Complex of transfer-messenger RNA and elongation factor Tu. Unexpected modes of interaction.

Transfer-messenger RNA (tmRNA) is a stable RNA in bacteria of 360 +/- 40 nucleotides that can be charged with alanine and can function as both tRNA and mRNA. Ribosomes that are stalled either in a coding region of mRNA or at the 3' end of an mRNA fragment lacking a stop codon are rescued by replacing their mRNA for tmRNA. Here we demonstrate that the interaction of tmRNA with the elongation factor Tu shows unexpected features. Deacylated tmRNA can form a complex with either EF-Tu.GDP or EF-Tu.GTP, the association constants are about one order of magnitude smaller than that of an Ala-tRNA.EF-Tu.GTP complex. tmRNA as well as Ala-tmRNA can be efficiently cross-linked with EF-Tu.GDP using a zero-length cross-link. The efficiency of cross-linking in the case of deacylated tmRNA does not depend on an intact CCA-3' end and is about the same, regardless whether protein mixtures such as the post-ribosomal supernatant (S100 enzymes) or purified EF-Tu are present. Two cross-linking sites with EF-Tu.GDP have been identified that are located outside the tRNA part of tmRNA, indicating an unusual interaction of tmRNA with EF-Tu.GDP.

Transfer-messenger RNA (tmRNA) is a stable RNA in bacteria of 360 ؎ 40 nucleotides that can be charged with alanine and can function as both tRNA and mRNA. Ribosomes that are stalled either in a coding region of mRNA or at the 3 end of an mRNA fragment lacking a stop codon are rescued by replacing their mRNA for tmRNA. Here we demonstrate that the interaction of tmRNA with the elongation factor Tu shows unexpected features. Deacylated tmRNA can form a complex with either EF-Tu⅐GDP or EF-Tu⅐GTP, the association constants are about one order of magnitude smaller than that of an Ala-tRNA⅐EF-Tu⅐GTP complex. tmRNA as well as Ala-tmRNA can be efficiently cross-linked with EF-Tu⅐GDP using a zero-length cross-link. The efficiency of cross-linking in the case of deacylated tmRNA does not depend on an intact CCA-3 end and is about the same, regardless whether protein mixtures such as the postribosomal supernatant (S100 enzymes) or purified EF-Tu are present. Two cross-linking sites with EF-Tu⅐GDP have been identified that are located outside the tRNA part of tmRNA, indicating an unusual interaction of tmRNA with EF-Tu⅐GDP.
Transfer-messenger RNA (tmRNA or 10 Sa RNA) 1 is a small, stable RNA of 360 Ϯ 40 nucleotides. The tmRNA is encoded by the ssrA gene (1). A knock-out of this gene has little effect on the growth of Escherichia coli cells at 37°C but suppresses growth at elevated temperatures (2). Additionally, this knockout strain is outcompeted by wild-type strains in a culture (3). The gene is present in all kingdoms of the bacterial domain, including the alpha-proteobacteria (4,5). This molecule is of special interest because it can function as both a tRNA and an mRNA. The 5Ј and 3Ј ends of the molecule can be folded into a tRNA-like structure with an amino acid-acceptor stem and a T⌿C-stem/loop that contains two modifications universally conserved in tRNAs (6). The tmRNA contains an identity element for the alanyl-tRNA synthetase and can be charged with alanine in vivo and in vitro (7). In addition, tmRNA has a messenger RNA sequence that encodes a decapeptide in E. coli (8,9), which is a recognition signal for some proteases (10).
