Molecular insights into protein synthesis with proline residues

Abstract

Proline is an amino acid with a unique cyclic structure that facilitates the folding of many proteins, but also impedes the rate of peptide bond formation by the ribosome. As a ribosome substrate, proline reacts markedly slower when compared with other amino acids both as a donor and as an acceptor of the nascent peptide. Furthermore, synthesis of peptides with consecutive proline residues triggers ribosome stalling. Here, we report crystal structures of the eukaryotic ribosome bound to analogs of mono‐ and diprolyl‐tRNAs. These structures provide a high‐resolution insight into unique properties of proline as a ribosome substrate. They show that the cyclic structure of proline residue prevents proline positioning in the amino acid binding pocket and affects the nascent peptide chain position in the ribosomal peptide exit tunnel. These observations extend current knowledge of the protein synthesis mechanism. They also revise an old dogma that amino acids bind the ribosomal active site in a uniform way by showing that proline has a binding mode distinct from other amino acids.

Introduction

Among the ~20 amino acids that comprise all proteins in a living cell, proline stands out as the only residue in which the side chain is covalently attached to the α‐amine, forming a rigid cyclic structure. This proline structure facilitates folding of many proteins by introducing rigid turns into the peptide chain and by setting borders of β‐sheets and α‐helices 1, 2. Also, peptides with consecutive proline residues fold into a characteristic polyproline helix (P_II‐helix), which constitutes a common protein–protein interaction motif and also endows proteins with unique mechanical properties 3, 4.

However, numerous studies show that, although the unique chemical structure of proline facilitates protein folding, it also markedly impedes the rate of peptide bond formation by the ribosome. When proline is used for protein synthesis—either as a peptidyl donor in ribosomal P‐site 5, 6, 7 or as a peptidyl acceptor in ribosomal A‐site 7, 8, 9—it reacts markedly slower than other amino acids. Most interestingly, the inhibitory effect of proline on protein synthesis becomes progressively stronger when peptides with consecutive proline residues must be produced by the ribosome 10, 11, 12. Thus, synthesis of peptides with three or more consecutive prolines provokes ribosome stalling 10, 11, 12.

In a living cell, ribosome stalling by polyproline sequences is resolved by a universally conserved translation factor, known as eIF5A in eukaryotes and EF‐P in bacteria 10, 11, 12. In eukaryotes, eIF5A alleviates ribosome stalling by contacting the acceptor stem of the P‐site tRNA, using a mechanism that is not yet fully understood 13, 14. The presence of eIF5A in eukaryotic cells enables synthesis of proteins containing polyproline motifs. This factor is essential because polyproline motifs are highly abundant in eukaryotic proteomes. Human proteome, for example, contains ~10,000 motifs with three or more consecutive proline residues, with some proteins having up to 27 consecutive prolines 1, 2.

Extensive kinetic studies of peptide bond formation with proline suggested that proline impedes the rate of protein synthesis by increasing entropy of peptide bond formation 7. Furthermore, cryoelectron microscopy analysis of ribosome complexes with stalling peptides revealed the position of proline residues in the ribosomal P‐site during translational stalling 15, 16, 17. These studies profoundly extended our understanding of protein synthesis chemistry with proline. However, the conformation of proline and its reactive groups in the peptidyl transferase center is still unknown and, therefore, it is unclear how proline slows down the rate of protein synthesis.

Seeking to address this question, we determined crystal structures of the yeast 80S ribosome, bound with prolyl‐ and diprolyl‐tRNA analogs. These initial structures provide high‐resolution snapshots of proline positioning in the active centers of the ribosome, such as the ribosomal A‐site and the nascent peptide tunnel. Our data illustrate that the unique chemical structure of proline cycle has a dramatic impact on the proline position in the ribosomal active centers, preventing proline from adopting an optimal position required for rapid protein synthesis.

Results and Discussion

Synthesis of aminoacyl‐ and peptidyl‐tRNA analogs and ribosome structure determination

To approach structural studies of protein synthesis with proline residues, we chemically synthesized aminoacyl‐ and dipeptidyl‐tRNA analogs (Fig EV1). These analogs comprise the A₇₃C₇₄C₇₅A₇₆ tRNA acceptor stem with a 3′‐amido (instead of the natural ester) linkage between A₇₆ and the C‐terminus of the corresponding amino acid or dipeptide moiety to impede hydrolytic cleavage. Based on a novel 3′‐prolylamino‐3′‐deoxyadenosine solid support (Fig EV1), we produced the ribosome substrates ACCA‐Pro and ACCA‐Pro‐Pro. For reasons of comparison, we additionally produced the ribosome substrates ACC‐puromycin and ACCA‐Leu‐Phe using previously described strategies 18, 19.

Figure EV1. Chemical synthesis of hydrolysis‐resistant ACCA‐Pro and ACCA‐Pro‐Pro conjugates.

The strategy is exemplified by the synthetic scheme for the ACCA‐Pro‐Pro conjugate.

