The Role of Ribosomal Protein L11 in Class I Release Factor-mediated Translation Termination and Translational Accuracy*

It has been suggested from in vivo and cryoelectron micrographic studies that the large ribosomal subunit protein L11 and its N-terminal domain play an important role in peptide release by, in particular, the class I release factor RF1. In this work, we have studied in vitro the role of L11 in translation termination with ribosomes from a wild type strain (WT-L11), an L11 knocked-out strain (ΔL11), and an L11 N terminus truncated strain (Cter-L11). Our data show 4-6-fold reductions in termination efficiency (kcat/Km) of RF1, but not of RF2, on ΔL11 and Cter-L11 ribosomes compared with wild type. There is, at the same time, no effect of these L11 alterations on the maximal rate of ester bond cleavage by either RF1 or RF2. The rates of dissociation of RF2 but not of RF1 from the ribosome after peptide release are somewhat reduced by the L11 changes irrespective of the presence of RF3, and they cause a 2-fold decrease in the missense error. Our results suggest that the L11 modifications increase nonsense suppression at UAG codons because of the reduced termination efficiency of RF1 and that they decrease nonsense suppression at UGA codons because of a decreased missense error level.

It has been suggested from in vivo and cryoelectron micrographic studies that the large ribosomal subunit protein L11 and its N-terminal domain play an important role in peptide release by, in particular, the class I release factor RF1. In this work, we have studied in vitro the role of L11 in translation termination with ribosomes from a wild type strain (WT-L11), an L11 knocked-out strain (⌬L11), and an L11 N terminus truncated strain (Cter-L11). Our data show 4 -6fold reductions in termination efficiency (k cat /K m ) of RF1, but not of RF2, on ⌬L11 and Cter-L11 ribosomes compared with wild type. There is, at the same time, no effect of these L11 alterations on the maximal rate of ester bond cleavage by either RF1 or RF2. The rates of dissociation of RF2 but not of RF1 from the ribosome after peptide release are somewhat reduced by the L11 changes irrespective of the presence of RF3, and they cause a 2-fold decrease in the missense error. Our results suggest that the L11 modifications increase nonsense suppression at UAG codons because of the reduced termination efficiency of RF1 and that they decrease nonsense suppression at UGA codons because of a decreased missense error level.
L11 is a highly conserved ribosomal 14.8-kDa protein located at the base of the L7/L12 stalk of the ribosome, which is essential for several steps in protein synthesis (1)(2)(3)(4)(5). L11 binds to the nucleotides 1051-1108 of Escherichia coli 23 S rRNA, commonly called the L11 binding region (L11BR) 2 (6), which constitutes the GTPase-associated center, an important sector of the bacterial ribosome, where all of the translational GTPases bind and hydrolyze GTP in the course of their action (7). This is also the site of action for the thiazole antibiotics thiostrepton and micrococcin (8 -10).
In addition to the GTPases, some other translational factors interact with L11 and the L11BR in functionally important ways. The class I release factors RF1 and RF2 belong to this group. RF1 recognizes the stop codons UAG and UAA, whereas RF2 recognizes the stop codons UGA and UAA in the A site of the ribosome (11). Recent cryo-EM studies with termination complexes containing RF2 (12,13) and RF1 3 show that these two factors acquire very similar overall conformations on the ribosome. Although they bind to the decoding center on the 30 S subunit and reach up to the peptidyltransferase center on the 50 S subunit to induce release of the nascent peptide chain, they interact with L11BR (12,13) and L11 3 with their flexible domain I. These observations are in line with previous suggestions, based on biochemical and genetic experiments, that there are interactions between L11 and the release factors (4, 14 -18).
The L11 protein consists of two domains, a tightly folded N-terminal domain (NTD), which is loosely connected to the large compact C-terminal domain (CTD). The CTD of L11 is in stable contact with the L11BR RNA, whereas the NTD can change its position and proximity with respect to the rest of the protein (19). Earlier biochemical and genetic studies indicate that depletion of L11 from the ribosome inhibits RF1-mediated translation termination on a UAG codon but somewhat facilitates RF2-mediated termination on a UGA codon (4,6). Antibodies against the NTD of L11 inhibit RF1-mediated termination, indicating that this part of L11 is crucial for RF1 binding (14). In a bacterial strain, containing only the CTD of L11, efficient UAG suppression was seen as in the complete L11 knocked-out strain. This result shows that the NTD of L11 is required for proper functioning of RF1 (18).
Here, we have studied in vitro how ribosomedepletion of L11 or its NTD affects RF1-and RF2-mediated release of the tetrapeptide fMet-Phe-Thr-Ile (MFTI) from pretermination ribosomes prepared from a cell free system with E. coli components of high purity (20,21). We have also studied the effects of L11 deletion or truncation on cognate and near cognate codon reading by tRNA. Taken together, our data on termination efficiency and missense error level explain why these L11 alterations lead to enhanced nonsense suppression at UAG codons, read by RF1, and decreased nonsense suppression at UGA codons, read by RF2 (4, 6).

