Transit of tRNA through theEscherichia coliRibosome

When bound to Escherichia coliribosomes and irradiated with near-UV light, various derivatives of yeast tRNAPhe containing 2-azidoadenosine at the 3′ terminus form cross-links to 23 S rRNA and 50 S subunit proteins in a site-dependent manner. A and P site-bound tRNAs, whose 3′ termini reside in the peptidyl transferase center, label primarily nucleotides U2506 and U2585 and protein L27. In contrast, E site-bound tRNA labels nucleotide C2422 and protein L33. The cross-linking patterns confirm the topographical separation of the peptidyl transferase center from the E site domain. The relative amounts of label incorporated into the universally conserved residues U2506 and U2585 depend on the occupancy of the A and P sites by different tRNA ligands and indicates that these nucleotides play a pivotal role in peptide transfer. In particular, the 3′-adenosine of the peptidyl-tRNA analogue, AcPhe-tRNAPhe, remains in close contact with U2506 regardless of whether its anticodon is located in the A site or P site. Our findings, therefore, modify and extend the hybrid state model of tRNA-ribosome interaction. We show that the 3′-end of the deacylated tRNA that is formed after transpeptidation does not immediately progress to the E site but remains temporarily in the peptidyl transferase center. In addition, we demonstrate that the E site, defined by the labeling of nucleotide C2422 and protein L33, represents an intermediate state of binding that precedes the entry of deacylated tRNA into the F (final) site from which it dissociates into the cytoplasm.

When bound to Escherichia coli ribosomes and irradiated with near-UV light, various derivatives of yeast tRNA Phe containing 2-azidoadenosine at the 3 terminus form cross-links to 23 S rRNA and 50 S subunit proteins in a site-dependent manner. A and P site-bound tRNAs, whose 3 termini reside in the peptidyl transferase center, label primarily nucleotides U2506 and U2585 and protein L27. In contrast, E site-bound tRNA labels nucleotide C2422 and protein L33. The cross-linking patterns confirm the topographical separation of the peptidyl transferase center from the E site domain. The relative amounts of label incorporated into the universally conserved residues U2506 and U2585 depend on the occupancy of the A and P sites by different tRNA ligands and indicates that these nucleotides play a pivotal role in peptide transfer. In particular, the 3-adenosine of the peptidyl-tRNA analogue, AcPhe-tRNA Phe , remains in close contact with U2506 regardless of whether its anticodon is located in the A site or P site. Our findings, therefore, modify and extend the hybrid state model of tRNA-ribosome interaction. We show that the 3-end of the deacylated tRNA that is formed after transpeptidation does not immediately progress to the E site but remains temporarily in the peptidyl transferase center. In addition, we demonstrate that the E site, defined by the labeling of nucleotide C2422 and protein L33, represents an intermediate state of binding that precedes the entry of deacylated tRNA into the F (final) site from which it dissociates into the cytoplasm.
Knowledge of the molecular events that take place during protein synthesis has been greatly influenced by studies demonstrating that aminoacyl-tRNA, peptidyl-tRNA, and deacylated tRNA are accommodated on the ribosome at the A, P, and E sites, respectively (reviewed in Ref. 1). The three-site model of translation has been modified to include three hybrid binding states, designated A/T, A/P, and P/E, which are adopted by tRNA during its passage through the ribosome (2).
Our earlier photoaffinity labeling studies, which focused pri-marily on the identification of ribosomal proteins contacted by the 3Ј end and anticodon of tRNA as it transits the ribosome, allowed us to propose a model for the arrangement of the A, P, and E-site tRNAs on the Escherichia coli ribosome (3,4). In this model, the tRNA molecules are positioned so that their 3Ј ends are directed toward the peptidyl transferase center of the 50 S ribosomal subunit, while their anticodons point toward the groove between the head and the body of the 30 S ribosomal subunit where they interact with complementary codons in the mRNA. Relative to the 50 S subunit interface, the A-site tRNA is located on the L7/L12 or "right" side, the E site tRNA is placed near the L1 protuberance on the "left" side, and the Psite tRNA occupies the space between them. A similar model was proposed by Noller et al. (5), who investigated the location of tRNA in ribosomal complexes by chemical footprinting.
