Conformation of methylated GGQ in the Peptidyl Transferase Center during Translation Termination

The universally conserved Gly-Gly-Gln (GGQ) tripeptide in release factors or release factor-like surveillance proteins is required to catalyze the release of nascent peptide in the ribosome. The glutamine of the GGQ is methylated post-translationally at the N5 position in vivo; this covalent modification is essential for optimal cell growth and efficient translation termination. However, the precise conformation of the methylated-GGQ tripeptide in the ribosome remains unknown. Using cryoEM and X-ray crystallography, we report the conformation of methylated-GGQ in the peptidyl transferase center of the ribosome during canonical translational termination and co-translation quality control. It has been suggested that the GGQ motif arose independently through convergent evolution among otherwise unrelated proteins that catalyze peptide release. The requirement for this tripeptide in the highly conserved peptidyl transferase center suggests that the conformation reported here is likely shared during termination of protein synthesis in all domains of life.


Results and Discussion
The E. coli 70S ribosome complexed with E. coli ArfA and RF2 on a truncated mRNA was used as a representative for the ribosomal complex involved in the co-translational quality control pathway. ArfA/RF2-dependent nonstop recognition is one of the three known pathways in bacteria for rescuing ribosomes that are translating mRNAs without a stop codon 4 . ArfA detects the presence of a truncated mRNA on the ribosome by inserting its positively charged C-terminal tail into the empty mRNA channel. Once bound to the ribosome, ArfA recruits RF2 and induces RF2 to its catalytically competent "open" conformation for releasing the nascent peptide [23][24][25][26][27] . The distance between the N 5 H amine to the O4′ of U2506 is 3.9 Å. The carbonyl oxygen of the glutamine sidechain is separated from the N 3 of A2451 and the 2′OH of A76 by 3.3 Å and 4.7 Å, respectively. 23S rRNA, RF2 and P-tRNA are coloured in teal, dark red and lemon. The coordinates for the overall view are from global refinement of the nonstop termination complex, and the coordinates for the zoomed view are from local refinement over the PTC region. (b) The crystal structure of the T. thermophilus 70S ribosome in complex with E-tRNA (pink), P-tRNA (lemon), mRNA with stop codon UGA (magenta) and T. thermophilus RF2 (dark red). Zoomed view (right panel) shows a 2mF obs -DF cal map (σ = 1.0) of the PTC. The methylated GGQ motif in the canonical termination complex adopts the same conformation as seen in the nonstop termination complex (a). Important distances between atoms of the glutamine and surrounding residues of the PTC are shown. The T. thermophilus ribosome complexed with T. thermophilus release factor RF2 and mRNA containing a UGA stop codon was used as a canonical termination complex. In both canonical and nonstop termination complexes, the orientation of the methylated glutamine sidechain assumes the same well-defined conformation (Fig. 2a), which differs significantly from previously determined conformations of unmodified GGQ motifs (Fig. 2b). The post-translationally added CH 3 group orients towards the nascent peptide exit tunnel, positioning its -N 5 Himino group close to the O4′ of the ribose in U2506. The carbonyl oxygen of the glutamine sidechain points towards the base of A2451 and A76 of the P-tRNA. The distances between the carbonyl oxygen of the glutamine to N 3 of A2451 and to the 2′OH of A76 are 3.3 Å and 4.7 Å, respectively (Fig. 1a).
Compared to various structures of unmodified glutamine reported previously ( Fig. 2b and Supplementary  Fig. S4), the N 5 -methylation stabilizes the glutamine conformation through increased van der Waals interactions with three important residues in the PTC: U2506, A2451 and A2452. Chemically, peptide release is achieved by hydrolysis of the ester bond between the nascent peptide and peptidyl-tRNA (P-tRNA) in the ribosome, and earlier kinetic data suggest that the water molecule required for this process needs to be activated for catalysis to take place 14,28 . As shown in Fig. 3a, the catalytic pocket created by polar groups, including N 3 in the base of A2451, 2′OH of A76, and the mainchain amide and sidechain carbonyl oxygen of the glutamine, is just right to accommodate one water molecule in the RNA-rich core of the PTC for the hydrolysis reaction. Based on the structure, one modeled water can be coordinated by hydrogen bonding with the 2′OH of A76, the sidechain carbonyl oxygen and mainchain amide of the glutamine, and in this way one lone pair of electrons remains available for nucelophilic attack on the carbonyl carbon in the ester bond at the P-site (Fig. 3b).
