Mimicry of canonical translation elongation underlies alanine tail synthesis in RQC

Summary

Aborted translation produces large ribosomal subunits obstructed with tRNA-linked nascent-chains, which are substrates of Ribosome-associated Quality Control (RQC). Bacterial RqcH, a widely-conserved RQC factor, senses the obstruction and recruits tRNA^Ala(UGC) to modify nascent-chain C-termini with a polyalanine-degron. However, how RqcH and eukaryotic homologs, despite their relatively simple architecture, synthesize such C-terminal tails in the absence of a small ribosomal subunit and mRNA has remained unknown. Here we present cryo-EM structures of Bacillus subtilis RQC complexes representing different Ala-tail synthesis steps. The structures explain how tRNA^Ala is selected via anticodon reading during recruitment to the A-site, and uncover striking hinge-like movements in RqcH leading tRNA^Ala into a hybrid A/P-state associated with peptidyl-transfer. Finally, we provide structural, biochemical and molecular genetic evidence identifying the Hsp15 homolog as a novel RQC component that completes the cycle by stabilizing the P-site tRNA conformation. Ala-tailing thus follows mechanistic principles surprisingly similar to canonical translation elongation.

eTOC Blurb

Incompletely-made nascent-chains are tagged for proteolysis by Ribosome-associated Quality Control (RQC). In bacterial RQC, this is mediated by C-terminal alanine-tailing. Filbeck et al. elucidate the structural basis for alanine-tailing using cryo-EM, discover Hsp15/RqcP as an essential factor, and reveal that C-terminal tailing follows similar principles to canonical translation elongation.

Graphical Abstract

Introduction

Translation elongation is a processive, GTP hydrolysis-dependent reaction whereby the ribosome ratchets along mRNA codons one at a time, recruiting aminoacylated tRNAs to the A-site and mediating successive rounds of transfer of the nascent polypeptide chain from the P-site tRNA to the A-site tRNA (Behrmann et al., 2015; Rodnina et al., 2017; Voorhees and Ramakrishnan, 2013). Sometimes, however, ribosomes stall and translation is aborted before protein synthesis is completed (Joazeiro, 2017, 2019). To prevent cytotoxicity, in addition to eliminating the associated mRNA (Schmidt et al., 2016), the nascent-chains produced by aborted translation are targeted for degradation by Ribosome-associated Quality Control (RQC) (Defenouillere and Fromont-Racine, 2017; Joazeiro, 2019). Central to RQC are the RqcH protein and its homologs, found in bacteria, archaea and eukaryotes (Burroughs and Aravind, 2014; Lytvynenko et al., 2019) (Figure 1A). In bacteria, RqcH acts redundantly with tmRNA/SsrA in the proteolytic tagging of ribosomal stalling products and is required for virulence of several pathogenic species (Lytvynenko et al., 2019). In eukaryotes, RqcH homologs (yeast Rqc2 and mammalian NEMF) are thought to support the function of the E3 ligase, Ltn1/Listerin (Defenouillere et al., 2013; Kostova et al., 2017; Lyumkis et al., 2014; Shao et al., 2015; Shen et al., 2015). Consistent with having an important role in protein quality control, mutations in both NEMF (Martin et al., 2020) and Listerin (Chu et al., 2009) cause neurodegeneration.

Figure 1. Overall molecular architecture of the bacterial RQC complex during tRNA^Ala(UGC) recognition.

(A) Domain architecture of B. subtilis RqcH. Note that the HhH domain is regarded as integral part of the NFACT-N domain in this work. Left, cartoon diagram. Right, segmented RqcH density from the structure shown in “B,” with individual domains colored as indicated. (B) Composite cryo-EM reconstruction of the bacterial RQC complex as seen from the ribosomal subunit interface. Density segments for the large ribosomal subunit (grey), the A-site (yellow) and P-site (magenta) tRNAs, Hsp15 (green) and RqcH (domains colored as in “A” are shown). In the right-hand panel, the density was sliced open to allow for an unobstructed view on the P-site tRNA-linked nascent chain (red). (C-E) Structural details of the RqcH-50S interactions, focused on the RqcH M domain and the ribosomal factor binding site (C), the NFACT-R domain and rRNA h38 (D) and the NFACT-N domain and rRNA h69 (E). The atomic model is shown in cartoon representation. Amino acid side chains and nucleotide bases central to the respective interactions are shown in full atomic representation. Small icons indicate the zoomed areas. See also Figures S1, S2, S3 and S4.

RqcH homologs act by first surveying large ribosomal subunits for obstruction with tRNA-linked nascent-chains (Joazeiro, 2019). Cryo-EM analyses of eukaryotic RQC complexes have shed light on how the obstruction is sensed (Lyumkis et al., 2014; Shao et al., 2015; Shen et al., 2015): the simultaneous binding of Rqc2 to the free 60S intersubunit surface and the exposed P-site tRNA indicates that these two components that should normally not be present together are associated. Next, RqcH homologs recruit an additional tRNA (tRNA^Ala in bacteria or tRNA^Ala and tRNA^Thr in yeast) for nascent-chain tagging with C-terminal tails (Lytvynenko et al., 2019; Shen et al., 2015). In bacteria, alanine (Ala) tails made by RqcH act as degrons recognized by proteasome-like proteases such as ClpXP (Lytvynenko et al., 2019). In eukaryotes, C-terminal Ala and threonine (Thr) tails (“CAT tails”) push lysine (Lys) residues out of the exit tunnel for ubiquitylation by Ltn1 (Kostova et al., 2017) and a subset of those sequences can play additional roles in stimulating protein aggregation (Choe et al., 2016; Defenouillere et al., 2017; Defenouillere et al., 2016; Yonashiro et al., 2016) or degradation (Sitron and Brandman, 2019).

Elegant studies recapitulating Rqc2-mediated CAT tail elongation in S. cerevisiae extracts supported the notion that C-terminal tailing is a small ribosomal subunit-independent, peptidyl transferase center-catalyzed reaction and found that drugs targeting either eEF1a or eEF2 GTPase, or a non-hydrolyzable GTP analog, had no effect on CAT tailing, while inhibiting canonical elongation (Osuna et al., 2017). However, exactly how C-terminal tails are synthesized in obstructed large ribosomal subunits by the small, architecturally-simple Rqc2-family proteins in the absence of a small ribosomal subunit and mRNA template has remained largely unknown. In particular, structures of RQC complexes depicting the components involved at various reaction steps remain scarce—for example, only one structure to date has captured tRNAs on both ribosomal P- and A-sites (Shen et al., 2015). Moreover, the limited resolution of this structure has precluded a detailed understanding of the A-site tRNA recruitment step. With the aim to shed light into mechanisms and principles underlying C-terminal tailing we set out to solve structures of the B. subtilis RQC complex along the reaction cycle.

Results

Molecular architecture of the bacterial RQC complex during tRNA^Ala(UGC) recognition

50S-RqcH complexes were isolated from B. subtilis by affinity purification with antibody directed against a C-terminal Flag-tag on RqcH (Figure S1A and (Lytvynenko et al., 2019)). Proteins co-immunopurified with RqcH-Flag were identified by mass spectrometry, which confirmed that components of the large ribosomal subunit were enriched (Figure S1B, Supplemental Data File 1).

The isolated RqcH-50S complexes were structurally characterized by cryo-EM single-particle analysis. In brief, 50S ribosomal subunits were automatically located on micrographs and computationally sorted for the presence and conformation of RqcH and tRNAs. RqcH-containing complexes were observed in two conformational states correlating with differing tRNA occupancy (Figure S2A). We first focused on the more abundant and structurally better-defined state, containing RqcH, an A-site tRNA, a nascent chain-linked P-site tRNA and, unexpectedly, a novel small globular protein assuming a central position in the complex. As this state depicts how the anticodon of the A-site tRNA is specifically read out by RqcH, we termed it “decoding state” (see below).

