The A-site Finger in 23 S rRNA Acts as a Functional Attenuator for Translocation*

Helix 38 (H38) in 23 S rRNA, which is known as the “A-site finger (ASF),” is located in the intersubunit space of the ribosomal 50 S subunit and, together with protein S13 in the 30 S subunit, it forms bridge B1a. It is known that throughout the decoding process, ASF interacts directly with the A-site tRNA. Bridge B1a becomes disrupted by the ratchet-like rotation of the 30 S subunit relative to the 50 S subunit. This occurs in association with elongation factor G (EF-G)-catalyzed translocation. To further characterize the functional role(s) of ASF, variants of Escherichia coli ribosomes with a shortened ASF were constructed. The E. coli strain bearing such ASF-shortened ribosomes had a normal growth rate but enhanced +1 frameshift activity. ASF-shortened ribosomes showed normal subunit association but higher activity in poly(U)-dependent polyphenylalanine synthesis than the wild type (WT) ribosome at limited EF-G concentrations. In contrast, other ribosome variants with shortened bridge-forming helices 34 and 68 showed weak subunit association and less efficient translational activity than the WT ribosome. Thus, the higher translational activity of ASF-shortened ribosomes is caused by the disruption of bridge B1a and is not due to weakened subunit association. Single round translocation analyses clearly demonstrated that the ASF-shortened ribosomes have higher translocation activity than the WT ribosome. These observations indicate that the intrinsic translocation activity of ribosomes is greater than that usually observed in the WT ribosome and that ASF is a functional attenuator for translocation that serves to maintain the reading frame.

involved in translocation. On the basis of these observations, we discuss the likely functional role the ASF plays in translocation.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Cultivation-E. coli ⌬7rrn strain TA542 (⌬rrnE ⌬rrnB ⌬rrnA ⌬rrnH ⌬rrnG::cat ⌬rrnC::cat ⌬rrnD::cat ⌬recA56/pTRNA66 pHKrrnC) (12) was kindly provided by Dr. Catharine L. Squires (Tufts University). The rescue plasmid pRB101 (13) was constructed by introducing the SacB gene and rrnB operon into pMW118 (Amp r ) (Nippon gene). The plasmid pRB102 (13) was constructed from pMW218 (Km r ) (Nippon gene) by inserting the rrnB operon only. The plasmid pHKrrnC in strain TA542 was replaced by pRB101 to generate strain NT101, which was used as the host cell to construct variant strains in this study. Cells were grown at 37°C in 2ϫ Luria-Bertani (LB) (2ϫ LB, 2% tryptone, 1% yeast Extract and 1% NaCl) medium; for solid medium, 1.5% agar was added. Antibiotics were added at the following concentrations when required: 40 g/ml spectinomycin, 100 g/ml ampicillin, and 50 g/ml kanamycin. To replace the plasmid in NT101, 5% sucrose was added to the LB medium. NT102 is a series of E. coli strains in which pRB101 has been replaced by pRB102 or its derivatives. A preculture aliquot (A 600 ϭ ϳ1.0 -1.2) was inoculated into 1.5 ml of 2ϫ LB medium. After being separated into five multiple aliquots, the growth rate of the NT102 variants was determined by measuring the A 600 every 30 min by using a Spectramax 190 plate reader (Molecular Device, Inc.).
After the PCR, the reaction mixture was treated with DpnI (New England Biolabs) to digest the template plasmid and then cleaned up with a QIA-Quick column (Qiagen). NT101 was then transformed with the various resulting plasmids, selected and confirmed as described (13,14). The sequences of the helices were confirmed by direct sequencing of the plasmids. The regions in the helices that were deleted are shown in gray in Fig. 1A.
Preparation of Ribosomes and Elongation Factors-Ribosomes were prepared from variant strains as described (15) with slight modifications. Each E. coli strain grown up to A 600 0.5 was harvested, ground with Al 2 O 3 , then dissolved in a buffer consisting of 20 mM Hepes-KOH (pH 7.6), 30 mM NH 4 Cl, 10 mM Mg(OAc) 2 , and 6 mM 2-mercaptoethanol. The lysate was subjected to ultracentrifugation to obtain the crude ribosome. 70 S tight-coupled ribosomes were then purified from the crude ribosome preparation by 14 -16 h of ultracentrifugation in a 6 -38% (w/v) sucrose density gradient (SDG) as described (16). Recombinant E. coli elongation factors EF-G and EF-Tu were prepared as described (17).