The main function of tmRNA in the cell seems to be the rescue of ribosomes that are arrested by truncated mRNA lacking their stop codon (8,9). These ribosomes carry a peptidyl-tRNA at the P site and either a truncated codon or no codon at all at the A site. The alanine-charged tmRNA enters the ribosomal A site, and the alanyl residue is transferred to the growing peptide. After translocation, translation switches from the truncated mRNA to the messenger part of tmRNA. After reading the codon sequence on tmRNA, the ribosome terminates at the stop codon encoded in tmRNA by the usual termination mechanism. This process has two consequences: (i) the ribosomes arrested with truncated mRNA are rescued and recycled for protein synthesis, and (ii) the truncated proteins are tagged with a signal peptide that is recognized by specific proteases. Recently, it was also demonstrated that stalling of a ribosome at rare codons can trigger tagging of polypeptides despite the presence of a substantial region of non-translated 3Ј-mRNA (11).
Aminoacyl-tRNAs enter the ribosome as a complex with the elongation factor Tu (EF-Tu) in the presence of GTP (12). Thermus thermophilus EF-Tu complexed with GTP protects Ala-tmRNA from spontaneous hydrolysis (13), as EF-Tu⅐GTP does with aminoacyl-tRNA. EF-Tu is therefore believed to be involved in the transport of Ala-tmRNA to the ribosome. An additional protein, the RNA-binding SmpB protein, binds with high affinity to tmRNA and is important for the tagging functions of tmRNA on the ribosome (14). Furthermore, the ribosomal protein S1 can interact with tmRNA (15), and three further proteins have been identified in the tmRNA⅐SmpB complex, the role of which is still unknown in relation to tmRNA function (16). Thus far, the interaction of EF-Tu with tmRNA has not been directly demonstrated.
In this research we aimed to identify one or more cellular components that interact with tmRNA under conditions optimal for protein synthesis. A cross-linking approach developed for the investigation of the ribosomal neighborhoods of mRNA and 5 S rRNA was then applied (17). This method utilizes the photosensitive 4-thiouridine that is incorporated randomly into the RNA under study. It is worth noting that results obtained by this approach with 5 S rRNA were in very good agreement with the recent structural data of the ribosome (18 -20; see Ref. 21). Here, we identified EF-Tu as the main protein that could be cross-linked to tmRNA in E. coli cell-free S100 extract (S100 enzymes). Surprisingly, deacylated tmRNA can form complexes with EF-Tu⅐GDP, and this unusual complex was characterized by cross-linking and footprinting experiments.

EXPERIMENTAL PROCEDURES
Transcription and Alanylation of tmRNA-The plasmid containing the tmRNA gene from E. coli under the control of a T7 promoter was a kind gift of Dr. Brosius. tmRNA transcript was synthesized using an in vitro T7 transcription system (22,32) in the presence of [␣-32 P]UTP using linearized plasmid as the DNA template. For the synthesis of sU-modified tmRNA transcripts, UTP was supplemented with 4-thio-UTP at different molar ratios (from 1:1 up to 1:100). The level of 4-thiouridine incorporation was estimated using the T2 fingerprinting technique (17). For example, one or two 4-thiouridine residues per molecule were incorporated at an sU:U ratio of 1:100, and tmRNA transcripts were purified as described for mRNAs (23). To create truncated tmRNAs, the plasmid was linearized with NcoI or XcmI yielding tmRNAs that lacked the CA-3Ј end or the ACCA-3Ј end, respectively. Complex Formation in the Presence of S100 Enzymes or EF-Tu for Cross-linking and Footprinting-E. coli tRNA-free S100 extract (S100 enzymes) was kindly provided by Dr. M. Dabrowski (MPI fü r Molekulare Genetik). EF-Tu his-tagged at the C terminus was isolated from E. coli as described previously (24). Complexes were formed by incubating tmRNA with the S100 enzymes for 10 min at 37°C under the following ionic conditions: 20 mM HEPES-KOH, pH 7.8 (0°C), 3 mM MgCl 2 , 150 mM NH 4 Cl, 2 mM spermidine, 0.05 mM spermine, 2 mM ATP, 0.2 mM GTP, 4 mM phosphoenolpyruvate, 0.05 g/l pyruvate kinase, and 4 mM 2-␤-mercaptoethanol. The final concentration of tmRNA was 0.1-1 pmol/l. Next, 9 l of S100 was added to 30 l of the reaction mix that was irradiated by mild (Ͼ330 nm) UV light for 5 min as described previously (25). Half of this sample was treated with proteinase K (1.4 g/l final concentration) for 15 min at 37°C, and both aliquots were loaded onto the same gel. Cross-linked and non-cross-linked tmRNAs were separated by 7.5% SDS-polyacrylamide gel electrophoresis (26) for 4 h at 100 V.
