Insights into the ribosome function from the structures of non-arrested ribosome nascent chain complexes

Abstract

During protein synthesis, the growing polypeptide threads through the ribosomal exit tunnel and modulates ribosomal activity by itself or by sensing various small molecules, such as metabolites or antibiotics, appearing in the tunnel. While arrested ribosome-nascent chain complexes (RNCCs) have been extensively studied structurally, the lack of a simple procedure for the large-scale preparation of peptidyl-tRNAs, intermediates in polypeptide synthesis that carry the growing chain, means that little attention has been given to RNCCs representing functionally active states of the ribosome. Here we report the facile synthesis of stably linked peptidyl-tRNAs—through a chemoenzymatic approach based on native chemical ligation—and use them to determine several structures of RNCCs in the functional pre-attack state of the peptidyl transferase center (PTC). These structures reveal that C-terminal parts of the growing peptides adopt the same uniform β-strand conformation stabilized by the intricate network of H-bonds with the universally conserved 23S rRNA nucleotides and rationalize how the ribosome synthesizes growing peptides having various sequences with comparable efficiencies.

Graphical Abstract

Editorial Summary:

Synthesis of peptidyl-tRNAs is challenging because there are no enzymes that can directly attach the desired peptide to tRNA. Now it has been shown that a chemoenzymatic approach based on native chemical ligation can be used for the semi-synthesis of peptidyl-tRNAs for structural/biochemical studies of arrested and non-arrested ribosome complexes.

Protein biosynthesis, also known as translation, is a key step in the gene expression pathway, and catalyzed by the ribosome – one of the most conserved and sophisticated molecular machines of the cell. The ribosome provides a platform for binding the messenger RNA (mRNA) and transfer RNAs (tRNAs). tRNAs serve as adaptor molecules and have two functional ends, one carrying the amino acid and the other end containing the anticodon that recognizes the mRNA codon. tRNAs bind to the ribosome in three places: A (aminoacyl), P (peptidyl), and E (exit) sites. The A site binds the incoming aminoacyl-tRNA (aa-tRNA), the P site retains the peptidyl-tRNA carrying the nascent polypeptide chain, and the E site binds deacylated tRNA before it dissociates from the ribosome. The ribosome is composed of two unequal subunits, small and large (30S and 50S in bacteria), which join together to form functional 70S ribosomes. The small subunit decodes genetic information delivered by mRNA, whereas the large subunit covalently links amino acids into a nascent protein, which is then threaded through the nascent peptide exit tunnel (NPET) that spans the body of the large subunit.

Although NPET was initially thought to be a passive conduit for the growing polypeptide chain, it became clear in recent years that it plays an active role in co-translational protein folding¹ as well as in the regulation of protein synthesis². For example, some of the most widely used classes of antibiotics, such as macrolides, bind in the NPET of the bacterial ribosome and interfere with the progression of some growing polypeptides through this tunnel^3-5. There are also many examples of small molecules and/or metabolites that modulate the rate of translation by binding in the NPET^2,6. The binding of an antibiotic or a small-molecule modulator to the NPET and the presence of a nascent peptide with a particular sequence results in a slowdown or even a complete arrest of such ribosome-nascent chain complexes (RNCCs) on the mRNA. This happens because the growing peptide adopts a particular conformation inside the NPET and establishes direct contacts with the walls of the tunnel and a small molecule present there at the same time². Moreover, the synthesis of some nascent chain sequences is intrinsically problematic for the ribosome, with polyproline stretches being the most striking example⁷.

Although molecular mechanisms underlying nascent chain-mediated ribosome stalling have attracted increasingly great attention in recent years, such events are rare and can occur with relatively few peptide sequences. In fact, evolution has built ribosomes to be able to synthesize most of the cellular proteins without the need for auxiliary factors. Furthermore, the ribosome remains an effective catalyst despite the existing chemical diversity of its multiple amino acid substrates as well as possible heterogeneity of growing peptide chain conformations in the NPET. Structural studies of arrested RNCCs can explain why ribosomes become inactive while trying to synthesize particularly problematic amino acid sequences^8-13. However, the complementary question – “What makes the ribosome a versatile catalyst capable of acting with comparable efficiency upon twenty different aa-tRNA substrates and plenty of possible nascent chain folds within the tunnel?” – always stayed beyond the scope of such studies and could be tackled by the structural analysis of non-arrested RNCCs or their mimics that represent a functionally active state of the ribosome.

In nearly all structural and biochemical studies of arrested RNCCs^8-18, peptidyl-tRNAs were prepared in cis by exploiting the natural peptidyl transferase activity of the ribosome to generate peptides attached to tRNAs in the P site. While certainly producing close-to-natural ribosome complexes, this approach allows preparation of homogenous samples of only the peptide-arrested RNCCs, which represent an inactive state of the PTC by definition. The inability to perform pairwise comparisons of structures of arrested vs. non-arrested RNCCs (e.g., with vs. without antibiotic or a small molecule; WT stalling motifs vs. mutants) hampers our profound understanding of the underlying ribosome stalling mechanisms. Capturing non-arrested nascent peptides in the pre-peptidyl transfer state remains challenging and requires either a kinetic control¹⁹ or usage of hydrolysis-resistant aa-tRNAs²⁰ to prevent transpeptidation reaction. To date, the only reported structures of the ribosome in the pre-peptidyl transfer state of the PTC contain a minimal possible P-site substrate, fMet-tRNA_i^Met, which allows visualizing only a single first amino acid of the nascent peptide but not a longer polypeptide chain within the ribosome tunnel^19,20. There are no available structures of the ribosome functional complexes containing both aa-tRNA in the A site and peptidyl-tRNA in the P site in the pre-peptidyl transfer state, which is in part due to the lack of a simple and reliable procedure for the large-scale preparation of peptidyl-tRNAs.

Alternatively, peptidyl-tRNAs could be prepared using a combination of synthetic and biochemical techniques and subsequently added in trans to ribosomes in order to form the desired RNCCs²¹. This approach is based on the in vitro translation-independent preparation of the peptidyl-tRNAs, allows incorporation of unnatural amino acids, fluorescent or stable isotope labels for FRET and NMR studies, and, in general, provides more flexibility and control over what is being charged to the tRNAs. In order to be more suitable for structural studies, especially for X-ray crystallography or NMR, the peptide moiety of the peptidyl-tRNA must be amide-linked to the CCA-end of the tRNA to prevent spontaneous hydrolysis and deacylation during the time-course of experiments^18,20. Although such hydrolysis-resistant tRNAs do not represent natural substrates per se, they have not only been shown to be structurally indistinguishable from native tRNA substrates¹⁹ but are also active in transpeptidation when placed in the A site and combined with native aa-tRNA in the P site^20,22, and, therefore, represent a reasonable approximation of the reactive state, which is suitable enough to allow mechanistic hypotheses to be formulated.

Here we report a chemoenzymatic approach based on native chemical ligation (NCL) reaction for the facile synthesis of both stable amide-linked as well as labile ester-linked peptidyl-tRNAs carrying peptide chains of the desired sequences that can be used in a wide range of structural and/or biochemical studies of both arrested and non-arrested RNCCs. We also report several structures of non-arrested RNCCs in the pre-attack state of the peptidyl transferase center (PTC) at the highest resolution available to date. These structures reveal a previously unknown role of the ribosome in stabilization of the growing polypeptide chain within the PTC and suggest an extended entropic trap model that mechanistically rationalizes how ribosome acts with comparable efficiency upon a multitude of possible nascent chain sequences. Moreover, unlike previous views that ribosome generates the α-helical conformation at the C-terminus of the nascent peptide, a detailed analysis of our structures suggests that, regardless of the sequence, the C-terminal non-proline residues of the nascent peptide emerge in the uniform zigzag β-strand conformation that is stabilized by the intricate network of H-bonds provided by the universally conserved 23S rRNA nucleotides.

RESULTS

NCL-based synthesis of non-hydrolyzable peptidyl-tRNAs.

