Communicated by Harry F. Noller, University of California, Santa Cruz, CA, August 29, 2006 (received for review July 24, 2006)
Protein synthesis requires the accurate positioning of mRNA and tRNA in the peptidyl-tRNA site of the ribosome. Here we describe x-ray crystal structures of the intact bacterial ribosome from Escherichia coli in a complex with mRNA and the anticodon stem-loop of P-site tRNA. At 3.5-Å resolution, these structures reveal rearrangements in the intact ribosome that clamp P-site tRNA and mRNA on the small ribosomal subunit. Binding of the anticodon stem-loop of P-site tRNA to the ribosome is sufficient to lock the head of the small ribosomal subunit in a single conformation, thereby preventing movement of mRNA and tRNA before mRNA decoding.
Protein synthesis, or translation, requires the ribosome to hold tRNAs in one reading frame of the mRNA. However, after each peptide bond is made, the ribosome must rearrange its contacts with mRNA and tRNA to allow translocation along the mRNA by a single 3-nt codon, with a concomitant movement of the cognate tRNAs. The ribosome controls the positioning of mRNA and tRNAs during translation through a number of direct intermolecular contacts (1). However, the molecular mechanisms by which the ribosome balances maintenance of the mRNA reading frame with rapid translation remain unclear (2).
Of the three tRNA binding sites on the ribosome [aminoacyl (A), peptidyl (P), and exit (E)], tRNA binds most tightly to the peptidyl-tRNA site (P site) (3), where the reading frame on the mRNA is maintained (4). Notably, the anticodon stem-loop (ASL) of tRNA, which binds in a cleft in the small (30S) ribosomal subunit, contributes a significant amount to the free energy of binding to the P site (5) in an mRNA-dependent manner (6, 7). The interactions among mRNA, tRNA, and the ribosome in the P site are only approximately known, having been determined at ≈5-Å resolution from x-ray crystal structures of the 70S ribosome (8). Structures of the 30S subunit at atomic resolution contain mimics of mRNA and the ASL of P-site tRNA that rely on crystal contacts between neighboring molecules (1, 9). However, the mRNA mimic does not form Watson–Crick base pairs to the ASL analog, and the ASL lacks a canonical 7-nt anticodon loop structure. We have now determined two structures of the intact 70S ribosome from Escherichia coli in a complex with mRNA and the ASL of initiator tRNAfMet (Fig. 1 A) at a resolution of 3.5 Å (Fig. 1 B and Table 1). These structures provide a more detailed view of how the ASL of P-site tRNA interacts with the intact ribosome.
Overall structure of intact E. coli 70S ribosome in a complex with mRNA and P-site ASL. (A) Secondary structures of the ASL and mRNA–ASL codon–anticodon base pairing. Numbering of mRNA and ASL nucleotides is indicated. (B) Anisotropic diffraction from crystals of the 70S ribosome complexed with a P-site ASL and mRNA. Completeness of the data as a function of resolution is plotted on the right axis, and the mean signal-to-noise of diffraction intensities, or I/σ(I), is plotted on the left axis. The dashed line represents a mean I/σ(I) of 2. (C) Conformational rearrangement of the 70S structure upon binding of mRNA and P-site ASL. Rearrangements of the 30S head positions in the apo-70S ribosome structures are indicated by difference vectors between phosphorous atoms (light blue) and Cα atoms (dark blue). In the ribosome complex, the ASL and mRNA are shown as space-filling models, and the 5′ to 3′ direction of mRNA is also indicated. Letters indicate the approximate alignments of the aminoacyl (A), peptidyl (P), and exit (E) tRNA binding sites at the subunit interface. Domains of the ribosome are labeled for the 30S head (Head), and 50S central protuberance (CP), as are ribosomal proteins in the small (S) and large (L) subunits.