Analysis of the Cross-linked Protein-The gel region containing the tmRNA-protein cross-link ( Fig. 1, line 2) was cut out and incubated in 300 l of a buffer containing 20 mM Tris-HCl, pH 7.8, at 0°C, 0.1% (w/v) SDS together with 5 l of RNase T1 (10 g/l) for 20 min at 55°C. The same volume of a buffer containing 0.1% SDS, 0.1% (v/v) 2-␤-mercaptoethanol, 300 mM sodium acetate, and 5 l of RNase T1 (10 g/l) was added and the resultant mixture was shaken overnight at 4°C. Proteins were precipitated by ethanol (5 volumes), and after a low speed centrifugation the pellet was dissolved in 10 l of ultrapure water and 1 l of RNase T1 (10 g/l) was added. This mixture was then incubated for 20 min at 55°C, diluted by an equal volume of loading buffer (250 mM Tris-HCl, pH 6.8, at 0°C, 8% (w/v) SDS, 40% glycerol, 400 mM dithiothreitol, and 0.04% bromphenol blue), and loaded onto a 7.5% protein gel (26). Protein bands were visualized by staining in a solution containing 0.1% (w/v) Coomassie Blue, 0.5% (v/v) acetic acid, 20% (v/v) methanol for 2 h. Destaining was achieved by washing the gel with 30% (v/v) methanol.
Tryptic Digestion and MALDI-Mass Spectrometry-Cross-linked proteins were digested in gel according to the method described in a previous study (27) using 100 ng of sequencing grade modified trypsin (Roche Molecular Biochemicals, Germany) per gel piece containing a band. The eluted peptides were purified on C-18 ZipTip micro columns (Millipore, Bedford, MA) and analyzed by matrix-assisted laser desorption/ionization mass spectrometric analysis (MALDI-MS) in the presence of ␣-cyano-4-hydroxycinnamic acid as matrix. MALDI-MS analysis was performed on a Reflex-II instrument (Bruker Daltonik, Bremen, Germany) in reflector mode. Samples for MALDI-MS were prepared by the dried droplet method on stainless steel targets precovered with thin matrix films. The determined peptide masses were submitted to the web-based data base searching programs, MS-Fit and Profound, to identify the protein (see Table II below).
Footprinting Analysis-Conditions of the modification reactions for footprinting analysis were taken from Moazed and Noller (28). For modification with CMCT (1-cyclohexyl-3-(2-morpholinoethyl)carbodiimide metho-p-toluene sulfonate), 25 l of a 42 mg/ml CMCT solution was added to either 25 l of tmRNA⅐EF-Tu⅐GDP complex mixture (2 pmol/l) obtained as described above (but without an energy regeneration system) or to 25 l of untreated control tmRNA in solution. This mixture was incubated at 37°C for 10 min, followed by immediate quenching of the modifying reagent with the appropriate stop solution (28). The modification positions were determined by a primer-extension (29) using primers complementary to the nucleotides 55-73, 149 -217, or 345-363 of tmRNA.
Determination of the Positions of Cross-links-The cross-linked tmRNA⅐EF-Tu⅐GDP complex were extracted from the gel as described above (under "Analysis of the Cross-linked Protein") and shaken overnight at 4°C. Samples were precipitated by ethanol (5 volumes), and after a low speed centrifugation the pellet was dissolved in 10 l of ultrapure water and treated with fresh proteinase K (1.4 g/l in the reaction mixture) for 15 min at 37°C. The extracted RNA was purified in a 8% polyacrylamide-urea gel, extracted, and dissolved in 5 l of Milli-Q water (Millipore) and analyzed by primer extension (29) using the primers described in the preceding paragraph.