Fragmented density and significantly lower local resolution for RqcH after the initial particle refinement suggested small-scale conformational plasticity between RqcH and the 50S subunit (Figure S3A). We therefore refined the 50S subunit and RqcH independently (Figure S2B), which dramatically improved local resolution and density quality for RqcH and directly-associated factors (Figure S3B). Using this approach, we obtained a cryo-EM density of the complex in the decoding state (Figure 1B) at overall resolutions of 2.5 Å and 3.2 Å for the 50S subunit and RqcH with the associated factors, respectively (Figure S3B). This allowed for detailed interpretation of the entire complex at a near-atomic level of detail. In particular, we could build accurate atomic models (Figure S3B, Table 1) for the A-site tRNA (Figure S4A), the P-site tRNA (Figure S4B) with the conjugated obstructing nascent chain (Figure S4C) and 98% of RqcH residues (Figure S4D), including the NFACT-R domain, which has not been identified in previous cryo-EM structures of eukaryotic RQC complex (Lyumkis et al., 2014; Shao et al., 2015; Shen et al., 2015), and functionally important protein segments that had not been resolved in previously reported X-ray structures of the homologous protein from Streptococcus suis, RqcH/FBPS (Musyoki et al., 2016) (see Methods). The only exception is a flexible linker of RqcH that connects the coiled-coil domain to the NFACT-R domain (residues 435–445). This almost complete atomic model of RqcH allowed us to rationalize the function of residues highly conserved among homologs (Figure S5). Several of these conserved residues appear to fulfil a structural role by stabilizing either RqcH domains’ core folds or the interface between the individual RqcH domains, and to mediate specific interactions with other components of the RQC complex as described below.

Table 1.

Cryo-EM data collection, refinement and validation statistics

Data collection and processing
Microscope	FEI Titan Krios
Voltage	300
Camera	Gatan K3
Energy filter	Quanta GIF
Energy filter slit width (eV)	30
Magnification (nominal)	81,000
Defocus range (μm)	−1.0 to −3.0
Calibrated pixel size	1.07
Exposure rate (e−/Å2)	44/40/54
Number of frames per movie	25/30/40
Automation Software	EPU
Number of micrographs	11,314
Initial particle images (no.)	5,869,410
Final particle images (no.)	138,382
Symmetry imposed	C1
	50S ribosomal subunit (decoding state)	non-ribosomal factors (decoding state)
Local resolution range (Å)	2.3 – 8.0	2.3 – 5.0
Map resolution at 0.143 FSC (Å)	2.56	3.22
Refinement
Software	PHENIX 1.18.2–3874	PHENIX 1.18.2–3874
Initial model used (PDB Code)	3J9W	5H3X, 5H3W, 1DM9
Masked resolution at 0.143/0.5 FSC (Å)	2.5/2.9	3.2/4.2
Correlation coefficient (CCmask)	0.8	0.64
Map sharpening B factor (Å2)	−52.37	−75.02
Model composition
Non-hydrogen atoms	86,654	10,827
Protein residues	2,977	775
Nucleotides	2,956	219
B factors (Å2)	min/max/mean	min/max/mean
Protein	1.39/45.80/19.47	17.46/42.38/30.89
Nucleotides	0.00/85.32/22.28	13.52/66.78/51.74
R.M.S deviations
Bond lengths (Å)	0.003	0.005
Bond angles (°)	0.819	0.976
Validation
MolProbity score	1.66	2.17
Clashscore	5.54	6.73
Rotamer outliers (%)	0.04	5.07
C-beta outliers (%)	0	0.14
CaBLAM outliers (%)	4.32	4.87
EMRinger score	2.82	1.34
Ramachandran plot
Favored (%)	94.7	96.18
Allowed (%)	5.3	3.82
Disallowed (%)	0	0

The overall molecular architecture of the complex is determined by the docking of RqcH onto the inter-subunit surface of the 50S ribosomal subunit via a tripartite interface, which becomes accessible only after subunit splitting. The first interaction is centred on the ribosomal factor-binding site including the universally conserved rRNA sarcin-ricin loop (SRL). The RqcH M domain and proximal regions of the coiled-coil domain interact with ribosomal rRNA helices h43, h44, h59, and ribosomal protein L11 (‘M domain contact’; Figure 1C). Notably, the highly conserved Phe377 (replaced with a Tyr in some homologs) in RqcH’s coiled coil-2 (cc2) stacks with rRNA bases A1113 and G1114 of h43. As for eukaryotic RQC complexes, binding of RqcH and translational GTPases to the factor-binding site is mutually exclusive. The second interaction site between RqcH and the 50S subunit involves the RqcH NFACT-R domain and rRNA h38 located close to the central protuberance of the 50S subunit (‘NFACT-R contact’; Figure 1D). Several surface-exposed and mostly conserved basic residues (Arg476, Arg480, Arg534, Lys537) of the NFACT-R domain project towards rRNA h38 and suggest the involvement of predominantly unspecific electrostatic interactions with the rRNA backbone. The third interaction site between RqcH and the 50S subunit involves the short rRNA h69, which is a central element in ribosomal subunit association. The NFACT-N domain interacts with rRNA h69 bases via hydrophobic interactions mediated by the highly conserved Met74 and Met77 residues, and with the rRNA h69 backbone via electrostatic interactions mediated by Lys81 (‘NFACT-N contact’; Figure 1E).

Collectively, this tripartite interface explains the stable interaction between RqcH and the 50S subunit underlying co-immunoprecipitation of the two components. At the same time, this interface appears to provide a versatile structural framework for the large-scale conformational rearrangements of RqcH during the Ala-tailing reaction cycle described in more detail below.

Coordination and selective reading of the A-site tRNA anticodon

The recruitment of tRNAs to the ribosomal A-site mediates the C-terminal-tailing activity of RqcH homologs (Shen et al., 2015). In our cryo-EM structure of the “decoding state”, the A-site tRNA is tightly coordinated within the RQC complex via several interactions. The 3’-CCA tail is accommodated in a canonical A-site position, interacting with the 50S rRNA (Figure 2A). The tRNA D- and T-loops are coordinated by a composite basic groove of RqcH mostly formed by the two coiled-coil helices (Figure 2B). The interface most critical for RqcH’s ability to specifically recruit tRNA^Ala involves the A-site tRNA anticodon region and the NFACT-N domain, including an extended interaction network tightly coordinating nucleotides U34-C38 (with the anticodon being U34-G35-C36). Nucleotide A37 and C38 directly following the anticodon are held in place by a loop of the NFACT-N domain (~145–155), which is one of the more variable regions among RqcH homologs (Figure 2C, Figure S5), and the strongly conserved Arg125 inserts into the anticodon loop further stabilizing it (Figure 2C).

Figure 2. Structural basis for tRNA^Ala(UGC) recruitment by RqcH.

(A) Interaction between the A-site tRNA CCA-tail and the rRNA. (B) D- and T-loop interaction of the A-site tRNA with the positively charged patch of the RqcH coiled coil. RqcH is shown in surface representation color-coded according to electrostatic charge (blue: positive; red: negative). (C) A-site tRNA anticodon-loop interaction with the NFACT-N domain. The anticodon nucleotides (U34-C36) are shown in orange. The NFACT-N domain is either shown in cartoon (left) or surface (right) representation. (D-E) Base-specific readout of the anticodon nucleotides G35 (D) and C36 (E). H-bonds relevant for the anticodon reading by RqcH are depicted as dotted lines. Unless stated otherwise, coloring and atomic model representation as in Figure 1. Small icons indicate the zoomed areas. See also Figure S5.

