60S conformation of UFL1 (22, 23) (posttermination state 3), the mostly unstructured PTC loop domain occupies the P-site, and the α-helical PTC loop is localized at the PTC (Fig. 2B). In contrast, in RQC state 3 (Fig. 3A), the P-site is occupied by the P-tRNA and NEMF, and the region of the PTC loop containing the small α-helical segment is delocalized and unresolved. In addition to the interface between UFL1 and NEMF formed by β-augmentation, we observed a short stretch of the disordered region of NEMF interacting with the concave surface of the UFL1 CTD, which we named the CTD-interacting motif of NEMF (Fig. 2, A and C, and fig. S3, G to I).

Fig. 3. ER-AP but not cytosolic AP clearance depends on interaction between NEMF and UFL1.

(A) Schematic of the ER-targeted stalling reporters contained the following features: mRuby; V5 epitope tag; T2A (Thosea asigna virus 2A) peptide bond skipping sequences; signal sequence (SS) from bovine preprolactin; hemagglutinin epitope tag; blue fluorescent protein (BFP); N-glycosylation sequon; polylysine (K20) stalling sequence; superfolder green fluorescent protein (GFP). The protein species produced by each stalling reporter are indicated. (B) Model of the ER-targeted stalling reporter on a 60S ribosome at the ER. When the ER-AP is still attached to the ribosome by a P-tRNA, the N-glycosylation sequon is retained in the ribosome exit tunnel and is inaccessible to the glycosylation machinery in the ER lumen. If the ER-AP is released from the 60S and P-tRNA, then it will enter the ER lumen, and the sequon will be glycosylated. (C) Position of NEMF-UFL1 β-augmentation interface (red box). (D and E) Residues mutated in VLT UFL1, shown as molecular model (D) or schematic representation (E). The vertical dashed line marks the site of the β-augmented residues. (F) Degradation of ER but not cytosolic APs depends on UFL1-NEMF interaction. UFL1-dependent degradation of ER-AP but not cytosolic AP in WT or UFL1^KO cells stably rescued with WT, but not mutant Δ79 (deletion of residues L395-E473) and VLT (V393A/L395A/T397A) UFL1-FLAG. (G) Quantification of cytosolic and ER-AP reporter intensities from data in (F). For ER-APs, the sum of the −Gly and +Gly bands are quantified. Data show mean V5 normalized fold change ± SD relative to UFL1^KO cells rescued with WT UFL1-FLAG, P value from the indicated comparison derived from two-way analysis of variance (ANOVA) of n = 3 biological replicates. ns, not significant. *P > 0.05, ****P > 0.0001.

Together, the structure of the E3^UFM1-RQC-60S shows that the UFMylation and RQC machineries can concomitantly engage the ER-AP containing 60S. The ability of E3^UFM1 to engage posttermination 60S and RQC-60S is enabled by the capacity of the UFL1’s CTD to adopt either of two alternate conformations. Notably, this observation solves the conundrum of P-tRNA-60S recognition by E3^UFM1 raised by the clash of UFL1’s CTD in the posttermination conformation with P-tRNA or NEMF.

UFL1-NEMF interface coordinates ER-RQC

To test the functional importance of the UFL1-NEMF interaction, we focused on the β-augmentation interface between UFL1 and NEMF, since it stabilizes the NFACT-C domain of NEMF in a rigid conformation (fig. S2C). We monitored the effect of mutations of key residues at this interface on turnover of cytoplasmic- and ER-targeted APs using modified versions of our previously described ER-targeted and cytosolic stalling reporters (Fig. 3, A to E) (17). Inclusion of an N-glycosylation sequon in the ER-targeted reporter directly upstream of the K20 stall sequence allows glycosylation status to distinguish ER-APs that are released into the ER lumen (sequon glycosylated) from those that remain ribosome bound (sequon sequestered in the exit tunnel not glycosylated) (Fig. 3, A and B). The normalized steady-state levels of the stalling reporters serve as a proxy for degradation (17). As expected, neither knockout of UFL1 nor rescue of UFL1^KO cells with UFL1 variants affected the cytosolic stalling reporter, as UFMylation is dispensable for cytosolic RQC (Fig. 3, F and G) (17). By contrast, knockout of UFL1 caused substantial accumulation of the glycosylated ER reporter (Fig. 3, F and G, “+Gly,” and fig. S5A), confirming our previous observation that UFL1 is essential for ER-AP degradation and that, in the absence of UFMylation, ER-APs are fully released into and are stabilized in the ER lumen (17). This defect was rescued by reexpression of WT UFL1, but not of a UFL1 variant (Δ79) that lacks the loop domain, or a variant harboring mutations predicted to disrupt the β-augmented NEMF interface on UFL1 (V393A/L395A/T397A; “VLT”) (Fig. 3, C to G, and fig. S5A). We conclude that this interaction serves to prevent ER-AP release into the ER lumen.