Chemicals and Buffers
ATP, GTP, and radioactive amino acids were purchased from Amersham Biosciences. Nonradioactive amino acids and other chemicals were from Sigma or Merck. All experiments were carried out in polymix buffer, containing 5 mM ammonium chloride, 95 mM potassium chloride, 0.5 mM calcium chloride, 8 mM putrescine, 1 mM spermidine, 5 mM potassium phosphate and 1 mM 1-4-dithioerythritol (22). . The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. prepared by in vitro T7 RNA polymerase transcription (23). The tRNA fMet purification and subsequent charging and formylation reactions were performed as described previously (24), and the bulk tRNA was purified as described previously (25). The tRNA Phe and tRNA GAG Leu isoacceptors used for missense error measurements were purified by BD-Sepharose chromatography.

Components of the in Vitro Translation
Bacterial Strains and Plasmids-The three different E. coli strains used for ribosome preparation (6,18) (Table 1) carry a chromosomal knock-out of the rplK gene, which encodes the L11 protein. In the FTP6063 and FTP6066 strains the knock-outs were complemented with the whole L11 gene or its CTD cloned in the p⌬CAT plasmid, which are referred to as pL11 and pL11Cter, respectively (18).
Ribosomes-Ribosomes from the bacterial strains were prepared using sucrose gradient zonal ultracentrifugation as described previously (26), with minor modifications. Prior to ribosome preparation we always checked the presence or absence of the whole gene or the gene for truncated L11 by PCR using primers specific for the upstream and downstream regions of rplK gene. We also had functional checks using methods described in Ref. 18. The ⌬L11 ribosomes had minor contaminations of RF1 and RF2 causing release of tetrapeptides during prolonged incubations. These contaminations were removed by splitting the ⌬L11 70 S ribosomes into 30 S and 50 S subunits, as described in Ref. 26, which were then used to reconstitute highly active 70 S ribosomes.
tRNA Synthetases and Other Translation Factors-The Ile-tRNA synthetase was isolated as described previously (27); the Phe-tRNA synthetase and the elongation factors EF-Tu, EF-Ts, EF-G were purified as described in Ref. 25; the Thr-tRNA synthetase was prepared according to Brunel et al. (28). The Met-tRNA synthetase was prepared as described in Ref. 24. RF1 and RF3 were purified following Ref. 27, and RF2-His-tagged was purified according to Ref. 29.

Formation of Ribosome Pretermination Complex
Pretermination ribosome complexes carrying fMet-Phe-Thr-Ile-tR-NA Ile in the P site and any one of the three stop codons UAA, UAG, and UGA in the A site were prepared as described previously (27,30,31). The peptide was labeled with either 3 H or 14 C isoleucine. The pretermination complexes were subjected to gel filtration using the Sephacryl S-300 matrix (Amersham Biosciences). The total ribosome concentration was determined from UV absorbance at 260 nm. The concentration of active release complex was estimated as the amount of radiolabeled tetrapeptide released after incubation with a release factor cognate to the A site codon present in excess over the ribosomes. The free tetrapeptide was identified in the supernatant after removal of the ribosomes and dropped off tRNAs by 5% trichloroacetic acid precipitation and centrifugation. The activity of different pretermination complexes was in the range from 60 to 85%.

