Colicin E3 cleavage of 16S rRNA impairs decoding and accelerates tRNA translocation on Escherichia coli ribosomes

The cytotoxin colicin E3 targets the 30S subunit of bacterial ribosomes and specifically cleaves 16S rRNA at the decoding centre, thereby inhibiting translation. Although the cleavage site is well known, it is not clear which step of translation is inhibited. We studied the effects of colicin E3 cleavage on ribosome functions by analysing individual steps of protein synthesis. We find that the cleavage affects predominantly the elongation step. The inhibitory effect of colicin E3 cleavage originates from the accumulation of sequential impaired decoding events, each of which results in low occupancy of the A site and, consequently, decreasing yield of elongating peptide. The accumulation leads to an almost complete halt of translation after reading of a few codons. The cleavage of 16S rRNA does not impair monitoring of codon–anticodon complexes or GTPase activation during elongation-factor Tu-dependent binding of aminoacyl-tRNA, but decreases the stability of the codon–recognition complex and slows down aminoacyl-tRNA accommodation in the A site. The tRNA–mRNA translocation is faster on colicin E3-cleaved than on intact ribosomes and is less sensitive to inhibition by the antibiotic viomycin.


Introduction
During times of stress, some strains of Escherichia coli produce the plasmid-encoded cytotoxin colicin E3. It is ã 60 kDa protein with three domains that are required for the toxin to enter and kill the cell. The receptor binding domain binds to the outer membrane protein, BtuB, of competing E. coli (Imajoh et al., 1982), and the translocation domain then interacts with the Tol complex of proteins involved in the maintenance of outer membrane stability (Garinot-Schneider et al., 1997;Bouveret et al., 1998). The cytotoxic domain then interacts directly with the bacterial inner membrane and is finally transported to the cytoplasm, possibly through the action of FtsH (Mosbahi et al., 2004;Walker et al., 2007). To prevent the producing cell from committing suicide, an inhibitor protein known as the immunity protein Im3 is expressed with the toxin. Im3 forms a complex with the cytotoxic domain of colicin E3 with a binding affinity of 10 -12 M and inactivates the RNase (Jakes and Zinder, 1974a;Kleanthous and Walker, 2001;Walker et al., 2003).
Once inside the cell, the E3 rRNase (cytotoxic) domain specifically cleaves the phosphodiester bond between A1493 and G1494 of 16S rRNA of the small ribosomal subunit (30S) and abolishes protein synthesis (Senior and Holland, 1971). The cleavage site is situated in the decoding site (A site) of the ribosome in helix 44 of 16S rRNA and is adjacent to two universally conserved adenine residues, 1492 and 1493, which are involved in cognate tRNA selection. In recent years, the colicin E3 rRNase domain has been isolated and purified, and its structure and function studied in detail. The crystal structure of the rRNase domain of colicin E3 in complex with Im3 was solved (Carr et al., 2000), providing insight into its function. The interaction between E3 and Im3 was also studied biochemically (Walker et al., 2003) and mutational analysis has identified residues which are presumed to take part in rRNase activity . The reaction requires the catalytic triad of histidine, glutamic acid and aspartic acid, which all act in general acid-base catalysis of the hydrolysis of the A1493-G1494 phosphodiester bond. Im3 was found to bind the rRNase domain at a position distant from the active site ('exosite'). The active site of colicin E3 is not blocked by the immunity protein; rather, it is believed that Im3 inhibits the E3 rRNase by steric and electrostatic occlusion, thereby abolishing colicin E3 rRNase binding to the ribosome.
Previous studies of the mechanism of colicin E3 action on the ribosome focused on the structure of ribosomes, experimental conditions required for cleavage to occur and the specific cleavage site (Boon, 1972;Bowman, 1972;Dahlberg et al., 1973;Kaufmann and Zamir, 1973;1975;Ohno-Iwashita and Imahori, 1977;Jakes and Zinder, 1974b). Very few investigations into the stage of translational arrest have been performed. Cloacin DF13, which targets Enterobacter cloacae and has a very similar mode of action as colicin E3, has been studied in more detail. It was reported that cloacin DF13-treated ribosomes were able to form initiation complexes, but the function of translation initiation factor 1 (IF1) was altered because it could no longer increase the rate of exchange between ribosome subunits or the rate and extent of IF3mediated dissociation of 70S ribosomes (Baan et al., 1978b). Although polypeptide chain elongation could take place once the 16S rRNA is cleaved, the binding of the aminoacyl-tRNA (aa-tRNA) to the A site of the ribosome was reduced (Baan et al., 1978a). The cleaved ribosomes exhibited slightly increased translational fidelity and produced polypeptides that were shorter than those on control ribosomes, although the length of the peptides was not estimated (Twilt et al., 1979). In the case of colicin E3, it is not clear whether the initiation complex can be formed once the cleavage has taken place. It was reported that colicin E3-treated ribosomes were unable to form initiation complexes (Tai and Davis, 1974;Tai, 1975), whereas the initiation complex was formed when natural mRNA was used (Kaufmann and Zamir, 1975). The effect of colicin E3 on other steps of translation was not studied.
Here we examine which elemental steps of translation are affected by colicin E3 cleavage, using a reconstituted in vitro system in which each step of translation can be followed in real time by rapid kinetics. Our data suggest that the inhibitory effect of colicin E3 originates from impaired decoding. In contrast, tRNA-mRNA translocation through the ribosome becomes faster upon cleavage, and other steps, including initiation and peptide bond formation, are not significantly affected.