Recently, cryo-electron microscopy and x-ray crystallography have permitted the visualization of tRNA molecules bound to the ribosome during different stages of protein synthesis at high resolution (6 -8). Stark et al. (7) have demonstrated that the arrangement of the A, P, and E site tRNAs in pre-and postranslocational ribosomes is very close to that predicted by Wower et al. (3,4) and Noller et al. (5). An alternative arrangement for E site-bound tRNA, designated as the E2 site, has been proposed on the basis of cryo-electron microscopy (6,9). This tRNA binding state most likely corresponds to the E site tRNA visualized by x-ray crystallography (8).
Many lines of evidence indicate that the 23 S rRNA is involved in essential ribosomal functions (10,11). Now that the position of tRNA molecules can be visualized relative to the structure of the ribosome, determining the topography of tRNA-23 S rRNA contacts on the level of individual nucleotides is crucial for interpreting the images provided by cryo-electron microscopy and x-ray crystallography as well as for constructing high resolution models of the different functional states of the ribosome. In the present work, we follow the movement of the 3Ј terminus of tRNA as it transits the E. coli ribosome during the elongation cycle of translation using photoreactive tRNA probes and characterize the interactions of tRNA with the 23 S rRNA at the nucleotide level.
Ribosomes and Poly(U) Templates-Tight-couple 70 S ribosomes, isolated from E. coli MRE 600 as described by Makhno et al. (17), bound 1.8 molecules of AcPhe-tRNA Phe per ribosome at 15 mM Mg 2ϩ . Poly(U) templates used in cross-linking experiments were prepared according to Kirillov et al. (18).
Isolation and Analysis of Cross-linked Complexes-Sites in the 23 S rRNA labeled by (2N 3 A76)tRNA Phe were identified by treatment with RNase H in the presence of selected pairs of oligodeoxyribonucleotides (23), RNase T1 protection experiments (24), and primer extension analysis (25,26). The oligodeoxyribonucleotides used in these assays were complementary to sequences within the 23 S rRNA encompassing nu-  (13). The labeling of proteins L27 and L33 was established earlier by immunological methods (3,4,13).

Binding and Cross-linking of Yeast (2N 3 A76)tRNA Phe to Ribosomal Complexes Representing Different Stages of the Elongation Cycle
Irradiation of 2-azidoadenosine with near-UV light induces very short cross-links to adjacent molecules and can therefore provide precise information on the molecular environment of the photoreactive base (27). Replacement of the 3Ј-terminal adenosine of yeast tRNA Phe with 2-azidoadenosine yields a photoreactive derivative, (2N 3 A76)tRNA Phe , which can be aminoacylated by yeast phenylalanyl-tRNA synthetase and bound to E. coli ribosomes (14). In the present work, we describe the preparation of 32 P-labeled (2N 3 A76)tRNA Phe derivatives and their incorporation into ribosomal complexes representing different stages in the translation elongation cycle. The photoreactive and non-photoreactive tRNA derivatives usually contained 3 H or 14 C labels in either the aminoacyl moiety or the 3Ј-terminal nucleotide to verify site occupancy and puromycin reactivity (Table I and "Experimental Procedures"). Identification of the cross-linking sites allowed us to delineate the path followed by the 3Ј terminus of the tRNA during its transit through the ribosome.
To simulate P-site complexes containing tRNA in the preand post-transpeptidation states of binding, Ac[ 14 C]Phe-[5Ј-32 P](2N 3 A76)tRNA Phe and [5Ј-32 P](2N 3 A76)tRNA Phe were individually bound to poly(U)-programmed 70 S ribosomes (Fig. 1,  a and b). Occupancy of the P site by Ac[ 14 G2069 a Some tRNA also bound to other sites. b Phe-tRNA bound to ribosomal A/T or R site as ternary complex with EF-Tu and GMPPNP.