X-ray crystallographic statistics and refinement
In the present 70S ribosome structures, both ribosomes in the crystallographic asymmetric unit bind mRNA and the ASL to the P site of the 30S subunit in an identical manner. We therefore averaged the electron density maps of both ribosomes to improve the effective resolution of the maps to ≈3.5 Å (see Materials and Methods). Given the nearly identical conformations of large portions of the two ribosomes, we describe the interactions in one of the ribosomes as representative of both. The crystals used to determine the present structures were grown in conditions highly similar to those reported previously (10). Therefore, we were able to compare the present structures to the apo-70S ribosome structures to determine the effects of mRNA and ASL binding to the ribosomal P site. Surprisingly, binding of mRNA and the P-site ASL completely reverses the swiveling of head of the 30S subunit observed in the previous structures in the absence of ligands (Fig. 1 C) (10). In contrast, there is no appreciable change in the global conformation of the large subunit that can be attributed to ASL binding. When compared with lower-resolution structures of the 70S ribosome, the conformation of the ribosome in the present structures adopts a pretranslocation conformation (2, 8). These structural results indicate that in the intact ribosome the position of the head of the small subunit is inherently dynamic in the absence of bound ligands and is free to rotate about the neck helix in 16S rRNA (10). The head of the small subunit is only stabilized in a single conformation upon binding mRNA and the ASL portion of P-site tRNA. In agreement with this idea, the neck helix in human ribosomes has also been observed to be dynamic in the absence of bound ligands (11).
In the 70S structures, the mRNA is held in place entirely by interactions with its phosphoribose backbone. This mode of binding contrasts with the ribosomal aminoacyl-tRNA site, where extensive minor groove interactions stabilize the codon–anticodon helix (9). The mRNA is threaded through the top of the major groove of helix 44 within 16S rRNA, in nearly the same conformation as was observed in the 30S subunit structures (9). The backbones of all three nucleotides in the P-site codon are coordinated by hydrogen bonds to universally conserved nucleotides in 16S rRNA (12). The phosphate of nucleotide +1 is within hydrogen-bonding distance of the exocyclic amine (N2) of G926 in 16S rRNA (Fig. 2 A), in agreement with protection of this nucleotide from chemical probes when P-site tRNA is bound (13). Nucleotide m3U1498 in 16S rRNA makes a number of contacts to the backbone of nucleotides +1 and +2. The phosphate of mRNA nucleotide +2 is within hydrogen-bonding distance of the ribose 2′-OH of m3U1498 in 16S rRNA, whereas the 2′-OH of nucleotide +1 hydrogen-bonds to the phosphate of m3U1498. Furthermore, the ribose of mRNA nucleotide +2 is stacked on the base of m3U1498 (Fig. 2 A). Two cytidines, m4Cm1402 and C1403, hydrogen-bond via their N4 amines to the nonbridging phosphate oxygens of mRNA nucleotide +3 (Fig. 2 A).
Interactions between the ribosome and mRNA in the P site. (A) Hydrogen bonds to the phosphates of nucleotides +1 and +3 of mRNA shown from the perspective of the 30S head. The position of G1401 and a fully hydrated Mg2+ have been removed for clarity. (B) Coordination of a fully hydrated Mg2+ to 16S rRNA and the backbone of mRNA, shown from the perspective of the subunit body. (C) View of (F obs − F calc) difference electron density in the electronegative pocket between the backbone of P-site mRNA and helix-44 nucleotides 1494–1498 in 16S rRNA. The position of A1493 is already adjusted to fit the electron density. (D) Nucleotides +4 through +6 of the mRNA, along with two Mg2+ ions, modeled into the electron density in C followed by refinement.
In addition to the above rRNA contacts, a single metal ion is coordinated to the major groove of the mRNA–ASL codon–anticodon helix (9). This specifically bound ion was not present in the apo-70S ribosome structures (10). The position of the density is most consistent with a fully hydrated Mg2+ ion, with the Mg2+ positioned ≈3.5 Å from the N7 and O6 of universally conserved nucleotide G1401 (12), which has been shown to be critical for mRNA binding (9, 14). Notably, all six of the modeled water ions coordinating the Mg2+ are within hydrogen-bonding distance of the major groove face and phosphate of G1401 in 16S rRNA and the phosphate oxygens of nucleotides +2 and +3 in the mRNA (Fig. 2 B).