The synthesis of non-hydrolyzable peptidyl-tRNAs is challenging because there are no enzymes that can directly attach desired peptides to a deacylated tRNA. So far, only one method for the preparation of non-hydrolyzable peptidyl-tRNA was established involving customized solid-phase synthesis of the precursors²³ that can be barely performed in a standard molecular biology laboratory precluding its wide use by researchers. Moreover, such an approach requires laborious tRNA engineering with DNA enzymes as well as enzymatic ligation steps to include the native tRNA nucleoside modifications²³. Thus, the utility of the in trans-RNCC reconstitution approach is greatly limited by the lack of a facile and reliable procedure for the large-scale synthesis of stable peptidyl-tRNAs.

To bridge this gap, we have established a simple method for the semi-synthesis of non-hydrolyzable peptidyl-tRNAs using the NCL reaction, which was originally developed for linking unprotected peptide fragments with each other under mild reaction conditions²⁴. NCL is based on a reaction between an activated C-terminal thioester of the first peptide fragment and the N-terminal cysteine residue of the second peptide fragment resulting in the formation of a native peptide bond between the two fragments. In the method that we report here, we used aminoacyl-tRNA charged with cysteine as the C-terminal reactant, while as the N-terminal reactant, we used a peptide with the desired sequence and activated with the thiobenzyl (TBZ) group, which is now available commercially as a C-terminal modification through the majority of vendors providing peptide synthesis (Fig. 1). Previously, the Micura group utilized NCL for the synthesis of short peptidyl-tRNA mimics comprising only the CCA-ends²⁵. However, those early procedures relied heavily on laborious and advanced organic synthesis methods to prepare the non-standard solid supports, essentially making them inaccessible to a general molecular biology laboratory. Moreover, flexizymes have been used in the past to charge 3’-NH₂-tailed tRNAs with a single amino acid²⁶, which in principle should work for peptides as well. However, successful applications of the flexizyme-based methodology for charging amino-tailed tRNAs with a peptide have not been reported to date. The advantage of our chemoenzymatic approach is that it utilizes only commonly available equipment, affordable chemicals, and universally available commercial peptide synthesis services and, therefore, could be employed virtually by any laboratory.

Figure 1 ∣. Synthesis of peptidyl-tRNA by native chemical ligation.

The scheme shows mechanism of NCL reaction between thiobenzyl-activated peptide and tRNA charged with cysteine, which could be connected to the 3’-OH of the terminal tRNA adenine nucleotide either via native ester link or via non-hydrolyzable amide link.

The overall procedure includes three steps (Fig. 2a): (i) tRNA-tailing to replace the 3’-terminal regular adenosine-3’-OH of the CCA-end with its amino-substituted adenosine-3’-NH₂ analog^27,28; (ii) enzymatic charging of the tailed 3’-NH₂-tRNA with cysteine by the aminoacyl-tRNA-synthetase^27,28; and (iii) native chemical ligation of the TBZ-activated peptide with cysteinyl-tRNA to yield the final product (Fig. 1). It would have been logical to use cysteine-specific tRNA^Cys to prepare non-hydrolyzable cysteinyl-NH-tRNA needed for the NCL. However, in our pilot experiments, we have found that methionine-specific aminoacyl-tRNA-synthetase (MetRS) can efficiently mischarge the tailed initiator 3’-NH₂-tRNA_i^Met with cysteine resulting in >90% yields (Fig. 2b, lane 2 vs. 3), which likely happens due to the compromised editing activity of this ARSase as a result of its inability to hydrolyze non-hydrolyzable aa-tRNA²⁹. The main advantage of using the initiator tRNA_i^Met instead of tRNA^Cys as the P-site substrate for the subsequent ribosome structural studies stems from its high affinity for the ribosomal P site, ensuring the proper binding of the tRNA body to the ribosome. Moreover, unlike tRNA^Cys, the procedure for large-scale preparation of tRNA_i^Met is well established^20,27,28,30.

Figure 2 ∣. Synthesis of full-length non-hydrolyzable peptidyl-tRNAs by native chemical ligation.

(a) Steps in preparation of non-hydrolyzable peptidyl-tRNAs. The first tRNA tailing step was performed via the exchange of natural 3’-OH-AMP moiety of tRNA with the 3’-NH₂-AMP by using the CCA-adding enzyme²⁷. The body of the tRNA is shown in green. (b, c) Electrophoretic separation of native E. coli tRNA_i^Met (lane 1) after (i) tRNA tailing reaction (lanes 2, 5, 6); (ii) MetRS-driven aminoacylation with cysteine (lanes 3, 7, 8); (iii) native chemical ligation reaction (lanes 4, 9, 10). In panel (c), samples were treated (+) or not treated (−) with proteinase K. Electrophoresis of the tRNA samples was performed in 20-cm long 8% PAAG with 7 M urea and stained with ethidium bromide. Note that the slower mobility of tRNA after the NCL (lane 3 vs. 4) and its sensitivity to the proteinase K treatment (lane 9 vs. 10) confirms the presence of the peptidyl moiety attached to the tRNA_i^Met.

For the first NCL trials, we have selected several model peptides of different lengths and sequences that would also be interesting to study beyond the scope of this work (Extended Data Fig. 1a). We should note that some of the peptide sequences in our list represent known ribosome stalling motifs, which, however, cause translation arrest only in the presence of a particular drug (such as PTC-targeting antibiotic chloramphenicol in the case of hns, or macrolide antibiotics in the case of ermDL gene products), but otherwise are translated normally and should be considered as non-arresting peptides in the absence of the corresponding drug molecules, thereby representing merely model cases for our structural studies. Using the TBZ-activated peptides (purchased from NovoPro Biosciences, China) and the Cys-NH-tRNA_i^Met as the N- and C-terminal fragments for the NCL reaction, respectively, we show that most of the selected peptides can be efficiently ligated to the Cys-NH-tRNA_i^Met with yields exceeding 50% (Fig. 2b, lane 3 vs. 4; Extended Data Fig. 1b). We used treatment with proteinase K, which hydrolyzes the amide bond between the peptide moiety and the tRNA³¹, to confirm the presence of the peptidyl moiety attached to the tRNA_i^Met (Fig. 2c, odd vs. even lanes). Even by using the previously reported conditions for peptide ligation²⁴, the efficiency of NCL was already sufficiently high (>50%). Nevertheless, by searching for optimal reactant concentrations, the types and combinations of catalysts and denaturing agents used for the NCL (see Online Methods), we have achieved >70% reaction efficiency for several peptides. Thus, by using the NCL approach, we have synthesized a set of non-hydrolyzable full-length peptidyl-tRNAs carrying various peptide sequences at their CCA-ends, which were then purified by reverse-phase chromatography and used directly for ribosome complex formation and structure determination.

We have also checked whether cysteine-specific tRNA^Cys in combination with the cysteinyl-tRNA-synthetase could be used for our NCL-based technique. Using the optimized conditions for aminoacylation reaction (see Online Methods), we have prepared non-hydrolyzable pentapeptidyl-tRNA with formyl-Met-Ser-Glu-Ala-Cys peptide moiety, designated as fMSEAC-NH-tRNA^Cys (Extended Data Fig. 2a). Moreover, by skipping the tRNA-tailing step, we were able to prepare regular ester-linked fMSEAC-O-tRNA^Cys (Extended Data Fig. 2b), suggesting that this method is not limited to the preparation of only non-hydrolyzable amide-linked peptidyl-tRNAs.

In trans addition of the peptidyl-tRNAs results in the formation of functionally relevant pre-peptidyl transfer ribosome complexes.

Next, to check whether the obtained full-length peptidyl-tRNAs can be used for the structural studies of the non-arrested RNCCs, we prepared a complex of Thermus thermophilus 70S ribosome programmed with 24-nt-long mRNA and also containing cognate full-length P-site fMSEAC-NH-tRNA_i^Met as well as A-site Phe-NH-tRNA^Phe, crystallized it, and determined its structure at 2.4Å resolution (Fig. 3a, b). The observed high-quality electron density maps for both A- and P-site tRNA substrates (Fig. 3c) allowed unambiguous modeling of the four out of five amino acid residues in the fMSEAC-peptide attached to the CCA-end of the P-site tRNA (Fig. 3d-f). We observed no electron density for the N-terminal formyl-methionine residue of the fMSEAC peptide, which is likely due to the lack of coordination with the surrounding nucleotides of the 23S rRNA resulting in its high flexibility.