Importantly, our structure explains the basis for RqcH’s binding selectivity towards tRNA^Ala (Lytvynenko et al., 2019) for generating Ala tails. All three nucleotides of the anticodon (U34-C36) are embedded into well-defined pockets in the NFACT-N domain (Figure 2C). G35 and C36 engage in specific H-bonding patterns with residues of the NFACT-N domain, providing a Watson-Crick-like reading of the anticodon explaining specificity for tRNA^Ala. Consistent with their phylogenetic conservation and requirement for C-terminal tailing activity (Burroughs and Aravind, 2014; Choe et al., 2016; Defenouillere et al., 2016; Lytvynenko et al., 2019; Shen et al., 2015; Yonashiro et al., 2016), residues Asp97 and Arg98 are crucially involved in coordinating G35 (Figure 2D). The universally conserved residues Glu121 and Arg154 are involved in reading C36 together with Asn151 (Figure 2E). We could not observe any base-specific interaction pattern for the binding pocket of the Wobble base U34. Instead, U34 is non-specifically coordinated by pi-cation interactions with Arg98 (Figure 2C). The RqcH-mediated anticodon reading is thus analogous to mRNA codon-mediated decoding in canonical translation elongation.

The B. subtilis Hsp15 homolog is a central structural component of the bacterial RQC complex

In the cryo-EM reconstruction of the decoding state, we made the surprising observation of an additional small globular protein at apparently stoichiometric association with the RQC complex. Despite its small size, this protein assumes a central position in the complex, interacting with all other components except the A-site tRNA, suggesting an important role in bacterial RQC. Aiming to identify this novel component, we revisited the mass spectrometry data (Figure S1B, Supplemental Data File 1). The molecular chaperone trigger factor, the ribosomal silencing factor RsfS and the homolog of the E. coli small heat shock protein Hsp15 were identified as main candidates among RqcH-Flag co-purified proteins. While trigger factor and RsfS could be excluded based on their shape, an X-ray structure of the region of E. coli Hsp15 that is conserved among bacteria (Staker et al., 2000) provided an almost perfect rigid body fit into the unoccupied globular density resolved in the cryo-EM reconstruction. We thus prepared an atomic model for B. subtilis Hsp15 and refined it against the cryo-EM density. The assignment was confirmed by excellent agreement of amino acid side chains resolved in the cryo-EM density with the atomic model of B. subtilis Hsp15 (Figure S4E).

Hsp15 assumes a central position in the RQC complex, simultaneously interacting with the 50S subunit, the RqcH NFACT-N domain and the peptidyl-tRNA in the P-site conformation (Figure 3A). Hsp15 binds to the stem of rRNA helices h68 and h69 via its ‘αL’ rRNA-binding motif, comprising two short α-helices with five basic residues (Arg2, Lys5, Arg15, Arg16, Lys20, Lys36) that interact unspecifically with the rRNA backbone (Figure 3B). Towards RqcH, a hydrophobic pocket of Hsp15 accommodates a surface-exposed hydrophobic residue in a loop of the NFACT-N domain (Ile110). This loop (108–112) and the hydrophobic character of Ile110 are conserved among bacterial RqcH homologs, but not in archaea or eukaryotes (Figure S5), which do not have obvious Hsp15 homologs. To accommodate the anticodon loop of the P-site tRNA, Hsp15 and the NFACT-N domain form a positively charged composite groove (Figure 3C), to which Hsp15 and RqcH contribute five (Lys8, Arg11, Lys14, Lys68, Lys69) and four (Arg23, Lys26, Arg40, Arg80) residues, respectively. Collectively, by directly binding to the 50S subunit, the P-site tRNA and the RqcH NFACT-N domain, Hsp15 engages in a comprehensive interaction network within the RQC machinery.

Figure 3. Hsp15 is a structural core component of the RQC complex.

(A) Zoomed view centered on Hsp15 and its molecular surroundings in the RQC complex. (B) Structural details of the Hsp15 interactions with the rRNA helices h68 and h69. (C) Hsp15 and the NFACT-N domain form a composite basic groove for P-site tRNA binding. Hsp15 and RqcH are shown in surface representation color-coded according to electrostatic charge (blue: positive; red: negative). Close-up view as indicated. See also Figure S5.

Hsp15 is required for Ala tail elongation

Having observed the extensive interactions of Hsp15 with other RQC complex components, we set out to determine whether Hsp15 is functionally required for RQC. We first examined relevant growth phenotypes caused by deletion of the B. subtilis Hsp15-encoding gene, yabO (we have renamed the gene rqcP, for ‘P-site tRNA stabilizing factor’). As reported for SsrA (Muto et al., 2000) and RqcH (Lytvynenko et al., 2019), Hsp15 was not essential for B. subtilis growth (Figures 4A, S6A, S6B). Whereas the simultaneous deletion of rqcP and rqcH did not cause an obvious growth defect, the ΔrqcP ΔssrA double deletion strain had a mild growth phenotype in rich media, like ΔrqcH ΔssrA (Lytvynenko et al., 2019) (Figures 4A, S6A, S6B). A hallmark of defective RqcH function is hypersensitivity to the translational inhibitor spectinomycin in a manner that depends on the simultaneous inactivation of SsrA (Lytvynenko et al., 2019). We thus next tested whether loss of Hsp15 function would have a similar effect. Remarkably, whereas ΔrqcP cells grew normally in presence of spectinomycin, ΔrqcP ΔssrA cells showed a marked synthetic growth defect, comparable to that observed for ΔrqcH ΔssrA cells (Figure 4A). ΔrqcP additionally phenocopied ΔrqcH regarding having synthetic interactions with ΔssrA in presence of the translational inhibitor erythromycin (Figure S6A) and under heat stress (Figure S6B), which can also increase ribosomal stalling (Moore and Sauer, 2007; Muto et al., 2000; Shin and Price, 2007).

Figure 4. Hsp15 is functionally required for Ala tailing.

(A) Hsp15 deficiency (ΔrqcP) phenocopies RqcH deficiency in B. subtilis. Cultures were spotted in 10-fold serial dilutions onto LB-agar plates containing or not spectinomycin at the indicated concentrations. (B) Hsp15 mediates degradation of ribosomal stalling products. Left, indicated strains were analyzed for expression of episomal reporters, GFP-nonstop (GFP-ns) and parental GFP. E.V., empty vector. Western blots (WB) against GFP or the FtsZ loading control are shown. Right, GFP-ns/FtsZ ratios for the indicated strains, determined from biological replicates. Error bars reflect standard error of the mean (SEM). (C) RqcH-Flag co-immunoprecipitated proteins from wild type (WT) or Hsp15-deficient (ΔrqcP) strains. Top, Coomassie staining. Bottom, anti-Flag immunoblot of whole cell lysate (WCL). (D) Extracted-ion chromatogram of samples analyzed using the PRM method (examples of raw data utilized to generate Table 2). The chromatograms show a GFP peptide (FSVSGEGEGDATYGK) and a selected Ala-tailed peptide (LmSGLFSAAAA; lowercase m represents oxidized Met) derived from GFP-ns, which was isolated from ΔssrA, ΔssrA ΔrqcH and ΔssrA ΔrqcP cells as indicated. See also Figure S6.

Since RQC and SsrA regulate the levels of ribosome stalling products in a redundant fashion (Lytvynenko et al., 2019), we asked whether Hsp15 also plays a role in this process. For this purpose, we analyzed deletion strains for expression levels of a reporter protein encoded by an mRNA lacking stop codons, which causes ribosomes to stall at the mRNA 3’ end (Lytvynenko et al., 2019). As reported, levels of GFP-nonstop (GFP-ns) were increased by inactivation of ssrA and even further by the simultaneous deletion of rqcH (Lytvynenko et al., 2019) (Figure 4B) although the effects were not as strong in the assays with an episomal reporter compared to previous experiments using strains with integrated reporters (Lytvynenko et al., 2019). Like rqcH, deletion of rqcP alone did not affect GFP-ns levels, but did so in the ΔssrA background (Figure 4B). In contrast, expression of the parental GFP control was unaffected by the gene deletions, alone or in combination (Figure 4B). Thus, Hsp15-null strains closely phenocopy RqcH-null strains in all respects examined, supporting the hypothesis that Hsp15 has an essential function in RQC.