Association Constants-Association constants (K a ) for tmRNA⅐EF-Tu⅐GDP/GTP complexes were measured by filtration of the complexes through a 0.45-m nitrocellulose membrane (Millipore). This method was a modification of that described previously (30). EF-Tu⅐GDP was incubated in a multiple of 3.33 l (32 pmol per 3.33-l volume unit) of reaction buffer containing 60 mM HEPES-KOH, pH 7.6, at 4°C, 10 mM Mg(acetate) 2 , 150 mM NH 4 Cl, and 1 mM dithiothreitol, and (in the case of complexes with GTP) 5 mM phosphoenolpyruvate, 30 M [␥-32 P]GTP (specific activity 3500 dpm/pmol) and 0.1 g/l pyruvate kinase. After incubation at 37°C for 10 min, the volume unit was increased to 10 l to maintain the ionic conditions, and up to 80 pmol of tmRNA was added. After 20 min at 0°C, 10 l was spotted onto a nitrocellulose filter presoaked in reaction buffer and washed with 100 l and subsequently three times with each 3 ml of reaction buffer. The filter was resolved in 5 ml of Filter-Count (Packard, Groningen, The Netherlands) and counted. The difference, cpm (binary complex EF-Tu⅐GTP/GDP in a 3.3-l volume unit) Ϫ cpm (ternary complex in a 10-l volume unit), gives the amount of ternary complex tmRNA⅐EF-Tu⅐GTP/GDP formed. Note that the binary complex is adsorbed to the filter, whereas the ternary complex goes through. The procedure of gel-shift experiments for measuring the formation of ternary complexes was as described previously (44).
EF-Tu was isolated as described (24). In the experiments with tmRNA, native EF-Tu from E. coli lacking a His-tag was used. Experiments monitoring the formation of ternary complexes in the presence of [ 14 C]Phe-tRNA and [ 14 C]Ala-tRNA, respectively, were performed as controls.

RESULTS
Preparation of tmRNA Containing the Photo-reagent 4-Thiouridine (sU)-tmRNA containing statistically two randomly distributed sU residues was synthesized by in vitro transcription using T7 RNA polymerase. The level of sU incorporation was controlled by RNase T 2 -fingerprints (25). The T7-transcript tmRNA could be charged with [ 14 C]alanine up to a maximal 40%. A similar value (20 -40%) has been found by other groups (6,14,30) and might indicate a conformational heterogeneity. The alanine residue could be efficiently attached to nascent peptide chains in vitro in the presence of the 3 mM Mg 2ϩ /polyamine system 2 (see Ref. 31 for the system). With regards to sequence and length, the homogeneity was determined using transcribed phosphorothioated tmRNA to be about 80% (not shown).
Cross-linking Experiments with S100 Enzymes-S100 enzymes are the post-ribosomal supernatants of bacterial cell extracts. All tRNAs were removed from the S100 enzymes by treatment with DEAE-Sepharose. The S100 enzymes were mixed with modified [ 32 P]tmRNA (sU) in the presence of GTP and a GTP regeneration system. Cross-links were induced by mild irradiation of the reaction mixture with UV light ( Ͼ 330 nm). SDS-gel electrophoresis revealed cross-linked products containing tmRNA by reduced mobility of the complex (see lane 2 in Fig. 1A; lane 1 contains the non-irradiated sample). This mobility shift was due to the formation of a complex of tmRNA with one or more proteins, because treatment with proteinase K cancelled the shift (lane 3).