1. Synthetic scheme overview. Letters indicate conditions: a) 1.25 equivalent of Fmoc‐Pro, 1.25 equivalent of O‐(benzotriazol‐1‐yl)‐N,N,N',N'‐tetramethyluronium hexafluorophosphate, 1.25 equivalent of 1‐hydroxybenzotriazole hydrate, 1.5 equivalent of N,N‐diisopropylethylamine in DMF, room temperature, 12 h, 76% for compound 2; b) 3 equivalent of adipic acid bis(pentafluorophenyl)ester, 1 equivalent of 4‐(N,N‐dimethyl‐amino)pyridine in N,N‐dimethylformamide/pyridine (1/1, v/v), room temperature, 1 h, 54% for compound 3; c) ˜1 equivalent of amino‐functionalized polystyrene support (GE Healthcare, Custom Primer SupportTM 200 Amino), pyridine, N,N‐dimethylformamide, room temperature, 1 day, loading: 45 μmol/g for solid support 4. DMT, 4,4′‐dimethoxytrityl; Fmoc, N‐(9‐fluorenyl)methoxycarbonyl; SPS, solid‐phase synthesis; AE HPLC, anion‐exchange high‐pressure liquid chromatography.
2. Anion‐exchange HPLC profiles of crude (top) and purified (middle) ACCA‐Pro‐Pro conjugate and LC‐ESI mass spectra of purified products (bottom). Anion‐exchange chromatography conditions: Dionex DNAPac PA‐100 (4 × 250 mm) column; temperature: 60°C; flow rate: 1 ml/min; eluant A: 25 mM Tris–HCl (pH 8.0), 6 M urea; eluant B: 25 mM Tris–HCl (pH 8.0), 6 M urea, 500 mM NaClO₄; gradient: 0–40% B in A within 25 min; UV detection at 260 nm.

Our choice of amido variants of aminoacyl‐tRNA analogs was based not only on the fact that these analogs are hydrolysis resistant and therefore, compatible with crystallization conditions, but also because they were shown to functionally mimic natural ribosome substrates. In particular, it was previously shown that amido derivatives of aminoacyl‐tRNA mimics adopt the same conformation in the peptidyl transferase center as natural aminoacyl‐tRNAs 20.

Having synthesized the ribosome substrates, we determined the crystal structures of these substrates bound to the Saccharomyces cerevisiae 80S ribosome. The ribosome structures bound with the A‐site ACCA‐proline (I/σ = 1 at 3.3 Å) or with ACC‐puromycin (ACCm²A‐methyl tyrosine) (I/σ = 1 at 3.25 Å) (Table EV1, Fig 1A and B) were determined after rapid soaking of preformed ribosome crystals in a 100 μM substrate solution (Materials and Methods). Similarly, the ribosome structures bound with P‐site ACCA‐proline‐proline (I/σ = 1 at 3.1 Å) and ACCA‐leucine‐phenylalanine (I/σ = 1 at 3.5 Å) were determined after soaking of ribosome crystals in 100 μM substrate solution supplemented with 300 μM the antibiotic sparsomycin. Sparsomycin was used to stabilize binding of the peptidyl‐tRNA analogs in the ribosomal P‐site 20, 21, 22 (Table EV1, Fig 1C and D).

Figure 1. Structures and electron density maps of the ribosome‐bound aminoacyl‐ and dipeptidyl‐tRNA analogs.

The refined models of four crystal structures of the ribosome bound to aminoacyl‐ or peptidyl‐tRNA analogs are displayed in their respective unbiased electron density F_o‐F_c maps (contoured at 2.5 σ). The maps were calculated using phases produced by rigid body refinement of the ligand‐free test structure put into the F_o dataset.

• A, B
  A‐site‐bound substrates ACC‐puromycin (A) and ACCA‐Pro (B).
• C, D
  P‐site‐bound substrates ACCA‐Leu‐Phe (C) and ACCA‐Pro‐Pro (D).

Data information: The insets in each panel indicate position of a ligand relative to the A‐site (in yellow) and P‐site (in green) of the ribosome; the actual binding site of each ligand is highlighted in red.

Our attempts to simultaneously bind aminoacyl‐ and peptidyl‐tRNA analogs to the A‐site and P‐site, respectively, resulted in datasets in which only aminoacyl‐tRNA analogs were observed in the A‐site, whereas the P‐site remained vacant. These results were similar to the previous studies of the ribosomal complexes with partial tRNA analogs 20, 21, 22. This possibly reflects a relatively weak affinity of the CCA moiety to the ribosomal P‐site 20, 21, 22.

Collectively, we obtained four ribosome crystal structures in which either the A‐site or the P‐site was occupied by a model aminoacyl‐ or dipeptidyl‐tRNA analog (Table EV1).

Proline in the ribosomal A‐site

Ribosome structures with the aminoacyl‐tRNA analogs bound to the A‐site have a similar resolution of ~3.3 Å (Table EV1). The electron density maps reveal the CCA fragment bound with the A‐loop of 25S rRNA, as well as aminoacyl moieties bound with the amino acid binding pocket of the ribosome (Fig 1A and B).