In E. coli, which does not have an RQC pathway, Hsp15 has been suggested to function in recycling of obstructed 50S subunits by stabilizing an otherwise flexible 50S-associated tRNA in the P-site conformation, presumably to enable a nascent-chain release factor to bind to the A-site (Jiang et al., 2009; Korber et al., 2000). In analogy, we thus asked whether in B. subtilis, Hsp15 was required for binding of RqcH. To test that, we immunoprecipitated (IP’ed) RqcH-Flag from WT or ΔrqcP cells and analyzed the extent of co-IP’ed 50S subunits (Lytvynenko et al., 2019). The results show that a comparable amount of 50S subunit proteins co-IP’ed with RqcH from both strains (Figure 4C), suggesting that Hsp15 does not play a critical role in RqcH recruitment.

We next tested the alternative possibility that Hsp15 functions in RQC during the Ala-tailing reaction step. For this purpose, we measured the extent of GFP-ns modification with Ala tails in the presence or absence of Hsp15 using a mass spectrometry (MS) approach (Parallel Reaction Monitoring, or PRM) as reported previously (Lytvynenko et al., 2019). In short, GFP-ns was isolated from ΔssrA, ΔssrA ΔrqcH and ΔssrA ΔrqcP strains; the ΔssrA background was selected to prevent the competing SsrA-tagging reaction. Detection of Ala-tailed species also required slowing down their degradation, which was accomplished by treating cells with the ClpP protease inhibitor, Bortezomib (Velcade) (Akopian et al., 2015; Lytvynenko et al., 2019). Results of the PRM analysis show that, in comparison to ΔssrA, the ΔssrA ΔrqcP strain had a nearly complete loss of Ala tail-modified peptides relative to reference GFP peptides (Figure 4D, Table 2). Moreover, this reduction was of similar extent to that observed for the ΔssrA ΔrqcH strain. These findings indicate that Hsp15 and RqcH are both required for the synthesis of Ala tails, which can account for the marked effects of rqcP deletion on RQC activity (Figures 4A,B).

Table 2. Ala-tailing is Hsp15 (rqcP)-dependent.

MS analyses of GFP-ns monitored specific Ala tail-modified peptides and GFP peptides using a parallel reaction monitoring (PRM) method. Area ratios were calculated relative to the GFP peptide FEGDTLVNR from the respective strain to indicate each peptide’s relative abundance in the different strains.

	Peptide	ΔssrA	ΔssrA ΔrqcH	ΔssrA ΔrqcP
Ala tail	LMSAAAAAA	2.5	0.0	0.0
	LMSAAA	2.8	0.0	0.0
	LmSGLFSAA	7.5	0.0	0.1
	LmSGLFSAAAA	2.1	0.0	0.0
GFP	FEGDTLVNR	100	100	100
	FSVSGEGEGDATYGK	25	22	16

An RQC complex structure with a hybrid state A/P-tRNA provides insights into the Ala-tail elongation mechanism

Aiming to obtain further insights into the role of Hsp15 in Ala tail synthesis, we set out to structurally characterize the second type of RqcH-containing 50S complexes observable in the cryo-EM dataset (Figures S2A, S2B), notably lacking Hsp15. This subpopulation of RQC complexes carried a hybrid state A/P-site tRNA, which is a required intermediate for peptidyl transfer during canonical translation elongation. This RQC “translocating state” thus depicts a critical step in the Ala-tailing reaction (Figure 5A). Even though resolution for our translocating state is lower compared to the decoding state (Figure S3C), large-scale structural rearrangements underlying translocation are clearly evident (Figure S7A,B), which are all linked to re-orientation of the short α-helix (Phe279-Gln290) inserted between the NFACT-N and cc1 domains of RqcH. In the decoding state, this short α-helix directly intersects with the trajectory that an A-site tRNA has to follow during translocation to the P-site (Figure 5B), thereby locking the A-site tRNA in the decoding state (hence, the “locking helix”). The locking helix, the cc1 helix and the C-terminal NFACT-N helix, which are organized in a kinked arrangement in the decoding state, straighten out in the translocating state (Figure 5B, Figure S7C) through the movement of two hinges (Phe277-Phe279 and Gln290-Gln292). Notably, this aligned helical structure is reminiscent of the X-ray structure of the S. suis RqcH/FBPS coiled-coil domain (Figure S7C) (Musyoki et al., 2016). In the decoding state, the locking helix is held in place by intermolecular contacts with the NFACT-R domain, including hydrophobic interactions between Phe279 (NFACT-N) and Ile565 (NFACT-R), as well as a salt bridge formed between Arg276 (NFACT-N) and Asp562 (NFACT-R) (Figure 5C). The NFACT-R domain becomes more mobile in the translocating state, as suggested by the fact that it is no longer visible (Figure 5A), raising the possibility that an allosteric mechanism is involved in releasing the locking helix.

Figure 5. Conformational changes in RqcH drive the A-site tRNA into a hybrid state associated with peptidyl transfer.

(A) Cryo-EM reconstruction of the bacterial RQC complex in the translocating state filtered according to local resolution. The NFACT-R domain not resolved in this state is indicated by a transparent oval. Coloring as in Figure 1. In the right-hand panel, the density was sliced open to allow for an unobstructed view on the hybrid state tRNA-linked nascent chain (red). (B) Comparison of RqcH conformations in the decoding (colored) and translocating (grey) states. Only the C-terminal helix of the NFACT-N domain, cc1 and the locking helix are shown. The tRNA is shown in the A-site conformation of the decoding state and the arrow indicates the direction of tRNA movement towards the P-site conformation. Asterisks denote the two hinges underlying the conformational change. (C) Structural details of the decoding state-specific interactions between the NFACT-R domain and residues in close proximity to the locking helix. Coloring and representation as in Figure 1. (D) RqcH conformational change from the decoding to translocating state visualized by trajectories linking the C_α−atoms in both conformations. The model of the decoding state is shown as reference. Same components as in “B”. The A-site and A/P-site tRNAs corresponding to the two conformations are depicted in dark and light grey, respectively, and shown only in the top view for clarity. The RqcH atomic model and trajectories are color-coded from blue (no motion) to red (high motion). The black spot indicates the axis around with the tRNA-NFACT-N body rotates. See also Figures S2, S3 and S7.

The striking rearrangement of the locking helix has several implications. The straightened-out helix moves the NFACT-N and M domains further away from each other, forcing the NFACT-N domain to tilt around its 50S-interaction site to accommodate the new helix arrangement (Figure 5D). Since the tight interaction between the NFACT-N domain and the A-site tRNA anticodon is retained during this motion, and since the locking helix is lifted high enough to no longer present an obstacle to tRNA movement, the A-site tRNA is pushed into the hybrid A/P state, with its CCA-tail accommodated in the canonical P-site of the 50S subunit. Defined density for the nascent chain linked to the hybrid state tRNA indicates that the peptidyl transfer reaction has already been completed in these complexes (Figure 5A). The A/P hybrid state is further stabilized by tRNA binding to the locking helix and to rRNA h38 near the central protuberance.

Collectively, the binding of Hsp15 to a P-site tRNA in the decoding state of the RQC complex (Figure 3C), the absence of Hsp15 in the translocating state (Figure 5A), the Hsp15 requirement for Ala-tail synthesis (Figure 4D, Table 2) and the previously reported function of E. coli Hsp15 (Jiang et al., 2009), suggest a model in which Hsp15 acts in driving the hybrid state tRNA to a classical P-site conformation, thus completing the Ala-tailing cycle.