Analysis of the Cross-linked Complexes-For identification of the cross-linked protein(s) the complex was purified from the S100-derived proteins that co-migrated with the tmRNA during gel electrophoresis. The complex seen in lane 2 of Fig. 1A was extracted from the gel and subjected to a second electrophoresis step under the same conditions, after the RNA had been digested with T1 RNase. Cross-linked proteins were expected to change their mobility upon the loss of the RNA moiety. Only one protein band was clearly seen in the gel after this treatment. The gel section containing the band of the cross-linked protein was isolated and subjected to tryptic digestion, and the resulting peptides were analyzed by MALDI (Table I). A data base search concerning the peptide masses revealed that all major peptides could be assigned as fragments of EF-Tu. The identified peptides covered 65% of the protein's sequence. It follows that EF-Tu is the major protein component of the cross-linked complex.
Complexes of tmRNA with Isolated EF-Tu⅐GTP/GDP-An intriguing observation from the experiments described above is the high yield of cross-linked tmRNA in the presence of EF-Tu⅐GTP and in the absence of alanylation. Under the conditions used here, a significant amount of transcribed tmRNA could not be alanylated in the presence of S100 enzymes, because alanine was not added to the reaction mixture. Therefore, it is possible that EF-Tu⅐GTP can form complexes with deacylated tmRNA.
The formation of tmRNA⅐EF-Tu complexes in the presence of either GTP or GDP has been examined by using nitrocellulose filtration methods (30). Indeed, binding was observed, and the association constants, K a , for either complex could be determined. Controls with Phe-tRNA were performed and, furthermore, the association constants with Phe-and Ala-tRNA were determined using band-shift assays.
Both constants with deacylated tmRNA (in the presence of GDP or GTP) were about 1 ϫ 10 6 M Ϫ1 , for the canonical ternary complexes Ala-tRNA⅐EF-Tu⅐GTP or that with Phe-tRNA up to 10 to 20 times higher values were observed (Table II). It follows that even EF-Tu⅐GDP can form a ternary complex with deacylated tmRNA. In contrast, a comparable complex even with a large excess of deacylated tRNA bulk (40-fold over the concentration of tmRNA) could not be detected with the applied nitrocellulose filtration technique, in agreement with literature data underscoring the specificity of the complex observed with deacylated tmRNA.
An intact CCA end is a prerequisite for aminoacylation of tRNA. Therefore, a rigorous test whether or not deacylated tmRNA can form a complex with EF-Tu⅐GDP would be an analysis of the question: Is an intact CCA-3Ј end of the tmRNA truly a prerequisite for complex formation? To address this question, tmRNA was transcribed lacking either the CA-3Ј or the ACCA-3Ј ends, and cross-linking experiments were performed with tRNA-free S100 enzymes as shown in Fig. 1A. The cross-linked products were isolated, and the protein species was analyzed by gel-electrophoresis after digestion of RNA with RNase T1. The protein observed was indeed co-migrating with purified EF-Tu (Fig. 1B). We conclude that the shortened versions of tmRNA formed complexes with EF-Tu⅐GDP as efficiently as did full-length tmRNA.
Next, we asked whether the cross-link between tmRNA and EF-Tu⅐GDP observed here exclusively occurs with deacylated tmRNA rather than with alanylated tmRNA. To this end, transcribed [ 32 P]tmRNA randomly containing about 5 (Ϯ2) 4-thiouridines per molecule was charged with [ 14 C]Ala with a yield of ϳ40% and was employed in a cross-linking experiment with EF-Tu⅐GDP (Table III). A 32 P/ 14 C ratio of 2.7 was found in the non-cross-linked control band, and a ratio of 2.4 in the crosslinked band indicating significant amounts of [ 14 C]Ala-tmRNA in the cross-linked product, although deacylated tmRNA was slightly over-represented. This finding demonstrates that EF-Tu⅐GDP forms a complex with both tmRNA and Ala-tmRNA almost equally well.
