A 16S rRNA–tRNA product containing a nucleotide phototrimer and specific for tRNA in the P/E hybrid state in the Escherichia coli ribosome

Abstract

Ribosome complexes containing deacyl-tRNA₁^Val or biotinylvalyl-tRNA₁^Val and an mRNA analog have been irradiated with wavelengths specific for activation of the cmo⁵U nucleoside at position 34 in the tRNA₁^Val anticodon loop. The major product for both types of tRNA is the cross-link between 16S rRNA (C1400) and the tRNA (cmo⁵U34) characterized already by Ofengand and his collaborators [Prince et al. (1982) Proc. Natl Acad. Sci. USA, 79, 5450–5454]. However, in complexes containing deacyl-tRNA₁^Val, an additional product is separated by denaturing PAGE and this is shown to involve C1400 and m⁵C967 of 16S rRNA and cmo⁵U34 of the tRNA. Puromycin treatment of the biotinylvalyl-tRNA₁^Val –70S complex followed by irradiation, results in the appearance of the unusual photoproduct, which indicates an immediate change in the tRNA interaction with the ribosome after peptide transfer. These results indicate an altered interaction between the tRNA anticodon and the 30S subunit for the tRNA in the P/E hybrid state compared with its interaction in the classic P/P state.

INTRODUCTION

The details of the ribosome interactions with the tRNAs are a critical issue in understanding the mechanism of the protein synthesis cycle (1–3). Photocross-linking has been of some use in monitoring the interaction of the tRNA in the tRNA P site with the 30S ribosome subunit. Specifically, the interaction between 16S rRNA (C1400) and tRNA^fMet was significantly altered by Initiation Factor 3 (IF3), as judged by the frequency of an intermolecular cross-link between C1400 and C35 (the 5′ nucleotide of the anticodon in Escherichia coli tRNA^fMet), and this is probably connected to the decoding preference change induced by IF3 in the initiation complex (4). The tRNA^Phe also forms an UVB light-induced intermolecular cross-link with 16S rRNA, again involving the 5′ anticodon nucleotide and C1400 (5).

However, the cross-linking efficiencies of tRNA^fMet and tRNA^Phe to E.coli ribosomes were only ∼1%, and it was reasoned that a tRNA with higher cross-linking efficiency should be more sensitive in detecting differences in the tRNA–subunit interactions. It was shown previously by the Ofengand group that up to 70% of non-enzymatically bound E.coli N-acetyl-Val-tRNA₁^Val could be cross-linked to ribosomes by irradiation with wavelengths >310 nm (6–9). This cross-link occurs by cyclobutane dimer formation between the 5-(carboxymethoxy) uridine at position 34 (cmo⁵U34) of tRNA₁^Val and C1400 of 16S RNA, and is dependent on a codon that leaves cmo⁵U34, which is the 5′ anticodon base, unpaired (7–10). The characterization of this cross-link, together with the determination of the tRNA location by electron microscopy, provided valuable information about the molecular organization of the P site in the 30S subunit (11). This was an important constraint in ribosome models prior to the X-ray structure and has been confirmed in the X-ray structures (12–14). Most of the previous experiments had been done with aminoacylated tRNAs because the [¹⁴C]valine label was used to track and analyze the photoproducts. Because of this, the results pertained to the tRNA in the classic P/P tRNA binding site in which both the anticodon end of the tRNA and acceptor end of the tRNA occupy positions in the 70S ribosome characteristic of peptidyl tRNA.

In addition to being highly efficient, the 16S rRNA(C1400) × tRNA₁^Val (cmo⁵U34) cross-link is activated by wavelengths, which should not produce other intramolecular RNA–RNA cross-links in the ribosome. To check this, a control experiment in which biotinylvalyl-tRNA₁^Val was bound to 70S ribosomes and irradiated with wavelengths >310 nm showed a single prominent product by gel electrophoresis analysis, in agreement with the properties described by the Ofengand group. However, when ribosome complexes containing deacyl-tRNA₁^Val were irradiated, an additional product with a much reduced electrophoresis mobility was seen in the gel electrophoresis experiment as well as the prominent product. The identification of the cross-linking sites of the new photoproduct in the 16S rRNA and tRNA and its correlation with the tRNA P/E hybrid binding state are described here.