We attempted to confirm these findings by rescuing NEMF^KO cells with NEMF variants harboring mutations in key residues on the β strand that contribute to this interface (F635A/I637A; “FI”). Reexpression of WT NEMF in NEMF^KO cells robustly rescued the defect in degradation of cytosolic, but not ER-targeted stalling reporters (fig. S5, B and C) (17), suggesting that NEMF^KO cells had adapted to the engineered deletion by becoming partially NEMF independent. All four NEMF^KO clones had elevated steady-state levels of uL24 UFMylation (fig. S5D), which could not be returned to WT levels by reexpressing NEMF (fig. S5E). The observation that transient knockdown of NEMF does not induce hyper-UFMylation (17) suggests that elevated UFMylation is an adaptive strategy to survive the selective impact of chronic NEMF depletion on ER-RQC and consistent with the high coessential dependency between UFMylation genes and NEMF (29) and the synthetic lethality between NEMF and UFM1 (17). Because of adaptations to the loss of ER-RQC in the NEMF^KO clones, we could not draw conclusions about the contribution of specific amino acid residues on the NEMF side of the UFL1 β-augmented interface in RQC. However, the UFL1 rescue experiments strongly support an essential role of this interaction in ER-RQC (Fig. 3, F and G). Further studies will be required to understand how uL24 hyper-UFMylation and perhaps additional undiscovered adaptive changes contribute to cell survival in the absence of NEMF.

In addition to considering a model in which the interaction between NEMF and UFL promotes ER-AP release, we also considered a model in which the direct physical interaction of NEMF with UFL1 could facilitate ER-RQC by promoting uL24 UFMylation. However, we ruled out this model because UFMylation is not impaired upon NEMF knockdown, even in ANM-treated cells, suggesting that UFM1 conjugation occurs independently of NEMF (17). Moreover, we observed similar levels of UFL1 and UFMylated uL24 cofractionated in ribosomal pellets from ANM-treated UFL1^KO cells rescued with both WT and VLT variants of UFL1 (fig. S6A). Therefore, disruption of the NEMF-UFL1 interface does not inhibit either E3^UFM1 binding or UFMylation in response to ribosome stalling. By contrast, deletion of the entire loop domain (Δ79), which includes the PTC loop and the NEMF β-augmented interface, or the PTC loop alone (Δ32) impairs UFMylation (fig. S6A), suggesting that this region contributes to the stable binding of E3^UFM1 to posttermination-60S and RQC-60S independently of the UFL1-NEMF β-augmented interface.

Cryo-EM reveals distinct snapshots of ER-RQC

Apart from the E3^UFM1-RQC-60S (RQC state 3) and RQC-60S (RQC state 4) structures shown above (Figs. 1 and 2), two additional RQC-60S structures were identified in our cryo-EM dataset from the ANM-challenged UFL1-pulldown sample (Fig. 4, A to D, and fig. S1). The first structure (RQC state 1; Fig. 4A) contains the SEC61 complex and NEMF bound to non-UFMylated P-tRNA-60S in the same conformation as observed in the E3^UFM1-RQC-60S structure (RQC state 3) (Fig. 1, B to D). While we cannot directly prove that this state is part of the UFMylation cycle, we speculate that it corresponds to recently split ER membrane–bound P-tRNA-60S to which NEMF has bound and has presumably begun synthesizing CAT tails, which is consistent with biochemical evidence showing that ER-AP CATylation can occur in the absence of UFMylation (17). Since all particles in this study were captured by UFL1 pulldown, the absence of visible density corresponding to UFMylation machinery in RQC states 1 and 4 possibly reflects that E3^UFM1, weakly associated in the absence of tight binding to UFMylated uL24, was either highly delocalized or had dissociated during sample preparation.

Fig. 4. UFMylation promotes SEC61 displacement and LTN1 binding.