The unusual complex of tmRNA⅐EF-Tu⅐GDP might represent a new type of RNA⅐EF-Tu interaction. To study this further, we analyzed this complex in the next experiments in more detail.
The Interaction of tmRNA with EF-Tu in the tmRNA⅐EF-Tu⅐GDP Complex-Cross-linking experiments were performed with sU containing tmRNA and purified EF-Tu⅐GDP. Around 20% of the input tmRNA was found in the cross-linked product in different experiments indicating a highly efficient cross-linking reaction. The complex corresponding to the upper band in Fig.  1A, lane 2, was isolated from the gel and used for primerextension analysis. Non-cross-linked tmRNA from the same irradiated sample, isolated from the same lane of the gel (but corresponding to the lower band) was used as a control. The results are shown in Fig. 2. Specific stops of the reverse transcriptase were observed at the positions C269 and C270 (Fig. 2A, lane X) and at A309 (Fig. 2B, lane X). Therefore, the corresponding cross-linking sites are U268 and U308, respectively. The bands seen in lane K ( Fig. 2A) are probably internal cross-linked sites, if they are stronger than the corresponding band in the X lane (cross-linked tmRNA). The bands corresponding to crosslink sites did not show a band at the same position in the control lane.
Bases of the tmRNA that revealed altered reactivity against the modifying reagent CMCT upon complex formation with EF-Tu⅐GDP were identified in footprinting experiments. CMCT reacts specifically with N3 of the uracil base and N1 of the guanine base. The most obvious changes in the reactivity of the nucleotide bases were detected in the helix 2 of tmRNA. The results are shown in Fig. 3. The reactivity of the base G315 is strongly decreased whereas that of U311 is only weakly reduced corresponding to reverse transcriptase stops at A316 and A312, respectively (Fig. 3, lane 3) as compared with tmRNA in solution (lane 4). Note the weak band at position 316 also in the control (lane 1 of Fig. 3) that was not always seen and was always significantly weaker than in the lane with the modified tmRNA. The two bases G315 and U311 are protected in the complex. The reactivity of the base U308 is increased, indicating that this base is more exposed in the complex than in isolated tmRNA in solution (lane 3, reverse transcriptase stops at position A309, compare with lane 4). U268 is present in a pseudoknot structure; it is only accessible to CMCT under semi-denaturing conditions, and V1 RNase strongly cleaves at this nucleotide (33,38). a Met-ox, oxidized methionine; Cys-am, acrylamide-modified cysteine.

TABLE II Association constants for EF-Tu and tmRNA and Ala-and Phe-tRNA
For nitrocellulose, binding constants were determined with nitrocellulose filtration method (see "Experimental Procedures"). For gel-shift, gel-shift experiments were performed as described previously (44). For Phe-tRNA, note that the constants are usually smaller (up to one order of magnitude) when measured via gel shift as compared to nitrocellulose filter measurement (45).   The cross-linking and footprinting data presented here were seen reproducibly in several independent experiments.

DISCUSSION
The elongation factor EF-Tu within ternary complex aminoacyl-tRNA⅐EF-Tu⅐GTP increases the affinity of an aminoacyl-tRNA (aa-tRNA) to the ribosomal A site by at least two orders of magnitude (34). EF-Tu interacts with the short arm of an aa-tRNA that comprises the acceptor stem and the T loop-stem structure (35). Hydrolysis protection of alanylated tmRNA by EF-Tu indicates that this factor also interacts with alanylated tmRNA at the end of the acceptor stem (13).