Although the ribosomal P‐site remains vacant in these crystal structures, previous studies showed that the P‐site tRNA binding has a relatively minor effect on the aminoacyl moiety conformation in the ribosomal A‐site 20, 22. Therefore, we used these structures to gain insight into a possible position of proline residue in the ribosomal A‐site during protein synthesis.

In the ribosome structure with the ACC‐puromycin (ACCm²A‐methyl tyrosine), the puromycin moiety has a nearly identical conformation to those observed in prokaryotic ribosomes in the pre‐attack state of the peptide bond formation 21, 23 (Figs 2A and EV2). In this conformation, the α‐amine is directed toward the peptidyl‐tRNA, whereas the amino acid chain flips away from the P‐site and remains bound to the conserved hydrophobic cavity of the ribosome, the A‐site cleft (Figs 2A and EV2). While bound to the A‐site cleft, the amino acid side chain remains physically separated from the actual site of the peptide bond formation by the conserved nucleotide A2820 of the yeast 25S rRNA (A2451 of the E. coli 23S rRNA).

Figure 2. Proline has an atypical side chain position in the ribosomal A‐site.

• A, B
  This figure compares conformation of aminoacyl residues of the two A‐site substrates: ACC‐puromycin (A) and ACCA‐Pro (B). To show position of the A‐site substrates relative to the P‐site, the model P‐site substrate tRNA‐Phe is shown as in the pre‐attack complex of the ribosome (PDB ID 1vy4). In both illustrations, the A‐site is viewed as through the peptide exit tunnel. In addition to S. cerevisiae residues numbering, E. coli numbering is shown in parentheses. Comparison of the two structures shows that the side chain of the methyl tyrosine is bound to the conserved hydrophobic A‐site cleft, whereas the side chain of the proline residue penetrates the actual site of peptide bond formation where it may interfere with alignment of the reacting groups and proper positioning of the attacking electron pair of the α‐amine.

Figure EV2. Evolution of the amino acid binding pocket in the ribosomal A‐site.

• A–D
  When aminoacyl‐tRNAs bind the ribosomal A‐site, the amino acid side chain is accommodated by the A‐site cleft—a hydrophobic cavity formed by universally conserved nucleotides A2820 and C2821 in the 25S rRNA in S. cerevisiae (A2451 and C2452 in E. coli 23S rRNA). The views compare the A‐site cleft of prokaryotic (A, C) and eukaryotic (B, D) ribosomes. In the vacant eukaryotic ribosome (A), the cleft is smaller compared to those in prokaryotes (B), which likely results from a A‐to‐U mutation of the adjacent rRNA nucleotide (U2822 in S. cerevisiae 25S rRNA). In the archaeal ribosome, the corresponding residue is represented by A2488 in H. marismortiu 23S rRNA (PDB ID 3cc2). In bacteria, it is represented by A2452 in E. coli 23S rRNA (PDB ID 4y4o). However, when the aminoacyl‐tRNA or its analog puromycin occupies the A‐site (C, D), the cleft adopts similar conformations in pro‐ and eukaryotic ribosomes. Consequently, the aminoacyl moiety has a similar position in the active site of prokaryotic and eukaryotic ribosome. Therefore, the structural organization of the A‐site cleft is highly evolutionary conserved, despite rRNA sequence variations between prokaryotes and eukaryotes.

In the ribosome structure with the ACCA‐Pro, the proline moiety has a strikingly different and until now unknown conformation of an amino acid in the ribosomal A‐site compared to the typically uniform binding of other amino acids. Unlike methyl tyrosine and other amino acids observed in the A‐site 24, the proline side chain does not occupy the A‐site cleft, but instead flips toward the ribosomal P‐site (Fig 2B). In this conformation, the side chain of the proline residue penetrates the actual site of the peptide bond formation (Fig 2B). It is likely that in this position, the side chain may prevent proper alignment of the A‐site and P‐site substrates in the active site of the ribosome.

Additionally, compared to other amino acids, the α‐amino group of the proline residue is displaced by ~1 Å from the ribosomal P‐site (Fig EV3). In this conformation, proline may have an unusual orientation of the reactive electron pairs in the α‐amine group: The tetrahedral electron pair geometry and limited flexibility of the proline cycle suggest that proline's electron pair should deviate from the favorable position required for optimal nucleophilic attack (Fig 2B).

Figure EV3. Position of the aminoacyl‐tRNA analogs in the active site of the ribosome relative to the peptidyl‐tRNA.

The panels show structures of the aminoacyl‐tRNA analogs docked into the structure of the H. marismortui 50S ribosomal subunit bound to a peptidyl‐tRNA analog (PDB ID 1vqn). Transparent spheres highlight the reacting groups—nitrogen of the amino group in the A‐site substrate and carbon of the carbonyl group in the P‐site substrate.

1. Position of ACC‐puromycin relative to the P‐site substrate.
2. Position of ACCA‐Pro‐Pro relative to the P‐site substrate.