(A to D) Cryo-EM density maps of the four main classes found in the ANM-challenged UFL1 immunoprecipitation. (A) RQC state 1, a SEC61-bound ER-RQC intermediate with NEMF bound to a P-tRNA. (B) RQC state 2, same as (A) but with the E3^UFM1 bound to uL24-UFMylated 60S. (C) RQC state 3, E3^UFM1-60S-RQC complex as shown in Fig. 1. Compared to (B), no SEC61 is visible, instead the DDRGK1 EBH is positioned below the tunnel exit and LTN1 is bound. (D) RQC state 4, same as (C) but lacking the E3^UFM1. (E) LTN1 binding to ribosomes at the ER depends on UFMylation. Microsomes derived from WT or UFC1^KO HEK293 cells were treated with ANM and analyzed by immunoblot. Quantification is from n = 3 biological replicates, fold change ± SD relative to WT cells, P value from the indicated comparison derived from one-way ordinary ANOVA of n = 3 biological replicates. **P = 0.0052. (F) Loss of UFM1 reduces ER-AP ubiquitylation. WT or UFM1^KO HEK293 cells transfected with SS^VgVK20 from (17) were treated with dimethyl sulfoxide (DMSO) or BTZ for 4 hours before isolation of Ub conjugates using pan-ubiquitin (TUBE2) agarose. Quantification is from n = 3 biological replicates from TUBE affinity purification in the presence of BTZ. Data show mean FLAG smear in elution/total ubiquitin smear in elution (fig. S8), fold change ± SD relative to WT cells, P value from the indicated comparison derived from unpaired t test, ***P = 0.0004. (G) Loss of UFSP2 accumulates ER-APs. Human embryonic kidney (HEK) 293 WT or UFSP2^KO cells were transfected with the ER-targeted stalling reporter and analyzed by immunoblot. For ER-APs, both the −Gly and +Gly bands are quantified. Data show mean V5 normalized fold change ± SD relative to WT cells, P value from the indicated comparison derived from unpaired t test of n = 3 biological replicates, **P = 0.0021.

In the next intermediate (RQC state 2; Fig. 4B), NEMF, P-tRNA, and SEC61 are all positioned as in RQC state 1, with the addition of UFMylated uL24 and the tightly bound E3^UFM1, which is fully resolved except for the EBH of DDRGK1, which is not yet positioned to dislocate SEC61, as in posttermination state 2 (22). RQC state 2 confirms that, even in the UFL1 CTD rotated conformation and with the β-augmented interface with NEMF (Fig. 1), E3^UFM1 is still able to catalyze uL24 modification. eIF6 was observed in RQC states 1 and 2 and a subset of RQC state 3 particles. We did not observe eIF6 in RQC state 4 particles.

RQC state 3 (Fig. 4C) is the best-resolved structure shown in Fig. 1. Its most notable features are the conspicuous absence of density corresponding to SEC61, which is replaced by the DDRGK1 EBH, and the presence of LTN1. As previously reported for posttermination 60S, SEC61 displacement occurs after conjugation of UFM1 to uL24, which stabilizes DDRGK1 EBH at the tunnel exit (Figs. 1 and 4C) (22, 23). We never observed particles in which RQC-60S was simultaneously bound to both SEC61 and LTN1. This mutually exclusive binding can be explained by the steric clash between the translocon (and the ER membrane) with the RING domain and adjacent regions of LTN1 (fig. S7), supporting the conclusion that displacement of P-tRNA-60S from SEC61 must temporally precede recruitment of LTN1. It is also likely that SEC61 engagement by 60S would sterically hinder not only LTN1 engagement but also access of an activated E2 ubiquitin–conjugating enzyme to lysine residues on ER-AP.

To test this model, we assessed the impact of disrupting UFMylation on the binding of endogenous LTN1 to ER-bound ribosomes in response to ANM-induced ribosome collisions. ANM treatment increased the amount of NEMF associated with microsomes from both WT and UFMylation E2–deficient (UFC1^KO) cells, confirming that UFMylation does not interfere with NEMF binding (Fig. 4E). By contrast, ANM treatment caused the amount of microsome-bound LTN1 to increase only in WT cells, suggesting that UFMylation-dependent displacement of 60S from ER translocons is a prerequisite for LTN1 engagement (Fig. 4E). We also observed that ER-AP ubiquitylation was significantly reduced in UFM1^KO compared to WT cells, consistent with impaired access of LTN1 to ER-AP in the absence of UFMylation (Fig. 4F and fig. S8).

RQC state 4 [Figs. 1F (right) and 4D] features 60S bound to P-tRNA, NEMF, and LTN1 in conformations similar to RQC state 3 but lacks densities corresponding to SEC61, E3^UFM1, and UFM1 on uL24, suggesting that it could possibly represent RQC-60S after translocon dissociation and uL24 deUFMylation. We found that ER-AP levels were strongly elevated in the absence of the deUFMylase, UFSP2, suggesting that deUFMylation may be necessary for ER-AP degradation (Fig. 4G). While we cannot provide direct evidence that the observed structures are directly linked in the same pathway, the suggested scenario represents the simplest and most plausible interpretation.