Cleavage of 16S rRNA by colicin E3
To test the extent and the specificity of 16S rRNA cleavage by colicin E3, 70S ribosomes were incubated for 30 min at room temperature with different concentrations of colicin E3, and the rRNA was extracted and analysed by denaturing gel electrophoresis (see Experimental procedures) ( Fig. 1). The treatment resulted in an rRNA fragment of 50 nucleotides, which did not appear in the absence of colicin E3 (lane 1). The appearance of the fragment was observed even at the lowest concentration of colicin E3 tested (Fig. 1); according to RT-PCR, cleavage of 16S rRNA was nearly complete (95%, data not shown). The cleavage was specific, as indicated by the distinct bands of the colicin E3 fragment, 5S rRNA and 16/23S rRNA. The same results were obtained when 70S initiation complexes carrying mRNA and [ 3 H]fMet-tRNA fMet in the P site were used (data not shown). In all following experiments, equimolar concentrations of colicin E3 and 70S ribosomes were used (as in lane 5, Fig. 1). Unless indicated otherwise, after colicin E3 treatment, Im3 was added to the reaction mixture to remove colicin E3 from the ribosome. This colicin E3-Im3 interaction is very strong with picomolar affinity (Walker et al., 2003), and Im3 completely removes colicin E3 rRNase from the ribosome, as measured with fluorescence-labelled colicin E3 (Lancaster, 2005).

Overall translation
Colicin E3-treated ribosomes were tested in a translation assay reconstituted from purified components (Buskiewicz et al., 2004) (Fig. 2), using a truncated mRNA coding for the first 75 amino acids of leader peptidase (Lep). The mRNA construct does not contain a stop codon at the end of the mRNA and termination of translation does not take place. Therefore, any effects of colicin E3 on translation would imply initiation or elongation of translation as targets for inhibition. Ribosomes treated with colicin E3 showed no detectable translation activity, while control ribosomes produced the expected amount of protein (maximum one peptide chain per ribosomes can be obtained), indicating that one or more elemental steps during initiation or elongation of translation are impaired by colicin E3 treatment. The complete inhibition of translation is consistent with practically complete cleavage of 16S rRNA indicated by RT-PCR analysis (data not shown).

Initiation
To identify the step that is impaired by the cleavage of 16S rRNA, colicin E3-treated ribosomes were tested for their ability to perform factor-dependent initiation (Fig. 3) (Calogero et al., 1988;Tomsic et al., 2000). Initiation with colicin E3-treated ribosomes was somewhat slower and less efficient, reaching~75% compared with control ribosomes. The effect was observed regardless whether colicin E3 was removed from the ribosomes by adding the immunity protein Im3 or not. However, the effect was moderate and cannot account for the complete block of translation after colicin E3 cleavage (Fig. 2), indicating that other steps of translation must be affected.