Figure 3 ∣. Structure of the 70S ribosome in complex with fMSEAC-peptidyl-tRNA.

(a, b) Structure of the T. thermophilus 70S ribosome containing full-length aa-tRNA (Phe-NH-tRNA^Phe) and peptidyl-tRNA (fMSEAC-NH-tRNA_i^Met) in the A and P sites, respectively, viewed from two different perspectives. The 30S subunit is shown in light yellow, the 50S subunit is in light blue, the mRNA is in teal, and the A-, P-, and E-site tRNAs are colored green, blue, and orange, respectively. The phenylalanyl and peptidyl moieties of the A- and P-site tRNAs are colored magenta and yellow, respectively. The direction of the view in panel b is indicated by the inset. (c) High-resolution (2.4Å) 2F_o-F_c electron difference Fourier map (blue mesh) of the ribosome-bound A- and P-site tRNAs. The refined models of tRNAs are displayed in their respective electron density maps contoured at 1.0σ. The entire bodies of the A- and P-site tRNAs are viewed from the back of the 50S subunit, as indicated by the inset. Ribosome subunits are omitted for clarity. (d-f) Close-up views (from different angles) of the CCA-end of the A-site tRNA carrying a phenylalanyl (magenta) moiety and the CCA-end of the P-site tRNA carrying an fMSEAC-peptidyl (yellow) moiety. Nitrogens are colored blue; oxygens are red. H-bonds are shown by black dotted lines. Note that the H-bond between the α-amino group and the 2’-OH of the A76 of the P-site tRNA is pivotal for optimal orientation of α-amine for an in-line nucleophilic attack onto the carbonyl carbon of the P-site substrate. Interactions of the MSEAC-peptide with the nucleotides of the 23S rRNA (light blue) are shown in (f).

Next, we aligned our structure containing fMSEAC-peptidyl-tRNA with the previously published structures of the 70S ribosome in the pre-attack state containing either the non-hydrolyzable amide-linked (Extended Data Fig. 3a)^20,28 or native ester-linked full-length aa-tRNAs in the A and P sites (Extended Data Fig. 3b)¹⁹. We found no major structural differences in the positions of the CCA moieties (Extended Data Fig. 3a, b) or most of the key 23S rRNA nucleotides around the PTC (Extended Data Fig. 3c, d), suggesting that this structure also represents a functionally relevant pre-attack state of the peptidyl-tRNA (Fig. 3e), even though the peptidyl-tRNA was added to the ribosome in trans. In particular, the orientation of the attacking α-amino group of the aa-tRNA relative to the carbonyl carbon of the P-site substrate is nearly identical between the structures harboring amide-linked vs. native ester-linked full-length aminoacyl- and peptidyl-tRNAs in the A and P sites, respectively (Fig. 3e; Extended Data Fig. 3a, b)²⁰, indicating that stable amide-linked aminoacyl- and peptidyl-tRNAs are adequate mimics of their native labile ester-linked counterparts. Our structural comparison reveals somewhat different conformations of highly flexible nucleotides A2062 and A2602 of the 23S rRNA (Extended Data Fig. 3d), which is likely due to their uncertain positions in the previous E. coli ribosome structure¹⁹ resulting from lower resolution and lack of cryo-EM density.

It is important to emphasize that using stable aminoacyl-/peptidyl-tRNAs allows trapping of the ribosome in the pre-peptidyl-transfer state because PTC is unable to cleave the amide bond connecting peptide to the tRNA so that peptidyl-tRNA is unable to donate its fMSEAC moiety to form a peptide bond with the A-site amino acid residue. This methodology of RNCC preparation is principally different from previous approaches used to produce stalled RNCCs, in which translation arrest happens not because substrates are chemically unreactive but because either the key nucleotides of the catalytic site are perturbed^9,32, or incoming aa-tRNA cannot efficiently accommodate into the A site¹¹, or the C-terminal part of the peptidyl-tRNA adopts non-productive conformation⁸. From this perspective, despite featuring unreactive substrate analogs in the PTC, our ribosome complexes represent non-arrested RNCCs containing full-length aminoacyl- and peptidyl-tRNAs, also providing the highest-resolution view of the PTC in the functional pre-attack state immediately before peptide bond formation.

An extensive H-bond network tightly coordinates growing peptide chains.

The structure of fMSEAC-peptidyl-tRNA in the ribosomal NPET reveals tight coordination of the PTC-proximal part of the nascent polypeptide by the elements of the ribosome. In particular, we observe four hydrogen bonds (H-bonds) between the nucleotides of the 23S rRNA and the three C-terminal amino acid residues of the nascent peptide, as well as an additional intramolecular H-bond between the C-terminal residues of the nascent peptide. The carbonyl and amide groups of the penultimate (−1) residue in the fMSEAC peptide (Ala4) are coordinated by the universally conserved 23S rRNA nucleotides G2061 and A2062, respectively (Figs. 3f; 4a, b, HB-1 and HB-2). At the same time, the N3 and O4 atoms of the nucleotide U2506 form two additional H-bonds with the carbonyl and amide groups of the (−2) residue of the fMSEAC peptide (Glu3), respectively (Figs. 3f; 4a, b, HB-3 and HB-4). Another interesting intramolecular H-bond is formed between the amide group of the ultimate residue (Cys) of the peptide and the carbonyl group of the (−2) amino acid (Glu3) (Figs. 3e; 4a, HB-5). Consistent with previous structural studies, the carbonyl group of the last amino acid residue of the peptide is coordinated by the nucleotide A2602 of the 23S rRNA via a water molecule (referred to as W2 in ref. ²⁰) that was previously proposed to play a key role in stabilizing the tetrahedral oxyanion intermediate of the transition state (Extended Data Fig. 4)^20,33. We also observe the two other tightly coordinated water molecules in the PTC (referred to as W1 and W3 in ref. ²⁰) that were suggested to take part in the formation of an intricate network of H-bonds, the proton wire, needed for efficient deprotonation of the attacking α-amine during the initial rate-limiting step and the subsequent breakdown of the transition-state intermediate (Extended Data Fig. 4)²⁰.

Figure 4 ∣. Tight coordination of the PTC-proximal part of the nascent peptide in the NPET.

(a, b) Close-up views of the interactions between the universally conserved nucleotides G2061, A2062, and U2506 of the 23S rRNA (light blue) and the main chain groups of the ribosome-bound fMSEAC-peptidyl-tRNA (yellow) in the ribosomal exit tunnel. Additional intramolecular H-bond is formed between the residues of the peptide. H-bonds are shown by black dotted lines and are annotated as HB-1 through 5. Nitrogen atoms are shown in blue, and oxygens are red. Note that all of these interactions involve only the main-chain atoms of the peptide, suggesting that, in principle, a peptide of any sequence (except for those containing prolines) should be coordinated the same way.

Ribosome catalyzes peptide bond formation by (i) stabilizing the transition-state intermediate and (ii) ensuring that the nucleophilic attack by the α-amino group of the aa-tRNA is coordinated with its efficient deprotonation during the rate-limiting step of the reaction^20,34,35. However, for the reaction to even start, the two reactants must be optimally positioned relative to each other in the PTC³⁶. The observed here tight coordination of the C-terminus of the growing peptide inside the NPET might not only (i) ensure the proper fixation of the carbonyl group of the peptidyl-tRNA in the PTC required for efficient attack by the nucleophile to occur but also (ii) prevent premature peptidyl-tRNA drop-off while the peptide remains relatively short^37,38. Although the contribution of the ribosome to the A- and P-site substrate stabilization was evident from the previous high-resolution structures^20,28,33, it was not known that ribosome stabilizes more than just the carbonyl group of the C-terminal residue of the growing peptide. While universally conserved and essential nucleotides A2451, A2602, and C2063 of the 23S rRNA have been suggested to participate in deprotonation of the attacking α-amino group²⁰, the individual roles of other important PTC nucleotides remained unclear. The observed here network of H-bonds formed by the universally conserved (G2061, A2062, U2506) and also functionally essential (G2061, U2506) nucleotides of the PTC^39-41 suggests their primary role in peptidyl-tRNA substrate stabilization. It is also evident from the structure that other nucleobases at positions 2061, 2062, and 2506 would be unable to form the same set of H-bonds rationalizing their evolutionary conservation.