DISCUSSION

Our structural and biochemical observations suggest a model for ER-RQC (Fig. 5). After translational stalling and 80S ribosome splitting, P-tRNA-60S remains bound to the SEC61 translocon and is recognized by NEMF, which CATylates the ER-AP (RQC state 1; Fig. 4A). Following CATylation, E3^UFM1, with the UFL1 CTD in the rotated conformation, binds to RQC state 1, where it forms a β-augmented interface with NEMF, giving rise to RQC state 2 (Fig. 4B). Further research will be needed to identify what, if any additional signal is required for E3^UFM1 to bind to RQC state 1.

Fig. 5. Model for E3^UFM1 and RQC cooperation in ER-RQC.

Stalled, translocon-docked ribosomes are split, yielding a translocon-engaged 60S subunit with a peptidyl-tRNA in the P-site, an ER-AP clogging the exit tunnel and the SEC61 translocon (RQC state 1). As in cytosolic RQC, binding of NEMF to the P-tRNA and empty A-site initiates TLT, forming a CAT tail (yellow circles) on the AP. Double headed arrows denote conformational flexibility of the NEMF NFACT-C domain. In RQC state 2, E3^UFM1 is bound with UFL1 CTD in the rotated conformation, forming a stabilizing interface with NEMF to pause TLT, and E3^UFM1 catalyzes the transfer of UFM1 to uL24. In RQC state 3, the EBH of DDRGK1 is stabilized at the tunnel exit, promoting dissociation of 60S from SEC61 and allowing LTN1 to position its RING domain near the tunnel exit to ubiquitylate the ER-AP. DeUFMylation allows E3^UFM1 to dissociate, producing RQC state 4 in which the NEMF NFACT-C domain regains its mobility and the P-tRNA-60S complex is no longer tethered to the ER membrane by E3^UFM1. DUF, DeUFMylation enzyme, UFSP2.

We hypothesize that this binding mode of UFL1 has two consequences for the CATylation activity of NEMF. First, the translocation of any deacylated tRNA from the P-site into the E-site is prevented by the presence of E3^UFM1 in the E-site. Second, the interaction between the UFL1 loop domain and the NEMF NFACT-C domain stabilizes NEMF’s conformation on the 60S subunit, as suggested by comparably higher local resolution (fig. S2C). As the β-augmentation interaction likely restrains the conformational flexibility of NEMF necessary for efficiently promoting the TLT cycle for CATylation (11), we propose that CATylation is stalled by the NEMF-UFL1 interaction. This, along with the observation that CAT tail length is unaffected by loss of UFMylation (17), suggests a model in which CAT tails are formed before E3^UFM1 binding to the RQC-60S, although it is possible that that CATylation may resume after translocon release and deUFMylation. Further study will be needed to determine whether the UFL1-NEMF interaction influences CAT tail composition or length, and how this could contribute to ER-AP degradation.

Disruption of this β-augmented interface results in lumenal accumulation of the glycosylated form of the ER-targeted reporter, pointing to premature peptidyl-tRNA cleavage and failed ER-RQC–mediated extraction of the ER-AP to the cytosol (Fig. 3, F and G). This supports a model in which the UFL1-NEMF β-augmented interaction halts CATylation, keeping NEMF in a conformation over the P-tRNA where it could potentially protect it from hydrolysis, delaying cleavage by nucleases or hydrolases such as ANKZF1 or Ptrh1 (30–32). Unfortunately, exactly how CATylation termination by ANKZF1 (or Ptrh1) is triggered and the exact role NEMF plays in termination is not understood. Previous reports (30–32) suggest that NEMF (Rqc2 in yeast) has an inhibitory effect on ANKZF1’s activity (Vms1 in yeast), in agreement with our model.

Transfer of UFM1 onto uL24 results in tight binding of E3^UFM1 to 60S and displacement of 60S from the SEC61 translocon (22, 23), enabling LTN1’s RING domain to access the ER-AP which was previously inaccessible (RQC state 3; Fig. 1, B to D). Critically, cleavage of the P-tRNA must occur after ubiquitylation of ER-AP to allow for extraction of ER-AP by an ubiquitin-dependent AAA+ adenosine triphosphatase like p97/VCP (12, 17). If the P-tRNA is cleaved before translocon release or ubiquitylation, then there is nothing to prevent the entire ER-AP from being pulled into the ER lumen, evading degradation by the UPS. CATylated APs are highly aggregation prone (15, 16, 33), making these APs likely to be problematic if retained in the lumen or secreted.