Discussion

Our results define the bacterial RQC complex as being minimally composed of an obstructed 50S subunit, RqcH and its Hsp15 cofactor, and elucidate key steps in Ala tail synthesis. It is remarkable that RqcH and Hsp15 appear to be sufficient for carrying out peptide synthesis with the large ribosomal subunit, even though both components are small, architecturally-simple proteins that bear no structural similarity with the ribosomal small subunit. Strikingly, the Ala tailing reaction and canonical translation elongation follow evolutionarily convergent principles and, from the perspective of the 50S subunit, the reaction intermediates are structurally almost undistinguishable (Figure S8). In the decoding state that follows initial RQC complex assembly, the nascent chain-linked tRNA is bound to Hsp15 and oriented in a typical P-site conformation, whereas an acceptor tRNA^Ala is specifically recruited to the A-site by RqcH-mediated anticodon reading. This step in Ala tail synthesis mimics mRNA codon-mediated decoding in canonical elongation (Figure S8A), but relies entirely on protein-RNA interactions instead of RNA-RNA base pairing. Next, a conformational change in RqcH drives the A-site tRNA^Ala to a hybrid A/P conformation, coupled with peptidyl transfer. This step copies forward ratcheting of ribosomal subunits in canonical elongation (Figure S8B). How tRNA translocation in Ala-tailing is initiated remains unclear. It is tempting to speculate that the NFACT-R domain, which sits adjacent to the A-site and shows a dynamic behavior between decoding and translocating states, may sense the occupancy of the A-site and trigger the release of the locking helix. Finally, we observe a strong requirement for Hsp15 to complete the Ala-tailing cycle. In an analogous mechanism to the reversal of ribosomal subunit ratcheting in canonical translation, Hsp15 drives the A/P hybrid tRNA to a P-site conformation. Because of the physical block imposed by the RqcH NFACT-N domain, Hsp15’s action must rely in one of at least two alternative pathways. If RqcH remains anchored to the ribosomal factor binding site via the M domain, further large-scale conformational rearrangements of RqcH could allow the tRNA to be moved below the NFACT-N domain towards the P-site. Alternatively, RqcH may dissociate entirely from the complex at this point with every Ala-addition cycle.

The Ala tailing reaction steps observed in this study indicate that the binding of RqcH would be mutually exclusive with EF-G and EF-Tu throughout the entire cycle, which implies that translational GTPases are not required in Ala-tailing. This is consistent with results of biochemical studies on CAT tailing in yeast extracts (Osuna et al., 2017) and, conceptually, with the reaction taking place in the absence of mRNA and a small ribosomal subunit. Thus, the energy consumed during the tRNA charging reaction is likely sufficient to drive the RqcH-mediated incorporation of Ala to the obstructing nascent chain.

Limitations

While this study provides mechanistic insights into the Ala-tailing elongation cycle, an important step for future work that we cannot directly address with our current data will be to clarify how Ala tail synthesis is terminated and the nascent chain released from the large subunit for degradation. It is likely that mechanisms previously suggested for eukaryotic RQC may also be relevant for prokaryotes. This includes the stochastic action of a dedicated release factor similar to the eukaryotic tRNA endonuclease Vms1/ANKZF1 (Izawa et al., 2017; Kuroha et al., 2018; Su et al., 2019; Verma et al., 2018; Yip et al., 2019; Zurita Rendon et al., 2018), a peptidyl-tRNA hydrolase (Kuroha et al., 2018), but conceivably also canonical termination factors (Chadani et al., 2012; Shimizu, 2012) or uncharged tRNA (Caskey et al., 1971). Whether, by similarity to the function proposed for E. coli Hsp15 (Jiang et al., 2009), also its B. subtilis homolog facilitates recruitment of a release factor to the complex remains to be investigated.

Our findings also open entirely new questions. In particular, does eukaryotic RQC have an Hsp15 counterpart? Although Hsp15 homologs are not obvious outside bacteria, other factors could perform its role. Alternatively, specific reaction requirements may be distinct in eukaryotes. Nonetheless, it is likely that the principles elucidated here will be generally applicable to RQC across organisms, as residues required for C-terminal tail synthesis by RqcH are universally conserved (Figure S5) and the RqcH-large ribosomal subunit complexes from bacteria and eukaryotes show remarkable overall similarity.

STAR METHODS

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Stefan Pfeffer (s.pfeffer@zmbh.uni-heidelberg.de).

Materials Availability

All unique/stable reagents generated in this study are available from the Lead Contact without restriction.

Data and Code Availability

Atomic coordinates have been deposited in the Protein Data Bank under accession codes PDB- 7AQC (decoding state) and PDB-7AQD (translocating state). Cryo-EM densities have been deposited in the Electron Microscopy Data Bank under accession codes EMD-11862 (decoding state; composite density from multibody refinement) and EMD-11864 (translocating state).

Experimental model and subject details

All strains were derived from B. subtilis 168 1A700 and were grown in LB media at 37°C. To generate strains deleted for rqcP we followed the procedure described by (Arnaud et al., 2004) and as reported in (Lytvynenko et al., 2019).

Method Details

Bacillus cultures

For plasmid transformation, overnight cultures were diluted to OD₆₀₀ = 0.1 and grown until OD₆₀₀ ≃ 1.5 in BMK media containing 60 mM K₂HPO₄, 40 mM KH₂PO₄, 2% dextrose, 3 mM trisodium citrate dihydrate, 0.08 mM ferric ammonium citrate, 15 mM potassium aspartate, 10 mM MgSO₄, 150 nM MnCl₂, 0.2 mM L-tryptophan, 0.05% yeast extract (Koo et al., 2017). 350 μl of culture were incubated with 1 μg plasmid DNA at 37°C, 160 rpm for 1 h followed by plating on agar in presence of chloramphenicol (5 μg/ml) or a combination of erythromycin (1 μg/ml) and lincomycin (25 μg/ml).

Constructs

DNA cloning was performed according to standard procedures using E. coli DH5α as described (Lytvynenko et al., 2019). To generate a construct for deleting rqcP (pMAD ΔrqcP), 1-Kb fragments flanking rqcP were amplified from bacterial chromosomal DNA. Gene-upstream and downstream fragments were cloned with the BamHI/SpeI and the SpeI/NcoI restriction sites, respectively. Double-digested fragments were ligated to each other through the SpeI ends and cloned into the BamHI/NcoI sites of the linearized pMAD vector.

Purification of RQC complexes for Cryo-EM

Aiming to capture complexes containing tRNA bound to the A-site, RQC complex purification was performed in presence of methanol, which has been reported to strengthen tRNA-A site interactions (e.g., (Ali et al., 2006)). In another experiment with the same purpose, but performed in the absence of methanol, immunopurified material was supplied with purified tRNA^Ala(UGC) in excess and yielded similar results.

2×2-l cultures were grown to a final OD₆₀₀ of ~1.2 then harvested and lysed as described (Lytvynenko et al., 2019) in sterile-filtered buffer containing 50 mM Tris/HCl pH 7.4 at 4 °C, 100 mM KCl, 10 mM MgCl₂, 10% methanol (v/v), 5% glycerol (v/v), 0.1% NP-40 (v/v), 1 mM DTT, 1x protease inhibitor and ribonuclease inhibitor. Lysates were clarified for 22 min at 17 krpm in a SS-34 rotor (RCF_avg = 22548.5) (Thermo Scientific) and the supernatant was incubated with 2×240 μl pre-washed anti-Flag M2 Affinity Gel for 2 h at 4°C on a turning wheel. Beads were collected by centrifugation and washed three times with IP buffer. Next, they were transferred to a Mobicol ‘F’ column with a filter of a 35-μm pore size and washed extensively by gravity flow with lysis buffer without glycerol and NP-40. RqcH complexes were eluted by addition of 50 μl lysis buffer without glycerol and NP-40 containing 0.3 mg/ml 3X Flag-peptide for 45 min at 4°C on a turning wheel and an additional 15 min standing on ice after a brief centrifugation. Afterwards the eluate was collected and centrifuged for 5 min at ~18000 RCF and 4 °C. After centrifugation, no pellet was visible and the supernatant was transferred to a fresh microcentrifuge tube. A portion of the eluate was collected for SDS-PAGE analysis and the remaining sample was directly used for cryo-EM grid preparation.