Thus, proline conformation in the ribosomal A‐site suggests that the poor reactivity of proline as a peptidyl acceptor originates from two factors: the unfavorable orientation of the α‐amine and its reactive electron pair as well as the binding of the proline side chain to the outside of the A‐site cleft of the ribosome. Both of these factors appear to prevent the optimal alignment of substrates in the peptidyl transferase center of the ribosome. Notably, despite proline's side chain not entering the A‐site cleft of the ribosome and adopting a highly unusual conformation, the α‐amine of proline remains accessible for the peptide bond formation and has a position in the peptidyl transferase center comparable to the ones observed for other amino acids, illustrating why proline remains reactive, although at an order of magnitude slower rate when it is compared to other proteinogenic amino acids 7, 8, 9.

Diprolyl peptide in the ribosomal nascent peptide exit tunnel

Ribosome structures with dipeptidyl‐tRNA analogs bound to the P‐site have a comparable resolution of 3.1–3.5 Å (Table EV1). The electron density maps reveal the CCA fragment bound to the P‐loop of 25S rRNA, as well as the dipeptidyl moieties at the entrance to the ribosomal nascent peptide exit tunnel (Fig 1C and D).

The antibiotic sparsomycin, which we used in this study to stabilize P‐site substrates on the ribosome, was described to deform the A76 sugar pocket and the carbonyl group in the P‐site peptidyl‐tRNA 20. Therefore, the crystal structures did not reveal the functional state of these critical groups in the P‐site ligands. However, sparsomycin does not affect the overall peptide chain conformation of the model dipeptides 20. Therefore, we used our structures to gain insights into how dipeptides Pro‐Pro and Leu‐Phe are positioned in the nascent peptide‐conducting tunnel.

In the ribosome structure with the ACCA‐Leu‐Phe ligand, the Leu‐Phe peptide enters the tunnel in a way similar to the short model peptides observed earlier 20, 22. The Phe residue appears to be partially disordered, reflecting conformational flexibility of this moiety in the peptide exit tunnel (Fig 3A).

Figure 3. Diprolyl peptide has a bent conformation in the nascent peptide exit tunnel of the ribosome.

• A, B
  This figure compares conformation of two dipeptides in the ribosome nascent peptide tunnel: Phe‐Leu peptide (A), and Pro‐Pro peptide (B). In both illustrations, the arrow points to the α‐amine of the N‐terminal residue. Comparison of the two structures reveals that the N‐terminus of the Phe‐Leu peptide is directed into the tunnel, whereas the N‐terminus of the Pro‐Pro peptide is directed toward the wall of the tunnel.

In the ribosome structure with ACCA‐Pro‐Pro ligand, the observed diprolyl peptide structure is bent: The N‐terminus of the diprolyl peptide is oriented toward the nascent tunnel wall instead of being directed directly into the tunnel (Fig 3B). It is important to note that this bent conformation reflects unique stereochemical constraints for the pyrrolidine ring of a proline residue (Fig EV4). Consistently with these constraints, the diprolyl peptide has parameters of the trans‐polyproline helix (P_II‐helix) (Fig EV4).

Figure EV4. The diprolyl peptide has a P_II‐helix geometry.

Unlike other polypeptides, polyproline peptides have a rigid fold in which proline residues lack freedom to rotate around the peptide bond. This is because cyclic structure of proline residues creates stereochemical constraints for polyproline peptides in which torsion angles ψ, φ, and θ should remain close to 150°, −75°, and 180° (for trans‐proline conformation), respectively. With these torsion angles, polyproline peptides inevitably fold into a characteristic secondary structure—P_II‐helix.

1. A Ramachandran plot, calculated for the diprolyl peptide, illustrates that the peptide has parameters of P_II‐helix. Red spot indicates torsion angles in the diprolyl peptide, labels show average values for the three most common elements of protein secondary structure—α‐helix, β structure, and P_II‐helix.
2. The diprolyl peptide structure shows positions and values of the major torsion angles. Semitransparent spheres represent van der Waals radii of the diprolyl peptide atoms.
3. Side‐by‐side structures of the diprolyl peptide, observed in this study, and an ideal P_II‐helix. Residues are numbered from the C‐ to N‐terminus.

The diprolyl peptide position in the ribosomal P‐site suggests that bending of a nascent peptide by consecutive proline residues may compromise the peptide passage along the tunnel, because proline‐proline bonds lack the rotatory freedom and do not allow the peptide chain to deviate from the polyproline helix fold (Fig EV5). If correct, this model may explain how residues, which are not located directly in the peptidyl transferase center (e.g., N‐terminal residues in the XPP stalling motifs), may dramatically influence the rate of peptide bond formation 25. Also, if this assumption is correct, then synthesis of polyproline peptides may originate in a way similar way to ribosome stalling by regulatory peptides 15, 16, 17, 26. Although future studies are required to understand this complex problem, our initial data indicate that ribosome stalling by polyproline‐containing peptides may originate in part from interactions between the nascent peptide and the tunnel.

Figure EV5. Structure of the diprolyl peptide in the peptide exit tunnel and a possible mechanism of ribosome stalling.

Views of the 80S ribosome/diprolyl‐tRNA analog structure.