SDG Profile Analysis to Determine Subunit Association-Ultracentrifugation of ribosome preparations was performed as described above for the preparation of the 70 S ribosomes. The SDG fractions were separated by a fractionator (Biocomp Instrument, Inc.), and the amounts of 50 S and 70 S in each fraction were estimated by measuring the A 260 . Association ratio was calculated as described (14). Total 50 S was calculated by adding the area of the 50 S peak to 63.4% of the 70 S peak. The fraction of 50 S in the 70 S fraction can be obtained by dividing 63.4% of 70 S peak by total 50 S.
Estimation of tRNA Binding to the A-and P-sites and Peptidyltransferase Activity-A-and P-site tRNA binding experiments were basically performed as described (14,19). Prior to the A-site binding, the P-site of poly(U)-programmed ribosomes (20 pmol) were occupied by deacyl-tRNA Phe (40 pmol). [ 14 C]Phe-tRNA Phe (15 pmol) was then added to the mixture and incubated at 37°C for 15 min to bind to the A-site. The reaction mixture was spotted onto a nitrocellulose filter (0.45 M, Advantech) and washed with a 5 ml buffer consisting of 50 mM Tris-HCl (pH 7.5), 6.5 mM MgCl 2 , 60 mM KCl, 1 mM dithiothreitol, and 0.5 mM spermine. The dried filter was then subjected to liquid scintillation counting. For P-site binding, 2.5 pmol of acetyl-[ 14 C]Phe-tRNA Phe (2.5 pmol) was mixed with 70 S ribosome (12.5 pmol) and poly(U) (2.5 g) and incubated at 37°C for 15min. An aliquot was then diluted twice, spotted and washed as described above. The ribosome-bound acetyl-[ 14 C]Phe-tRNA Phe was quantified by liquid scintillation counting. Afterward, puromycin (final concentration: 2 mM) was added to the rest of the reaction mixture to examine peptidyltransferase activity. After 2 h on ice, the reaction was stopped by adding an equal volume of solution containing 250 mg/ml MgSO 4 and 0.3 M NaOAc (pH 5.5). acetyl-[ 14 C]Phe-puromycin was extracted by 1 ml of ethyl acetate, and 700 l of the upper layer was suspended in Ultima Gold (PerkinElmer Life Sciences) and counted by a liquid scintillation counter.
Translocation Assay-Toeprinting assays were used to determine the extent of translocation (22). Ribosomes (400 nM), tRNA f Met (800 nM), and ASL6 Phe (6.0 M) were prepared individually in buffer containing 80 mM potassium cacodylate (pH 7.2), 20 mM MgCl 2 , 150 mM NH 4 Cl, and 3 mM 2-mercaptoethanol. The ribosomes were activated and preprogrammed as described previously (23,24). ASL6 Phe was then added to the preprogrammed ribosome mix and A-site binding was allowed to occur for 30 min at 37°C, followed by 20 min on ice. Sample mixtures were left at room temperature for 10 min. EF-G and GTP were prepared in binding buffer and added (final concentrations of 80 nM and 2 M per reaction, respectively) to bring the final reaction volume to 25 l. Sample mixtures were incubated with EF-G at room temperature. 2-l aliquots were removed and placed on ice before the addition of EF-G for a background translocation control, and at each time point thereafter (30 s and 2, 5, 10, 30, and 60 min). Avian myeloblastosis virus-reverse transcriptase was added to each aliquot and primer extension was carried out at 37°C for 10 min. Extension products corresponding to the pre-and post-translocation complexes were separated on a 10% denaturing polyacrylamide gel and visualized by autoradiography. A PhosphorImager (GE Healthcare) was used to quantify the intensity of the bands seen on the gel. Total counts within each lane were the sum of the counts for the pre-and post-bands. The fraction of translocated ribosomes is equal to the counts for the post-band divided by the total counts. All toeprint data shown are the result of at least three independent experiments.