RqcH-Flag immunoprecipitation to monitor 50S binding

The procedures were performed as described (Lytvynenko et al., 2019), with minor modifications. 1-l cultures were grown in LB medium supplemented with 5 μg/ml chloramphenicol at 37°C, 120 rpm to OD₆₀₀ ≃ 1.5, then cells were harvested, frozen in liquid nitrogen and stored at −80°C. Cells were disrupted on ice by sonication (10 cycles 30 s each) in lysis buffer containing 50 mM Tris-HCl pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 5% glycerol, 0.1% NP-40, 1 mM DTT and 1x protease inhibitor (Complete ULTRA Tablets, EDTA free; Roche). Lysates were cleared by centrifugation at 17,000 rpm in an SS-34 fixed-angle rotor for 22 min at 4°C, and the supernatant was incubated for 1.5 h at 4°C on a turning wheel (Labinco) with 100 μl pre-washed anti-Flag M2 Affinity Gel. Beads were collected by centrifugation, and after 3 washes with 20 ml lysis buffer were transferred to a Mobicol ‘F’ filter with a 35-μm pore size (MoBiTec) and washed by gravity flow. Proteins were eluted with 200 μl lysis buffer containing 0.2 mg/ml 3X Flag-peptide for 45 min at 4°C on a turning wheel. Eluates were precipitated with 10% TCA, washed twice with acetone and separated on NuPAGE^™ 4–12% Bis-Tris gels (Invitrogen, USA). Gels were stained with Coomassie InstantBlue^™ (Expedeon).

Immunoblotting

Cultures were grown at 37°C, 120 rpm in LB medium supplemented with 5 μg/ml chloramphenicol to OD₆₀₀ = 0.8–1.0, then pellets were harvested, frozen in liquid nitrogen and stored at −80°C. Cells were disrupted with a FastPrep-24^™ 5G instrument (MP Biomedicals) in lysys buffer containing 50 mM Tris-HCl pH 7.4, 100 mM NaCl, 5 mM MgCl₂, 5% glycerol, 0.1% NP-40, 1 mM DTT and 1x protease inhibitors. Proteins were separated on a NuPAGE^™ 4–12% Bis-Tris gel and transferred to a 0.45 μm PDVF membrane using the Invitrogen Mini Blot Module (Invitrogen, USA). Membranes were sequentially incubated with mouse monoclonal anti-GFP antibody or rabbit polyclonal anti-FtsZ antibody (a kind gift of Jeff Errington), and HRP-conjugated secondary antibodies. To develop the signal, membranes were treated with enhanced chemiluminescence substrate kit (ECL, GE Healtcare). Images were acquired with a GE ImageQuant LAS 4000 instrument and analyzed using Fiji (Schindelin et al., 2012) and Prism 8 (GraphPad).

Growth assays

Strains were grown at 37°C in LB medium to OD₆₀₀ = 1.0–1.5, then diluted to OD₆₀₀ =0.1 and 10-fold serial dilutions spotted on LB-agar plates containing or not the indicated drugs and grown for 16 h at 37°C or at the indicated temperature.

GFP immunoprecipitation for mass spectrometry analysis

Procedures were performed as described (Lytvynenko et al., 2019), with minor modifications. 1-l cultures were grown, harvested and disrupted as for RqcH-Flag immunoprecipitation, with the exception that during growth cultures were treated with 20 μM bortezomib for 3.5 h. After clearing by centrifugation at 17,000 rpm in an SS-34 fixed-angle rotor for 22 min at 4°C, lysates were incubated with 50 μl pre-washed GFP-Trap agarose beads for 1.5 h at 4°C in a turning wheel. Beads were collected by centrifugation and washed 3 times with lysis buffer, then transferred to a Mobicol ‘F’ filter with a 35-μm pore size and washed by gravity flow. To elute GFP, beads were incubated with 100 μl 0.2 M Glycine pH 2.5 for 15 min at 4°C, followed by pH neutralization with 1 M Tris pH 10.4. Proteins in the eluate were precipitated with TCA 10%, washed twice with acetone and separated on NuPAGE^™ 4–12% Bis-Tris gels. Protein bands were identified by gel staining with Coomassie InstantBlue^™, and excised for mass spectrometry analysis.

Mass spectrometry (MS) characterization of RQC preparation

Procedures were performed essentially as described previously (Lytvynenko et al., 2019) with the exception that peptide separation was performed on as EASY PepMapTM RSLC C18 column (2μm, 100Å, 75 μm x 50cm, Thermo Scientific, San Jose, CA), and ions were created with an EASY Spray source (Thermo Scientific, San Jose, CA) held at 50°C using a voltage of 1.9 kV. All flow rates were 250 nL/min on this column and peptides were eluted using a gradient of 5%–25% solvent B (80/20 acetonitrile/water, 0.1% formic acid) in 180 min, followed by 25%–44% solvent B in 60 min, 44%–80% solvent B in 0.10 min, a 5 min hold of 80% solvent B, a return to 5% solvent B in 0.10 min, and finally a 20 min hold of solvent B. Tandem mass spectra were searched as previously described (Lytvynenko et al., 2019) and the protein and peptide identification results were also visualized with Scaffold v 4.7.1 (Proteome Software Inc., Portland OR), a program that relies on various search engine results (i.e.: Sequest, X!Tandem, MASCOT) and which uses Bayesian statistics to reliably identify more spectra (Keller et al., 2002). Proteins were accepted that passed a minimum of two peptides identified at 1% peptide and protein FDR, within Scaffold.

Parallel reaction monitoring method (PRM) to determine the relative abundance of MS-identified peptides across different strains

Procedures were performed essentially as described (Lytvynenko et al., 2019). Proteins were in-gel digested. A positive control sample (GFP-ns isolated from bortezomib-treated ΔssrA strain) was analyzed with a data-dependent acquisition (DDA) 2-h LC-MS/MS run to determine the retention times of Ala-extended C-terminal peptides. Subsequently, a parallel reaction monitoring method (PRM) (Peterson et al., 2012; Rauniyar, 2015) was created under identical LC conditions to monitor these Ala peptides from different experimental samples (in triplicates) in a targeted manner. GFP peptides (FEGDTLVNR and FSVSGEGEGDATYGK) were also included in the scheduled inclusion target mass list for normalization.

Cryo-EM grid preparation and data acquisition

Holey carbon grids (Cu R2/1; 200 mesh; Quantifoil) were glow-discharged for 20 s under oxygen atmosphere using a Solarus 950 plasma cleaner (Gatan, Inc.). EM-grids were prepared at ambient temperature and humidity using a Vitrobot Mark IV (Thermo Fisher). 4 μl of sample were applied onto the grids, blotted with Whatman filter paper Nr.1 for 3 s, and plunged into liquid ethane cooled by liquid nitrogen.

Data were acquired in three sessions on a Titan Krios TEM (Thermo Fisher/FEI) using the EPU software package (Thermo Fisher). The microscope was operated at 300 kV using an energy-filtered K3 camera (Gatan) in dose fractionation mode. Datasets were acquired at an objective pixel size of 1.07 Å/px and a defocus range of −0.5 to −2. The cumulative dose of 43 e/A² (dataset 1), 40 e/A² (dataset 2) and 54 e/A² (dataset 3) was distributed over 25 (dataset 1), 40 (dataset 2) and 30 (dataset 3) frames. For each selected hole, the defocus was adjusted automatically prior to acquisition of four micrograph movie stacks per hole.