MATERIALS AND METHODS

Preparation of ribosomes, mRNA and tRNA

E.coli ribosomes

Washed 70S E.coli ribosomes were prepared according to Makhno et al. (15). The 30S and 50S ribosomal subunits were separated by centrifugation of 70S ribosomes on sucrose gradients in T₂₀A₂₀₀M₃ buffer (20 mM Tris–HCl, pH 7.5, 200 mM NH₄Cl, 3 mM MgCl₂ and 6 mM 2-mercaptoethanol) and were concentrated by sedimentation and dissolved in activation buffer (20 mM Tris–HCl, pH 7.5, 200 mM NH₄Cl, 20 mM MgCl₂ and 6 mM 2-mercaptoethanol). The 70S ribosomes were formed just before use by re-association of 30S and 50S subunits in activation buffer for 30 min at 37°C.

tRNA aminoacylation and purification

Ten A₂₆₀ units of E.coli deacyl-tRNA₁^Val (Sigma) were aminocylated according to Rheinberger et al. (16) using E.coli tRNA synthetase (Sigma). Val-tRNA₁^Val was incubated in 10 mM EZlink-ss-Biotin (Pierce Chemical) in 1 ml of 600 mM HEPES–HCl, pH 8.3 for 2 h on ice (17). Following biotinylation, the sample was ethanol precipitated twice, dried, dissolved in H₂O and purified by high-performance liquid chromatography on a PRP-1 reverse-phase column (Hamilton) with a 0–25% acetonitrile gradient in 0.1 M NaOAc and 10 mM MgCl₂, pH 5.1.

tRNA labeling

For 5′ labeling, tRNA₁^Val was treated with calf intestinal alkaline phosphatase and then 5′ end-labeled with [γ-³²P]ATP and T4 polynucleotide kinase (MBI Fermentas, Hanover, MD). For 3′ labeling, tRNA₁^Val was reacted with [5′-³²P]pCp using T4 RNA ligase (18) or alternatively, it was 3′-³²P-labeled using Bacillus stearothermophilus tRNA nucleotidyltransferase (CCA-adding enzyme). For this, B.stearothermophilus CCA-adding enzyme was purified and stored according to Yue et al. (19). The conditions of Wolfson and Uhlenbeck (20) were used for the exchange reaction: 500 pmol tRNA₁^Val were incubated in 400 μl total volume containing 100 mM glycine, NaOH, pH 9.0, 10 mM Mg²⁺, 50 μM CTP, 50 μM PPi, 150 μCi [α-³²P]ATP and 1.44 μg B.stearothermophilus CCA-adding enzyme for 8 min at 55°C, followed by the addition of 4 U of yeast pyrophosphatase and incubation for 30 s at 37°C. The tRNA sample was phenol extracted, ethanol precipitated and stored in 50 mM NaOAc, pH 7.0.

mRNA analog

The mRNA analog used in the experiments was made by in vitro transcription using the conditions for high yield of RNA (18). It had 54 nt containing the sequence GGCGAUAACACUCAGGAGAUAAUAAAUGGUUACAGCUGAUCAAUCGUGCAUCCA, where the Shine–Dalgarno sequence is underlined and GUU for the tRNA₁^Val is in boldface.

Irradiation conditions

Preparative samples contained 200 pmol ribosomes, 1000 pmol mRNA and 400 pmol tRNA in 200 μl total volume in T₂₀A₇₀M₇E.₃₅ buffer. Samples were incubated for 30 min at 37°C and stored on ice for 10 min before irradiation. Ribosome complexes were irradiated at 4°C in a stirred quartz cuvette using a 1 kW Xenon light (Oriel Corp.) filtered by one layer of mylar plastic (A₃₁₀ ≥ 2.5, A₃₁₅ = 1.4). The irradiation time was 60 min except as described in the time-course experiment.

Gel electrophoresis and separation of photoproducts

After irradiation, RNA samples were purified by incubation with proteinase K, phenol extraction and ethanol precipitation, and were usually purified on agarose gels to isolate 16S-sized RNA (18). The RNA was ³²P-labeled in those experiments that did not contain ³²P-labeled tRNA. For this, after extraction from the agarose gel, RNA was incubated with calf intestinal alkaline phosphatase followed by incubation with T4 PNK and [γ-³²P]ATP (18). RNA was separated on polyacrylamide gels containing 3.55% acrylamide, 0.05% bisacrylamide, 8.3 M urea in BTBE buffer (21). Gel slices containing cross-linked molecules were identified using a PhosphorImager plate and RNA was isolated by centrifugation through a 2 ml cushion consisting of 2 M CsCl, 0.2 M EDTA, pH 7.4, for 16 h at 40 000 r.p.m. (21,22). Analytical polyacrylamide gels were run under the same conditions and were dried and exposed on PhosphorImager plates.