Construction of Ribosome Variants with Shortened Bridge-
forming Helices-To analyze the functional importance of the ASF in 23 S rRNA, we employed E. coli strain NT101, which lacks all chromosomal rRNA operons and is rescued by the plasmid pRB101, which harbors the rrnB and sacB genes (13,14). Manipulation of the bridge-forming helices in 23 S rRNA was carried out on plasmid pRB102, which encodes rrnB and has the same replication origin as pRB101 but bears a different selection marker. We constructed ribosome variants with shortened bridge-forming helices, namely, H38/ASF, H34 and H68, which form the B1a, B4, and B7a bridges, respectively (Fig.  1AB). The deletions were intended to disrupt each intersubunit bridge. Thus, H38/ASF in 23 S rRNA was shortened by deleting 26 bases (⌬876 -901) from the tip and connecting both ends by inserting the GAGA tetra loop; this yielded the H38d22 variant (Fig. 1A). During the writing of this paper, Dontsova's group reported a study with a deletion mutant of the ASF that lacks 34 bases (⌬872-905) (25). Consequently, we constructed a comparable variant (H38d34) for comparison. As controls, variant strains with shortened H34 (H34d) and H68 (H68d) in 23 S rRNA were also constructed (Fig. 1A). Thus, H34 was shortened by 4 bases (⌬709 -710,721-722), while H68 was shortened  OCTOBER 27, 2006 • VOLUME 281 • NUMBER 43 by 47 bases (⌬1845-1895, GGAA insertion). NT101 was transformed with each pRB102 variant and selected by kanamycin and sucrose. Each plasmid displaced the rescue plasmid pRB101 to generate cells showing sucrose resistance and ampicillin sensitivity. The resultant strains can thus survive with only a ribosome variant with a shortened rRNA helix. This indicates that the various deletions in the bridge-forming helices did not abolish minimal ribosomal function.

Truncation of Helix 38 in 23 S rRNA Elevates Translocation Activity
A Shortened ASF Does Not Affect Subunit Association-The growth rate of H38d22 and other strains was measured (Table  1). Judging from the doubling time, each deletion in the bridgeforming helix did affect the growth rate to some extent. As shown in Table 1, H38d22 showed only a small reduction in the growth rate (58.6 min versus 52 min for the WT strain), whereas the H68d variant showed a severe growth defect (108 min). The H38d34 and H34d variants showed intermediate reductions in the growth rate (67.1 and 61.7 min, respectively). Thus, the mutations in the intersubunit bridges decreased ribosome functionality, which indicates that these bridges do play some functional roles.
The efficacy of subunit association in each ribosome variant was measured by SDG centrifugation. The association ratio was determined by measuring the amounts of 70 S and free 50 S. When 10 mM Mg 2ϩ was used, 71% of all the 50 S subunits in the WT strain were incorporated into tight-coupled (TC) 70 S ribosomes (Table 1). H38d22 (70%) and H38d34 (67%) had similar TC ratios. In contrast, the H34d and the H68d variants had lower TC ratios (40 and 54%, respectively). The extreme reduction in the growth rate of the H68d variant may thus be explained by its weak subunit association. In any case, it appears that disruption of bridge B1a by truncating the ASF does not affect subunit association.
ASF-shortened Ribosomes Show Enhanced ϩ1 Frameshift Activity-Since the ASF directly interacts with A-site tRNA, we examined whether truncation of the ASF would perturb the accuracy of mRNA decoding. To do this, we measured the ability of ribosomes to translate ␤-galactosidase mRNAs containing engineered frameshift sites or stop codons. Thus, each strain bearing ribosomes with shortened H38 was transformed with a series of reporter plasmids encoding ␤-galactosidase bearing either N-terminal ϩ1 or Ϫ1 frameshift sites or premature stop codons (20). An E. coli strain with error prone mutations (C912G and G885U) in 16 S rRNA that results in significant read-through frequency for the UGA and UAG stop codons (26) was employed as a positive control. In addition, an E. coli strain with high fidelity mutations (C912G and G888U) served as a positive control for the ϩ1 frameshift phenotype (26). As shown in Table 2, the H38d22-and H38d34-bearing variants showed enhanced ϩ1 frameshift activity compared with the WT strain, while their read-through activity of UGA and UAG codons was relatively lower than that of the WT. H38d22 showed higher ϩ1 frameshift activity than H38d34. However, concerning Ϫ1 frameshift activities, both H38d22 and H38d34 did not exhibit significant differences from the WT. These results demonstrated that ASF truncation results in enhanced ϩ1 frameshift activity.