Cryo-EM data processing

All image processing steps were performed using Relion 3.0-beta (Zivanov et al., 2018) and are summarized in Figure S2, which also provides detailed information on the number of micrographs, the number of particles, the shape of masks for particle sorting, estimated resolution, etc. Initial 2D image processing and classification steps were performed separately on each dataset as follows: Beam-induced motion was corrected based on a 5×5 grid using MotionCor2 (Zheng et al., 2017) as implemented in Relion 3.0-beta. Estimation of the contrast transfer function (CTF) was performed using gCtf (Zhang, 2016) as implemented in Relion 3.0-beta. Only micrographs with an estimated resolution of better 3.0 Å were retained for the analysis. 840 particles were located manually in micrographs of dataset 1, extracted with a box size of 128×128 px (3.21 Å pixel size) and subjected to 2D classification. 2D classes depicting a 50S ribosome served as templates for autopicking. Autopicked particles were extracted with a box size of 128×128 px (3.21 Å pixel size) and subjected to an initial round of 3D classification to remove false-positive particles. Classes depicting 50S ribosomes with high-resolution features were selected and merged. The retained particles were subjected to a 3D auto-refinement run, which served as a basis for subsequent 3D classification runs without sampling. One to two rounds of focused classification per dataset with a binary mask encompassing the A-site tRNA, the P-site tRNA and RqcH were used to identify particles representing either the “decoding state” or translocating state (Figure S2A).

For the decoding state, the retained particles from all three datasets were merged and extracted at full spatial resolution with a box size of 384×384 px (1.07 Å pixel size) based on re-centred coordinates. Particles were subjected to 3D auto-refinement, per-particle CTF-Refinement (including beam tilt estimation) and Bayesian particle polishing. The polished particles were subjected to another round of 3D auto-refinement, which served as a basis for multibody refinement, in which the bulk of the 50S subunit (segment 1) and all non-ribosomal components plus the ribosomal factor binding site (segment 2) were treated as two independently moving bodies.

For the translocating state, particles representing 50S subunits with a hybrid state A/P-site tRNA from all three datasets were merged and subjected to 3D auto-refinement, which served as a basis for two subsequent runs of 3D classification without sampling, firstly focused on the RqcH coiled-coil domain and secondly on the NFACT-N domain. The retained particles were extracted at full spatial resolution with a box size of 384×384 px (1.07 Å pixel size) based on re-centred coordinates and subjected to a 3D auto-refinement. Subsequent multibody refinement did not result in improved densities, likely because the A/P-site tRNA-RqcH density segment was too flexible or did not provide sufficient signal for refinement independent from the 50S subunit.

All final cryo-EM densities were subjected to post-processing including automatic B-factor sharpening and/or to local resolution analysis and filtering based on Relion’s implementation of local post-processing.

Atomic modelling of the “decoding state”

The atomic model for B. subtilis RqcH was prepared based on X-ray structures of individual S. suis RqcH/FBPS domains (PDB-5H3X, PDB-5H3W) (Musyoki et al., 2016). The X-ray structures were fitted as rigid bodies in UCSF Chimera (Pettersen et al., 2004) into the RqcH density segment at 3.2 Å resolution obtained from multibody refinement. Amino acid residues were mutated to correspond to the B. subtilis sequence and at the same time fitted into the cryo-EM density in Coot (Emsley et al., 2010). Lys435 – Pro445 were not resolved and were therefore not included in the model. The same approach was taken to create an atomic model for B. subtilis Hsp15 based on the X-ray structure of Escherichia coli Hsp15 (PDB-1DM9) (Staker et al., 2000) and to create atomic models for B. subtilis tRNA^Ala(UGC) in the A- and P-site conformations based on a B. subtilis tRNAs included in PDB-3J9W (Sohmen et al., 2015). For the P-site tRNA, nucleotides 16–18 and 32–37 (anticodon loop) were not resolved and therefore not included in the model. For the large ribosomal subunit, an atomic model of a B. subtilis 50S ribosome (PDB-3J9W) based on a high-resolution cryo-EM structure (Sohmen et al., 2015) was fitted into the 50S density segment at 2.6 Å resolution obtained from multibody refinement and adjusted where required in Coot.

All atomic models were combined in UCSF Chimera and relaxed using MDFF simulation in ISOLDE (Croll, 2018). Subsequently, the model was refined in real space against a composite density including the two segments from multibody-refinement using Phenix (Liebschner et al., 2019), with the initial model as a reference. Real space refinement was based on restraints generated in “Phenix Refine” and weighted by a factor of 0.6. Weighting for nonbonded restraints was set to a factor of 2000. The resulting model was validated in Phenix.

Flexible fitting for the “translocating state”

Atomic coordinates for RqcH and the A-site tRNA obtained from the decoding state were used Atomic coordinates for RqcH and the A-site tRNA obtained from the decoding state were used as a starting model for flexible fitting into the respective density segments of the translocating state, resolved locally to 8–10 Å (Figure S3C). Residues for the NFACT-R domain were removed, because it was not visible in the cryo-EM density for the translocating state. The rest of the model was split into three segments that were individually docked into the density of the translocating state as rigid bodies using UCSF Chimera: 1) the A-site tRNA, 2) the NFACT-N domain and 3) the coiled-coil domain plus the M domain. Next, rigid bodies for flexible fitting were defined based on their secondary structure and domain organization as analyzed in Ribfind (Pandurangan and Topf, 2012). These bodies were: 1) the NFACT-N domain, 2) the locking helix, 3) the coiled-coil 1 helix, 4) the coiled-coil 2 helix and 5) the M domain. The model was subsequently fitted into the density using two consecutive FlexEM (Topf et al., 2008) runs. The orientations of the coiled-coil 1 and coiled-coil 2 helices were slightly adjusted in UCFS Chimera between the two runs.

Analysis and presentation of cryo-EM densities and atomic models

All cryo-EM densities and atomic models were presented using UCSF Chimera (Pettersen et al., 2004) or ChimeraX (Goddard et al., 2018). Molecular surfaces, electrostatic surface analysis and H-bond analysis were performed in UCSF Chimera. For the visualization of conformational changes in RqcH (Figures 5C,D), the atomic models for the decoding and translocating states were aligned according the 50S subunit. For the vector representation in Figure 5D, trajectories between the Cα atoms in the two conformational states were computed and colour-coded according to their root mean square deviation in PyMol (Delano, W. L. PyMOL: an open-source molecular graphics tool). For comparison of tRNA orientations occurring in the Ala tailing reaction and canonical translation elongation in Figure S8, atomic models for the RqcH-50S complexes were aligned according to the 50S rRNA with atomic models of bacterial ribosomes including either a hybrid state A/P-site tRNA (PDB-6WDF) (Loveland et al., 2020) or canonical A-site and P-site tRNAs (Crowe-McAuliffe et al., 2018) (PDB-5JTE).

Multiple Sequence Alignment and analysis

Sequences from (Lytvynenko et al., 2019) were aligned using Clustal Omega (Sievers et al., 2011) and visualized in Jalview (Waterhouse et al., 2009).

Quantification and Statistical Analysis

Forms of quantitation are specified under each method when applicable. Errors bars represent standard error of the mean (SEM).