1. Unbiased F_o‐F_c electron density map, contoured at 3 σ, reveals the diprolyl‐tRNA analog in the ribosomal P‐site. In the peptide exit tunnel, the diprolyl peptide has a conformation in which the N‐terminal residue (Pro −1) is directed toward the wall of the tunnel (residues G2403 and A2872). In this conformation, further attachment of proline residues to the nascent peptide would result in a steric clash between the nascent peptide and the ribosomal peptide exit tunnel.
2. To illustrate the steric clash, a hypothetical proline residue Pro −2 (in magenta) is attached to the N‐terminus of the diprolyl peptide. The resulting triprolyl peptide illustrates how a characteristic and rigid fold of polyproline peptides—P_II‐helix—may potentially create an obstacle for the nascent peptide passage through the ribosomal tunnel.

The observation that diprolyl peptide has parameters of the P_II‐helix may have important implications for studies of co‐translational protein folding. The diprolyl peptide structure suggests that P_II‐helices are formed simultaneously with the formation of peptide bonds between consecutive proline residues. Because P_II‐helix is the third most common secondary structure element after β‐sheets and α‐helices 27, it will be interesting to study whether P_II‐helices can indeed be formed in the ribosomal tunnel and how it passes along the tunnel during protein synthesis.

Perspectives

Beyond insights into protein synthesis with proline, our data revise old dogmas suggesting that amino acids bind the ribosomal active site in a uniform way 24. Proline's position in the ribosomal A‐site illustrates that side chain structure may have a tremendous impact on amino acid positioning in the ribosomal active site. Proline position also demonstrates that amino acids have more than a single mode by which to bind the peptidyl transferase center.

In summary, our study provides ribosome structures which describe proline binding within the functional centers of the ribosome: the ribosomal A‐site and the peptide tunnel. In the future, this study may impact not only the field of protein synthesis, but also the emerging field of synthetic biology in which synthetic amino acids are used for protein synthesis to expand the chemistry of living systems 28, 29, 30. The knowledge of how amino acid's structure affects its binding to the ribosome may enable rational design of unnatural ribosome substrates and remodeling of the ribosome active sites to enable rapid protein synthesis with unnatural substrates.

Materials and Methods

Synthesis of hydrolysis‐resistant aminoacyl‐ and peptidyl‐tRNA analogs

As the presence of STM1 protein in the P‐site of our yeast ribosome crystals blocks access of full‐length tRNAs 31, we produced partial aminoacyl‐tRNA and dipeptidyl‐tRNA analogs, using A₇₃C₇₄C₇₅A₇₆ RNA moiety and introduced a 3′‐amido linkage between A₇₆ and the C‐terminus of the diprolyl moiety to prevent hydrolysis of this analog during crystallization. While eukaryotic tRNA^Pro typically contains C rather than A at residue 73, this residue does not interact with the ribosome and, consistently, does not affect the tRNA position on the ribosome.

The ACC‐puromycin conjugate was produced as previously described 32. The ACCA‐Leu‐Phe conjugate was produced according to the following references 33, 34. The ACCA‐Pro‐Pro and ACCA‐Pro conjugates were produced as outlined in Fig EV1 and as described below:

Compound 2: N⁶‐[(Di‐n‐butylamino)methylene]‐3′‐[N‐(9‐fluorenyl)methoxycarbonyl‐L‐prolyl] amino‐3′‐deoxy‐5′‐O‐(4,4′‐dimethoxytrityl)‐D‐adenosine. Fmoc‐protected proline (84 mg, 0.25 mmol) was dissolved in 3 ml DMF followed by addition of O‐(benzotriazol‐1‐yl)‐N,N,N',N'‐tetramethyluronium hexafluorophosphate (HBTU, 94 mg, 0.25 mmol), 1‐hydroxybenzotriazole hydrate (HOBt, 38 mg, 0.25 mmol), and N,N‐diisopropylethylamine (DIPEA, 50 μl, 0.29 mmol). After 3 min of activation, 3′‐amino‐N⁶‐[(di‐n‐butylamino)methylene]‐3′‐deoxy‐5′‐O‐(4,4′‐dimethoxytrityl)‐D‐adenosine 1 35 (135 mg, 0.19 mmol, in 1 ml DMF) was added and the mixture was stirred overnight at room temperature. Then, the solvent was evaporated, the residue dissolved in CH₂Cl₂ and washed consecutively with half‐saturated aqueous NaHCO₃ solution, 5% citric acid solution, and saturated NaCl solution. The organic layer was dried (Na₂SO₄) and evaporated, and the crude product was purified via SiO₂ chromatography yielding 150 mg of compound 2 as white foam (76%). TLC (6% MeOH in CH₂Cl₂) R_f = 0.4. ¹H NMR (300 MHz, CDCl₃) δ 9.02 (s, 1H, HC = N(6)), 8.44 (s, 1H, HC(2)), 8.09 (s, 1H, HC(8)), 7.73 (m, 2H, HC(ar)), 7.54 (d, 1H, J = 7.3, HC(ar)), 7.33–7.16 (m, 14H, HC(ar)), 6.98 (s, 1H, HN(3′)), 6.75 (d, 4H, J = 7.2), 6.00 (s, 1H, HC(1′)), 4.83 (s, 1H, HC(2′)), 4.70 (m, 1H, HC(3′)), 4.37 (m, 3H, HC(4′) and OCH₂(Fmoc)), 4.22 (m, 2H, HC(9, Fmoc) and HC(α, Pro)), 3.75 (s, 6H, 2xOCH₃(DMT)), 3.71 (m, 2H, N(CH₂CH₂CH₂CH₃)₂), 3.57 (m, 2H, CH₂(Pro)), 3.43 (m, 4H, H₂C(5′) and N(CH₂CH₂CH₂CH₃)₂), 2.16–1.91 (m, 4H, 2xCH₂(Pro)), 1.66 (m, 4H, N(CH₂CH₂CH₂CH₃)₂), 1.35 (m, 4H, N(CH₂CH₂CH₂CH₃)₂), 0.94 (m, 6H, N(CH₂CH₂CH₂CH₃)₂). ¹³C NMR (75 MHz, CDCl₃) δ 158.6 (HC = N(6)), 152.7 (C(2)), 140.0 (C(8)), 130.2 (C(ar)), 128.3–125.2 (C(ar)), 120.1 (C(ar)), 113.2 (C(ar)), 91.2 (C(1′)), 83.2 (C(4′)), 74.7 (C(2′)), 68.0 (OCH₂(Fmoc)), 63.8 (C(5′)), 60.8 (C(α, Pro)), 55.3 (2xOCH₃), 53.5 (C(3′)), 52.0 (N(CH₂CH₂CH₂CH₃)₂), 47.3 (CH(Fmoc)), 46.5 (HC(γ, Pro)), 45.3 (N(CH₂CH₂CH₂CH₃)₂), 31.1 (N(CH₂CH₂CH₂CH₃)₂), 29.4 (CH₂, Pro), 25.0 (CH₂(Pro)), 20.3 and 19.9 (N(CH₂CH₂CH₂CH₃)₂), 14.0 and 13.8 (N(CH₂CH₂CH₂CH₃)₂). ESI‐MS (m/z): [M+H]⁺ calcd for C₆₀H₆₇N₈O₈, 1027.51; found 1,027.50.

Compound 3: N⁶‐[(Di‐n‐butylamino)methylene]‐3′‐[N‐(9‐fluorenyl)methoxycarbonyl‐L‐prolyl] amino‐3′‐deoxy‐5′‐O‐(4,4′‐dimethoxytrityl)‐2′‐O‐[1,6‐dioxo‐6‐(pentafluorophenyloxy)hexyl]‐D‐adenosine. To a solution of compound 2 (200 mg, 0.19 mmol) in DMF and pyridine (2.0 ml each) was added DMAP (24 mg, 0.19 mmol) and bis(pentafluorophenyl) adipate (298 mg, 0.61 mmol). The mixture was stirred for one hour followed by evaporation of the solvents. The crude product was purified via SiO₂ chromatography yielding 139 mg of compound 3 as white foam (54%). TLC (20% acetone in CH₂Cl₂) R_f = 0.4. ¹H NMR (300 MHz, CDCl₃) δ 8.98 (s, 1H, HC = N(6)), 8.47 (s, 1H, HC(2)), 8.01 (HC(8)), 7.74 (d, 2H, J = 7.4, HC(ar)), 7.55 (d, 2H, J = 7.2, HC(ar)), 7.40–7.14 (m, 15H, HN(3′) and HC(ar) and CDCl₃), 6.75 (d, 4H, J = 7.7, HC(ar)), 6.12 (d, 1H, J = 2.8, HC(1′)), 5.80 (m, 1H, HC(2′)), 5.14 (q, 1H, HC(3′)), 4.48 (b, 1H, HC(9, Fmoc)), 4.23 (m, 4H, HC(4′) and HC(α, Pro) and OCH₂(Fmoc)), 3.75 (s, 6H, 2xOCH₃(DMT)), 3.70 (m, 2H, N(CH₂CH₂CH₂CH₃)₂), 3.52–3.36 (m, 6H, H₂C(5′) and CH₂(Pro) and N(CH₂CH₂CH₂CH₃)₂), 2.51 and 2.36 (s, 2H, OOCH₂CH₂CH₂CH₂COO), 2.02–1.80 (m, 4H, 2xCH₂(Pro)), 1.62 (m, 8H, OOCH₂CH₂CH₂CH₂COO and N(CH₂CH₂CH₂CH₃)₂), 1.35 (m, 4H, N(CH₂CH₂CH₂CH₃)₂), 0.93 (m, 6H, N(CH₂CH₂CH₂CH₃)₂). ¹³C NMR (75 MHz, CDCl₃) δ 158.6 (HC = N(6)), 153.0 (C(2)), 141.4 (C(8)), 130.3 (C(ar)), 128.4‐125.0 (C(ar)), 120.2 (C(ar)), 113.3 (C(ar)), 86.8 (C(1′)), 82.7 (C(4′)), 75.2 (C(2′)), 68.0 (OCH₃(Fmoc) and CH(Fmoc)), 63.5 (C(5′)), 60.1 (C(α, Pro)), 55.3 (2xOCH₃)), 52.0 (N(CH₂CH₂CH₂CH₃)₂), 50.8 (N(CH₂CH₂CH₂CH₃)₂), 47.2 (HC(γ, Pro)), 45.3 (N(CH₂CH₂CH₂CH₃)₂), 33.2 (OOCH₂CH₂CH₂CH₂COO), 32.9 (OOCH₂CH₂CH₂CH₂COO), 31.5 (N(CH₂CH₂CH₂CH₃)₂), 31.1 (N(CH₂CH₂CH₂CH₃)₂), 27.4 (H₂C(Pro)), 24.9 (H₂C(Pro)), 20.3 (N(CH₂CH₂CH₂CH₃)₂), 19.9 (N(CH₂CH₂CH₂CH₃)₂), 14.0 (N(CH₂CH₂CH₂CH₃)₂), 13.8 (N(CH₂CH₂CH₂CH₃)₂). ESI‐MS (m/z): [M+H]⁺ calcd for C₇₂H₇₄F₅N₈O₁₁, 1,321.54; found 1,321.41.