During the elongation phase of protein synthesis, the ribosome oscillates between the pre-and post-translocational states. Pre-translocational complexes were formed by the nonenzymatic binding of Ac[ 14 C]Phe-[5Ј-32 P](2N 3 A76)tRNA Phe to the A site of poly(U)-programmed 70 S ribosomes in which the P site was filled with deacylated E. coli [ 14 C]tRNA Phe (see Fig.  1d) or of Ac[ 3 H]Phe-tRNA Phe to the A site of poly(U)-programmed 70 S ribosomes in which the P site was filled with deacylated [5Ј-32 P](2N 3 A76)tRNA Phe (see Fig. 1e). Comparison of the latter state with that depicted in Fig. 1a was expected to show whether the binding of tRNA to the A site influences the arrangement of tRNA in the P site.
E site complexes were formed by binding deacylated [5Ј-32 P](2N 3 A76)tRNA Phe to poly(U)-programmed ribosomes under conditions in which both the A and P sites were blocked with Ac[ 14 C]Phe-tRNA Phe (see Fig. 1f) (29).
The finding that poly(U)-programmed E. coli ribosomes can bind three molecules of deacylated tRNA Phe in vitro was crucial for the development of the three-site model of the ribosome, which suggests that tRNAs bind successively to the A, P, and E sites during the elongation cycle (30 -33). Therefore, we examined the pattern of cross-linking in the presence of saturating amounts of deacylated [5Ј-32 P](2N 3 A76)tRNA Phe in addition to the complexes described above (Fig. 1g).
Each tRNA-ribosome complex was irradiated with 300-nm lamps to induce cross-linking (3). Separation of the tRNAribosome complexes into 30 S and 50 S subunit fractions by centrifugation at low Mg 2ϩ concentration demonstrated that the tRNA derivatives cross-linked exclusively to the 50 S sub-unit. Subsequent analysis of the covalent tRNA-50 S subunit complexes on a sucrose gradient containing SDS showed that the cross-links are distributed between the 23 S rRNA and the 50 S subunit proteins. As we have previously reported (3,4,34,35), A and P site-bound (2N 3 A76)tRNA Phe and its aminoacylated derivatives primarily label protein L27, whereas E sitebound (2N 3 A76)tRNA Phe labels proteins L33 and L1.

Determination of tRNA Cross-linking Sites on the 23 S rRNA
The sites to which (2N 3 A76)tRNA Phe cross-links on the 23 S rRNA were determined by a combination of three approaches, RNase H cleavage, RNase protection, and primer extension, as outlined in Fig. 2.

Excision of Cross-linked rRNA Sequences by Cleavage with RNase H
Covalent [5Ј-32 P](2N 3 A76)tRNA Phe -23 S rRNA complexes were digested with RNase H in the presence of pairs of 15-mer oligodeoxyribonucleotides complementary to sequences located approximately 100 -250 nucleotides apart in the primary structure of the 23 S rRNA (for a list of oligonucleotides see "Experimental Procedures"). Cleavage of the rRNA at sites that bracket the cross-link releases a fragment, which is tagged by covalently attached [5Ј-32 P](2N 3 A76)tRNA Phe or its derivatives and can be readily detected by denaturing PAGE (Fig. 1). Such "scans" of the covalent tRNA-23 S rRNA complexes revealed five labeled segments within domains IV and V of the 23  (segment H4), and 1910 -2024 (segment H5). The specific pattern of cross-linking depended on the conditions under which the photoreactive tRNAs were bound to the ribosome (Fig. 1).