Analysis of cross-linking sites

Cross-linking sites in the rRNA were identified by reverse transcription reactions on the isolated RNA fractions using a set of oligonucleotide primers according to the procedure of Wilms et al. (21).

tRNA samples were incubated in hydrolysis buffer (25 mM Na₂CO₃, pH 9.0 and 0.5 mM EDTA) for 3 min at 90°C for partial hydrolysis; RNase T₁ and RNase U₂ sequencing was performed on control and cross-linked tRNA as described previously (23). All samples were immediately cooled to 0°C and analyzed using gel electrophoresis on denaturing 12% polyacrylamide gels.

RESULTS

Products from irradiation of 70S complexes containing Val-tRNA₁^Val and tRNA₁^Val

Biotinyl-[¹⁴C]Val-tRNA₁^Val or tRNA₁^Val were non-enzymatically bound to 70S ribosomes programmed with an mRNA analog, and the complexes were irradiated with wavelengths >310 nm to induce cross-linking. The mRNA analog contained a Shine–Dalgarno sequence and the codon GUU for tRNA₁^Val; the GUU sequence results in a U-cmo⁵U34 mismatch in the wobble position of the codon–anticodon interaction and should allow high reactivity of the cmo⁵U34 nucleotide (9). In the first experiment, tRNA₁^Val 3′-labeled with [5′-³²P]pCp, unlabeled tRNA₁^Val, or biotinylated-[¹⁴C]Val-tRNA₁^Val were non-enzymatically bound to 70S ribosomes. Empty 70S ribosomes and the complexes were irradiated with a xenon lamp using a mylar filter to block wavelengths <310 nm. After irradiation, RNA was extracted from each of the complexes by treatment with proteinase K and phenol extraction, and 16S-sized RNA was then separated from 23S and 5S using agarose gel electrophoresis. The samples (except for the sample already containing ³²P-labeled tRNA) were treated with phosphatase and were 5′-³²P-labeled with T4 polynucleotide kinase. All samples were then separated on a denaturing 3.6% polyacrylamide gel, which separates molecules with different inter- or intra-molecular cross-links (Figure 1) (4,21). Two bands were expected: one corresponding to the linear, non-cross-linked 16S rRNA and a second, containing 16S rRNA with C1400 cross-linked to cmo⁵U34 of tRNA₁^Val (9).

Figure 1.

Gel electrophoresis separation of RNA cross-linked products. Denaturing polyacrylamide electrophoresis of RNA from 70S complexes irradiated with wavelengths >310 nm. The 70S complexes were either irradiated empty (lane 1), or with the GUU mRNA analog and [3′-³²P]pCp-labeled tRNA₁^Val (lane 2), unlabeled tRNA₁^Val (lane 3) or biotinylated-[¹⁴C]Val-tRNA₁^Val (lane 4). RNA in lanes 1, 3 and 4 were 5′-³²P-labeled after purification of 16S-sized RNA by agarose gel electrophoresis.

The tRNA-dependent band (band 2) attributable to the C1400/tRNA cross-link is the most prominent new feature of the electrophoresis pattern in all of the samples containing the tRNA₁^Val (Figure 1, lanes 2–4). However, in complexes formed with deacyl-tRNA₁^Val (Figure 1, lanes 2 and 3), an extra band (band 1) is present in the top part of the gel, indicating an RNA species that, based on its electrophoretic mobility, was expected to contain a covalent link between distant nucleotides in the 16S rRNA primary structure. An additional band of lower intensity, band 1a, is present when deacyl-tRNA is non-enzymatically bound to the ribosome.