We propose that deUFMylation-mediated release of E3^UFM1 in the transition to RQC state 4 restores mobility to the NEMF NFACT-C domain, enabling peptidyl tRNA cleavage by ANKZF1 or other nucleases/hydrolases. Therefore, deUFMylation is another critical step in choreography of ER-RQC, by ensuring that P-tRNA cleavage can only occur after release of the SEC61 translocon from P-tRNA-60S. This coordination ensures that ER-APs that are released upon cleavage of the P-tRNA are efficiently degraded by cytosolic proteasomes instead of being translocated into the ER lumen.

MATERIALS AND METHODS

Mammalian cell culture

Human embryonic kidney (HEK) 293 cells (American Type Culture Collection) were maintained in Dulbecco’s modified Eagle’s medium (DMEM)–high glucose (Cytiva) supplemented with 10% fetal bovine serum. Cell lines were grown in a humidified incubator at 37°C and 5% CO₂. All cell lines were routinely tested for mycoplasma infection using a polymerase chain reaction (PCR) mycoplasma detection kit according to the manufacturer’s instructions (ABM Inc.).

Mammalian cell transfections

For reporter transfections, HEK293 cells were transfected using Lipofectamine 3000 (Thermo Fisher Scientific, L3000015) according to the manufacturer’s instructions. Transfected cells were cultured for 24 to 48 hours before being processed for downstream analysis.

Plasmids

Plasmids and DNA constructs were generated using standard PCR and site-directed mutagenesis techniques using NEBNext High-Fidelity PCR Master Mix (New England Biolabs) and verified by sequencing. Lentiviral vectors for the expression of UFL1 were generated from a modified pLVX vector with an EF1ɑ promoter and a blasticidin selection marker. All lentivirus packaging vectors were obtained from Addgene. To generate stalling reporter sequences, mRuby-V5-2xT2A-SS-HA-SBP-BFP-K20 was ordered as a gene block (Genewiz) and inserted into a pcDNA3.1 parent vector containing a cytomegalovirus promoter and green fluorescent protein downstream of the insertion. The cytosolic variant of the reporter was generated by standard subcloning methods. SS^VgVK20 was reported previously (17).

Cell line generation

Previously reported UFL1^KO and UFSP2^KO HEK293 cells were used for this study (17, 20, 22). Stable UFL1-FLAG expressing HEK293 cells were generated through the lentiviral transduction of UFL1^KO cells. Lentivirus was used to produce stable cell lines through transfection of HEK293T cells with third-generation packaging plasmids (pRSV, pMDL, and pVSVG) and a lentiviral vector containing either WT UFL1-FLAG, UFL1-FLAG Δ32 (deletion of residues K417-V448), UFL1-FLAG Δ79 (deletion of residues L395-E473), or UFL1-FLAG V393A/L395A/T397A using TransIT-LT1 transfection reagent (Mirus) according to the manufacturer’s instructions and grown for 72 hours before collection of the viral supernatant. The supernatant (medium) containing the viral particles was collected and filtered through a 0.45-μm syringe filter and used fresh. UFL1^KO HEK293 cells were infected by reverse transduction; cells were resuspended in the viral supernatant containing polybrene (8 μg ml⁻¹) and plated. The viral supernatant was removed, and cells were provided fresh DMEM and grown for about 72 hours before selection with blasticidin for ~2 weeks, after which cells were used as a polyclonal line.

AP accumulation assay

The ribosome stalling reporter (0.5 or 1 μg of plasmid DNA) was transfected 48 hours before cell collection. Whole cell lysates were prepared in radioimmunoprecipitation assay (RIPA) lysis buffer [50 mM tris (pH 7.6), 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS] with protease inhibitor cocktail (complete, EDTA-free protease inhibitor cocktail; Roche) and 1 mM phenylmethylsulfonyl fluoride (PMSF). The total protein concentration was determined for each sample using a Pierce BCA Protein Assay kit (23225). Normalized samples were analyzed by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and hemagglutinin immunoblotting to detect AP produced. The number of biological replicates for each experiment is listed in the legends; bar graphs in Fig. 3G show the means and SD, and significance was determined using two-way analysis of variance (ANOVA).

Glycosidase treatment

HEK293 cells were collected and lysed in RIPA lysis buffer [50 mM tris (pH 7.6), 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS] with protease inhibitor cocktail and 1 mM PMSF.Total protein concentration was determined for each sample with the Pierce BCA Protein Assay Kit (23225) according to the manufacturer’s instructions. Protein concentrations were normalized between samples. Samples were denatured and Endo H treated following the manufacturer’s protocols (New England Biolabs Inc., P0702L). Reactions were incubated at 37°C for 1 hour and then analyzed by SDS-PAGE and immunoblotting.