Protection of Cross-linked Sequences from RNase by Oligonucleotide Hybridization
To delineate the sites of cross-linking more narrowly, [5Ј-32 P](2N 3 A76)tRNA Phe -23 S rRNA complexes were subjected to further RNase H digestions using pairs of oligodeoxynucleotides that differed from those used in the initial scan. Products of these digestions, denoted F fragments, were fractionated by denaturing PAGE (Figs. 2 and 3a). This step was followed by RNase protection analysis in which complementary oligodeoxynucleotides were hybridized to the F regions of the 23 S rRNA and the resulting heteroduplexes were digested with RNase T1. Protected fragments that retained the 32 P label derived from cross-linked [5Ј-32 P](2N 3 A76)tRNA Phe moieties, designated T fragments, delimited the sites of cross-linking to sequences of the 23 S RNA ranging from 10 to 20 nucleotides in length (Figs. 2 and 3b).

Analysis of Cross-linking Sites by Primer Extension Analysis
Final identification of the cross-linking sites was carried out using the primer extension method (26). For this purpose, fragments of approximately 250 nucleotides spanning each of the five cross-linking sites in the 23 S rRNA were excised from the appropriate [5Ј-32 P](2N 3 A76)tRNA Phe -23 S rRNA complexes with RNase H, separated by polyacrylamide gel electrophoresis as in Fig. 1 and subjected to primer extension analysis (Fig. 3, c and d).

Identity of Nucleotides to Which tRNA Is Cross-linked in the 23 S rRNA
The Cross-linked Nucleotide in Segment H1 Is U2585-When tRNA-23 S rRNA complexes containing tRNA cross-links to H1 were digested with RNase H in the presence of oligonucleotides 2, 2a, and 3, all of which resulted in the excision of 3Ј-terminal segments of 23 S rRNA, only the longest was tagged by [5Ј-32 P](2N 3 A76)tRNA Phe (Figs. 2 and 3a). The cross-link site must therefore be located within fragment F1, which encompasses nucleotides 2567-2590. RNase T1 protection analysis showed that oligonucleotide 10 protected [5Ј-32 P](2N 3 A76)tRNA Phetagged subfragment T10, which corresponds to nucleotides 2570 -2592 (Figs. 2 and 3b). The only new primer extension stop within this sequence is at U2586 (Figs. 3c and 4), one nucleotide before the cross-link site.
The Cross-linked Nucleotide in Segment H2 Is U2506 -Covalent tRNA-23 S rRNA complexes, in which [5Ј-32 P] (2N 3 A76)tRNA Phe labeled segment H2, were cleaved with RNase H in the presence of oligonucleotides 3, 3a, 3b, and 4. Inspection of the radioactively labeled sequences released in this assay revealed that the cross-link site is located within fragment F2a, which spans nucleotides 2499 -2539 (Fig. 2). RNase T1 protection experiments revealed that subfragment T11, which corresponds to nucleotides 2488 -2524, retained the [5Ј-32 P](2N 3 A76)tRNA Phe tag (Figs. 2 and 3b). Primer extension analysis identified the cross-linked nucleotide as U2506 (Fig. 3d). When a similar analysis was carried out on tRNA-23 S rRNA complexes prepared from ribosomes in the absence of poly(U), two sites of cross-linking were identified within segment H2. One of them was located within fragment F2a and corresponded to U2506, whereas the other was found to be in fragment F2b, which encompasses nucleotides 2446 -2498 (Fig.  2). The site of attachment within F2b was determined by primer extension to be C2452.
The Cross-linked Nucleotide in Segment H3 Is C2422-RNase H cleavage of all tRNA-23 S rRNA complexes in which tRNA was cross-linked to H3 in the presence of the oligonucleotide pairs 4/4a and 4a/5 demonstrated that the site of cross-linking is located in fragment F3, which encompasses nucleotides 2360 -2445 (Fig. 2). This portion of the complex was further analyzed by RNase T1 protection using oligonucleotides 14, 15, 16, and 17. Because oligonucleotides 14 and 15 protected two overlapping [5Ј-32 P](2N 3 A76)tRNA Phe -labeled subfragments, denoted T14 and T15 in Fig. 2, the site of crosslinking must be located between nucleotides 2415 and 2436. The nucleotide covalently attached to the tRNA was determined to be C2422 by primer extension analysis.