Nucleotide U1498 in16S rRNA, which is universally conserved but does not form a base pair (3-methyl-uridine, m3U, in E. coli), packs against the phosphate and ribose of mRNA nucleotide +2, as described above (Fig. 2 A). The base of m3U1498 also stacks on another universally conserved nucleotide, G1497, and creates an electronegative pocket at the top of helix 44 immediately adjacent to the phosphate of mRNA nucleotide +3 and the phosphates of 16S nucleotides 1494–1496. Intriguingly, this pocket is occupied by electron density that runs parallel to the entire P-site codon, with one end stacking on base A790 in 16S rRNA and the other approaching the position where mRNA nucleotide +4 would be positioned (Fig. 2 C). Structural refinement revealed that the electron density corresponds to nucleotides +4 through +6 in the short mRNA, along with two Mg2+ ions (Fig. 2 D). One of the Mg2+ ions is coordinated to the phosphate of nucleotide +6 and is juxtaposed to the major groove face of G1497 and m3U1498, whereas the other bridges the phosphates of mRNA nucleotide +5 and 16S rRNA nucleotides 1494–1495.
The tight folding of the 3′ end of the short mRNA into the pocket formed by A790, the top of helix 44, and the codon–anticodon interaction in the P site comes as a surprise. As noted above, all of the residues lining this pocket are universally conserved (12). The combination of an exposed purine base (A790) and the electronegative pocket within which a Mg2+ ion is coordinated (G1497 and m3U1498 and the phosphates of nucleotides 1496–1498) suggests that the site occupied by mRNA nucleotide +6 could bind free nucleotides in certain circumstances. A number of cyclic nucleotides that serve as second messengers (cAMP, cGMP, and cyclic diguanylate) are used in bacteria and eukaryotes to regulate cell physiology (15, 16). It will be interesting to test whether these nucleotides can regulate translation by binding to this site.
In contrast to the pyrimidine/pyrimidine base pairs seen in the 30S crystal structures (9), the wider codon–anticodon helix formed between the mRNA and P-site ASL pulls the ASL ≈2 Å deeper into the P-site binding pocket. In addition, the binding pocket is ≈1.3 Å more closed when compared with that in the 30S subunit (Fig. 3 A and B). The deeper binding of the ASL in the P site leads to stacking of the ribose and base of nucleotide 34 in the anticodon with both G966 and C1400 in 16S rRNA, respectively. Both of these nucleotides in 16S rRNA have been directly implicated in P-site tRNA binding (14, 17). Notably, deeper binding of the ASL in the P site does not appreciably change the interactions between the 30S subunit and the ASL that clamp the P-site tRNA in position closer to the small and large (50S) subunit interface. These contacts involve universally conserved nucleotides A790 on the platform and nucleotides G1338 and A1339 in the 30S head (9). The phosphate of A790 in 16S rRNA, which caps an RNA U-turn in the platform of the small subunit, forms a hydrogen bond to the 2′-OH of the ribose of ASL nucleotide 38. As observed in the 30S subunit crystal structures, G1338 forms a type II “A-minor” interaction with nucleotide 41 of the ASL (9, 18). A1339 forms a type I A-minor motif with the G30–C40 base pair of the ASL (Fig. 3 C) (18).