Interestingly, unlike poorly active U2506 or G2061 mutants, substitutions of A2062 have only modest effects on GFP translation in vitro⁴¹, suggesting dispensability of A2062 for the translation of at least this particular protein (GFP). Furthermore, in vitro translation of the short ErmCL peptide or the ribosomal protein L12 tolerates even the total absence of the nucleobase at this position⁴². However, A2062 is critical for ribosome stalling on the same ErmCL leader sequence (and also ErmAL) in response to macrolide antibiotics as well as for natural arrest on SecM^42,43, pointing to the importance of this nucleotide for cellular functions. Due to its high flexibility, this nucleotide was suggested to be involved in the allosteric regulation of the PTC⁴³. These data altogether allow us to hypothesize that, by forming H-bond-2 (Fig. 4) with nascent peptides, nucleotide A2062 performs “fine-tuning” of the PTC function required either for efficient transpeptidation or efficient translation arrest of particular nascent chain sequences.

Nascent peptides emerge in the uniform β-strand conformation at the PTC.

In contrast to many published cryo-EM structures of various RNCCs, most of which represent arrested (and thus inactive) ribosome complexes, in which PTC is unable to catalyze transpeptidation, our structure provides the first snapshot of the PTC in the functional pre-attack state with both A- and P-site substrates bound. Interestingly, Ramachandran plot analysis reveals that the three C-terminal peptide residues of fMSEAC-peptidyl-tRNA appear in a β-strand conformation (Extended Data Fig. 5a).

Comparisons of our structure with the available highest-resolution cryo-EM structures of several arrest peptides revealed that the overall trajectories of the C-terminal segments of SpeFL (2.7Å)¹², ErmDL (2.9Å)¹¹, or VemP (2.9Å)¹³ peptides in the NPET are similar to the path of the fMSEAC peptide (Extended Data Fig. 6a-c)^11-13. The differences in the paths of the peptides become more prominent as we move further away from the C-terminus. We have also compared this structure of the full-length peptidyl-tRNA with the recent structures of pre-peptidyl transfer complexes featuring short tripeptidyl-tRNA analogs in the P site and ACC-Pmn in the A site (Extended Data Fig. 6d-f)⁴⁴. These peptidyl-tRNA mimics contained only the ACCA tetranucleotide corresponding to the 3’-terminus of a full-length tRNA that was attached to one of the three tripeptides via a non-hydrolyzable bond to yield ACCA-IAM, ACCA-ITM, or ACCA-IFM conjugates. Despite the missing tRNA body, these analogs exhibited not only the same exact placement of all the nucleotides but also highly similar peptide trajectories with nearly identical positions of main-chain and Cβ atoms in comparison to the structure with fMSEAC-peptidyl-tRNA (Extended Data Fig. 6d-f)⁴⁴. Because structures with the short tRNA analogs also represent non-arrested pre-peptidyl transfer ribosome complexes and also contain peptides with various sequences, it is tempting to suppose that any three C-terminal residues (except for prolines) of the growing peptide would always emerge in the same uniform β-strand-like conformation that is strictly enforced by the PTC via multiple H-bonds described above. In principle, there is enough space in the pockets next to the side chains of residues at position −2 (Glu in MSEAC) or position 0 (Cys in MSEAC) to accommodate any of the proteinogenic amino acids. However, the side chain of the residue at position −1 (Ala in MSEAC) points towards the confined space of the ribosomal A site raising the possibility that a large side chain at this position might not be able to fit snugly, resulting in sequence-specific alterations of the peptide path in the NPET and a different secondary structure.

To address this possibility and check whether or not bulky residue in the penultimate position of the nascent peptide can affect its trajectory in the NPET, we synthesized and purified peptidyl-tRNAs carrying formyl-Met-Arg-Cys (fMRC) and formyl-Met-Thr-His-Ser-Met-Arg-Cys (fMTHSMRC) peptides (Extended Data Fig. 1b, lanes 2 and 5) with the largest possible amino acid side chain (Arg) in the penultimate position using our NCL-based approach and determined their structures in complex with the ribosome. Except for the C-terminal Leu-to-Cys substitution, which does not affect activity⁴⁵, both of these sequences represent short and long versions of the macrolide-dependent arrest peptide ErmDL, respectively^11,32,45. Appearance of either of these peptide sequences in the NPET results in translation arrest and ribosome stalling, but only in the presence of a macrolide antibiotic and only if combined with lysyl- or arginyl-tRNA in the A site⁴⁵. Otherwise, these sequences are translated normally and do not cause translation arrest.

Crystals containing T. thermophilus 70S ribosome in complex with either the P-site fMRC-NH-tRNA_i^Met or fMTHSMRC-NH-tRNA_i^Met and the A-site Phe-NH-tRNA^Phe diffracted to 2.5Å and 2.3Å resolution, respectively (Table 1). The obtained experimental electron density maps for the corresponding peptidyl-tRNAs allowed unambiguous placement of all of their peptide residues (Fig. 5a-c), including the N-terminal residues for the longer fMTHSMRC-heptapeptide (Fig. 5d, e). Most importantly, both peptides appear to be coordinated by the same exact network of H-bonds (Fig. 5c) as the three C-terminal residues of the fMSEAC-peptide (Fig. 3f). Moreover, the side chains of penultimate arginines in both peptides establish electrostatic interactions with the phosphate of G2505, further anchoring these peptides in the NPET (Extended Data Fig. 7a, b). Surprisingly, but despite being longer than fMSEAC, for which only the four C-terminal residues could be resolved, all residues of the fMTHSMRC-heptapeptide are visible in the electron density map. Most likely, this is due to the prominent π-π stacking interaction of the histidine residue with the nucleobase of A2062 resulting in the overall higher rigidity of the entire ribosome-bound peptide (Extended Data Fig. 7c).

Table 1 ∣.

X-ray data collection and refinement statistics.

	70S ribosome complex with A-site Phe-NH-tRNAPhe, and P-site fMSEAC-NH-tRNAiMet PDB entry 8CVJ	70S ribosome complex with A-site Phe-NH-tRNAPhe, and P-site fMRC-NH-tRNAiMet PDB entry 8CVK	70S ribosome complex with A-site Phe-NH-tRNAPhe, and P-site fMTHSMRC-NH-tRNAiMet PDB entry 8CVL
Data collection
Space group	P212121	P212121	P212121
Cell dimensions
a, b, c (Å)	210.47 x 451.32 x 627.74	210.47 x 451.32 x 627.74	209.43 x 447.70 x 618.27
α, β, γ (°)	90.0, 90.0, 90.0	90.0, 90.0, 90.0	90.0, 90.0, 90.0
Resolution (Å)	175-2.40 (2.46-2.40)a	191-2.50 (2.56-2.50)a	187-2.30 (2.36-2.30)a
R merge	16.2 (139.5)	18.7 (150.7)	14.6 (141.0)
I / σI	8.96 (1.00)b	7.99 (0.93)c	6.54 (0.83)d
Completeness (%)	98.1 (95.5)	99.6 (98.6)	97.5 (94.0)
Redundancy	4.94 (4.80)	5.79 (4.88)	3.44 (3.40)
Refinement
Resolution (Å)	2.40	2.50	2.30
No. reflections	2,248,924	2,012,292	2,467,334
Rwork / Rfree	24.6/29.5	23.5/28.4	23.7/28.1
No. atoms
Protein	91,020	91,008	91,070
Ligand/ion	203,081	203,087	203,042
Water	4,076	4,118	3,983
B factors
Protein	51.8	51.0	55.9
Ligand/ion	50.0	49.0	53.6
Water	35.5	34.5	40.0
R.m.s. deviations
Bond lengths (Å)	0.004	0.004	0.004
Bond angles (°)	0.872	0.851	0.861

Values in parentheses are for the highest-resolution shell.