The Cross-linked Nucleotide in Segment H4 Is G2069 -RNase H digestion of tRNA-23 S rRNA complexes containing tRNA cross-linked to segment H4 in the presence of oligonucleotides pairs 5/6, 5/6a, 5/7, and 6a/7a showed that the crosslink is located in fragment F4, between nucleotides 2024 and 2091 (Fig. 2). RNase T1 protection experiments utilizing oligonucleotide 18 localized the cross-linking site to the sequence encompassing nucleotides 2058 -2083 (Fig. 2). Primer extension analysis demonstrated that G2069 was the residue covalently linked to tRNA.
The Nucleotide Sequence Cross-linked to tRNA within Seg- ment H5-When oligonucleotide pairs 7/8 and 7a/8a were used to direct RNase H cleavage of tRNA-23 S rRNA complexes containing tRNA cross-linked to segment H5, the site of tRNA attachment was found to be delimited by nucleotides 1910 -1990 within fragment F5 (Fig. 2). Hybridization to oligonucleotides 19, 20, and 21 revealed that only oligonucleotide 21 protects the 23 S rRNA region tagged by [5Ј-32 P](2N 3 A76)tRNA Phe from RNase T1 digestion; the site of cross-linking must therefore be located between nucleotides 1922 and 1929 (see Fig. 2). Unequivocal identification of the cross-linked nucleotide within this sequence proved impossible owing to stops in all lanes at C1924 and A1927. We suggest that the stop at A1927 obscures the cross-link of tRNA to U1926, a highly conserved nucleotide that is protected from chemical modification by P site tRNA (2).
Cross-linking in the Presence of Excess tRNA-When three deacylated (2N 3 A76)tRNA Phe molecules were bound simultaneously to poly(U)-programmed ribosomes and then subjected to UV irradiation, cross-links were observed in segments H1, H2, H3, H4, and H5 (Fig. 1g). Subsequent analysis revealed that the tRNA mainly labeled nucleotides U2506, U2585, and C2422, with smaller amounts cross-linked to nucleotides U1926 and G2069. Because these five nucleotides are exactly the same as those labeled by (2N 3 A76)tRNA Phe molecules, or their derivatives, placed individually at either the A, P, or E sites, we found no evidence for additional, nonspecific crosslinks when tRNA is present in excess. DISCUSSION Defining the Nucleotides of the 23 S rRNA At the Peptidyl Transferase Center-Incorporation of 2-azidoadenosine into tRNA yields a photoreactive derivative, (2N 3 A76)tRNA Phe , which can be aminoacylated, will bind to ribosomes, is able to participate in peptide bond formation and, when irradiated with near-UV light, can form 2-to 3-Å cross-links to other  H1 and H2. C, U, A, and G are sequencing lanes. XL denotes 23 S rRNA fragment labeled by photoreactive tRNA derivative. Lane K 2 is the corresponding free (i.e. non-cross-linked) 23 S rRNA fragment isolated from the same digest. Lane K 1 is the corresponding 23 S rRNA fragment isolated from non-irradiated ribosomes. The stop signal corresponding to the nucleotide preceding the cross-link site is indicated by an arrow, nucleotide symbol, and its number in the 23 S rRNA sequence. macromolecules in its immediate vicinity (3,14). In our earlier studies, (2N 3 A76)tRNA Phe was primarily used for the identification of ribosomal proteins that serve as topographical markers of the A, P, and E sites on the E. coli ribosome. When bound to the A and P sites, (2N 3 A76)tRNA Phe labeled mainly protein L27 (14,34). In contrast, E site-bound (2N 3 A76)tRNA Phe crosslinked exclusively to protein L33 (4).