Compound 4: DMTO‐rA^3′‐NH‐Pro‐NHFmoc solid support. Compound 3 (130 mg, 0.09 mmol) was dissolved in dry DMF (5.0 ml), and pyridine (20 μl) was added. To this solution, amino‐functionalized support (GE Healthcare, Custom Primer Support™ 200 Amino, 500 mg) was added, and the suspension was agitated for 20 h at room temperature. Subsequently, the beads were collected on a Büchner funnel and washed with DMF, methanol, and CH₂Cl₂. For capping of unreacted amino groups, the beads were treated with a mixture of solution A (0.2 M phenoxy acetic anhydride in THF, 10 ml) and solution B (0.2 M N‐methyl imidazole, 0.2 M sym‐collidine in THF, 10 ml) and agitated for 10 min at room temperature. The suspension was filtrated again, the beads were washed with THF, methanol, and CH₂Cl₂ and dried under vacuum. Loading of the support was 45 μmol/g.

Solid‐phase peptide synthesis on the solid support 4

In a frit‐fitted syringe, the solid support 4 (80 mg) was soaked with dry DMF (2 ml, 30 min). For deprotection of the Nα‐Fmoc group, the solid support was treated two times with piperidine solution (20% in DMF, 1.5 ml, 8, 12 min) and subsequently washed with DMF (3 × 2 ml). Coupling was performed by treating the solid support for one hour with a mixture of Fmoc‐proline solution (0.4 M in DMF, 500 μl), activator solution (HBTU and HOBt, 0.6 M in DMF, 750 μl), and DIPEA (140 μl). This step was performed twice. Then, the solid support was washed with DMF (3 × 2 ml) and acetonitrile (3 × 2 ml) and vacuum‐dried.

RNA synthesis

The ACCA moiety was assembled on an ABI 392 Nucleic Acid Synthesizer following standard synthesis protocols. Detritylation (120 s): dichloroacetic acid/1,2‐dichloroethane (4/96); coupling (120 s): phosphoramidites (0.1 M in acetonitrile, 130 ml) were activated with benzylthiotetrazole (0.3 M in acetonitrile, 180 μl); capping (2 × 10 s, Cap A/Cap B = 1/1): Cap A: phenoxyacetic anhydride (0.2 M in THF), Cap B: N‐methyl imidazole (0.2 M), sym‐collidine (0.2 M) in THF; oxidation (20 s): I₂ (0.2 M) in THF/pyridine/H₂O (35/10/5). Amidites, benzylthiotetrazole, and capping solutions were dried over activated molecular sieves (4 Å) overnight.

Final product

Deprotection of the 3′‐diprolyl‐ACCA conjugate. A) Fmoc deprotection. In the ABI synthesis column, the solid support was treated with a solution of 20% piperidine in acetonitrile (10 ml, 10 min), washed with acetonitrile, and dried. B) Acyl deprotection and cleavage from the solid support. For the conjugates synthesized on solid support 4, the beads were transferred into an Eppendorf tube and equal volumes of methylamine in ethanol (8 M, 0.5 ml) and methylamine in H₂O (40%, 0.5 ml) were added. After 6 h shaking at room temperature, the supernatant was filtered and evaporated to dryness. C) 2′‐O‐TOM deprotection. The obtained residue was treated with TBAF·3 H₂O in THF (1 M, 1 ml) overnight at room temperature. The reaction was quenched by the addition of triethylammonium acetate (TEAA) (1 M, pH 7.4, 1 ml). After reducing the volume of the solution, it was applied on a size‐exclusion chromatography column (GE Healthcare, HiPre 26/10 desalting, 2.6 × 10 cm, Sephadex G25). By eluating with H₂O, the conjugate‐containing fractions were collected and evaporated to dryness, and the residue was dissolved in H₂O (1 ml). Analysis of the crude products was performed by anion‐exchange chromatography on a Dionex DNAPac PA‐100 column (4 × 250 mm) at 60°C. Flow rate: 1 ml/min; eluent A: 25 mm Tris–HCl (pH 8.0), 6 M urea; eluent B: 25 mM Tris–HCl (pH 8.0), 0.5 M NaClO₄, 6 M urea; gradient: 0–60% B in A within 45 min or 0–40% B in A within 30 min, UV detection at λ = 260 nm.