Consistent with our structural observations, N 5 -methylation of the glutamine promotes peptide release on stop codons during canonical translation termination (Table 1 and Supplementary Fig. S5), as well as on nonstop mRNAs by ArfA/RF2 20 . Methylation of the GGQ sequence marginally affects the K 1/2 for binding of the release factors to their cognate codons (Table 1). However, the same covalent modification enhances the k cat of peptide release for both RF1 and RF2. The rate enhancement for RF1 is modest (~5-fold), whereas the enhancement for RF2 is ~20-fold. Finally, comparison of the kinetics between the cognate and near-cognate complexes suggests the coupling between correct binding of a stop codon and catalysis is less tight for RF2 compared to RF1 (Table 1 and Supplementary Fig. S5), consistent with earlier biochemical work 29 . Notably, the orientation of the glutamine sidechain in our experimentally determined structure agrees well with earlier theoretical structural models 11,17,30 , but does not support an alternative model suggested from a recent crystal structure of the heterologous canonical In both structures, the carbonyl oxygen of the glutamine side chain orients toward the base of A2451 and P-tRNA. It's remains to be understood whether the different conformations of U2506 between the two structures have mechanistic implications regarding the function of this residue in canonical versus nonstop termination. Backbone differences of G250 and G251 in the GGQ motif may be due to sequence variation between RF2 in E. coli and T. Thermophilus. In E. coli RF2, an alanine preceeds the GGQ, whereas the preceeding residue is proline in T. thermophilus RF2. (b) Side chains of the unmethylated glutamine in the GGQ motif determined by previous structural investigations assume different conformations in the PTC. Unmethylated glutamine displays conformational diversity (grey) compared to the structures determined in this study (dark red); N 5 -methylation of the glutamine stabilizes the side chain orientation. P-tRNA is shown as sticks in lemon. The 23S of the ribosome was used for alignment. The 70S termination complexes with RF2 and stop codon UAA (①, PDB: 4V67) 10 ; with RF1 and stop codon UAG (②, PDB: 4V7P) 11 ; with RF1 on stop codon UAA (③, PDB: 4V63) 12 ; with RF2 on stop codon UGA (④, PDB: 4V5E) 13 ; with RF2 on stop codon UAA (⑤, PDB: 4V5J) 9 .
termination complex of T. thermophilus ribosome and E. coli RF1 22 . In the structure of this heterologous canonical termination complex, the post-translationally added methyl group was reported to point to the base of A2451. Instead of the carbonyl oxygen, the -N 5 H-was modeled to face A76 of the P-tRNA, although the electron density to support the model fitting was less convincing. Furthermore, in our structure, an ~180-degree rotation around the C δ -C γ bond in the glutamine swaps the positions of carbonyl and methyl-NH groups, but the methyl group resides outside of the density. A refinement of this rotamer with Refmac or Phenix showed that the overall fitting of this alternative glutamine conformation to the density is not as good as our structure shown in Fig. 1  (Supplementary Fig. S6). Lastly, it's worth mentioning that sequences of bacterial release factors from different species vary, yet these proteins resemble each other structurally. Therefore, it remains questionable whether the conformation from a heterologous termination complex is functionally pertinent, as suggested by a decreased rate constant for peptide release compared to the homologous complex (Jin Lab unpublished data), and also seen from one heterologous termination complex recently 25 . In this study, the E. coli ArfA can recruit T. Thermophilus RF2 but fails to induce the RF2 into the "open" or catalytically competent conformation in the T. Thermophilus ribosome.
Taken together, the post-translational modification helps to stabilize the glutamine conformation in the PTC of the ribosome, and orients the carbonyl oxygen of the glutamine to coordinate a water nucleophile in the hydrolysis reaction. Therefore, precise structural orientation of the glutamine sidechain conferred by N 5 methylation allows for better accommodation of the water molecule during peptide release. The GGQ tripeptide has evolved to catalyze the hydrolysis reaction in the PTC of the ribosome. The highly conserved nature of this peptide motif and the ribosomal PTC suggest the conformation of the GGQ reported in this study will most likely be adopted by eukaryotic class I release factor eRF1 and other release factor-like surveillance proteins in higher organisms.