Various lines of evidence indicate that interactions of the 3Ј terminus of tRNA with 23 S rRNA are important both for peptide bond formation and for the movement of tRNA through the ribosome. Footprinting of tRNA-ribosome complexes demonstrated that, as the 3Ј-terminal adenosine of tRNA transits the E. coli ribosome, it protects from chemical modification a small number of nucleotides in the 3Ј-half of the 23 S rRNA (36,37). According to these studies, the 3Ј-terminal adenosine of Asite tRNA Phe protects G2553, whereas that of P-site tRNA Phe protects U2506, U2584, and U2585. All of these nucleotides are located within or near the central loop of domain V. Several lines of evidence indicate that this region of the 23 S rRNA is located in proximity to proteins L2 and L27, which are positioned on the 50 S subunit interface in the valley between the L1 ridge and the central protuberance (3,38,39). At the same time, the 3Ј-terminal adenosine of the E site tRNA Phe protects G2112, G2116, and C2394. The two E site guanosines are located in proximity to the binding site for protein L1 (40), whereas C2394 is close to the site at which L33 cross-links to 23 S rRNA (41).
To learn more about the path taken by the 3Ј terminus of tRNA as it moves through the ribosome during protein syn-thesis, we have analyzed a number of cross-linked (2N 3 A76)tRNA Phe -ribosome complexes representing different stages of the elongation cycle of translation. Inasmuch as the chemistry of photoaffinity labeling and chemical footprinting are different (27,42), we expected that the use of azidoadenosine-substituted tRNA probes would enable us to further define the sites that accommodate the 3Ј-terminal adenosine of deacylated, aminoacyl-, and peptidyl-tRNAs during polypeptide synthesis. In this report, we identify five nucleotides of the 23 S rRNA that are labeled by (2N 3 A76)tRNA Phe as it traverses the E. coli ribosome (Fig. 4). The pattern of cross-linking is distinct for each tRNA binding site or state that we have investigated. The site specificity of these cross-links is substantiated by the fact that cross-linking in the presence of excess (2N 3 A76)tRNA Phe leads to the labeling of all five of these nucleotides, but no others. Labeling of a sixth nucleotide, C2452, was observed only in the absence of mRNA.
Patterns of tRNA-rRNA Cross-linking At the Ribosomal P, A, and R Sites-Deacylated (2N 3 A76)tRNA Phe bound at the P site labels mainly U2585 and, to a lesser extent, U2506 (Fig. 1a and Table I). Similar results were obtained by Kirillov et al. (43), although we have not observed the cross-link to nucleotides C2601/A2602 that they reported. In contrast, AcPhe-(2N 3 A76)tRNA Phe at the P site labels U2506, U2585, and G2069, with the cross-link to U2506 predominating ( Fig. 1b and Table I). The effect of the AcPhe group on the cross-linking pattern indicates that the AcPhe moiety, an analogue of the peptidyl group, influences either the positioning or the mobility of the 3Ј-terminal residue of P site tRNA relative to the 23 S rRNA and is consistent with cryo-electron microscopic observations (44). The labeling of U2506 and U2585 by P site-bound (2N 3 A76)tRNA Phe indicates that these two nucleotides are close to one another and are likely to play an important role at the peptidyl transferase center as suggested by earlier studies (36,37,(45)(46)(47)(48). The proximity of G2069, which is methylated at N7, to the 3Ј-terminal adenosine of P site-bound tRNA lends weight to an earlier proposal that a highly conserved structural motif consisting of U2438, A2439, m 7 G2069, and m 1 A2071 lies very close to the catalytic site of peptidyl transferase (49). Furthermore, the labeling of C2452, in addition to U2506 and U2585, by (2N 3 A76)tRNA Phe bound to the P site in the absence of poly(U) suggests that codon-anticodon interaction, which occurs on the 30 S subunit, can also affect the position or orientation of the 3Ј end of the tRNA on the 50 S subunit.