Band 1 is present both when the tRNA is 3′-³²P-labeled before irradiation (lane 2) and when the RNA is 5′-³²P-labeled after purification of 16S-sized RNA on an agarose gel (lane 3), hence it is not connected just to the addition of the [5′-³²P]pCp to the 3′ end of tRNA prior to irradiation. In addition, in another experiment that used 3′-labeled tRNA by an exchange reaction at the 5′ phosphate of the terminal A with CCA-adding enzyme, the same band 1 photoproduct was seen. Therefore, the appearance of the band is associated specifically with the deacylated state of the tRNA in the P site. There is no change in the frequency of bands 1, 1a and 2 if proteinase K treatment is omitted prior to phenol extraction. This is consistent with the cross-links involving direct RNA–RNA reactions without incorporation of ribosomal proteins.

The position of band 1 is nearly identical to that of a 16S rRNA with an internal cross-link between C967 and C1400 (21). This indicates a cross-link involving nucleotides distant in the 16S rRNA primary sequence is formed in some way. Importantly, however, bands 1, 1a and 2 are seen in the autoradiogram when tRNA is ³²P-labeled (Figure 1, lane 2), indicating that tRNA is present as a covalent part of the photoproducts in those bands.

Reverse transcription analysis of the 16S rRNA cross-linking sites in the bands 1 and 2 RNA

Reverse transcription experiments were done on RNA purified by 3.6% PAGE to verify the identity of band 2 and to determine the 16S rRNA cross-linking sites in the band 1 RNA. Deacyl-tRNA₁^Val was 3′ labeled with [5′-³²P]pCp and non-enzymatically bound to 70S complexes and after irradiation 16S-sized RNA was separated on a denaturing 3.6% polyacrylamide gel (Figure 2A). RNA from bands 1 and 2 was isolated for further analysis. A set of primers was used for reverse transcription to identify cross-linking sites. Since the RNA had been pre-purified based on its content of covalent cross-links, fraction- specific reverse transcription stops should be present to indicate the cross-linking sites. The only reverse transcription stop seen in the band 2 RNA and not in control RNA corresponds to a cross-link at C1400 (Figure 2B, lane 6), as expected from the results of Prince et al. (9). Band 1 RNA also contained the stop corresponding to a cross-link at 16S rRNA C1400 (Figure 2B, lane 5), although the intensity of the stop in this sample is ∼50-fold less than the intensity of the stop in the band 2 RNA, reflecting the relative frequency of the two products.

Figure 2.

Reverse transcription analysis of 16S rRNA cross-linking sites. (A) Preparative denaturing PAGE of RNA from 70S ribosomal complexes containing tRNA₁^Val 3′-labeled with [5′-³²P]pCp and GUU mRNA analog irradiated with wavelengths >310 nm. RNAs from bands 1 and 2 were isolated for reverse transcription analysis. Linear 16S RNA is not seen on this gel because only the tRNA is labeled. (B) Reverse transcription of RNA from bands 1, 2 and linear 16S RNA in the region 1393–1409. RNA from bands 1 and 2 both produce reverse transcription stops at C1401/C1400, indicating a cross-link at C1400. (C) Reverse transcription of RNA from bands 1, 2 and linear 16S in the region 958–975. A double reverse transcription stop is seen in the RNA from band 1 at A968/C967 (lane 5), indicating a cross-link at C967. This reverse transcription stop is absent in RNA from band 2 and linear 16S (lanes 11 and 12) and only the intrinsic reverse transcription stop at C967 due to m²G966 is seen.

An additional double reverse transcription stop at nucleotides A968/m⁵C967 is seen in the band 1 RNA (Figure 2C, lane 5). This is distinctly different from the single intrinsic reverse transcription stop on m⁵C967, which is due to the post-transcriptional methylation of m²G966. The pattern of stops at A968/m⁵C967 in the band 1 RNA is characteristic of cross-linking on nucleotide m⁵C967 (5). The band 2 RNA did not show the A968 reverse transcription stop corresponding to cross-linking at m⁵C967 (Figure 2C, lane 11). RNA from the lower intensity band 1a was also isolated; reverse transcription stops indicated cross-linking sites at C967 and C1400 were also seen (data not shown). This suggests a close relationship between band 1a and 1; however, it was not possible to isolate enough of the band 1a RNA to analyze it further.

Identification of the cross-linking in tRNA in bands 1 and 2

One explanation for the involvement of C1400 and m⁵C967 in the band 1 RNA was that, there were two intermolecular cross-links to tRNA, one between C1400 and cmo⁵U34, and another between m⁵C967 and some other nucleotide in the tRNA. Therefore, RNA sequencing reactions and alkaline hydrolysis were done to sequence tRNA from bands 1 and 2 from both ends to determine the cross-linking site(s).