Methods
Purification of ribosomes, tRNAs and proteins. Escherichia coli ribosomes from MRE600 strain and Thermus thermophilus ribosomes from K8 strain were purified as described 13,31 . E. coli tRNA fMet and tRNA Phe were prepared by in vitro transcription 9,20 . Release factors from E. coli and T. thermophilus were cloned and overexpressed in BL21 (DE3) cells and purified by affinity and ion-exchange chromatography 9,20 . E. coli and T. thermophilus release factors with GGQ motif methylated (RF1 m and RF2 m ) were obtained by co-expressing the release factors and their cognate methyltransferase (PrmC) gene together and the methylation status of these proteins was identified by mass spectroscopy as described 20,27 . mRNAs with AUG or UUC in the P-site and different cognate or near-cognate stop codons in the A-site were purchased from Dharmacon (GE, Thermo Scientific). Sequences of the mRNAs are listed as the following with the P-site codon underlined: Peptide release assay. Aminoacylation and formylation of tRNA fMet were done as described 20 . Translation termination complexes containing E. coli 70S ribosomes, f-[ 35 S]-Met-tRNA fMet and mRNAs with different cognate or near-cognate stop codons were programmed and the catalytic rate constants k cat and binding constants K 1/2 were measured as described 3,20,32 . Electron microscopy, data collection and image processing. Ribosomal complexes were formed by incubating ribosomes with mRNA (6 μM), tRNA fMet , ArfA and RF2 m (12 μM each) in buffer A (20 mM Hepes-KOH, pH 7.5, 15 mM magnesium acetate, 150 mM potassium acetate, 4 mM β-mercapthoethanol, 2 mM spermidine, 0.05 mM spermin) for 30 min at 37 °C 33 . Sample preparation for cryoEM followed the methods as described 34,35 . Data was collected in vitreous ice using a JEOL 3200FS transmission electron microscope operating at 300 keV equipped with a K2 summit direct electron detector (Gatan) 27 . Micrographs from the same nonstop complex 27 were combined to achieve the best possible data quality. 2,107 micrographs showing Thon rings beyond 3.5 Å were used. The drifts of movie frames were corrected using MotionCor2 36 . Magnification anisotropy was corrected with mag_distortion_estimate and mag_distortion_correct 37 . CTFFIND4 38 was used to determine the contrast transfer function.
Subsequently, a total of 268,364 particles were extracted and subjected to reference-free 2D classification in Relion 39,40 to remove non-ribosomal particles. 3D classification was carried out over 221,911 particles to remove ribosomes without ligands bound ( Supplementary Fig. S1). A total of 177,529 particles yielded from the 3D classification were refined using the cryoEM map of the same nonstop complex (EMD-8505) 27 as an initial reference. Statistical particle-based movie correction and radiation-damage weighing 41 were then used to obtain the polished particles. Finally, a focused classification with signal subtraction (FCwSS) 42 using a soft-edged mask over P-tRNA and RF2 was performed. After additional rounds of 2D and 3D classifications, a total of 143,372 particles were subjected to refinement and yielded a global 3D reconstruction with an overall resolution of 3.24 Å.
A masked 3D refinement with partial signal subtraction was performed 42,43 to obtain a better resolved map for the PTC region of the ribosome. A mask over the PTC was applied, and the final refinement led to a reconstruction with an overall resolution of 3.10 Å.
Resolutions were reported base on the gold-standard Fourier shell correlation (FSC) of 0.143 criterion 44,45 . Final map was sharpened by applying a negative B-factor estimated using Relion 40,44 . Local resolution was estimated using ResMap 46 .
Model building and refinement. Coordinates for the same nonstop complex obtained from an earlier study (PDB: 5U4I) 27 were used as starting models for the model building and refinement. Models were fitted into the maps using Chimera 47 and built according to the map using Coot 48 .
Refinement was carried out in Refmac5.8 49 with an optimized B-factor of −50. Molecular restraints for the secondary structure, RNA sugar pucker, base stacking and base-pair geometry were generated by ProSMART 50 and LIBG 51 . Cross-validation against overfitting was done on two half maps where the FSC was monitored 51 , and the final model was validated using MolProbity 52 . Refinement statistics were summarized in Supplementary  Table S1.
Canonical translation termination complex formation and crystallization. Canonical termination complexes of the 70S ribosome and RF2 m from T. thermophilus on mRNA SD-PPhe-UGA with a stop codon UGA were formed in buffer B (5 mM Hepes pH 7.5, 10 mM Mg(OAc) 2 , 50 mM KCl, 10 mM NH 4 Cl and 6 mM β-mercaptoethanol). Crystallization of the complex was carried out as described 9,13 . Briefly, 2.8 mM Deocy Big Chap (Hampton Research) was mixed with the formed termination complex immediately before the crystallization. Crystals were grown at 19 °C in 100 mM Tris-HAc pH 7.12, 200 mM KSCN, 3.9-4.2% (w/v) PEG20K, 3.9-4.2% (w/v) PEG550MME and were cryoprotected by using 4.6% PEG20K and PEG550MME (4.6%, 10.95%, 17.3%, 23.65% and 30%). Crystals were harvested and plunge frozen in liquid nitrogen. Data collection, model building and refinement. X-ray diffraction data were collected at 100 K at the 24-IDC beamline at Northeastern Collaborative Access Team (NE-CAT) in the Advanced Photon Source, Argonne National Laboratory. Data were processed using XDS 53 . The apo T. thermophilus 70S ribosome from PDB code 4V51 54 was used as the starting model. Refinement was carried out using CNS 55 and Phenix 56 . The ligands including mRNA, E-tRNA and partial P-tRNA without the CCA-end, and partial RF2 without the GGQ motif were first built into unbiased difference density. After the refinement, the sigma-A weighted difference Fourier maps clearly showed the orientation of the methylated glutamine. The CCA end of the P-tRNA and the GGQ m motif were built lastly. Refinement statistics were summarized in Supplementary Table S2