Sucrose cushion sedimentation

Cells were collected and lysed in 1% Triton lysis buffer [20 mM tris (pH 7.5), 150 mM NaCl, 5 mM MgCl₂, and 1% Triton X-100] with protease inhibitor cocktail, 1 mM PMSF, and 1 mM dithiothreitol (DTT). Total protein concentration was determined for each sample with the Pierce 600 nm Protein Assay Reagent according to the manufacturer’s protocol. Samples were centrifuged at 100,000 rpm for 1 hour at 4°C through a 1 M sucrose cushion in 1% Triton lysis buffer. Pellets were washed once with ice-cold H₂O and resuspended in 1× Laemmli buffer containing 2-mercaptoethanol 5% (v/v) by heating at 100°C for 5 min. Samples were analyzed by SDS-PAGE and immunoblotting.

SDS-PAGE and immunoblotting

Proteins were denatured in 1× Laemmli buffer containing 2-mercaptoethanol 5% (v/v) by heating at 100°C for 5 min. Samples were then separated by SDS-PAGE [12% tris-glycine gels or “4 to 20% Mini-PROTEAN TGX” (Bio-Rad)] and transferred in a semidry transfer to nitrocellulose following the manufacturer’s protocol (Bio-Rad). Nitrocellulose membranes were blocked in Intercept (tris-buffered saline) Blocking Buffer to reduce nonspecific antibody binding and incubated with primary antibodies diluted in PBS-T containing 0.1% Tween 20 and 5% bovine serum albumin. Immunoreactivity was detected using fluorescent IRDye secondary antibodies and scanning by Odyssey imaging (LI-COR Biosciences). Band intensities were quantified by Image Studio Lite software (LI-COR Biosciences).

Polyubiquitin affinity capture and identification of polyubiquitylated ER-APs

HEK293 WT or UFM1^KO cells were transfected with 2 μg of SS^VgVK20 stalling reporter for 48 hours and treated with or without 1 μM BTZ for 4 hours. Cells were harvested by centrifugation at 1000g for 5 min, washed thrice with phosphate-buffered saline (PBS), and lysed in 250 μl of lysis buffer [50 mM Hepes (pH 7.35), 150 mM NaCl, 1% NP-40, Roche protease inhibitor cocktail, and 1 mM PMSF] supplemented with 40 mM N-ethylmaleimide, 10 mM iodoacetamide, and 50 μM PR-619. Lysates were clarified by centrifuging at 13,000g for 10 min and incubated with immobilized tandem ubiquitin binding entity (TUBE2 agarose, LifeSensors, catalog no. UM-402) for 16 hours at 4°C by rotating (0.5 mg of whole cell lysates added to 20 μl of TUBE2-agarose). The agarose was washed twice with high salt buffer [50 mM Hepes (pH 7.35), 500 mM NaCl, and 0.5% NP-40] and once with low salt buffer [50 mM Hepes (pH 7.35), 150 mM NaCl, and 0.5% NP-40] by shaking at 4°C for 10 min. Polyubiquitin conjugates were eluted by boiling beads in the presence of ~50 μl of 2× SDS-PAGE sample buffer. The inputs and eluates were analyzed by immunoblotting for endogenous ubiquitin and FLAG to detect ER-APs.

Preparation of rough microsomes from HEK293 cells

HEK293 WT, UFC1^KO, or UFSP2^KO cells (20) grown to ~80% confluency in 15-cm plates were treated with 0.2 μM ANM for 20 min. Microsomal membranes were isolated using a protocol that has been previously described and adapted for HEK293 cells (34–36). Briefly, cells from each 15-cm plate were harvested in 5 ml of ice-cold PBS by pipetting, centrifuged for 5 min at 800g, and resuspended in 2.5 ml of lysis buffer [10 mM Hepes-NaOH (pH 7.4), 250 mM sucrose, 2 mM MgCl₂, and 0.5 mM DTT] containing protease inhibitor cocktail (complete, EDTA-free protease inhibitor cocktail; Roche) and 1 mM PMSF. Cells were homogenized using a chilled and equilibrated isobiotec cell homogenizer (six single passes, 18 μm clearance) on ice, and lysate was cleared twice (1500g for 3 min at 4°C). Microsomes were pelleted at 10,000g for 10 min at 4°C, resuspended in 300 μl of microsome buffer [10 mM Hepes-NaOH (pH 7.4), 250 mM sucrose, 1 mM MgCl₂, and 0.5 mM DTT] containing protease inhibitor cocktail, PMSF, and RNaseOUT (Thermo Fisher Scientific), and pelleted again (10,000g for 10 min at 4°C). The membrane pellets were resuspended in microsome buffer and adjusted to a final concentration of 4 mg/ml.