Supplementary Material

1

Supplemental Data File 1 Mass spectrometry analysis of RqcH-Flag preparation, Related to Figure 1. Proteins co-immunoprecipitated with RqcH-Flag in two different preparations were identified by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Numbers refer to unique peptide counts. In red font, proteins of the 50S subunit; in blue font, proteins of the 30S subunit.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal αnti-GFP antibody	Roche	Cat#11814460001
Rabbit polyclonal αnti-FtsZ antibody	Jeff Errington	N/A
Bacterial and Virus Strains
B. subtilis Wild-type 168 1A700 (WT) trpC2	BGSC	N/A
ΔrqcH trpC2	Lytvynenko et al., 2019	N/A
ΔssrA trpC2	Lytvynenko et al., 2019	N/A
ΔrqcP trpC2	This work	N/A
ΔrqcH ΔssrA trpC2	Lytvynenko et al., 2019	N/A
ΔrqcH ΔrqcP trpC2	This work	N/A
ΔssrA ΔrqcP trpC2	This work	N/A
ΔrqcH ΔssrA ΔrqcP trpC2	This work	N/A
WT, empty vector trpC2 WT, pHT01 Cmr	This work	N/A
WT, RqcH-FLAG trpC2 WT, pHT01-RqcH-FLAG Cmr	This work	N/A
ΔrqcP, RqcH-FLAG trpC2 ΔrqcP, pHT01-RqcH-FLAG Cmr	This work	N/A
WT, GFP trpC2 WT, pHT01 GFP Cmr	This work	N/A
ΔrqcH, GFP trpC2 ΔrqcH, pHT01 GFP Cmr	This work	N/A
ΔssrA, GFP trpC2 ΔssrA, pHT01 GFP Cmr	This work	N/A
ΔrqcP, GFP trpC2 ΔrqcP, pHT01 GFP Cmr	This work	N/A
ΔrqcH ΔssrA, GFP trpC2 ΔrqcH ΔssrA, pHT01 GFP Cmr	This work	N/A
ΔrqcH ΔrqcP, GFP trpC2 ΔrqcH ΔrqcP, pHT01 GFP Cmr	This work	N/A
ΔssrA ΔrqcP, GFP trpC2 ΔssrA ΔrqcP, pHT01 GFP Cmr	This work	N/A
ΔrqcH ΔssrA ΔrqcP, GFP trpC2 ΔrqcH ΔssrA ΔrqcP, pHT01 GFP Cmr	This work	N/A
WT, GFP-ns trpC2 WT, pHT01 GFP-ns Cmr	This work	N/A
ΔrqcH, GFP-ns trpC2 ΔrqcH, pHT01 GFP-ns Cmr	This work	N/A
ΔssrA, GFP-ns trpC2 ΔssrA, pHT01 GFP-ns Cmr	This work	N/A
ΔrqcP, GFP-ns trpC2 ΔrqcP, pHT01 GFP-ns Cmr	This work	N/A
ΔrqcH ΔssrA, GFP-ns trpC2 ΔrqcH ΔssrA, pHT01 GFP-ns Cmr	This work	N/A
ΔrqcH ΔrqcP, GFP-ns trpC2 ΔrqcH ΔrqcP, pHT01 GFP-ns Cmr	This work	N/A
ΔssrA ΔrqcP, GFP-ns trpC2 ΔssrA ΔrqcP, pHT01 GFP-ns Cmr	This work	N/A
ΔrqcH ΔssrA ΔrqcP, GFP-ns trpC2 ΔrqcH ΔssrA ΔrqcP, pHT01 GFP-ns Cmr	This work	N/A
Chemicals, Peptides and Recombinant Proteins
GFP-Trap® agarose beads	Chromotek	Cat#gta-10
Anti-FLAG® M2 Affinity Gel	Sigma-Aldrich	Cat#A2220
3X FLAG® Peptide	Sigma-Aldrich	Cat#F4799
RNase OUT™ Ribonuclease Inhibitor	Thermo Fisher	Cat#10777–019
Protease inhibitor mixture	Roche	Cat#04693159001
Bortezomib	Alfa Aesar	Cat#J60378
Erythromycin	Serva	Cat#21208.02
Spectinomycin	Serva	Cat#35294.01
Deposited Data
Atomic coordinates ‘decoding state’ of bacterial RQC complex	This work	PDB-7AQC
Cryo-EM density ‘decoding state’ of bacterial RQC complex	This work	EMD-11862
Atomic coordinates ‘translocating state’ of bacterial RQC complex	This work	PDB-7AQD
Cryo-EM density ‘translocating state’ of bacterial RQC complex	This work	EMD-11864
Atomic coordinates of Streptococcus suis FBPS N-terminal domain	Musyoki et al., 2016	PDB-5H3X
Atomic coordinates of Streptococcus suis FBPS C-terminal domain	Musyoki et al., 2016	PDB-5H3W
Atomic coordinates of Escherichia coli Hsp15	Staker et al., 2000	PDB-1DM9
Atomic coordinates of the Bacillus subtilis MifM-stalled ribosome	Sohmen et al., 2015	PDB-3J9W
Atomic coordinates of Escherichia coli elongating ribosome	Loveland et al., 2020	PDB-6WDF
Atomic coordinates of Escherichia coli ErmBL-stalled ribosome	Crowe-McAuliffe et al., 2018	PDB-5JTE
Oligonucleotides
rqcP BamHI upstream_fw: TTTTTTTGGATCCTCATATGTCAGGTTTACGACC	This work	N/A
rqcP SpeI upstream_rv: AAAAAAACTAGTATCTATGATCTCCTCTCAGTTTC	This work	N/A
rqcP SpeI downstream_fw: TTTTTTTACTAGTCCCCACCTCATACAATGCAG	This work	N/A
rqcP NcoI downstream_rv: AAAAAAACCATGGAACCGTGTGATGGTCGTTC	This work	N/A
Recombinant DNA
pMAD ampr, ermr	Arnaud et al., 2004	N/A
pMAD ΔrqcP ampr, ermr, ΔrqcP	This work	N/A
pHT01 ampr, Cmr, lacI, repA	MoBiTec	PBS001
pHT01 RqcH-FLAG PrqcH RqcH-FLAG TrqcH	Lytvynenko et al., 2019	N/A
pHT01 GFP PrqcH FLAG only HA-Smt3 TrqcH, P43 GFP TtrpA	Lytvynenko et al., 2019	N/A
pHT01 GFP-ns PrqcH FLAG only HA-Smt3 TrqcH, P43 GFP-ns TtrpA	Lytvynenko et al., 2019	N/A
Software and Algorithms
ImageJ	Schindelin et al., 2012	https://imagej.nih.gov/ij/
EPU 2.6	ThermoFisher Scientific	N/A
Relion 3.0 beta	Zivanov et al., 2018	https://www3.mrc-lmb.cam.ac.uk/relion; RRID:SCR_016274
UCSF MotionCor2	Zheng et al., 2017	https://emcore.ucsf.edu/ucsf-software; RRID:SCR_016499
gCtf	Zhang, 2016	https://www2.mrc-lmb.cam.ac.uk/research/locally-developed-software/zhang-software/; RRID:SCR_016500
UCSF Chimera	Pettersen et al., 2004	http://plato.cgl.ucsf.edu/chimera/; RRID:SCR_004097
UCSF ChimeraX	Goddard et al., 2018	https://www.cgl.ucsf.edu/chimerax/; RRID:SCR_015872
Coot 0.9	Emsley et al., 2010	http://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/; RRID:SCR_014222
ISOLDE	Croll, 2018	https://isolde.cimr.cam.ac.uk/
Phenix 1.18.2	Liebschner et al., 2019	https://www.phenix-online.org/; RRID:SCR_014224
Ribfind	Pandurangan and Topf, 2012	http://ribfind.ismb.lon.ac.uk/
FlexEM	Topf et al., 2008	http://topf-group.ismb.lon.ac.uk/flex-em/
PyMol 2.4.0	Delano, W. L. PyMOL	https://pymol.org/2/; RRID:SCR_000305
Clustal Omega	Sievers et al., 2011	https://www.ebi.ac.uk/Tools/msa/clustalo/; RRID:SCR_001591
Jalview 2.11.2.1	Waterhouse et al., 2009	http://www.jalview.org/; RRID:SCR_006459
PowerPoint 16	Microsoft	N/A
Word 16	Microsoft	N/A
Other
Holey carbon cryo-EM grids Cu R2/1 200 mesh	Quantifoil Micro Tools	N/A

Highlights.

• Structural basis for nascent chain modification with Ala tails in bacterial RQC

• Specific recruitment of tRNA^Ala to A-site by RqcH-mediated tRNA anticodon reading

• RqcH-mediated formation of an A/P-tRNA state associated with peptidyl transfer

• Hsp15/RqcP completes the cycle by stabilizing the P-site tRNA conformation