^aDiffraction data from a single crystal were used to obtain the structure.

^bI/σI = 2 at 2.60Å resolution.

^cI/σI = 2 at 2.75Å resolution.

^dI/σI = 2 at 2.50Å resolution.

Figure 5 ∣. Structure of the 70S ribosome in complex with fMRC- and fMTHSMRC-peptidyl-tRNAs.

(a-e) Close-up views of the 2F_o-F_c electron difference Fourier map (blue mesh) of the ribosome complexes containing P-site fMRC-NH-tRNA_i^Met (a-c, blue with peptide highlighted in orange) or fMTHMRC-NH-tRNA_i^Met (d-e, navy with peptide highlighted in crimson) and A-site Phe-NH-tRNA^Phe (green). Note that fMRC and fMTHSMRC peptides appear to be coordinated by the same network of H-bonds as the three C-terminal residues of the fMSEAC-peptide shown in Fig. 3f. (f) Superpositioning of the current three ribosome structures in complex with the full-length tRNAs carrying either fMSEAC (yellow), fMRC (orange), or fMTHSMRC (crimson) peptides in the P site. Note that the path of a peptide in the exit tunnel is not affected by the size of amino acid in the penultimate position.

Remarkably, the presence of bulky arginine side chain in the penultimate position of fMRC or fMTHSMRC peptides does not affect their paths in the NPET (Fig. 5f), supporting the hypothesis that, regardless of the sequence, the C-terminal non-proline residues of the nascent peptides emerge in the same uniform zigzag β-strand conformation (Extended Data Fig. 5b, c) that is stabilized by the intricate network of H-bonds provided by the universally conserved nucleotides G2061, A2062, and U2506 of the 23S rRNA. This experimental finding contradicts the previous view that ribosome, by default, generates α-helical conformation at the C-end of the nascent peptide, which was inspired by theoretical calculations⁴⁶. The observed fixation of the PTC-proximal part of the peptide in the NPET is likely to be important for the efficient catalytic function of the PTC and/or to prevent possible peptidyl-tRNA drop-off.

DISCUSSION

The main finding of this study is the intricate network of H-bonds that tightly coordinate the C-terminal part of the nascent peptide in the PTC of the ribosome. All of these interactions involve only the main-chain H-bond donor and acceptor atoms of the nascent polypeptide and, therefore, are sequence-independent, which is critical for the ribosome’s ability to translate virtually any sequences. While the existence of these H-bonds does not necessarily mean that they are required for efficient transpeptidation, there are several studies that speak in support of this hypothesis. For example, it has been demonstrated that the reactivity of a peptidyl-tRNA is modulated by the length of its nascent peptide chain: peptidyl-tRNAs with longer peptide chains react with puromycin (PMN) more rapidly than those with shorter ones^47-49. Based on this data, it has been hypothesized that the differences in the peptide bond formation rates with peptides of different lengths are due to tighter fixation of the longer nascent chains and, hence, better positioning of the reactive groups in the PTC⁴⁸. Analysis of our structural data rationalizes this hypothesis. Indeed, while formyl-methionine forms only a single H-bond out of five possible peptide-stabilizing H-bonds, dipeptides and tripeptides can form three and five H-bonds, respectively, being more tightly restrained in the transpeptidation-competent conformation. As evident from our structures, having more than three amino acid residues in a growing peptide chain does not provide any additional stabilization suggesting that the rate of transpeptidation should reach a plateau at ~3-aa peptide length, a prediction which is in excellent agreement with the previous biochemical data⁴⁹.

Interesting is the inability of proline residues to form some of these H-bonds that can rationalize the apparent redundancy of these interactions involving not just one but the three C-terminal residues of the growing peptide (positions 0, −1, and −2). Despite a number of previous structural and functional studies^7,50,51, there is no definitive answer to the fundamental question of why ribosome is unable to polymerize more than two consecutive prolines. In this case, ribosome stalling occurs due to the slow peptide bond formation between peptidyl-Pro-Pro-tRNA^Pro and Pro-tRNA^Pro located in the A site^7,50. Previous biochemical studies suggest that the poor reactivity of Pro-containing peptides stems from steric, rather than chemical, properties of this imino acid⁵². However, the mechanistic understanding of why peptidyl-Pro-Pro-tRNAs are especially inactive as P-site substrates is still lacking. In silico modeling reveals that appearance of a single proline residue at any of the three C-terminal positions of the growing peptide chain results in a loss of some but not all of the peptide-restraining H-bonds (Extended Data Fig. 8), which provide incomplete yet sufficient stabilization of the conformation of the nucleophile acceptor, the carbonyl group, for the transpeptidation reaction to occur. This finding is corroborated by the biochemical data showing a significantly slower (~700 fold) rate of peptide bond formation with the C-terminal proline of the peptidyl-tRNA⁴⁸. However, the presence of two consecutive prolines in a growing polypeptide chain results in the loss of at least four out of five H-bonds (Extended Data Fig. 9a) that most certainly negatively affects the efficiency of transpeptidation reaction because poorly stabilized diprolyl-tRNA substrate with wobbling carbonyl group is unlikely to be a good peptidyl donor during the nucleophilic attack. What is more striking is that, unlike any other residue, modeling of the proline in the penultimate position of the growing peptide shows a substantial steric clash with the nucleotide A2062 of the 23S rRNA (Extended Data Figs. 8b, 9a), which is locked in place through symmetric trans A-A Hoogsteen base pair with the residue A2503 suggesting that, during the passage of proline residues, the growing peptide must either deviate to the side or deflect A2062 to avoid this steric hindrance. Comparison of our model with the available structure of the ribosome-bound diprolyl-tRNA analog confirmed our prediction showing an alternate path of the Pro-Pro-containing peptide in the tunnel (Extended Data Fig. 9b)⁵³. Nevertheless, having two consecutive prolines in the peptide sequence does not result in a complete ribosome stalling⁷, most likely because the unstabilized wobbling peptidyl-tRNA could still randomly visit the productive conformation(s), and transpeptidation can still occur, albeit at a much slower rate⁷. However, ribosome stalls completely when the next incoming amino acid is yet another proline residue⁷. Most likely, this happens because the non-optimal position of the attacking A-site nucleophile is now combined with the poorly stabilized carbonyl group of the diprolyl-tRNA substrate in the P site. Thus, it becomes obvious that elongation factor P (EF-P), which is recruited to the ribosome every time two or more consecutive prolines need to be polymerized, provides stabilization of the otherwise wobbling diprolyl-tRNA substrate in the P site consistent with its previously proposed role in translation as an entropy-decreasing factor^7,50,51.

The efficient protein synthesis requires tight coordination of A- and P-site substrates during all peptide bond formation events³⁶, including the first one. Analysis of our structure also provides insight into the possible role of formylation of initiator tRNA in bacteria. Superpositioning of the ribosome structure containing formylated initiator fMet-tRNA_i^Met in the P site with the structure of fMSEAC-peptidyl-tRNA reveals nearly identical positions of the carbon and oxygen atoms in the formyl group and those in the carbonyl group of the penultimate residue in the growing peptide chain (Extended Data Fig. 10). Similar to the carbonyl group, the formyl group is coordinated by the same nucleotide G2061 of the 23S rRNA and, thus, provides additional stabilization to the entire methionine residue attached to the P-site tRNA (Extended Data Fig. 10). In other words, during the first round of elongation, when only a single amino acid (and not yet a peptide) is present in the P site, the formyl group mimics a dipeptide and provides added coordination to the P-site substrate that otherwise would be available only to a dipeptidyl-tRNA. Consistent with our observation, the inability of non-formylated methionine residue to form H-bond with the G2061 results in a 5-fold decrease in the transpeptidation reaction rate⁵⁴.