ASF-shortened Ribosomes Exhibit Enhanced Translation Activity at Limited EF-G Concentrations-During EF-G-catalyzed translocation, bridge B1a becomes disrupted and the relative orientation of the ASF and S13 alters significantly (8). To further determine the function of the ASF in translation, we examined the translational activity of the ASF-shortened ribosome H38d22 in an in vitro translation reaction with varying concentration of EF-G (0 -125 nM). The initial velocity of polyuridine (poly(U))-dependent polyphenylalanine (poly(Phe)) synthesis of each ribosome variant is plotted in Fig. 2A. H38d22 showed higher poly(U)-poly(Phe) synthesis than the WT but only at limited EF-G concentration. In contrast, both the H34d and H68d variants showed lower activity than the WT throughout the full range of EF-G concentration. At higher EF-G concentration (200ϳ600 nM), H38d22 and WT showed equivalent translational activity (data not shown, see Fig. 2B). Moreover, with regard to poly(U)-poly(Phe) synthesis in the presence of sufficient levels of EF-G (600 nM) and varying concentrations of EF-Tu (Fig. 2B), H38d22 showed similar translational activity to the WT. We also found the H38d34 variant had higher translational activity than the WT (Fig. 2C), although to a lesser degree than H38d22. Thus, ASF truncation enhances translation activity at limited EF-G concentration.
The GTPase activity of EF-G is known to be highly stimulated by vacant ribosomes (27,28). We thus next examined the ability of each ribosome variant to stimulate EF-G-related GTPase activity. As shown in Fig. 2D, the initial GTPase activity was elevated more efficiently by H38d22. In contrast, the H34d and H68d variants showed a comparable GTPase activity with WT ribosome.
ASF Truncation Stimulates Translocation-Since the higher translocation activity of the ASF-shortened ribosome was

of E. coli strains bearing ribosome variants and their subunit association
The doubling time of each strain was calculated from five independent aliquots. All variants were assessed together except for H38d34, whose doubling time was measured independently from the other variants. The doubling time of its WT control was 49.8 Ϯ 0.9. The subunit association ratio was calculated from the SDG profiles in the presence of 10 mM magnesium ion. The height of each peak in the SDG profile was measured in the UV trace at 260 nm. The amount of 50 S fraction that was contained in 70 S was calculated as being 63.4% of the 70 S peak. Total 50 S levels were calculated by adding the height of the 50 S peak to the 50 S fraction in the 70 S peak. The association ratio was then obtained by dividing the 50 S fraction in 70 S by the total 50 S levels.  observed only at limited EF-G concentrations, we next assessed the single round translocation of the ASF-shortened ribosomes compared with the WT ribosome. Prior to performing the translocation assay, we assessed the tRNA binding and peptidyl transfer activity of H38d22. Non-enzymatic binding of phenylalanyl (Phe)-tRNA Phe to A-site of H38d22 and wild type ribosomes was performed. As shown in Fig. 3A, H38d22 and WT ribosomes did not differ significantly in the binding affinity of Phe-tRNA Phe to the A-site. Next, P-site tRNA binding of H38d22 was compared with the wild type ribosome. H38d22 and WT ribosomes showed a similar binding affinity of acetyl-phenylalanyl (AcPhe)-tRNA Phe to the P-site (Fig. 3B). Then, puromycin was added to estimate peptidyl transfer reaction of H38d22. About 20% of AcPhe was released from the ribosome as AcPhe-puromycin in this reaction. By counting AcPhepuromycin released from the ribosomes, no significant difference in the reactivity of puromycin was observed (Fig. 3C). These results revealed that H38d22 has similar activities for tRNA binding and peptidyl transfer reaction to the wild type ribosome. We next measured the translocation activity of the ASF-shortened ribosome by performing a toeprinting assay. In this assay, the P-site is prefilled with deacyl-tRNA fMet and the A-site is occupied by the anticodon stem loop analog, ASL6 Phe . Preprogrammed ribosomes were incubated with EF-G to induce translocation or without EF-G as a control. The pre-and post-translocation mixes were subjected to primer extension analysis to measure the distance between the ribosome and the primer, which hybridizes to the downstream point of the mRNA. As shown in Fig. 4, both H38d22 and H38d34 showed faster translocation than the WT ribosome in the presence of both low (Fig. 4A) and high (Fig. 4B) concentrations of EF-G. While the initial velocity of translocation for both ASF-shortened and WT ribosomes was accelerated by high concentrations of EF-G (Fig. 4B), the enhanced activity of the ASF-shortened ribosomes was maintained. These results suggest that ASF truncation enhances translocation at each step. Taken together with the poly(U)-poly(Phe) synthesis results ( Fig. 2A), it appears that H38/ASF in the WT ribosome acts as a negative regulator of EF-G turnover during translocation.