Cross-linking patterns derived from complexes containing peptidyl-tRNA analogues bound under A site conditions were of particular interest. As indicated above, A site-bound Phe-(2N 3 A76)tRNA Phe labeled both U2585 and U2506, although cross-linking to U2585 predominated (see Table I). In contrast, A site-bound AcPhe-(2N 3 A76)tRNA Phe preferentially labeled U2506 (see Fig. 1d and Table I). More strikingly, P site-bound (2N 3 A76)tRNA Phe labeled only U2585 when the A site was filled with AcPhe-tRNA Phe (compare Fig. 1e and 1a; see also Table I). The latter observation shows that the position of the 3Ј end of tRNA at one ribosomal site can be influenced strongly by the tRNA derivatives that occupy other sites. In addition, these data demonstrate that the 3Ј end of tRNA in AcPhe-tRNA Phe , a peptidyl-tRNA analogue, is in close contact with U2506 regardless of whether its anticodon is located in the A site or P site (compare Fig. 1d and 1b; see also Table I).
Incoming Phe-tRNA Phe in a ternary complex with EF-Tu and a non-hydrolyzable GTP derivative, is stabilized in a binding state designated as the A/T state (2), which can be equated with the ribosomal R or recognition site proposed earlier (28). When part of the ternary complex, Phe-(2N 3 A76)tRNA Phe labeled nucleotides U2585, U2506, and U1926 (see Fig. 1c and Table I whereas the same tRNA derivative at the A site labeled U2585 and U2506 (see Table I). From these observations, we conclude that the 3Ј end of Phe-(2N 3 A76)tRNA Phe in the ternary complex contacts the peptidyl transferase center even though it cannot participate in peptide bond synthesis, perhaps because of restrictions imposed by the presence of EF-Tu (50). Alternatively, as EF-Tu triggers the transition from the post-translocational state to the pre-translocational state, labeling of U1926 may be indicative of a change in the conformation of 23 S rRNA in the neighborhood of the 3Ј-end of the incoming tRNA.
The cross-linking experiments reported here, together with chemical footprinting data (36), demonstrate the proximity of U2506 and U2585 to the 3Ј-terminal adenosine of P site-bound deacylated and peptidyl-tRNAs as well as to A site-bound aminoacyl-and peptidyl-tRNAs. These and other potential contacts between the 3Ј end of tRNA and the 23 S rRNA are depicted in Fig. 5. Taken together, the above results indicate that there is a complex interplay between nucleotides U2585 and U2506 of the 23 S rRNA and nucleotide A76 of the tRNA, which is of potential mechanistic importance in tRNA-ribosome interaction, peptide bond formation, and translocation. We first note that Phe-(2N 3 A76)tRNA Phe can cross-link to U2585 and U2506 at the R site as well as the A site and that labeling of U2585 is more intense in both cases. Coincident with peptide bond formation, the 3Ј terminus of the tRNA becomes more closely associated with U2506, as evidenced by the preferential labeling of U2506 by the peptidyl-tRNA analogue, AcPhe-(2N 3 A76)tRNA Phe , and by the simultaneous exclusion of this residue from labeling by deacylated (2N 3 A76)tRNA Phe at the P site. Under these conditions, AcPhe-(2N 3 A76)tRNA Phe is in a state that corresponds to the hybrid A/P binding site (2). The 3Ј end of the tRNA moiety remains in the same position upon subsequent translocation of the tRNA anticodon. With AcPhe-tRNA Phe still in the A/P state, the 3Ј end of the deacylated tRNA, previously in the P/P state, assumes a position from which it can label only U2585 and protein L27 while its anticodon remains anchored in the P site. This pattern of cross-linking suggests that the 3Ј end of deacylated tRNA does not immediately leave the peptidyl transferase center. Our findings thus differ from the conclusion that deacylated tRNA "reaches" into the E site to adopt the P/E state of binding after transpeptidation, which was inferred from chemical footprinting of tRNA-ribosome complexes (2). The pattern of labeling manifested by the A and P site-bound (2N 3 A76)tRNA Phe derivatives suggests that the 3Ј-terminal adenosine may be "sandwiched" between U2506 and U2585, where it can readily access either nucleotide during peptide bond formation (35).