Lack of the machinery for formylation of the initiator tRNA in eukaryotes could have been compensated in the course of evolution by acquiring a dedicated protein factor, such as eukaryotic initiation factor 5A (eIF-5A), that provides added stability to the first methionine residue resulting in a more efficient transpeptidation. Although there is compelling evidence that both eukaryotic/archaeal eIF-5A and bacterial EF-P are needed primarily for the synthesis of polyproline stretches in proteins^7,55,56 and both carry functionally important hyper-modified lysine residues that encroach upon the P-site substrate in the PTC^53,57, we cannot exclude the possibility that eIF-5A also contributes to the first peptide bond formation as suggested by early reports^58-60, and it is only later that its general role in elongation and termination has also been revealed^55,56. These roles are not mutually exclusive, and it is conceivable that while diverging from a common ancestor, eIF-5A retained both of the original functions (facilitation of the first peptide bond formation and enhancement of transpeptidation as well as nascent chain hydrolysis of problematic substrates during both elongation and termination, respectively), whereas EF-P retained only the second function delegating the first one to the initiator tRNA formylation machinery.

Curiously, the number of observed H-bonds established by the ribosome with the C-terminal part of the growing peptide is in striking contrast to the lack of any visible H-bonds with the peptide’s N-terminal residues, suggesting that regardless of the length, only the C-terminal residues of the growing peptide need to be well-coordinated for efficient protein synthesis and/or to prevent a rare spontaneous dissociation of the peptidyl-tRNA from the translating ribosome known as drop-off⁶¹. Translation of particular sequences (especially in the presence of macrolide antibiotics) or polymerization of consecutive prolines as well as non-proteinogenic amino acids (such as β-, D-, or N-methyl- amino acids), all of which are poor substrates for transpeptidation, promotes frequent peptidyl-tRNA drop-off events^62,63. Recently, EF-P was shown to efficiently prevent peptidyl-tRNA drop-offs not only on polyproline sequences but also on other problematic P-site substrates, pointing to its more versatile function in bacteria than thought before^63-65. Although the molecular mechanism of this phenomenon is yet to be uncovered, it is tempting to suggest that peptidyl-tRNA drop-off occurs when the C-terminal residues of a nascent chain are unable to establish most of the five H-bonds observed in our non-stalled RNCCs (Fig. 4), such as in the case with C-terminal Pro-Pro (Extended Data Fig. 9a) or D-Ala-D-Ala sequences⁶³. We hypothesize that the presence of multiple prolines, D-amino acids, or other challenging nascent chain sequences results in the inability of the PTC to coordinate it well. This, in turn, leads to poor reactivity of the P-site substrate and, if a nascent chain is short, increases the chances of a peptidyl-tRNA drop-off.

While cysteine is one of the rarest amino acids in bacterial proteomes⁶⁶ and, therefore, is unlikely to appear in the ultimate (C-terminal) position, many of the ribosome stalling peptides, such as ErmDL⁴⁵, preserve their full functionality even after the replacement of the C-terminal residue with cysteine and can be studied using our approach. The apparent limitation of our procedure is a requirement for the ultimate amino acid of the peptide chain to always be cysteine, dictated by the chemistry of NCL reaction. However, the same limitation also exists with the widely used peptide/protein-only NCL, where multiple approaches have been utilized to overcome this limitation and extend the NCL reaction to other amino acids⁶⁷. For example, desulfurization of cysteine to alanine can be performed under mild conditions^68,69 that are unlikely to affect tRNA stability and, in principle, can be used to extend our methodology beyond Cys in the last position of the peptide.

CONCLUSIONS

In summary, here we present a simple, affordable, and fast NCL-based method for the synthesis of peptidyl-tRNAs carrying nascent peptide chains of the desired sequences (except for the C-terminal Cys) that can be used in a wide range of structural and/or biochemical studies, especially those that are focused at understanding the role of the nascent peptide sequence in the regulation of protein synthesis in response to various environmental cues, such as small molecules and antibiotics. By providing 2.3-2.5Å resolution X-ray crystal structures of the RNCCs featuring hydrolysis-resistant aminoacyl-tRNA in the A site and peptidyl-tRNAs in the P site, we demonstrate that synthetic full-length peptidyl-tRNAs can efficiently be introduced into the ribosome in trans and yet represent a functionally significant pre-attack state of the PTC. The availability of such peptidyl-tRNAs opens avenues for the structural studies of RNCCs under both stalling and non-stalling conditions in parallel, including rationalization of the context-specific mechanisms of action of many ribosome-targeting antibiotics such as those that target the PTC or the NPET. Moreover, the detailed analysis of our structural data suggests an answer to the question of why ribosome is unable to polymerize more than two consecutive proline residues without the need for a dedicated facilitator (such as EF-P or eIF-5A). Furthermore, here we provide important insights into the possible role of formylation of the initiator tRNA in the protein synthesis in bacteria. Finally, we propose a hypothesis that the C-terminal segments of all nascent peptides emerge in the same uniform β-strand conformation that is yet to be verified experimentally as more structures of non-arrested RNCCs will become available with the advent of this method for facile synthesis of full-length non-hydrolyzable peptidyl-tRNAs by native chemical ligation.

ONLINE METHODS

Preparation of full-length tRNA_i^Met and tRNA^Cys.

All reagents and chemicals were obtained from MilliporeSigma (USA). Wild-type deacylated initiator tRNA_i^Met was overexpressed and purified from E. coli as described previously³⁰.

Cysteine-specific tRNA^Cys was purified essentially as described previously for glycine-specific tRNA^Gly with modifications⁷⁰. In brief, tRNA^Cys (GCA anticodon) encoding sequence from E. coli was cloned into the pBSTNAV plasmid vector using EcoRI and PstI restriction sites resulting in pBSTNAV-Cys vector for expression of tRNA^Cys. For large-scale preparation of tRNA^Cys, tRNA-expressing E. coli cells were grown overnight in LB media with 100 μg/μL ampicillin. The cells were harvested and resuspended in buffer containing 1 mM Tris-HCl pH 7.4, 10 mM Mg (CH₃COO)₂, and treated with 0.5 volumes of acidic phenol pH 4.5. The aqueous phase was then precipitated by the addition of two volumes of ethanol and NaCl to 500 mM final concentration. The resulting pellet was resuspended in 1M NaCl, and the soluble fraction was precipitated with ethanol again. The pellet was resuspended in 200 mM Tris-HCl pH 9.0, and the resulting solution of total tRNA was incubated for 2 hours at 37°C to promote deacylation of any remaining aa-tRNAs. Deacylated total tRNA was precipitated with ethanol, resuspended in buffer A (40 mM sodium phosphate buffer pH 7.0), and subjected to anion exchange chromatography on an 8-ml MonoQ column (10/100 GL, GE Healthcare) using 300-ml 0-100% linear gradient of buffer B (40 mM sodium phosphate buffer pH 7.0, 1M NaCl) for elution. Fractions containing the desired tRNA^Cys were identified by hybridization with the complementary oligonucleotide AGACGGATTTGCAATCCGCTACATAACC resulting in a gel-mobility shift⁷⁰ as well as by aminoacylation with cysteine that also results in a characteristic shift in the electrophoretic mobility (Extended Data Fig. 2b, lane 1 vs. 2). All fractions containing tRNA^Cys were pooled together, precipitated with ethanol, resuspended in buffer C (20 mM NH₄CH₃COO pH 5.5, 400 mM NaCl, 10 mM MgCl₂, 1 mM EDTA, 1% methanol), and subjected to reversed-phase chromatography on 20-ml C5 column (C5-5, 250×10 mM, Discovery BIO Wide Pore, Supelco) using 300-mL 0-60% linear gradient of buffer C supplemented with 40% methanol. Fractions containing tRNA^Cys were identified, pooled together, and precipitated with ethanol as before. The final tRNA^Cys preparation was resuspended in 10 mM NH₄CH₃COO pH 5.5 and assessed for purity and ability to accept cysteine using denaturing PAGE and aminoacylation assay, respectively. The efficiency of aminoacylation of the final tRNA^Cys preparation was estimated to be greater than 90%.

Tailing of tRNA_i^Met and tRNA^Cys.