The tRNA₁^Val was either 5′-³²P-labeled with T4 polynucleotide kinase after dephosphorylation or was 3′-³²P-labeled with the CCA-adding enzyme. The tRNA was non-enzymatically bound to 70S ribosomes programmed with the GUU mRNA analog and the complexes were irradiated. To minimize loss, the purification of 16S rRNA from the 23S and 5S rRNAs by agarose gel electrophoresis was omitted and total RNA was separated on denaturing 3.6% polyacrylamide gels.

In the experiments in which tRNA₁^Val was 5′-³²P-labeled, the alkaline hydrolysis ladders from the tRNA in bands 1 and 2 are interrupted at the same position (Figure 3A, lanes 2 and 3). The last linear fragment before the cross-linking site ends at U33, based on the pattern of RNase U2 digestion of [³²P]RNA from band 2, and RNase U2 and RNase T1 digestion of control ³²P-labeled tRNA, obtained by photoreversal of the band 2 RNA; therefore, the first cross-linked nucleotide encountered from the 5′ end is cmo⁵U34. Even though the 5′-labeled tRNA was purified by 8% PAGE before complex formation and irradiation, there is a shadow on the bands in all lanes of the sequencing reactions. This shadow leads to an extra band with 3-fold lower intensity after position 33 in tRNA from both bands 1 and 2. However, shadows are present in all lanes, including the sequencing lanes and including the band 2 RNA lanes, which is known to involve only cmo⁵U34 (9), so the shadows must be an artifact.

Figure 3.

RNA sequencing reactions to identify cross-linking sites in tRNA. (A) Analysis of cross-linking site in 5′-³²P-labeled tRNA₁^Val. tRNA₁^Val, 5′-³²P-labeled by treatment with alkaline phosphatase followed by labeling with T4 PNK and [γ-³²P]ATP, was non-enzymatically bound to 70S ribosomes programmed with GUU mRNA analog and the complex was irradiated with >310 nm. RNA was extracted and separated on 3.6% denaturing polyacrylamide gels to isolate RNA from bands 1 and 2. Lanes 2 and 3 contain the partial alkaline hydrolysis products from the RNA from bands 2 and 1 and show that the last linear fragment determined from the 5′ end has its terminus at U33. Controls include RNase U2 (U2) digestion of band 2 (lane 1), and partial alkaline hydrolysis, RNase T1 (T1) and RNase U2 digestion of 5′-labeled tRNA₁^Val (lanes 4–6). (B) Analysis of cross-linking site in 3′-³²P-labeled tRNA₁^Val. tRNA₁^Val, 3′ end-labeled at the 5′ phosphate of the terminal A by incubation with B.stearothermophilus CCA-adding enzyme and [α-³²P]ATP, was non-enzymatically bound to 70S ribosomes programmed with GUU mRNA analog and the complex was irradiated with >310 nm. RNA was extracted and RNA from bands 1 and 2 was isolated. RNA from bands 1 and 2 was subjected to partial alkaline hydrolysis and run on denaturing 12% polyacrylamide gels (lanes 4 and 3); this shows that the last linear fragments have their termini at A35. Controls are RNase T1 and RNase U2 digestion of band 2 (lanes 1 and 2), and partial alkaline hydrolysis, RNase T1 and RNase U2 digestion of 3′ labeled tRNA₁^Val (lanes 5–7). Note that since RNase U2 and RNase T1 enzymes cut on the 3′ side of A and G residues respectively, the pattern produced by these enzymes are displaced by one nucleotide when compared with the partial alkaline hydrolysis ladder.

In the sequencing of 3′-labeled tRNA₁^Val, the alkaline hydrolysis ladders are again interrupted in the same place for both bands 1 and 2 (Figure 3B, lanes 3 and 4). Based on the pattern of RNase T1 and RNase U2 digestion of band 2, and RNase U2 and RNase T1 digestion of tRNA₁^Val, obtained by photoreversal of the band 2 RNA, the identity of the last linear fragment determined from the 3′ end is A35; therefore, the cross-linking site is cmo⁵U34 in both cases. Unlike the sequence from the 5′ end, there are no shadows of individual bands, leading to a complete absence of intensity at position 34. There is a band compression somewhere in the sequence between A27 and A35 because a nucleotide is missing between A27 and A35 in the partial alkaline hydrolysis ladder. However, the position of A35 is completely clear and the sequence of the tRNA above A27 and below A35 corresponds to the tRNA sequence (with the exception of reduced RNase T1 recognition of G42–46 and reduced RNase U2 recognition of N⁶-methyladenosine 37). Therefore, we conclude that sequencing from either end of the tRNA identifies cmo⁵U34 as the single site of cross-linking in both band 1 and band 2 RNA.