Decoding
The elongation step of protein synthesis comprises decoding of the mRNA codon in the A site, peptide bond formation and tRNA-mRNA translocation from the A site to the P site. To test whether the E3-promoted cleavage and/or the presence of colicin E3 in the A site affect tRNA binding, ribosomes were incubated with ternary complexes consisting of elongation-factor Tu (EF-Tu)·GTP·[ 14 C]Phe-tRNA Phe , and the amount of [ 14 C]Phe-tRNA Phe bound to the A site was determined by nitrocellulose filtration   (Fig. 4A). Control ribosomes bound [ 14 C]Phe-tRNA Phe rapidly with more than 90% efficiency. Ribosomes treated with colicin E3 followed by removal of the RNase with the immunity protein Im3 showed equally efficient A-site binding, although at a reduced rate. When colicin E3 was not removed from the ribosomes by immunity protein Im3, not only the rate of A-site binding was strongly affected, but the binding was also much less efficient, reaching a plateau at about 40%. This effect is in agreement with the notion that colicin E3 and [ 14 C]Phe-tRNA Phe compete for the A site (Kaufmann and Zamir, 1973). For further experiments, only ribosomes treated with colicin E3 followed by removal of E3 with immunity protein Im3 were used.
The EF-Tu-dependent A-site binding of aa-tRNA involves several elemental steps, such as docking of the ternary complex EF-Tu·GTP·aa-tRNA to the ribosome, formation of the codon-recognition complex, GTP hydrolysis by EF-Tu and accommodation of aa-tRNA in the A site . GTP hydrolysis -as measured using an indicator reaction monitoring free Pi released from EF-Tu -is not inhibited by colicin E3 cleavage of 16S rRNA (Fig. 4B). Given that GTP hydrolysis is equally efficient on colicin E3-treated and control ribosomes, the preceding steps, i.e. binding of the ternary complex to the ribosome and codon recognition, are also not impaired appreciably. However, this does not exclude that the affinity of ternary complex binding to the ribosome may be altered. This was tested using the EF-Tu mutant EF-Tu(H84A) which is deficient in GTP hydrolysis (Daviter et al., 2003). Because GTP hydrolysis is blocked, ternary complex binding is reversible and -provided that the complex does not dissociate during nitrocellulose filtration -the equilibrium dissociation constant (K d) of the codonrecognition complex can be measured. Titration of 70S initiation complexes with EF-Tu(H84A)·GTP·[ 14 C]Phe-tRNA Phe gives Kd = 3 Ϯ 0.4 nM for control ribosomes, consistent with previous measurements (Daviter et al., 2003). In contrast, no significant binding of ternary complex to initiation complexes containing colicin E3-treated ribosomes can be detected within the concentration range   (Fig. 4C), indicating that the stability of the codon-recognition complex is strongly decreased upon colicin E3 cleavage. It should be noted that nitrocellulose filtration is a non-equilibrium method, implying that unstable complexes may dissociate upon washing of the filters.
Another effect on A-site binding was observed when the rate of fMetPhe dipeptide formation was monitored by quench-flow (Fig. 4D). Colicin E3-treated ribosomes were 15-fold slower in dipeptide formation (0.1 s -1 compared with 1.5 s -1 on control ribosomes), and the yield of dipeptide was reduced to about 70% compared with control ribosomes. It is unlikely that the peptidyl transfer reaction per se was affected by the cleavage of 16S rRNA, because the peptidyl transferase centre is located on the 50S subunit and the rate of fMet-tRNA fMet reaction with the aa-tRNA analogue puromycin (Pmn) was not altered (about 0.5 s -1 ; data not shown). The decreased rate of fMetPhe formation is therefore likely to be caused by a decreased rate of Phe-tRNA accommodation in the A site, whereas the lower final level indicated drop-off of part of the Phe-tRNA Phe from the ribosome during the accommodation phase.
Once [ 14 C]Phe-tRNA Phe was accommodated and peptide bond formation had taken place, the stability of the fMetPhe-tRNA Phe binding was similar on colicin E3-treated and control ribosomes (Fig. 4E). This was tested by the addition of a slight excess of initiation complexes to ternary complexes to ensure full binding of the ternary complex and fMetPhe-tRNA formation. The resulting complex was incubated at 37°C and samples taken at different time points were analysed by nitrocellulose filtration. Control and colicin E3-treated ribosomes showed the same slow loss of fMet[ 14 C]Phe-tRNA Phe from the A site (Peske et al., 2004), i.e. E3 cleavage did not destabilize peptidyl-tRNA binding in the A site.

Translocation
The effect of colicin E3 cleavage on translocation was tested first by quench-flow, monitoring the reaction of peptidyl-tRNA with Pmn (Katunin et al., 2002a). Pretranslocation complexes with f[ 3 H]Met[ 14 C]Phe-tRNA Phe in the A site and deacylated tRNA fMet in the P site were mixed with excess EF-G and Pmn. Peptidyl-tRNA in the A site does not react with Pmn (Fig. 5A). However, as soon as  (Semenkov et al., 2000). Same symbols as above.
f[ 3 H]Met[ 14 C]Phe-tRNA Phe is translocated to the P site, the reaction with Pmn takes place rapidly [10 s -1 with fMetPhe-tRNA Phe at the present conditions (Katunin et al., 2002b)], resulting in the formation of f[ 3 H]Met[ 14 C]Phe-Pmn, which can be determined by (high-performance liquid chromatography) HPLC. The rate of translocation on control ribosomes was about 10 s -1 (see below); therefore, the overall rate of coupled translocation and Pmn reaction is expected to be about 5 s -1 . It should be noted that a rapid Pmn reaction is indicative of full translocation of tRNAs and mRNA on both ribosomal subunits, because partially translocated peptidyl-tRNA [in the hybrid A/P state (Moazed and Noller, 1989)] reacts with Pmn 500 times more slowly (Sharma et al., 2004). The same rate of translocation, 5 Ϯ 1 s -1 , was found on colicin E3-treated and untreated ribosomes (Fig. 5A). This suggests that the 16S rRNA cleavage does not significantly impair translocation. However, given the limited rate of the indicator Pmn reaction, a moderate stimulation or inhibition of translocation is not excluded by these results.
Rapid kinetics of translocation can be followed by the fluorescence of fluorophores attached either to the tRNA (Robertson and Wintermeyer, 1981;Robertson et al., 1986;Rodnina et al., 1997;Pan et al., 2007) or to the 3′ end of the mRNA (Studer et al., 2003;Peske et al., 2004). The complexes of colicin E3-treated ribosomes with fluorescence-labelled mRNA were of insufficient stability for kinetic experiments (data not shown). Therefore, single-round translocation was measured using proflavin in fMetPhe-tRNA Phe (Prf16/17) as reporter group, and the fluorescence changes during translocation were monitored by fluorescence stopped-flow as described previously (Rodnina et al., 1997). On control ribosomes, translocation resulted in a fluorescence increase taking place at a rate of 10 Ϯ 2 s -1 (Fig. 5B), as seen in previous experiments (Rodnina et al., 1997). The time-course of translocation on colicin E3-cleaved ribosomes showed very fast translocation with a rate of 110 Ϯ 10 s -1 , implying that translocation is accelerated by the cleavage in 16S rRNA. Furthermore, the antibiotic viomycin, which completely blocks tRNA movement on native ribosomes (Rodnina et al., 1997), does not abolish translocation on the ribosomes with the colicin E3 cleavage. In fact, on colicin E3-cleaved ribosomes, the end point of the Pmn reaction was the same (data not shown) in the absence and presence of viomycin, whereas the reaction rate was 4.8 Ϯ 0.1 s -1 in the presence of viomycin (Fig. 5B).
To determine whether the GTPase activity of EF-G was affected by colicin E3 cleavage, Pi release from EF-G was measured using a limiting amount of ribosomes and excess EF-G as described (Savelsbergh et al., 2003) (Fig. 5C). At these conditions, the first round of Pi release is observed as a burst phase (reflected in a fluorescence increase lasting up to 0.1 s), followed by multiple rounds of GTP hydrolysis, and Pi release (further fluorescence increase). That colicin E3 cleavage had no effect on EF-G turnover was substantiated by measuring the GTPase activity of EF-G using catalytic amounts of factor and excess ribosomes (Fig. 5D). The initial velocities of multiple-turnover GTP hydrolysis were identical on colicin E3-treated and control ribosomes, k cat = 4 Ϯ 1 s -1 .