DISCUSSION
H38/ASF in 23 S rRNA is a component of the intersubunit bridge B1a. In this study, we analyzed the role H38/ASF plays in translation by constructing E. coli strains that bear the ASFshortened ribosomes. We found these strains were all healthy and had a normal growth rate. They did show moderate ϩ1 frameshift activity, but the ASF truncation did not affect subunit association. This was unlike the other ribosomes with shortened H34 or H68 bridge-forming helices, which showed reduced association. These observations suggest that the ASF and bridge B1a are unlikely to play an important role in subunit association, rather, they probably have another function in translation. By using an in vitro translation reaction, we then discovered that the extent of translocation by H38d22 ribosome is greater than that of WT ribosome when EF-G is present FIGURE 2. In vitro translation and EF-G-related GTPase stimulatory activity of ribosome variants with shortened bridge-forming helices. A-C, initial velocity of poly(U)-poly(Phe) synthesis. In A, the EF-G concentrations were titrated and EF-Tu was present at 600 nM. In B, the EF-Tu concentrations were titrated and EF-G was present at 600 nM. In C, the productivity of poly(U)-poly(Phe) synthesis of ASF-shortened ribosomes performed with 10 nM EF-G for 10 min was determined. D, EF-G-related GTPase activation stimulated by ribosome variants with shortened bridge-forming helices. The P i released from [␥-32 P]GTP was resolved by TLC and quantified by a fluoro imager (Fujifilm BAS5000). The WT and H34d, H68d, and H38d22 variants are represented by black squares, white squares, black circles, and white circles, as in A and B.  OCTOBER 27, 2006 • VOLUME 281 • NUMBER 43 at limited concentration, although both translate equally well when more EF-G (200 -600 nM) is present. This implies that the ASF-shortened ribosome interacts better with EF-G during the elongation cycle or turns it over more quickly than the WT ribosome. Supporting this is that H38d22 stimulated EF-G-related GTPase activity significantly better than WT ribosome, although its molecular event is different from the case in actual translocation. Single round translocation followed by toeprinting analysis then clearly demonstrated that the ASF-shortened ribosomes show greater translocation activity than WT ribosome even when high concentrations of EF-G (2 M) were present. Notably, at this EF-G concentration, ASF-shortened and WT ribosomes did not differ in an in vitro translation activity (data not shown). These observations together suggest that the more efficient translation of the ASF-shortened ribosomes at limited EF-G concentration is due to a more rapid turnover of EF-G. As EF-G and the ASF have not been observed to interact directly upon EF-G binding (3,8), a plausible interpretation of our data is that truncation of ASF results in a lowering of the energy required to convert a ribosome from a pre-translocation to a post-translocation state. Furthermore, we found that although the ASF interacts with the elbow region of A-site tRNA (1), ASF-truncation did not affect A-site tRNA binding. In addition, H38d22 showed normal P-site tRNA binding and peptidyl transfer activity. These observations together thus suggest that H38/ASF is mainly involved in EF-G-catalyzed translocation. However, since the ASF-shortened ribosomes showed ϩ1 frameshift activity, it is possible that the truncation of the ASF may also affect tRNA positioning during translocation on the ribosome. Nevertheless, the observations in this study provide evidence that the intrinsic potential of translocation is actually greater than is observed in the WT ribosome, which indicates the presence of a negative regulatory mecha-nism. Thus, H38/ASF appears to be a functional region that acts as a negative regulator of translocation.