Time course for cross-link formation

A complex of deacyl-tRNA₁^Val and 70S ribosomes were irradiated for increasing times with >310 nm light to determine the time course of formation of the bands 1 and 2 photoproducts. Gel electrophoresis of 5′-³²P-labeled 16S-sized RNA is shown in Figure 4A. The band 2 photoproduct reaches a maximum value by ∼40 min of irradiation (Figure 4B). The band 1 product appears less slowly at early times but continues to increase through 80 min, the longest time of irradiation (Figure 4B).

Figure 4.

Time course of appearance of bands 1 and 2 photoproducts. (A) Complexes containing deacyl-tRNA₁^Val and the GUU mRNA analog were irradiated with wavelengths ≥310 nm for increasing times and 16S-sized RNA was isolated and 5′-³²P-labeled before electrophoresis on denaturing 3.6% polyacrylamide gels. The bands 1 and 2 products and the linear 16S rRNA are indicated. (B) The radioactivity in bands 1, 2 and the linear 16S rRNA parent band was determined using a phosphoimager; the amount of RNA in bands 1 and 2 were expressed as percentage of the total at each time point. The uncertainty bars indicate standard deviations for the values from three experiments.

Photoreversal test of the bands 1 and 2 products

To determine whether the photoproducts in bands 1 and 2 were photoreversible, [3′-³²P]pCp-labeled tRNA₁^Val was non-enzymatically bound to 70S ribosomes, complexes were irradiated and the RNA from bands 1 and 2 was isolated. The RNAs were then re-irradiated with a germicidal (254 nm) lamp or with wavelengths >310 nm for 1, 2 and 5 min (Figure 5). In the case of the band 2 RNA, the cross-link is very responsive to 254 nm light, and all of the 3′ labeled tRNA is released from 16S RNA after 5 min. This is consistent with a cyclobutane dimer structure present in the 16S rRNA(C1400) × tRNA(cmo⁵U34) cross-link. However, band 1 RNA, is insensitive to 254 nm irradiation, indicating that this product does not contain a cyclobutane bridge. Therefore, this product probably does not have the same structure as the phototrimer described previously by Wang (24), which contained a cyclobutane bridge between two of the three thymines.

Figure 5.

Photoreversibility tests for RNA from bands 1 and 2. RNA isolated from bands 1 and 2 were irradiated with 254 nm light and >310 nm light to determine if tRNA would be released from the cross-link due to photoreversal. RNA isolated from bands 1 and 2 was re-dissolved in 1 mM Tris, pH 7.5, 0.1 mM EDTA and were irradiated with a 254 nm lamp or the mylar-filtered xenon light (λ > 310 nm) at room temperature with stirring for the indicated times. The resulting samples were then analyzed on a denaturing 4–10% composite polyacrylamide gel so that both the 16S RNA and tRNA could be resolved. The locations of bands 1, 2 and tRNA are indicated.

Puromycin reaction of the biotinyl-Val-tRNA₁^Val–ribosome complex results in the formation of the band 1 photoproduct

Since irradiation of 70S complexes containing deacyl-tRNA₁^Val produces the additional band 1 photoproduct compared with complexes containing amino-blocked val-tRNA₁^Val under the same conditions, the novel photoproduct might be formed by the tRNA in the P/E state as opposed to the P/P state (25). To test this, biotin-[¹⁴C]Val-tRNA₁^Val was non-enzymatically bound to the P-site of GUU programmed 70S ribosomes and half of this complex was immediately irradiated at 4°C, while the other half was treated with 1 mM puromycin for 10 min at 37°C before irradiation at 4°C. The puromycin reaction was found to be complete in 5 min at 37°C as determined by the ethylacetate extraction assay (26). It is well established that puromycin, which binds to the A-site of 50S subunits as a mimic of the 3′ end of aminoacyl-tRNA, reacts with the peptide of the tRNA in the 50S P-site via peptidyl transferase activity of the 50S subunit. As a consequence of this, the tRNA is deacylated. From tRNA footprinting experiments, this newly deacylated tRNA occupies the P-site of the 30S subunit and the E site of the 50S subunit (P/E) (25). If the band 1 photoproduct is indicative of tRNA in the hybrid P/E site, reaction of biotin-[¹⁴C]Val-tRNA₁^Val with puromycin should yield a complex capable of forming the phototrimer.