Tailing of tRNA_i^Met and tRNA^Cys (replacement of the 3’-terminal A76 nucleotide carrying 3’-OH group with the one carrying 3’-NH₂) was performed as described previously with minor modifications²⁷. Briefly, deacylated tRNA_i^Met or tRNA^Cys (40 μM) were incubated at 37°C for 1 hour in a buffer containing 100 mM glycine-NaOH pH 9.0, 1 mM DTT, 1 mM pyrophosphate, 1 mM 3’-NH₂-ATP (Axxora, USA), and 10 μM of the CCA-adding enzyme from E. coli. The reaction was terminated by the addition of EDTA to 20 mM, treated with 1:1 phenol-chloroform mixture (pH 8.0), and precipitated with ethanol. The resulting tRNA pellet was dissolved in 20-40 μL of 10 mM NH₄CH₃COO pH 5.5 and desalted via Sephadex G-25 (MilliporeSigma, USA) spin columns (20-40 μL of tRNA solution per 500 μL of G-25 media).

Aminoacylation of tailed 3’-NH₂-tRNA_i^Met and 3’-NH₂-tRNA^Cys and native tRNA^Cys.

The following conditions were used for aminoacylation of tRNAs:

• For aminoacylation of the tailed 3’-NH₂-tRNA_i^Met with cysteine, 40 μM tRNA was incubated at 25°C for 80 minutes in a buffer containing 100 mM HEPES-KOH pH 8.2, 20 mM MgCl₂, 7.5 mM KCl, 1 mM DTT, 10 mM ATP, and 1 mM cysteine together with 1 mg/ml methionine-specific aminoacyl-tRNA-synthetase (MetRS) from E. coli.

• For aminoacylation of the tailed 3’-NH₂-tRNA^Cys with cysteine, 40 μM tRNA was incubated at 37°C for 16 hours in a buffer containing 100 mM HEPES-KOH pH 7.5, 20 mM MgCl₂, 30 mM KCl, 10 mM DTT, 10 mM ATP, 10 mM ascorbic acid, 1 mM EDTA, and 10 mM cysteine together with 1 mg/ml cysteine-specific aminoacyl-tRNA-synthetase (CysRS) from E. coli.

• For aminoacylation of the untailed (native) 3’-OH-tRNA^Cys with cysteine, 40 μM tRNA was incubated at 37°C for 30 min in a buffer containing 40 mM HEPES-KOH pH 7.5, 12.5 mM MgCl₂, 10 mM KCl, 10 mM DTT, 2.5 mM ATP, and 2 mM cysteine together with 0.2 mg/ml CysRS.

Aminoacylation reactions were terminated by the addition of EDTA to 30 mM concentration and then treated with phenol and precipitated with ethanol. The pellet was dissolved in a buffer containing 5 mM NaCH₃COO, 1 mM DTT, 0.1mM EDTA to a final tRNA concentration of 20 μg/μL. Products were separated by acidic PAGE with 7M urea.

Preparation of peptidyl-tRNAs.

For native chemical ligation, dry thioester-activated peptides fMSEA-TBZ, fMR-TBZ, or fMTHSMR-TBZ (98% purity) carrying thio-benzyl group as the C-terminal modification (NovoPro Biosciences, China) were dissolved in a buffer containing 1M HEPES-KOH pH 7.4 and 6M guanidine-HCl to obtain 50 mM final concentration. Next, 5 μL of the peptide solution were mixed with 15 μL of 400 mM 4-mercaptophenylacetic acid (MPAA) titrated to pH 7.0 with NaOH, and 20 μL of either non-hydrolyzable amide-linked Cys-NH-tRNA_i^Met or Cys-NH-tRNA^Cys or native ester-linked Cys-O-tRNA^Cys. Also, TCEP pH 6.8 was added to the reaction mixture to 100 mM final concentration. The NCL reaction mixture was incubated for 16 hours at room temperature. The NCL products were purified by HPLC on a 1.7-mL C4 reversed-phase column (Proteo 300, 100x4.6 mm, Higgins Analytical) using 20-mL 0-60% linear gradient of 40% methanol in a buffer containing 20 mM NH₄CH₃COO pH 5.5, 400 mM NaCl, 10 mM MgCl₂, 1 mM EDTA, 1% methanol, 10 mM β-mercaptoethanol. Before applying onto the C4 column, the NCL mixture was subjected to a buffer-exchange procedure using Amicon Ultra 10K centrifugal filter units (MilliporeSigma, USA). The fractions from the C4 column that contained the desired peptidyl-tRNAs were pooled, ethanol-precipitated, and dissolved in a buffer containing 10 mM NH₄CH₃COO pH 5.5, 5 mM DTT to 200 μM final concentration. The aliquots were flash-frozen in liquid nitrogen and stored at −80°C until further use in crystallization experiments.

Proteinase K treatment and gel electrophoresis of peptidyl-tRNA.

For proteinase K treatment, 1 μM of the C4-purified fMSEAC-NH-tRNA_i^Met was incubated with 0.4 mg/mL proteinase K (MilliporeSigma, USA) in a buffer containing 10 mM Tris-Acetate pH 8.0, 10 mM Mg(CH₃COO)₂ and 40 mM NH₄CH₃COO for 1 hour at 25°C. The reaction was terminated by the addition of 2 volumes of formamide electrophoresis buffer containing 10% β-mercaptoethanol and heating for 2 minutes at 95°C. For separation of the NCL products, we used denaturing 8% (19:1) PAGE in the presence of 7 M urea (20-cm long, 0.4-mm thick gels). Gels were stained with ethidium bromide and visualized with Alpha Imager (Alpha Innotek, USA).

X-ray crystallographic structure determination.

Wild-type 70S ribosomes from Thermus thermophilus (strain HB8) were prepared as described previously^20,71-73. Synthetic mRNA with the sequence 5’-GGC-AAG-GAG-GUA-AAA-AUG-UUC-UAA-3’ containing Shine-Dalgarno sequence (underlined) followed by methionine (AUG) and phenylalanine (UUC) codons was obtained from Integrated DNA Technologies (USA). Non-hydrolyzable aminoacylated Phe-NH-tRNA^Phe was prepared as described previously²⁸ using optimized 3’-NH₂-tailing and aminoacylation procedures described above²⁷. Complexes of the wild-type Thermus thermolilus 70S ribosomes with mRNA and hydrolysis-resistant aminoacyl (A-site Phe-NH-tRNA^Phe) and peptidyl (P-site fMSEAC-NH-tRNA_i^Met, or fMRC-NH-tRNA_i^Met, or fMTHSMRC-NH-tRNA_i^Met) tRNAs were formed as described previously for deacylated⁷³ or aminoacylated tRNAs^20,28. Functionally-relevant binding of aminoacyl- and peptidyl-tRNAs to the A and P sites, respectively, was achieved without the use of auxiliary initiation or elongation factors. Collection and processing of the X-ray diffraction data, model building, and structure refinement were performed as described in our previous publications^{20,28,44,72,73}. The statistics of data collection and refinement are compiled in Table 1. All figures showing atomic models were rendered using PyMOL Molecular Graphics System software (version 1.8.6, Schrödinger, www.pymol.org). Extended Data Fig. 5 showing Ramachandran plots was rendered using UCSF ChimeraX software (version 1.3)⁷⁴.

Extended Data

Extended Data Fig. 1. Native chemical ligation of Cys-NH-tRNA_i^Met with TBZ-activated peptides of various lengths and sequences.

(a) General information about the model peptides chosen for this study: macrolide-sensing ErmDL stalling peptide^11,32; macrolide-resistance peptide MFLRC⁷⁵; chloramphenicol-sensitive protein H-NS⁷⁶; proline-rich antimicrobial peptide apidaecin⁷⁷. ERY, erythromycin or other macrolides; CHL, chloramphenicol; RF, release factor. Peptidyl-tRNA products used in the structural studies are highlighted in grey. (b) Electrophoretic separation of crude NCL reaction mixtures. Indicated TBZ-activated peptides and Cys-NH-tRNA_i^Met were used as N- and C-terminal reactants, respectively. Electrophoresis was performed in 20-cm long 8% PAAG with 7 M urea and stained with ethidium bromide. Note the slower mobility of tRNA after the NCL that is also proportional to the ligated peptide length (lanes 2-8 vs. 1).

Extended Data Fig. 2. Native chemical ligation of non-hydrolyzable or native Cys-tRNA^Cys with TBZ-activated fMSEA-peptide.