Truncation of Helix 38 in 23 S rRNA Elevates Translocation Activity
Dontsova and colleagues (25) have recently also characterized the effect of truncating the ASF (⌬872-905). They observed that their ASF truncation, which is more extensive than ours (34 bases versus 22 bases in our study), led to growth defects and less efficient subunit association. In addition, their ASF-shortened ribosome showed no difference in EF-G-related GTPase stimulation and translocation activity. The discrepancy between these results and ours may be due to several reasons. First, our genetic system is different from theirs. They used an E. coli strain lacking all chromosomal rrn operons but bearing a multicopy plasmid (ColE1 origin) containing the rrnB operon under the control of P L promoter. We used a strain (TA542 derivative) that also lacked all chromosomal rrn operons but that was rescued by a low copy number plasmid (pSC101 origin) containing the rrnB operon under the original P1P2 promoter. Given that E. coli has seven rrn operons in its genome, the complementation of the rrn operon with its original promoter by using a plasmid with a pSC101 origin may result in wild type copy numbers of the operon in the cell. Supporting this is that the doubling time of our strain (about 50 min) is somewhat shorter than that of theirs (87 min). Second, as mentioned above, the ASF of Dontsova and colleagues (25) had a longer truncation. When we constructed ribosomes bearing this longer truncation ourselves (H38d34), we found that our original mutant H38d22 had higher translocation activity while H38d34 ribosomes induced a lower growth rate (67.1 min) and a less efficient poly(U)-poly(Phe) activity at a limited EF-G concentration (Fig. 2C). However, H38d34, like H38d22, also showed efficient single round translocation (Fig. 4). Thus, despite the variations in phenotype induced by the truncation of the ASF by 34 as opposed to 22 bases, our observations indicate that ASF-truncation in general enhances translocation.
It has been observed that when ribosomes are treated with the thiol-specific reagent pCMB (p-chloromercuribenzoate), spontaneous translocation occurs (29,30), which suggests that the ribosome itself has all the equipment needed for translocation. The ribosomal protein S12 in 30 S was proposed to be the target of pCMB (31,32), and indeed, Green and colleagues (33) showed that ribosomes depleted of S12 and/or S13 exhibit spontaneous translocation. Since S13 interacts with H38/ASF to form the B1a bridge, and since S13 depletion also disrupts bridge B1a, the molecular mechanism that allows the ASFshortened ribosome to undergo faster translocation may be related somehow to the mechanism involved in spontaneous translocation. However, since the ASF and S13 both have other ribosomal functions, this relationship cannot be easily elucidated.
In eukaryotes, H38 is longer and forks into two stems, while the forked stem-loop is located in the back (solvent side) of the CP, the major stem-loop of H38 interacts with the A-site and forms bridge B1a, similar to what is found in eubacteria (34). Thus, bridge B1a and the ASF are broadly conserved in eubacteria, archea and eukaryotes. In animal mitochondrial ribosomes, on the other hand, H38 is very short and cryo electron microscopy of the bovine mitochondrial ribosome has failed to detect bridge B1a (35). Instead, the mitochondrial ribosome has another structural protrusion named the P-site finger (PSF) near the P-site region. Since ratchet-like rotation upon EF-G binding is not observed in the bovine mitochondrial ribosome (Agrawal, personal communication), translocation in mammalian mitochondrial ribosomes may involve distinct molecular mechanisms.
The swiveling of the head region of 30 S that was observed in the two crystal structures of E. coli ribosomes (7), coupled with the ratchet-like rotation that was observed to occur upon EF-G binding in cryo-EM structures (3), has revealed a part of molecular mechanics of translocation. Analysis of the two structures that comprise the pre-and post-translocation state shows that this structural change is associated with the event that the ASF alters its interacting partner from S13 to S19. Our biochemical observations described here suggest that ASFtruncation seems to cause the looseness of this partner switch as well as EF-G positioning, thereby accelerating translocation. Consequently, we propose that the ASF acts as a functional attenuator that regulates translocation for the purpose of maintaining translational fidelity.