Statistical analysis

Data are represented as means ± SD unless otherwise stated. The number of independent replicates performed for each experiment is indicated in the figure legends. Western blot band intensities were quantified using Image Studio Lite version 5.2.5 (LI-COR Biosciences) and normalized to translation/expression normalizer (mRuby-V5).

Affinity purification of UFL1-bound ribosomes

Large ribosomal subunits bound to E3^UFM1 and the RQC complex were purified essentially as described before (22) with the main difference being that cells were treated with ANM before harvesting for lysis. Following procedures previously described in (22), for purification, HEK293 FlpIn TRex cells with a plasmid expressing C-terminally 3× Flag-tagged UFL1 were grown to 50% confluency, and protein expression of 3xFlag-UFL1 was induced by tetracycline (1 μg/ml). At 22 hours following induction, cells were treated with 200 nM ANM for 20 min before being collected and washed twice with PBS by centrifugation at 127g for 10 min. Cells were then resuspended in lysis buffer [150 mM potassium acetate (KOAc), 20 mM Hepes (pH 7.5), 5 mM magnesium chloride (MgCl₂), 5% glycerol, 1% digitonin, 1 mM DTT, 0.5 mM sodium fluoride (NaF), 0.1 mM sodium vanadate (Na₃VO₄), and complete EDTA-free protease inhibitor (Roche)] and lysed by sonicating 4 × 10 s with 20 s on ice in between (Branson Sonifier 250). The lysate was clarified by centrifugation at 3166g for 15 min and at 36,603g for 20 min and then incubated with M2 anti-Flag agarose beads (Sigma-Aldrich) on a rotating wheel for 120 min at 4°C. Beads were washed twice with washing buffer [150 mM KOAc, 20 mM Hepes (pH 7.5), 5 mM MgCl₂, 0.1% GDN, 1 mM DTT, 0.5 mM NaF, 0.1 mM Na₃VO₄, and complete EDTA-free protease inhibitor (Roche)] and then once more using final buffer [150 mM KOAc, 20 mM Hepes (pH 7.5), 5 mM MgCl₂, 1 mM DTT, and 0.1% GDN]. Beads were transferred onto a 1-ml Mobicol (MoBiTec), washed with 5 ml of final buffer, and then incubated with final buffer containing 40 μg of 3C protease for 60 min at 4°C. Following elution, the ribosomes were pelleted through a sucrose cushion [20 mM Hepes (pH 7.5), 150 mM potassium acetate, 5 mM MgCl₂, 0.1% GDN, and and 1 M sucrose] by centrifugation at 100,000 rpm for 1 hour using a TLA 120.2 rotor, after which the pellet was resuspended in final buffer and used for cryo-EM sample preparation and NuPAGE gel analysis.

Electron microscopy and image processing

Following procedures previously described in (22), 3.5 μl of the sample was applied to Quantifoil R3/3 holey carbon grids with 2-nm continuous carbon coating, blotted for 3 s, and then plunge frozen in liquid ethane using a Vitrobot Mark IV. Data collection was performed at 300 keV using a Titan Krios equipped with a SelectrisX Energy Filter and a Falcon4i direct electron detector at a pixel size of 0.727 Å and a defocus range of −0.5 to −3.5 μm and 40 e⁻ per Ų total dose. Gain correction, alignment, and summation of movie frames were performed using MotionCor2 (37) with 20 EER frames grouped into one fraction, producing 40 fractions with 1 e⁻ per Ų dose per fraction. Contrast transfer function (CTF) parameters were estimated using CTFFIND4 (38). Collected micrographs were automatically filtered for CTF resolution (maximum 6 Å) and astigmatism (maximum 8), resulting in a total of 45,093 micrographs being selected.

Following procedures previously described in (22), particle picking was performed using crYOLO (39), with a total of 1,412,867 particles picked. Following two-dimensional (2D) classification in cryoSPARC (40), 1,027,683 ribosomal particles were selected. The particles were then 3D classified in Relion (41, 42) using a soft mask around the 40S subunit, resulting in a small subset featuring a previously described pre-60S harboring NMD3, LSG1, and ZNF622 (43), a mixture of stalled or hibernating 80S ribosomes, and a major subset featuring various 60S states. Here, extra density was already visible for E3^UFM1 and for RQC factors. Thus, the 60S class was classified further using a mask around the binding site for RQC factors NEMF and LTN1. This revealed three distinct classes featuring RQC subunits, two of which were at sufficient resolution to be further classified.