Pentapeptide synthesis
The results shown so far suggest that the strongest effect of colicin E3 cleavage is on A-site binding of ternary complex or aa-tRNA, whereas the effects on initiation, peptide bond formation and translocation are either small or cannot explain the inhibition of translation. To identify the step at which translation is halted, an assay was devised to monitor the stepwise synthesis of a pentapeptide. Initiation complexes with f[ 3 H]Met-tRNA fMet in the P site were prepared on an mRNA coding for the pentapeptide fMetValTyrIlePhe (Fig. 6A). In the first sample, initiation complexes were incubated with a 1.1 excess of [ 14 C]Val-tRNA Val and unlabelled Tyr-tRNA Tyr , Ile-tRNA Ile and Phe-tRNA Phe in the presence of EF-Tu·GTP and EF-G·GTP and the binding of f[ 3 H]Met-and [ 14 C]Val-tRNA was measured by nitrocellulose filtration. The fraction of ribosomes containing Val-tRNA Val in the A site was determined by the amount of 14 C radioactivity per initiation complex (Fig. 6B). The results show that [ 14 C]Val-tRNA Val binding to the A site was reduced by 50% on colicin E3-cleaved compared with control ribosomes. Because binding of Phe-tRNA to the A site was not appreciably inhibited (Fig. 4A), the reduction of Val-tRNA binding in the presence of the other ternary complexes may indicate that competition between ternary complexes inhibits the reaction, although variations in abilities of individual aa-tRNAs to bind to cleaved ribosomes cannot be excluded. The second sample contained the same components, except that now Tyr was 14 C-labelled to determine whether the dipeptide was translocated to allow the next aa-tRNA to enter the A site. The experiment was continued for the following two amino acids in the pentapeptide, i.e. with labelled Ile and then Phe (Fig. 6A) to determine if colicin E3-cleaved ribosomes could produce tetra-and pentapeptide, respectively. The results for the binding of the second ternary complex, Tyr-tRNA Tyr , show again that translocation is not inhibited by colicin E3 cleavage. If an inhibition of translocation played a role in translation inhibition, then very little binding of Tyr-tRNA Tyr would have been observed. However, 40% of colicin E3-cleaved ribosomes were able to bind Tyr-tRNA Tyr to produce, at least, a tripeptide. In the next steps, even less [ 14 C]Ile and [ 14 C]Phe were bound to the ribosome, and only about 15 (Ϯ5)% of [ 14 C]Val or [ 14 C]Phe were found in the pentapeptide fraction. By extrapolating these results to longer peptides, one can expect that essentially no A. Experimental approach. Initiation complexes (1 mM) were mixed with four different purified aa-tRNAs (1.1 mM each) in the presence of EF-Tu (1.65 mM), EF-G (2 mM), GTP (1 mM), phosphoenolpyruvate (3 mM) and pyruvate kinase (0.1 mg ml -1 ). In each cycle, only one aa-tRNA was 14 C-labelled. B. Binding of aa-tRNA (grey bars, monitored by nitrocellulose filtration) and formation of the pentapeptide (black bars, analysed by HPLC) on colicin E3-cleaved ribosomes, normalized to the respective values obtained with control ribosomes (white bars). About 80% of control ribosomes bound each aa-tRNA and formed the pentapeptide fMetValTyrIlePhe. Labels indicate the position in the peptide of the labelled amino acid being analysed in each row. nd, not determined.
peptides longer than about eight amino acid residues will be formed when the ribosomes are cleaved by colicin E3. The progressive decrease in the amount of ribosomebound aa-tRNA shows that defects in A-site binding, particularly in the presence of competing ternary complexes, is the major reason for the halt of protein synthesis.