After the puromycin reaction, there was a 10- to 20-fold higher level of band 1 photoproduct from irradiated tRNA–ribosome complexes compared with complexes irradiated before puromycin reaction (Figure 6, lane 3 versus lane 2). Biotin-[¹⁴C]Val-tRNA₁^Val non-enzymatically bound to 70S ribosomes formed the band 1 photoproduct to a low extent in this experiment (Figure 6, lane 2); the amount was subsequently related to the amount of deacylated tRNA₁^Val that was present in the tRNA recovered from the complexes. This was present either because of incomplete purification of the biotin-[¹⁴C]Val-tRNA₁^Val or deacylation during incubation at 37°C used to form the complexes. The band 1a photoproduct does not appear after puromycin reaction. The amount of the band 2 photoproduct is not different in the RNA from the complexes irradiated before puromycin treatment compared with those irradiated after puromycin treatment and there is no change in its electrophoretic mobility.

Figure 6.

Analysis of RNA after irradiation of complexes of biotinylvalyl-tRNA₁^Val and 70S ribosomes after puromycin reaction. Biotinylated-[¹⁴C]Val-tRNA₁^Val was non-enzymatically bond to 70S ribosomes programmed with GUU mRNA analog. Half of the resulting complexes were irradiated with >310 nm light (lane 2), while the other half was first treated with puromycin for 10 min followed by irradiation with >310 nm light (lane 3). Lane 1 contains RNA from complexes of [3′-³²P]pCp-labeled tRNA₁^Val non-enzymatically bound to GUU programmed 70S ribosomes and irradiated with >310 nm for comparison.

DISCUSSION

The three nucleotides, tRNA(cmo⁵U34), 16S rRNA(m⁵C967) and 16S rRNA(C1400) have been identified as cross-linking sites in the band 1 photoproduct (Figure 7). This photoproduct occurs specifically in ribosome complexes containing deacylated tRNA, so it is correlated with tRNA in the P/E hybrid state. The electrophoretic mobility of this product is nearly the same as that of a 16S rRNA with an internal cross-link between m⁵C967 and C1400 that results from irradiation with UVB light in the empty ribosome (5,21). However, that cross-link is not observed with wavelengths >310 nm (Figure 1) and is not expected in the presence of tRNA–ribosome complexes (5). In addition, tRNA is present in the band 1 product and the only cross-linking site in the tRNA is at the cmo⁵U34 nucleotide. These properties of the product indicate a phototrimer in which cmo⁵U34 mediates the connection between m⁵C967 and C1400.

Figure 7.

Location of photoreactive nucleotides in the 16S rRNA and tRNA₁^Val secondary structures. The middle part of the E.coli 16S rRNA secondary structure (35) and the whole tRNA₁^Val structures are shown. The arrows point to the cmo⁵U(34) nucleotide in the tRNA and the C1400 and m5C967 nucleotides in the 16S rRNA.

The main types of products formed between pyrimidines after UVB irradiation are dimers containing a cyclobutane bridge between the C5 and C6 bonds of each base, dimers containing a 4-6 bond between the bases and hydrated monomers that arise from direct photohydration or from reversal of dimers (27–29). In the case of the cyclobutane pyrimidine dimer, it is fully photoreversible into its constituent parts by re-irradiating the dimer with 254 nm light (27,29). Additionally, a thymine phototrimer has been reported to form upon 254 nm UV irradiation of thymine in a frozen aqueous solution (24). The thymine phototrimer was partially photoreversible by 254 nm irradiation and this was consistent with its structure, which contained a cyclobutane dimer and a single 6-4 bond joining the second and the third nucleotides of the trimer (24). The absence of any photoreversal in the RNA from band 1 when irradiated at 254 nm argues that none of the covalent bonds connecting the three participants is of the cyclobutane type. On the other hand, the RNA from the band 2 photoproduct was readily reversed in our hands, consistent with its previous characterization of containing the cyclobutane bridge (8).