The first subset featured NEMF in the absence of LTN1, as well as the SEC61 translocon. 3D classification with a mask around the tRNA binding sites followed by another round of classification around the E3^UFM1 binding site led to two final classes that were both refined to resolutions of 2.8 and 3.0 Å, respectively, and used for interpretation. The first, dubbed here as RQC state 1 (see also Fig. 4A), featured only NEMF, a P-site tRNA, and the SEC61 translocon (NEMF-SEC61-60S). The second (RQC state 2; Fig. 4B) also showed not only NEMF, a P-site tRNA, and SEC61 but also E3^UFM1 ligase (E3-NEMF-SEC61-60S) in a conformation identical to the previously described posttermination state 2 (posttermination state 2) (22), with the exception of the UFL1 C terminus, which was flipped back and the PTC loop was remodeled.

The second subset featured both NEMF and LTN1 but no SEC61. 3D classification around the E3^UFM1 binding site revealed two classes lacking the E3 and featuring only NEMF, LTN1, and a P-site tRNA, with minor conformational differences of NEMF between the two (NEMF conformations 1 and 2). Both classes were refined, but only the larger and better resolved “NEMF conformation 2” class (2.8 Å) was used for interpretation, here termed RQC state 4 (Figs. 1F and 4D). A third class resulting from 3D classification with the E3^UFM1 mask featured NEMF, LTN1, a P-tRNA, and the entire E3^UFM1, in a conformation similar to posttermination state 3 (posttermination state 3). As a fourth class, we identified a minor state featuring NEMF, LTN1, a P-tRNA, and partial occupancy of the UFL1 CTD in flipped conformation but lacking UFM1, thus likely representing a state where the E3^UFM1 only partially dissociated from the RQC-60S. The class featuring both RQC and E3 complexes was classified further, this time with a mask around the tRNA binding sites. This revealed two states, one featuring only a P-site tRNA and another featuring an A-site tRNA with partial occupancy for a P-site tRNA. For the latter class, however, we were unable to reach sufficient local resolution for the ligands due to the small particle number. The former (featuring only a P-site tRNA) was further refined, and multibody refinement in RELION (44) was used to achieve higher local resolution for the bound ligands. Multibody refinement was done by splitting the particle in three parts—the 60S core, LTN1 alone, and E3^UFM1 ligase together with NEMF and the P-tRNA. The focused refined maps were then combined into a composite map using Phenix (45), resulting in the best-resolved E3-RQC-60S map (RQC state 3; Fig. 4C). Postprocessing via DeepEMhancer (46) was used to assist with model building and interpretation of the density map. All important steps of image processing are summarized in fig. S1.

Model building and refinement

Following procedures previously described in (22), the best-resolved state (RQC state 3) was used to generate a model for the ER-RQC substrate. The previously generated model for the 60S-bound UFM1 E3 ligase (22) [Protein Data Bank (PDB) identifier 8ohd] was used as a template for the ribosomal backbone and E3 ligase. The template was largely unchanged with the exception of the UFL1 C terminus, which was refitted into the rotated conformation using Coot (47) and Isolde (48). In addition, the NEMF-interacting region (residues 390 to 399) was built using a combination of de novo modeling and AlphaFold Multimer (26) predictions. The extra density around the UFL1 CTD was identified via de novo modeling as NEMF residues 776 to 791, which was then confirmed via AlphaFold3 (27) prediction using the CTD and the region NEMF around said residues (770 to 800).

Following procedures previously described in (22), AlphaFold2 models were used for NEMF, LTN1, as well as ribosomal proteins uL10 and uL11. The proteins were initially rigid-body fitted using Coot (v.0.9.8.92) and then fine-tuned using Isolde. The NFACT_N domain of NEMF and parts of the coiled-coil were primarily rigid-body docked due to insufficient resolution, whereas most of the middle domain and a large part of the NFACT_C domain could be fitted at a side-chain resolution. Unlike in the yeast homolog Rqc2, we did not observe density for the hook domain. For LTN1, parts of the N- and C-terminal regions could be fitted at a side-chain level; however, most of the HEAT repeats constituting the backbone could only be rigid-body fitted. The RING domain of LTN1 was also rigid-body fitted. Ribosomal proteins uL10 and uL11 were docked into the density, with minimal adjustments necessary to fit the well resolved regions. A model for the alanyl-tRNA was generated using AlphaFold3 and fitted into the density map.

The complete model was refined using Phenix and then fine-tuned using Isolde. Figures of the model and densities were generated using ChimeraX (49).

Supplementary Materials

The PDF file includes:

Figs. S1 to S8

Table S1

Legend for dataset S1

Other Supplementary Material for this manuscript includes the following:

Dataset S1