Effect of colicin E3 on translation
Colicin E3 catalyses the cleavage of the phosphodiester bond between nucleotides 1493 and 1494 of 16S rRNA of E. coli ribosomes. Cleavage leads to a complete inhibition of protein synthesis in the cell and to cell death. Previous experiments, performed both in vivo and in cell lysates, have shown that protein synthesis is abolished when colicin E3 is present (Bowman et al., 1971;Senior and Holland, 1971;Walker et al., 2003). Using a translation system reconstituted from purified components for translation initiation and elongation (Buskiewicz et al., 2004), we show here that protein synthesis is abolished at the elongation step of translation. The initiation step is affected very little, as it is slighty slower and 30% less efficient with ribosomes treated with colicin E3. Similar effects were observed earlier with ribosomes cleaved by cloacin DF13 and attributed to an impaired function of IF1 (Baan et al., 1978b). Inhibition of IF3 or IF2 abolishes initiation on 70S ribosomes; therefore, a much larger effect on initiation would be expected if 16S rRNA cleavage would affect the functions of the latter two factors directly. The known activities of IF1 are: (i) to increase the binding of IF2 and fMet-tRNA fMet to the 30S subunit and (ii) to mediate the function of IF3 in the control of initiation fidelity. In fact, the inhibition of initiation by cloacin DF13 could be overcome by the addition of excess IF3 (Baan et al., 1978b), suggesting that the IF1-IF3 interaction was impaired by 16S rRNA cleavage. However, the effects on initiation are too small to account for the observed complete shutdown of protein synthesis. Thus, the overall effect on translation must be due to the inhibition of step(s) during elongation.
The elongation step of protein synthesis entails EF-Tudependent binding of aa-tRNA to the A site, peptide bond formation and translocation. Our data show that peptide bond formation and translocation are not inhibited by colicin E3 cleavage (Fig. 5A). The main defect is observed during decoding. When colicin E3 remains bound to the A site in the absence of Im3, the amount of aa-tRNA that binds to the A site is reduced by 60%, probably because of the competition between aa-tRNA and colicin E3 for binding to the decoding centre. When colicin E3 is removed from the ribosomes after the cleavage, the effects are more subtle. The stability of the codon-recognition complex is dramatically reduced and the rate of aa-tRNA accommodation in the A site is decreased 200-fold. When a specific ternary complex is present in large excess, colicin E3 cleavage does not prevent aa-tRNA binding and, most, but not all, of initially bound aa-tRNA accommodates properly to participate in peptide bond formation (Fig. 4D). In the presence of a mixture of different ternary complexes, which compete for the A site, the effect of colicin E3 cleavage is more pronounced. However, also in the latter case, the decrease in aa-tRNA binding did not exceed a factor of two in one round of elongation (Fig. 6). However, the accumulation of moderate binding and/or accommodation defects over several rounds of elongation leads to almost complete inhibition in the synthesis of peptides of five amino acids or longer. These results suggest that amino acid incorporation is reduced to zero after the translation of about eight codons. The conclusion is that within the cell, only short peptides would be produced by colicin E3-cleaved ribosomes, leading to cell death.