Patterns of tRNA-rRNA Cross-linking At the Ribosomal E and F Sites-In previous studies, we demonstrated that the E site is topographically distinct from the A and P sites, because (2N 3 A76)tRNA Phe bound to the E site of poly(A)-programmed 70 S ribosomes cross-links exclusively to protein L33 (4). In contrast, when (2N 3 A76)tRNA Phe is bound to the E site of poly(U)-programmed ribosomes, nucleotide C2422 of the 23 S rRNA is labeled in addition to protein L33 ( Fig. 1f and Table I). Concurrent labeling of C2422 and protein L33 agrees well with the earlier finding that a neighboring nucleotide, C2427, crosslinks to L33 when ribosomes are irradiated with UV light (38). Finally, deacylated (2N 3 A76)tRNA Phe bound to the ribosomal E site in the absence of mRNA exclusively labels proteins L1 and L33 (35). These results suggest that the presence of a cognate codon in the E site influences the positioning of tRNA, in accord with the observation that the affinity of tRNA Phe for the E site increases slightly in the presence of poly(U) (30,51). The crosslinking patterns described above may correspond to the different E-type states of binding that have been visualized by cryoelectron microscopy (6,7,9). This conclusion is also consistent with footprinting data on E site tRNA (36), because chemical protection of G2112, G2116, A2169, and C2394 can be correlated with at least two E-type binding states. Because the first three nucleotides are located in the vicinity of the L1 binding site, they are most likely protected by the variant of E site tRNA whose 3Ј end contacts protein L1. In our experiments, this type of E site binding was observed only in the absence of mRNA. Interestingly, a similar placement of E site tRNA, designated as the E2 site, was inferred from cryo-electron microscopy of poly(U)-programmed ribosomes to which three deacylated tRNA molecules were bound simultaneously (6). On the other hand, C2394 is most likely protected by tRNA in the E-type binding state observed in post-translocational complexes (7). As shown in this report, (2N 3 A76)tRNA Phe in the latter binding state labels C2422, which is adjacent to C2394 in the secondary structure of the 23 S rRNA (Fig. 4).
Our cross-linking data strongly suggest that protein L1, along with nucleotides G2112, G2116, and A2169, constitute markers for a new site, which we propose to call the F (final) site (Fig. 4). We define the F site as the area on the ribosome from which tRNA dissociates (or is ejected) into the cytoplasm. Before reaching the F site, the tRNA is associated with an intermediate site, the E site, for which protein L33 and the C2422 residue serve as markers. According to our observations, the binding of tRNA to the E site is influenced by the presence or absence of a cognate codon and may involve interactions with the 30 S subunit (Ref. 4 and present work). In this light, most of the earlier published data relating to the E site probably reflect the properties of tRNA bound at the F site, which is completely mRNA-independent.
Together, our results require the following modifications to the hybrid-state model of tRNA-ribosome interaction proposed by Moazed and Noller (2). In particular, we show that the 3Ј end of the deacylated tRNA that is formed after transpeptidation does not immediately progress to the E site but remains temporarily in the peptidyl transferase center. When and how FIG. 5. Proposed arrangement of the 3 ends of the A and P site-bound tRNAs in the peptidyl transferase center. The sketch illustrates interactions of U2505, U2585, G2252, and G2553 of the 23 S rRNA with ϪCCA OH termini of tRNA as suggested by this work and earlier studies (47,(52)(53)(54), and proximity of 3Ј ends of tRNA to protein L27 (shaded area). The A and P site tRNAs are shown in the S configuration as indicated by the preponderance of the experimental data (55). it moves to the E site is at this time a matter of conjecture. Given that E site binding is influenced by the presence of a cognate codon, we suggest that this event is dependent on or triggered by translocation. Moreover, we propose that the E site does not correspond to the final site of tRNA-ribosome interaction, but that it represents an intermediate state of binding for tRNA moving toward the F site, from which it dissociates into the cytoplasm.