Purification of the 3′‐diprolyl‐ACCA conjugate

The crude conjugate was purified on a semipreparative Dionex DNAPac PA‐100 column (9 × 250 mm) at 60°C with flow rate of 2 ml/min (for eluents, see above). Fractions containing the conjugate were loaded on a C18 SepPak Plus cartridge (Waters/Millipore), washed with 0.1–0.15 M (Et₃NH)⁺ ${\text{HCO}}_3^{-}$, H₂O, and eluted with H₂O/CH₃CN (1:1). Conjugate‐containing fractions were evaporated to dryness and dissolved in H₂O (1 ml). The quality of the purified conjugate was analyzed by analytical anion‐exchange chromatography (for conditions, see above). The molecular weight of the synthesized conjugate was confirmed by LC‐ESI mass spectrometry (see Fig EV1B). Yields were determined by UV photometrical analysis of conjugate solutions. The final compound was dissolved in water to achieve ~50 mM concentration for stock solutions and later used for soaking.

Ribosome purification, crystallization, crystal treatments

80S ribosomes from the yeast S. cerevisiae were purified, crystallized, and treated essentially as previously described 14, 31. Ribosome complexes, containing ACCA analogs, were formed by soaking with 100 μM of the corresponding compound and 300 μM of sparsomycin for ~2 h at 4°C in a buffer containing 80 mM Tris‐acetate pH 7.0, 70 mM KSCN, 10 mM Mg(OAc)₂, 20% v/v glycerol, 15% w/v PEG 20,000, 6.5 mM spermidine, 7.5 mM NH₄OAc, 1.4 mM deoxy‐big‐chap, 2 mM DTT before the transfer into a cryoprotecting buffer containing 80 mM Tris‐acetate pH 7.0, 70 mM KSCN, 10 mM Mg(OAc)₂, 18% v/v glycerol, 20% w/v PEG 20,000, 6.5 mM spermidine, 7.5 mM NH₄OAc, 20% w/v PEG 6,000, 2 mM Os(NH₂)₆Cl₃.

Data collection and processing

Diffraction data were collected from crystals cooled to 90°K using 0.1^o oscillation range and the beam‐line Proxima 1 at the Synchrotron Soleil (France). We used a data collection strategy developed at Swiss Light Source Synchrotron (Switzerland), which exploits the unique features of the single photon counting pixel detector PILATUS 6M 31, 36, 37. During data collection, the beam was attenuated to 3–10% of its maximum flux. This was done to markedly reduce radiation damage and collect a highly redundant dataset from multiple spots of each crystal. Then, data were processed and reduced by the XDS suite 38, yielding the statistics displayed in Table EV1.

Because of highly attenuated beam, this data collection strategy results in unconventionally high R_meas values when it is compared to CCD‐detector type data collection strategies (no beam attenuation, no fine ϕ‐slicing, low data redundancy). However, it provides more accurate values for reflections' intensities and anomalous signal when single photon counting pixel detectors are used for data collection 37, 39.

Structure determination, refinement, validation, and analyses

The structures were determined by molecular replacement using the vacant yeast 80S ribosome structure (pdb 4v88) as a search model and then subjected to refinement using Phenix.refine 40. Restraints for ACCA analogs and sparsomycin were generated with JLigand 41 and ReadySet from the Phenix suite 40.

Ligands building, fitting, remodeling of ribosomal binding sites and analysis of Ramachandran plots were performed using Coot 42. Ribosome structure and ligand geometry (torsion angles, bond lengths, and bond angles) were refined using Phenix.refine 40. Crystallographic statistics are reported in Table EV1. Ligands geometry was validated with the software Mogul from the CCDC package 43. Compared to the original model of S. cerevisiae ribosome (pdb 4v88), conformation of several rRNA nucleotides was corrected in both monomers for 25S rRNA residues U766, A806, U922, G1152, U2211, C2278, G2283, C2726, A2401, A2872 and A2971 and 16S rRNA residues G163, G337, A542, and A811 in 18S rRNA, and metal ions were modeled de novo. The Ramachandran plot was calculated using Molprobity 44. Figures were prepared using PyMOL (Schrodinger LLC).

Author contributions

SM, MY, TED, and RM devised experiments; LR and SN produced tRNA conjugates, JM crystallized ribosomal complexes and performed the soaking; JM and SM performed crystallographic data collection and processing, SM and JM analyzed crystallographic data; SM, JM, LR, SN, B‐SS, GY, TD, RN and MY wrote and commented on the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Review Process File

EMBO Reports (2016) 17: 1776–1784