Implications for the mechanisms of decoding and translocation
In addition to its importance for bacterial physiology, colicin E3 provides a tool to obtain insight into the dynamic properties of the decoding centre. The colicin E3 cleavage occurs in helix 44 of 16S rRNA next to two adenine residues, 1492 and 1493, which are universally conserved in all organisms. Upon codon recognition by cognate aa-tRNA, A1492 and A1493 change their positions to interact with the minor grove of the first two base pairs of the codon-anticodon complex in a fashion which is specific for Watson-Crick geometry, but does not depend on sequence (Ogle et al., 2001). The local conformational changes in the decoding site result in a global rearrangement of the head and shoulder of the 30S subunit, which assume a closed configuration upon formation of the correct codon-anticodon duplex (Ogle and Ramakrishnan, 2005). Most likely, 16S rRNA cleavage between residues 1493 and 1494 does not impair the ability of the ribosome to select cognate codon-anticodon complexes, because the ribosomes with cleaved 16S rRNA were found to be highly accurate in aa-tRNA selection (Twilt et al., 1979). Similarly, we observed, by HPLC analysis, that upon synthesis of pentapeptides (Fig. 6), the last amino acid in the sequence (Phe) was incorporated exclusively at the position defined by the respective cognate codon and was not found in dipeptides where it could have been misincorporated at a valine codon (data not shown). This implies that A1492 and A1493 in the cleaved 16S rRNA retain the ability to monitor the codonanticodon complex. Furthermore, Pi release from EF-Tu was unaffected, indicating that also the preceding steps of ternary complex binding to the ribosome, i.e. codon reading and GTP hydrolysis, proceeded with essentially unchanged rates irrespective of the cleavage. This supports the notion that after colicin E3 cleavage, A1492 and A1493 retain the ability to interact with the codonanticodon complex, because disruption of the A-minor interactions results in slow GTP hydrolysis (Gromadski et al., 2006;Cochella et al., 2007). Furthermore, the global closing of the 30S subunit may also be expected to take place, as the restriction of the flexibility within the 30S subunit, e.g. by binding streptomycin (Carter et al., 2000), renders GTPase slow even on cognate codons (Gromadski and Rodnina, 2004a).
On the other hand, our data show that colicin E3 cleavage abolishes the stabilization of cognate aa-tRNA in the codon recognition complex, which is observed with intact ribosomes (Ogle et al., 2002;Daviter et al., 2003;Gromadski and Rodnina, 2004b), or during aa-tRNA accommodation into the A site. Interestingly, the main destabilizing effect was observed with the codonrecognition complex or during aa-tRNA accommodation in the A site. In both complexes, the codon-anticodon interaction comprises the major contact of aa-tRNA to the ribosome, while the 3′ end of the tRNA is either still bound to EF-Tu or moves through the ribosome. Essentially no effect on the stability of peptidyl-tRNA binding to the ribosome was observed at later stages, i.e. after the accommodation of aa-tRNA in the A site, possibly as a result of compensatory tRNA contacts outside the decoding region.
The adverse effects of 16S rRNA cleavage on GTPase activation and the stabilization of the codon-recognition complex imply that the two reactions are not necessarily coupled through rearrangements at the decoding site. This is supported by earlier observations indicating that the binding of paromomycin to the decoding centre, which stabilized the flipped-out conformation of A1492 and 1493 even in the absence of cognate codon-anticodon interaction (Ogle et al., 2001), induced rapid GTP hydrolysis in a ternary complex bound to a near-cognate codon . At the same time, the stability of the respective codon-recognition complexes was hardly changed Ogle et al., 2002).
The structural reason for the reduced stability of codonrecognition and proofreading complexes with colicin E3-cleaved ribosomes is unclear. One possible explanation may be a change in the structure of the bridge 2a, which is formed by elements of 16S rRNA (helix 44, positions 1494-1495, 1408-1410) and 23S rRNA (helix 69, positions 1913-1914, 1918) (Yusupov et al., 2001), and is one of the contact points between the ribosome and aa-tRNA in the A site (Yusupov et al., 2001;Stark et al., 2002;Valle et al., 2003). If the cleavage between positions 1493 and 1494 changes the structure of the contact site at position 1494, the formation or dynamics of the bridge may be impaired, thus inhibiting the correct interaction with aa-tRNA, which would explain the low stability of the complex. Deletion of helix 69 of 23S rRNA, which destroys the 50S side of bridge 2a, did not impair poly(Phe) synthesis, but increased the fidelity of aa-tRNA selection by a factor of two (Ali et al., 2006), which reminds of the fidelity effect of 16S rRNA cleavage (Twilt et al., 1979). The effect of helix 69 deletion on aa-tRNA binding during codon recognition and proofreading was not examined (Ali et al., 2006).
Another possible explanation for the low stability of the codon-recognition and proofreading complexes involves the notion that the tRNA is mechanically distorted upon decoding (Stark et al., 2002;Valle et al., 2003;Yarus et al., 2003;Daviter et al., 2005). In the codon recognition complex, the aa-tRNA undergoes a conformational change (Stark et al., 2002;Valle et al., 2003), which presumably is coupled to or required for GTPase activation (Rodnina et al., 1994a;Piepenburg et al., 2000;Cochella and Green, 2005). The distorted tRNA conformation is stabilized by the contacts of the tRNA with the ribosome and EF-Tu (Stark et al., 2002;Valle et al., 2003), and conformational rigidity of the decoding centre may be required to maintain the high stability of the complex. The cleavage between positions 1493 and 1494 may render the decoding site less rigid and therefore unable to withstand the strain exercised by the mechanic tension in the tRNA molecule. The resulting relaxation in the tRNA structure may then lead to a loss of contacts with the ribosome, thereby decreasing the stability of the complex. During initial selection of aa-tRNA, weaker binding of cognate aa-tRNA in the A site has little effect on decoding, because the reaction is driven by the rapid forward reactions of GTPase activation and GTP hydrolysis (Rodnina and Wintermeyer, 2001). However, a very low stability of near-cognate complexes, in combination with a low rate of GTPase activation (Gromadski and Rodnina, 2004b), may result in hyperaccurate ribosomes, as has been reported (Twilt et al., 1979).
Furthermore, compromised mechanical properties of the decoding site may be responsible for the loss of a portion of aa-tRNA at the proofreading stage, when the 3′ end of aa-tRNA moves from its position on EF-Tu into the peptidyl transferase centre of the ribosome (Gromadski and Rodnina, 2004b). This step requires tight interactions of the tRNA with the ribosome and entails conformational changes of aa-tRNA (Sanbonmatsu et al., 2005); increased flexibility of the decoding centre may impair the movement. In fact, aa-tRNA accommodation is slowed down 15 times by colicin E3 cleavage, and a relatively high proportion (30%) of cognate aa-tRNA is lost during proofreading (Fig. 4D). Thus, the cleavage of 16S rRNA between positions 1493 and 1494 does not appear to impair monitoring of codon-anticodon complexes or GTPase activation, but affects the stability of aa-tRNA binding at critical stages of decoding and its ability to move within the ribosome, probably by increasing the flexibility or altering the positions of elements at the decoding region.
Another reaction which is affected by the cleavage in 16S rRNA is translocation, which was 10 times faster and less sensitive to inhibition by viomycin on colicin E3-treated than on intact ribosomes. Translocation is catalysed by EF-G and entails the movement of the tRNA-mRNA complex through the ribosome. Our data suggest that colicin E3 cleavage accelerated the movement of tRNA through the ribosome, whereas the rate of Pi release from EF-G after GTP hydrolysis was not changed. On native ribosomes, the rates of Pi release and tRNA movement are limited to 30 s -1 by a ribosome rearrangement termed unlocking (Savelsbergh et al., 2003). In contrast, tRNA translocation on the cleaved ribosomes is faster (110 s -1 ) than unlocking (and translocation) on untreated ribosomes. This suggests that the cleavage of 16S rRNA leads to uncoupling of tRNA movement from unlocking. In other words, the step which usually limits the rate of tRNA movement during translocation appears to depend on the structure and/or dynamics of the decoding site.
The ribosomes cleaved by colicin E3 are capable of translocation even in the presence of viomycin, although at a lower rate than in the absence of the antibiotic, whereas on intact ribosomes, the translocation is completely abolished by the drug. There are two (nonexclusive) mechanisms by which viomycin may block translocation: it greatly stabilizes tRNA in the A site (Peske et al., 2004) and blocks ribosomes in the ratcheted state with tRNAs in their hybrid A/P and P/E positions (Ermolenko et al., 2007). Both mechanisms may lead to an inhibition of translocation by stabilizing, respectively, the ground state of the pre-translocation complex or an intermediate which is transient otherwise. However, both the stability of peptidyl-tRNA in the A site (Fig. 4E) and the ratcheting movements of the ribosome (see Discussion above) are not altered by colicin E3 cleavage, and it is therefore unlikely that the lowered sensitivity to viomycin is due to differences at these stages. Alternatively, viomycin may restrict the movements of the ribosome in a way which abolishes not only intersubunit movements, but also the flexibility within the 30S subunit. Cleavage of 16S rRNA may increase the mobility of the 30S subunit head and render the ribosomes flexible even in the presence of viomycin, thus allowing for slow but complete translocation in the presence of the antibiotic. While the structural consequences of the 16S rRNA cleavage remain speculative, the present data suggest that alterations at the decoding centre that affect the conformational flexibility are crucial for translocation and determine the rate of tRNA movement through the ribosome.