Ribosome interactions with the P-site ASL. (A) Closing of the P-site cleft when compared with the 30S ribosome structures (9). Changes in the distance between C1400 and G966 and between G966 and ASL are indicated by arrows and distances. The position of the very C terminus of protein S9 is indicated. The view is from the aminoacyl-tRNA site in the small subunit. (B) Stereoview of the averaged (3F obs − 2F calc) difference electron density for the 30S contacts to the ASL shown in A. Electron density for mRNA nucleotides +4 through +6 has been removed for clarity. The density indicated by an asterisk is disconnected from that for protein S9 and therefore has not been assigned. (C) Minor groove interactions among G1338, A1339, and the P-site ASL. The view is from the right in A, i.e., from the perspective of the 50S subunit.
Two ribosomal proteins in the small subunit, S13 and S9, have been shown in Thermus thermophilus 30S subunit structures to interact with P-site tRNA through their C termini (9). Although neither the C terminus of S13 nor the C terminus of S9 is essential for viability (19, 20), their presence may contribute to the efficiency and selectivity of translation initiation (21). In the E. coli ribosome neither protein has a highly ordered C terminus when compared with other proteins in the 30S subunit. The C terminus of S13, which is 10 aa shorter than in T. thermophilus, extends toward the minor groove of the top three base pairs of the ASL (Fig. 1 A). However, the C-terminal 5 aa of protein S13, including three lysines, are disordered in the present structures. In E. coli, protein S13 therefore does not make specific contacts to the ASL part of P-site tRNA but may simply contribute electrostatic stabilization to P-site tRNA binding. The highly conserved C terminus of protein S9, for which only the backbone is visible in the present structures, points toward the major groove face of the RNA U-turn at the tip of the anticodon loop near nucleotides 33–37 (Fig. 3 A) (9). The fact that the side chains of the C-terminal residues of protein S9 are not well ordered suggests that protein S9 may also contribute primarily electrostatic interactions with P-site tRNA.
Although the above contacts among the ASL, mRNA, and 30S subunit are similar to those reported previously in the 30S structures (9), the position of the small subunit head leads to a different positioning of the ASL with respect to the intact ribosome. In the present structures the 70S ribosome adopts a closed conformation of the 30S head when compared with isolated 30S subunit structure with an empty aminoacyl-tRNA site (9). This closing involves a tilting of the 30S head toward the central protuberance of the 50S subunit (Fig. 4). Closing of the small subunit head seen in the present structures is consistent with rearrangements in the small subunit observed in solution upon subunit association (22).
Closing of the 30S head domain in the 70S ribosome when compared with the 30S subunit structure. The structure of Protein Data Bank entry 1J5E was used for comparisons (9). The direction of closing of the small subunit head is indicated by an arrow; i.e., the axis of rotation is perpendicular to the plane of the image. The P-site ASL (green) and mRNA (orange) are also shown. The position of full-length tRNA in the P site as modeled based on the ASL positions is shown in outline form, with the arrow indicating the change in direction in going from the 30S to 70S structures.
Based on the ASL analog position in the 30S subunit structures, an intact tRNA projected from its orientation would clash with the central protuberance of the large ribosomal subunit. In contrast, the ASL observed in the 70S ribosome structures would extend the P-site tRNA in the correct direction to dock in the P-site cleft without these clashes (Fig. 4). This difference in ASL positioning may explain the need for closing of the head domain when the 30S subunit docks with the 50S subunit. Tilting of the 30S head domain toward the central protuberance in the 50S subunit also results in closing of the P-site cleft (Fig. 3 A and B). Closing of the P-site cleft in the absence of A-site tRNA likely increases the number of stabilizing interactions between the codon–anticodon helix in the P site and the ribosome, i.e., with G966. This closing may therefore help to minimize frame-shifting on the mRNA before mRNA decoding when the aminoacyl-tRNA site is empty.