The phototrimer is probably formed in a two-step reaction. The prominent photoproduct observed here is the cross-link between C1400 and cmo⁵U34. Because no simple cross-link is seen between m⁵C967 and cmo⁵U34, the order of formation of the trimer is likely to be the addition of C1400 and cmo⁵U34, followed by subsequent addition of m⁵C967. The C1400 × cmo⁵C34 dimer is not photoreversible upon irradiation with wavelengths of >310 nm for 5 min and this is evidence of its low absorbance at this wavelength. Thus, it is likely that the second reaction, which results in the formation of the trimer, is limited by the absorbance of the intermediate and is not an indication of the geometry between m⁵C967 and C1400<>cmo⁵U34, nor is it an indication of the fraction of the deacyl-tRNA complexes that are capable of undergoing the second photoreaction.

These results indicate that there is likely to be a structural difference in the 30S subunit when the tRNA is in the P/E hybrid state and when the tRNA is in the P/P state. The arrangement between m⁵C967 and the tRNA anticodon must be altered in some way in the P/E state compared with the P/P state. The X-ray structure in this region indicates that the base of nucleotide m²G966 is the closest part of the H31 end loop to the anticodon of the tRNA in the P site (11,13). The identification of m⁵C967 in the present study was made on the basis of reverse transcription, because of the small amount of the cross-linked band 1 product. However, the appearance of the reverse transcription stop is the same as seen earlier in the m⁵C967 × C1400 cross-link which was confirmed by RNA sequencing (5). Reconciliation between the C967 × C1400 cross-link and the X-ray structure is possible, since it is not unlikely that the conformation of the empty ribosome is different from the conformation of the ribosome with a P-site bound tRNA. The present results further indicate the conformational complexity of this region—there is some significant difference in the structure with tRNA in the P/E versus P/P states and for the P/E state, this is likely to be different than the conformation seen in the X-ray structure.

An alternative interpretation of the results is related to the issue of the tightness of the tRNA–ribosome interaction. Intramolecular 16S rRNA cross-links are correlated to favorable geometrical arrangements between the participating nucleotides but also require local conformational flexibility (W. Huggins, S. K. Ghosh, K. Nanda and P. Wollenzien, unpublished data). In this view, the formation of the phototrimer could be due to changes in the conformational flexibility in the 30S subunit if the structure of the complex with deacyl-tRNA in the P/E state is somehow less restrained than the structure with peptidyl-tRNA in the P/P state.

It was proposed in 2003 that deacyl-tRNA exists naturally in the P/P state and that binding of the translation factor EFG is necessary for the formation of the P/E tRNA hybrid state on the ribosome (30,31). However Semenkov et al. (32) and more recently Sharma et al. (33) have shown that after peptide bond formation the peptidyl tRNA is reactive with puromycin, indicating EFG independent movement of the tRNA acceptor ends on the 50S ribosomal subunit before translocation. Thus, the Semenkov et al. (32) and Sharma et al. (33) experiments are consistent with the scheme originally proposed by Moazed and Noller (25). The occurrence of the phototrimer product after puromycin reaction with the biotinyl-Val-tRNA₁^Val is additional evidence for immediate changes in the tRNA interactions after peptide bond formation. In this case, the changes even involve the 30S subunit and indicate the distinct nature of the P/P and P/E states.

One complicating factor in comparing the results from these different experiments is that they frequently have been done under different conditions: the experiments described by Moazed and Noller (25), Semenkov et al. (32), Sharma et al. (33) and in this report, were done in conventional buffers (K⁺, NH₄⁺ and Mg²⁺ cations), the experiments by Zavialov and Ehrenberg (30) and Valle et al. (31) and recently by Blanchard et al. (3) have been done in different versions of polymix buffers (K⁺, NH₄⁺, Ca²⁺, Mg²⁺, putrescine and spermidine) and earlier experiments by Agrawal et al. (34) were done in a polyamine buffer (K⁺, NH₄⁺, Mg²⁺ supplemented with spermidine and spermine). This is relevant because of the report that the P/P and P/E binding states are in equilibrium, and this depends on the buffer conditions (34). The ability to easily quantify the amount of tRNA in the P/E hybrid state through observation of the phototrimer product will allow an independent assessment of the P/E versus P/P equilibration question.