A productive interaction of a ternary complex aa-tRNA⅐EF-Tu⅐GTP with the A site results in GTP cleavage and, in most cases, an incorporation of the aminoacyl residue into the nascent polypeptide chain. Such a productive interaction has to fulfill a requirement that cannot be accomplished by the tmRNA; a cognate or at least a near-cognate anticodon is a prerequisite, whereas a non-cognate ternary complex cannot productively interact with the A site. However, tmRNA lacks an anticodon, yet one of its features is that, in the absence of a codon-anticodon interaction, its alanyl residue is incorporated into the nascent polypeptide chain. It follows that either the EF-Tu interaction with tmRNA differs in important aspects from that with aa-tRNA and/or additional factors are involved in the productive interaction of tmRNA with the A site. Therefore, we set out to determine molecular partner(s) of tmRNA by applying a cross-linking approach.
sU is a well known cross-linking reagent used in studies of RNA⅐RNA and RNA⅐protein interactions (17). This reagent can be easily incorporated into RNA molecules by in vitro T7transcription, and incorporation of sU residues does not significantly affect the structure of the RNA molecules nor their biological activity (36). Also, mild UV irradiation used for the activation of sU does not destroy RNA, proteins, or their activity. For example, it is known that substitution of ϳ20% of U by sU in 5 S rRNA does not affect its ability to be reconstituted into active 50 S ribosomal particles (25). In our case, tmRNA with about 5 (Ϯ2) sU residues per molecule was used to avoid multiple cross-linking.
tmRNA⅐protein complex formation in S100 extract was demonstrated by band-mobility shift analysis. The high efficiency of the cross-linking reaction, with about 20% of the input tmRNA, indicated that a specific complex is formed in the S100 fraction with a significant affinity. The cross-linked protein was found to be EF-Tu. Under the conditions of complex formation, alanine was not added, so the amount of charged tmRNA present was negligible. This means that only uncharged tmRNA participated in the cross-linking to EF-Tu. The tmRNA without the 3Јterminal CA or ACCA could be cross-linked as efficiently as intact transcribed tmRNA, confirming the observation that alanylation of tmRNA is not required for complex formation. The specific complex tmRNA⅐EF-Tu formed in a mixture of tmRNA and S100 enzymes; surprisingly, the formation of this complex did not depend on the presence of GTP, i.e. the same yield of cross-linked products was obtained in the presence of GDP. At this point we did not know whether the presence of EF-Tu is sufficient for the complex formation or whether there is another protein such as SmpB that helps the formation of the complex of deacylated tmRNA with EF-Tu. The protein SmpB has been reported to interact directly with tmRNA (14). However, we did not identify this protein in the cross-linked complex. Finally, control experiments demonstrated that purified EF-Tu⅐GDP is sufficient for complex formation, as shown with both crosslinking experiments and determination of binding constants that were similar in the presence of GDP or GTP (Table II).
The specificity of our assays is demonstrated by the fact that, with deacylated tRNA and EF-Tu⅐GTP or with aminoacyl-tRNA and EF-Tu⅐GDP, no complexes were observed. This contrasts with deacylated tmRNA and EF-Tu⅐GDP either with or without an intact CCA-3Ј end. This raises the question, what about the specificity of the tmRNA complexes? Could it be that this complex is an artifact only involving an inactive conformer of deacylated tmRNA that has no physiological relevance? The answer is no, if we accept that the alanylated form represents an active conformer of tmRNA. As described under "Results," Ala-tmRNA can participate in cross-link formation with EF-Tu⅐GDP in the same way as deacylated tmRNA. That the cross-link (i) occurs in high yields of 20% of the input tmRNA, (ii) involves distinct nucleotides, and (iii) can be demonstrated also with Ala-tmRNA strongly indicates the physiological relevance of the complex. We further note that the complexes described here have been formed under near in vivo conditions (see Ref. 37 for discussion).
Another point concerns the use of transcribed tmRNA: the two modifications found in native tmRNAs (m5U342 and ⌿347) are lacking but have minor effects upon the overall stability of tmRNA and might destabilize the acceptor stem (6). At present, it is not known whether the absence of these modifications influences the formation of this unusual complex with EF-Tu.