(a, b) Electrophoretic separations of deacylated, aminoacyl- and peptidyl-tRNA^Cys before and after NCL reactions. Amino-tailed 3’-NH₂-tRNA^Cys (a, lane 1) or native 3’-HO-tRNA^Cys (b, lane 1) were first aminoacylated using cysteine-specific aminoacyl-tRNA-synthetase (lanes 2) and then ligated with activated fMSEA-TBZ peptide for various times (lanes 3-5 in A and 3-7 in B) to yield either non-hydrolyzable fMSEAC-NH-tRNA^Cys (a) or native ester-linked fMSEAC-O-tRNA^Cys (b) peptidyl-tRNAs.

Extended Data Fig. 3. Comparison of the structures of fMSEAC-peptidyl-tRNA with aminoacylated full-length tRNAs.

(a, b) Superpositioning of our 70S ribosome structure carrying Phe-NH-tRNA^Phe (green) and fMSEAC-NH-tRNA_i^Met (blue) in the A and P sites, respectively, with the previously reported structures of ribosome-bound full-length aminoacyl-tRNAs featuring either non-hydrolyzable amide linkages (a, PDB entry 6XHW²⁸) or native ester bonds (b, PDB entry 6WDD¹⁹) between the amino acid moieties and the ribose of nucleotide A76 of A- and P-site tRNAs. All structures were aligned based on domain V of the 23S rRNA. (c, d) Comparisons of the positions of key 23S rRNA nucleotides around the PTC in the same structures. Note that there are no significant differences in the positions of A- or P-site substrates or the PTC nucleotides indicating that the amide-linked aminoacyl and peptidyl-tRNAs represent functionally meaningful analogs of native ester-linked tRNAs.

Extended Data Fig. 4. Tightly coordinated water molecules in the pre-attack state of the peptidyl transferase center.

(a-c) Close-up views of the 2F_o-F_c electron difference Fourier map (black mesh) for water molecules W1, W2, and W3 (yellow; nomenclature from ²⁰) in the pre-peptidyl-transfer complex structures featuring Phe-NH-tRNA^Phe in the A site (omitted for clarity) and fMSEAC-NH-tRNA_i^Met (a, yellow), fMRC-NH-tRNA_i^Met (b, orange), or fMTHSMRC-NH-tRNA_i^Met (c, crimson) in the P site. H-bonds are shown by black dotted lines.

Extended Data Fig. 5. Ramachandran plots for the peptide moieties of the ribosome-bound peptidyl-tRNAs.

(a-c) Diagrams of the phi vs. psi angles for amino acid residues in fMSEAC-NH-tRNA_i^Met (a, yellow), fMRC-NH-tRNA_i^Met (b, orange), or fMTHSMRC-NH-tRNA_i^Met (c, crimson) in the P site. Favored regions are shown in green; allowed regions are light green. Due to the absence of fMet1 residue in the structure of fMSEAC-peptidyl-tRNA, phi angle for the subsequent Ser2 residue cannot be determined.

Extended Data Fig. 6. Comparison of the structures of fMSEAC-peptidyl-tRNA with the structures of other ribosome-bound peptidyl-tRNAs.

Superpositioning of the 70S ribosome structure containing A-site Phe-NH-tRNA^Phe (green) and P-site fMSEAC-NH-tRNA_i^Met (blue) with the previously reported structures of stalled RNCCs carrying full-length peptidyl-tRNAs (a-c) or non-stalled RNCCs carrying short non-hydrolyzable tripeptidyl-tRNA analogs (d-f). Individual panels show comparisons of the fMSEAC-tripeptidyl-tRNA with the following peptides: (a) SpeFL (dark blue, PDB entry 6TC3¹²); (b) ErmDL (teal, PDB entry 7NSO¹¹), (c) VemP (cyan, PDB entry 5NWY¹³); (d) MAI-tripeptide (yellow, PDB entry 7RQB⁴⁴); (e) MTI-tripeptide (orange, PDB entry 7RQA⁴⁴), (f) MFI-tripeptide (red, PDB entry 7RQC⁴⁴). All structures were aligned based on domain V of the 23S rRNA. Note that the overall path of the MSEAC peptide in our structure is similar to the trajectories of the other peptides in the NPET.

Extended Data Fig. 7. Additional interactions of the side chains of fMRC- and fMTHSMRC-peptidyl-tRNAs with the ribosome.

(a, b) Close-up views of the electrostatic interactions between the side chain of the penultimate Arg residue of the P-site fMRC-NH-tRNA_i^Met (a, blue with peptide highlighted in orange) or fMTHMRC-NH-tRNA_i^Met (b, navy with peptide highlighted in crimson) and the phosphate of nucleotide G2505 of the 23S rRNA. H-bonds are shown by black dotted lines. (c) Stacking interactions between the aromatic side chain of His3 of fMTHMRC-peptidyl-tRNA and A2062 nucleobase of the 23S rRNA.

Extended Data Fig. 8. Proline residues in the nascent peptide are unable to form stabilizing H-bonds in the NPET.

(a-c) In silico modeling of proline residues at ultimate (a, Cys5Pro), penultimate (b, Ala4Pro), or pen-penultimate (c, Glu3Pro) positions of the fMSEAC peptide chain. Note that besides its inability to form most of the peptide-stabilizing H-bonds, proline in the penultimate and pen-penultimate positions clashes with nucleotides A2062 and U2506 of the 23S rRNA, respectively. Geometrically possible and impossible H-bonds are shown by black and white dotted lines, respectively.

Extended Data Fig. 9. Proline residues alter the path of the nascent peptide in the NPET.

(a) In silico modeling of the two consecutive proline residues at ultimate (Cys5Pro) and penultimate (Ala4Pro) positions of the fMSEAC peptide chain. Note that, due to the side chains, the diproline-containing peptide cannot adopt a conformation possible for other peptides in the NPET and must re-orient. (b) Comparison of the previous structure of ribosome-bound short diprolyl-tRNA analog (green, PDB entry 5DGV⁵³) with the in silico-modeled diprolyl-containing tRNA based on the structure of MSEAC-peptidyl-tRNA (red). Note that in order to avoid a steric clash with the A2062, the diprolyl moiety of the nascent peptide deviates to the side (black dashed arrows) and, thus, has an alternative path in the NPET.

Extended Data Fig. 10. Formylation of the first methionine residue provides additional stability to the initiator tRNA substrate in the P site.

Superpositioning of the previous structure of ribosome-bound initiator fMet-NH-tRNA_i^Met (navy with the fMet moiety in red, PDB entry 6XHW²⁸) with the new structure of fMSEAC-peptidyl-tRNA (blue with the peptide moiety highlighted in yellow) viewed from two opposite sides (a, b). H-bonds are shown by black dotted lines. Note that the positions of carbon and oxygen atoms in the formyl group and those in the carbonyl group of the penultimate residue in the nascent peptide chain are nearly identical, ensuring formation of the same H-bond with the exocyclic amino group of the G2061 residue.

DATA AVAILABILITY STATEMENT

Coordinates and structure factors were deposited in the RCSB Protein Data Bank with accession codes:

• 8CVJ for the T. thermophilus 70S ribosome in complex with mRNA, aminoacylated A-site Phe-NH-tRNA^Phe, peptidyl P-site fMSEAC-NH-tRNA_i^Met, and deacylated E-site tRNA^Phe;

• 8CVK for the T. thermophilus 70S ribosome in complex with mRNA, aminoacylated A-site Phe-NH-tRNA^Phe, peptidyl P-site fMRC-NH-tRNA_i^Met, and deacylated E-site tRNA^Phe;

• 8CVL for the T. thermophilus 70S ribosome in complex with mRNA, aminoacylated A-site Phe-NH-tRNA^Phe, peptidyl P-site fMTHSMRC-NH-tRNA_i^Met, and deacylated E-site tRNA^Phe.

All previously published structures that were used in this work for model building and structural comparisons were retrieved from the RCSB Protein Data Bank: PDB entries 6XHW, 6WDD, 6TC3, 5NWY, 5DGV.