Ribosomes were purified from E. coli strain MRE600 as reported (10), with slight modifications. S1 was depleted from intact 70S ribosomes by using a PolyU Sepharose column (23). Eluted ribosomes were concentrated by pelleting in a Beckman Ti45 rotor (Beckman Coulter, Fullerton, CA) at 158,000 × g for 8 h. The ASL of tRNAfmet (5′-UCGGGCUCAUAACCCGA-3′) and mRNA (5′-pAUGUUU-3′) from Dharmacon (Lafayette, CO) were used in 5- to 10-fold excess over ribosomes in crystallization drops. Ribosomes were crystallized at 4°C by using microbatch 96-well plates and buffers containing 11% 2-methyl-2,4-pentanediol, 2% PEG 8000, 27 mM MgCl2, 255 mM NH4Cl, 125 mM KCl, 1 mM spermine, 0.5 mM spermidine, 10 mM Tris (pH 7.5), 0.25 mM EDTA, and 3.5 mM 2-mercapthoethanol. These conditions are similar to those used previously with apo-70S ribosomes (10). Ribosome crystals were stabilized with crystallization buffer containing additional 2-methyl-2,4-pentanediol, PEG 400, and MES (pH 7.0) to allow cryocooling of the crystals to liquid nitrogen temperatures. Diffraction data were measured from 13 crystals cooled to 100 K by using 0.25° to 0.3° oscillations at the SIBYLS (12.3.1) beamline at the Advanced Light Source. The crystals diffracted anisotropically to 3.2 Å (Fig. 1 B). Data were reduced by using Denzo/Scalepack (24) and Truncate (25), yielding the statistics shown in Table 1.
The two copies of the 70S ribosome in the crystallographic asymmetric unit were positioned by using rigid-body refinement of the apo-70S ribosome structures, followed by rigid-body refinement of domains within the ribosome: 30S head, 30S body, 30S platform, 50S body, the L1 arm, and the L11 arm (10). Density-modified electron density maps were generated by using the program Pirate (26). CNS was used to generate (F obs − F calc) and (3F obs − 2F calc) difference electron density maps (27), with the phases derived from Pirate-based density modification (26). Additional restraints used in torsional dynamics and used in modeling magnesium ions are as described in ref. 10. In the structure refinement it was determined that large portions of the small and large ribosomal subunits could be restrained by noncrystallographic symmetry. Based on the nearly identical conformation of most of the two ribosomes, including the entire 30S subunits, it was possible to average the electron density for the two ribosomes (28). Averaging dramatically improved the overall quality of the electron density maps by removing anisotropic streaking apparent in the nonaveraged maps. The averaged (3F obs − 2F calc) difference electron density maps have features consistent with a resolution of 3.5 Å, as shown in Fig. 3 B.
Comparisons to atomic resolution structures of the isolated 30S subunit, to tRNAs, and to structural models of the intact ribosome were carried out by least-squares superposition in the program O (29), generally using ribose C1′ positions in nucleotides. A subset of phosphorous atoms identified as nearly invariant in position were also used for comparisons to the T. thermophilus 30S subunit (30). The full-length P-site tRNA used for superposition with the ASLs was taken from the T. thermophilus 70S ribosome (8). A full list of least-squares superpositions of ribosomes and tRNAs is given in Table 2.
Atoms used in ribosome and tRNA superpositions
Figures were made by using the programs Ribbons (31) and PyMol (32).
We thank K. Frankel, S. Classen, G. Meigs, and J. Holton for significant help with data measurement at the SIBYLS beamline at the Advanced Light Source. We also thank J. Doudna and H. Noller for helpful comments on the manuscript. This work was funded by National Institutes of Health Grant GM65050 (to J.H.D.C.), National Cancer Institute Grant CA92584 (for the SIBYLS beamline), and Department of Energy Grant DE-AC03-76SF00098 (to J.H.D.C., and for the SIBYLS beamline).
Author contributions: V.B. and J.H.D.C. designed research; V.B., W.Z., R.D.P., and J.H.D.C. performed research; V.B., W.Z., R.D.P., and J.H.D.C. analyzed data; and V.B., W.Z., R.D.P., and J.H.D.C. wrote the paper.
The authors declare no conflict of interest.
Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 2I2P, 2I2T, 2I2U, and 2I2V).