Buffers and reagents
All experiments were carried out in buffer A (50 mM pH 7.5,70 mM NH4Cl,30 mM KCl,7 mM MgCl2) at 37°C. In vitro translation was performed in buffer B (50 mM Tris-HCl, pH 7.5, 70 mM NH4Cl, 30 mM KCl, 3 mM MgCl2). GTP, phosphoenolpyruvate and pyruvate kinase were from Roche Diagnostics. Radioactive compounds were from MP Biomedicals or Hartmann Analytics. All other chemicals were from Merck.

Ribosomes, mRNAs, tRNAs and factors
70S ribosomes from E. coli MRE 600 were prepared as described   -purified), EF-Tu, EF-G, EF-Tu(H84A) and initiation factors were prepared as described (Rodnina et al., 1994a,b;1999;Daviter et al., 2003). MFTI-mRNA with the sequence GGCAAGGAGGUAAAUA AUGUUCACGAUU where the start codon is underlined was purchased from Curevac (Germany). The mRNA used for the pentapeptide synthesis was 134 nt long, had the coding sequence indicated in Fig. 6A and was prepared by T7 RNApolymerase transcription. The preparation of mRNA coding for 75 N-terminal amino acids of leader peptidase (Lep75 mRNA) is described elsewhere (Bornemann et al., 2008).

Expression and purification of colicin E3 and immunity protein Im3
Colicin E3 rRNase was overexpressed as a complex with His-tagged immunity protein Im3. After purification of the complex by Ni-NTA affinity chromatography, colicin E3 was separated from Im3 by denaturation with 6 M guanidinium chloride where the Im3 remains bound to the Ni-NTA resin while the E3 rRNase is eluted. Isolated colicin E3 was refolded by dialysis against water, followed by dialysis against 50 mM potassium-phosphate (pH 7.0). Final purification was by gel filtration on a Superdex 75 column (Amersham-Pharmacia-Biotech) in the same buffer. Purified colicin E3 was dialysed against water, lyophilized and stored at -20°C (Walker et al., 2003). Im3 was extracted from the overexpression cell culture by sonication and the cell debris removed by centrifugation. Ammonium sulphate was added to the cell lysate up to 45% saturation. The supernatant was dialysed against 50 mM Tris-HCl pH 7.5, 1 mM EDTA, 1 mM DTT and 50 mM KCl. Im3 was purified further by anion exchange chromatography on DE-52 (50 ml column) using a gradient of 50 mM-1 M KCl in the same buffer. Gel filtration on Superdex 75 (Amersham-Pharmacia-Biotech), equilibrated with 50 mM potassium phosphate, pH 7, was used for final purification. Purified Im3 was lyophilized in 50 mM potassium phosphate, pH 7, and stored at -20°C.