It is conceivable that EF-Tu⅐GTP delivers Ala-tmRNA to the A site, because it protects the Ala-tmRNA from hydrolysis as an aminoacyl-tRNA (13). However, the existence of a stable EF-Tu⅐GDP complex with tmRNA was unexpected and required closer inspection. In particular, we assessed possible contacts of tmRNA with EF-Tu in tmRNA⅐EF-Tu⅐GDP complexes using a combination of chemical footprinting and photoaffinity cross-linking techniques. Footprinting data identified one region of tmRNA protected from chemical modifications in the complex. The protected nucleotide bases (U311, G315) belong to helix 2 (nomenclature follows that of Refs. 38 and 39). This helix is the most conserved element of the secondary structure of tmRNAs (40), thus indicating the significance of its role, which is still to be identified. The crosslink to EF-Tu takes place at U308, which is within a threenucleotide bulge (Fig. 4), and this arrangement might be related with the increased reactivity toward CMCT observed at U308 upon complex formation (Fig. 3). This nucleotide is conserved in ϳ90% of known tmRNAs, and the next nucleotide, G307, is completely conserved in all known sequences (40). It follows that the interaction of EF-Tu⅐GDP with deacylated tmRNA outside the tRNA module at helix 2 ( Fig. 4) is strikingly different to that of EF-Tu⅐GTP in a canonical ternary complex with aminoacyl-tRNA.
Nucleotide residues of tmRNA protected by EF-Tu⅐GDP from chemical modification are near or at the ACCGA sequence (positions 312 through 316) that is also found in the sequence of five RNA-aptamers, which were selected for effective binding to both EF-Tu⅐GTP and EF-Tu⅐GDP from T. thermophilus (41). Domain II of EF-Tu was proposed to interact with these aptamers. This sequence is also present in the longest universally conserved sequence of the rRNAs, namely the ␣-sarcin loop of 23 S rRNA. It was shown that EF-Tu protects several nucleotide residues adjacent to ACCGA sequence from chemical modification (42), although binding of EF-Tu to the ␣-sarcin loop was never demonstrated directly. However, the ACCGA sequence is not conserved in tmRNAs, and the corresponding sequence in T. thermophilus reads, for example, AGGCC.
U268, another nucleotide residue of tmRNA found to be cross-linked to EF-Tu⅐GDP, belongs to the pseudoknot pK4 in tmRNA (Fig. 3). This structural element is present in many tmRNAs, although it is not absolutely conserved and can be destroyed without loss of tagging functions (31). This nucleotide residue is a part of a double-stranded element. It is known that 4-thiouridine residues participating in Watson-Crick base pairing could hardly form a cross-link after photo-activation. Its participation in the UV-inducible cross-linking reaction indicates, therefore, that interaction of EF-Tu with this tmRNA region alters its conformation. Taken together, the cross-linking data suggest that both helix 2 (together with the bulged nucleotides) and pseudoknot pK4 are folded around EF-Tu in the tmRNA⅐EF-Tu⅐GDP complex.
The functional role of the interaction of EF-Tu⅐GDP with uncharged tmRNA is unknown. EF-Tu may provide the proper folding of tmRNA and protect it from degradation. tmRNA is constitutively synthesized in the cell; however, it is required only under specific conditions. Hence, the bulk of tmRNA has to be stored in the cell, probably in a complex with EF-Tu⅐GDP/ GTP. Another not necessarily alternative possibility is that the EF-Tu⅐GDP complex may be involved in the release of tmRNA from the E site of the ribosome after completion of the tagsequence synthesis or in the protection of the tRNA module of tmRNA during the decoding of the mRNA module of this molecule. Note that, after reading two or three of the codons present on the mRNA module of tmRNA, the tRNA moiety of tmRNA has left the ribosome calling for protection by a EF-Tu⅐GDP/GTP complex as long as the translation of the tmRNA continues. It is noteworthy that eukaryotic eEF-1␣⅐GDP can form a stable complex with deacylated tRNA that might be involved in transporting the deacylated tRNA from the E site to the corresponding synthetase (43).