Analysis of colicin E3 cleavage of 16S rRNA
Ribosomes were incubated in buffer A with colicin E3 (for concentrations see text) at room temperature for 30 min. The cleavage reaction was stopped by the addition of Im3 fol-lowed by ethanol precipitation in 0.3 M sodium acetate, pH 5.2. After washing with 70% ethanol, the pellets were re-suspended in 3 M sodium acetate, pH 5.2, containing 0.5% SDS, and incubated for 10 min at 37°C. Proteins were removed by extracting the solution three times with watersaturated phenol followed by chloroform extraction. The rRNA was ethanol-precipitated, re-suspended in water and analysed by denaturing polyacrylamide gel electrophoresis.

Biochemical assays
Ribosomes (1 mM) were incubated in buffer A with colicin E3 (1 mM) (E3-treated ribosomes) or without the rRNase (control ribosomes) for 30 min at room temperature. Im3 (1.1 mM) was added to each sample and incubated for another 5 min. These ribosomes were used for the biochemical assays.
Total translation was assayed as described (Buskiewicz et al., 2004). Ribosomes (0.5 mM) were mixed in buffer B with Lep75 mRNA (2 mM), initiation factors (0.85 mM), GTP (1 mM) and f[ 3 H]Met-tRNA fMet (1 mM) to form initiation complexes. Ternary complex was prepared in buffer B with putrescine (6.4 mM), phosphoenolpyruvate (3 mM), GTP (1 mM) and EF-Tu (50 mM) and incubated for 15 min at 37°C. Total aa-tRNA (21 mM) containing [ 14 C]Leu-tRNA Leu was added to EF-Tu, and the mixture was added to initiated ribosomes. After incubation at 37°C for various times, samples were removed and the amount of protein synthesized was determined by TCA precipitation and quantifying radioactivity.
To synthesize pentapeptides, 70S ribosomes (1 mM) were initiated with mRNA coding for MetValTyrIlePhe (3 mM) as described above and [ 14 C]Val-tRNA Val ·GTP· EF-Tu, Tyr-tRNA Tyr ·GTP·EF-Tu, Ile-tRNA Ile ·GTP·EF-Tu, Phe-tRNA Phe ·GTP·EF-Tu (1.1 mM each), EF-G (2 mM) and GTP (1 mM) were added; the reaction was incubated for 10 min at 37°C. A total of 10 pmol of ribosome complex was removed and analysed by nitrocellulose filtration or HPLC peptide analysis. The experiments were carried out in parallel with 14 C-labelled Tyr, Ile or Phe instead of Val in the presence of the other four unlabelled aa-tRNAs.

Rapid kinetics
Fluorescence stopped-flow measurements were performed and data analysed as described previously (Rodnina et al., 1997;Savelsbergh et al., 2003). For translocation experi-ments, pre-translocation complexes were prepared by mixing initiation and ternary complexes, resulting in ribosomes carrying deacylated tRNA fMet in the P site and fMetPhe-tRNA Phe (Prf16/17) in the A site. Pre-translocation complexes (0.1 mM) were mixed rapidly with an equal volume of EF-G (2 mM after mixing) and GTP (1 mM) in buffer A at 37°C. Proflavin fluorescence was excited at 470 nm and measured after passing a cut-off filter (KV500, Schott).
Pi release from EF-G and EF-Tu was measured using the fluorescence change of phosphate-binding protein labelled with MDCC (2.5 mM) (Brune et al., 1994;Savelsbergh et al., 2003). For EF-G, equal volumes of pre-translocation complex (0.1 mM final) and EF-G (2 mM) with GTP (200 mM) in buffer A were mixed at 37°C. In the case of EF-Tu, equal volumes of initiation complex (0.1 mM final concentration) and EF-Tu·GTP·Phe-tRNA Phe (2 mM final concentration) in buffer A were mixed in buffer A at 37°C. MDCC fluorescence was excited at 425 nm and measured after passing a cut-off filter (KV450, Schott).
Kinetics of peptide bond formation were measured by quench-flow using a Kintek RQF-3 apparatus. Purified 70S initiation complexes (0.4 mM) were mixed with a threefold excess of ternary complex, then the reaction was quenched by 0.8 M KOH (Gromadski and Rodnina, 2004b). Samples were hydrolysed for 30 min at 37°C and neutralized with acetic acid. The amount of fMetPhe formed was determined by reversed-phase HPLC (Lichrospher 100 RP-8, Merck). The rate of translocation was determined by the reaction of translocated peptidyl-tRNA with Pmn, as monitored by quench-flow (Katunin et al., 2002a). Pre-translocation complexes (0.25 mM) were mixed with EF-G (3 mM) and Pmn (10 mM); the reactions were stopped and samples treated as described above. The amount of f[ 3 H]Met[ 14 C]Phe-Pmn was analysed by reversed-phase HPLC.