Reaction Scheme for Class I Release Factors
Peptide release by RF1 or RF2 in the absence or presence of RF3 in large excess can be described by Reaction 1. RF is either RF1 or RF2. RC pre MFTI and RF:RC pre MFTI are pretermination ribosome complexes carrying peptidyl-tRNA MFTI in the P site with and without a bound RF, respectively. RF:RC post and RC post are posttermination ribosome complexes carrying a deacylated tRNA in the P site with and without a bound RF, respectively. The rate constants for RF association to and dissociation from the pretermination ribosome are k a and k d , respectively. The rate constants for RF association to and dissociation from the post-termination ribosome in the absence or presence of RF3 in large excess are q a and q d , respectively, and the rate constant for ester bond hydrolysis is k c . The rate-limiting step of RF recycling is dissociation from the post-termination ribosome and, hence, the factor recycling rate is approximated by q d . The k cat /K m value ("efficiency") for termination is defined as the association rate constant k a multiplied by the probability that RF binding to the pretermination ribosome leads to ester bond hydrolysis, as shown in Equation 1.

Determination of the Relative RF Efficiency in Translation Termination
The pretermination complexes with MFTI peptidyl-tRNA at the P site and any one of the three stop codons UAA, UAG, or UGA at the A site were made with ribosomes containing the full-length L11 (WT-L11), only C terminus of L11 (Cter-L11), and no L11 (⌬L11). To compare the k cat /K m values for termination at the WT-L11, Cter-L11, and ⌬L11 complexes programmed with a particular stop codon, these were allowed to compete in pairs against each other for RF1-or RF2-dependent peptide release. One competing pretermination complex carried [ 3 H]Ile and the other [ 14 C]Ile on the P site-bound peptidyl-tRNA for quantification of the amount of peptide released from each complex. Reaction mixes were prepared by combining the competing pretermination complexes to a final active concentration of 100 nM each. Preincubation of the complexes for 1 min was followed by addition of RF1 or RF2 to final concentrations between 0 and 400 nM or between 0 and 180 nM, respectively, in a reaction volume of 25 l. After a 20-s incubation at 37°C, a time sufficient for complete single round but prohibitive of  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 multiple round RF termination (27,32), the reaction was quenched by the addition of 600 l of ice-cold 5% trichloroacetic acid. The amount of released MFTI tetrapeptide in the supernatant was quantified by radiometry by utilization of the Beckman Coulter LS6500 Multi Purpose Scintillation Counter, which enabled simultaneous measurement of both 3 H and 14 C radioisotopes in separate windows that were corrected for minimal interference (27). The time evolution of the competing reactions are described by Equations 2 and 3,

Role of L11 in Class I Release Factor-mediated Termination
where Division of Equation 5 by Equation 6 and integration over the 20-s incubation time leads to Equation 7, where P c is the amount of tetrapeptide released at a given concentration of release factor, P max is the total amount of releasable tetrapeptide, and 1 Ϫ (P c /P max ) is the peptide fraction remaining on peptidyl-tRNA in the pretermination complex after the incubation.

Determination of the Rate Constant for Ester Bond Cleavage (k c ) Induced by RF1 or RF2
Pretermination complexes (RC pre MFTI ) made from WT-L11, Cter-L11, and ⌬L11 ribosomes carrying MFTI-tRNA at the P site and different stop codons at the A site were prepared separately as described earlier. Peptide hydrolysis was initiated by rapid mixing of the RC pre MFTI complex with the release factor cognate to the stop codon in a quench flow instrument (Chemical-Quench-Flow model QF-3, KinTek Corp.). The postmixing concentration of the termination complex was 200 nM, and the postmixing release factor concentrations were 0.5, 1.0, 2.0, and 3.0 M. The reaction was quenched after varying incubation times with trichloroacetic acid at a final concentration of 5% trichloroacetic acid. The extent of peptide release was quantified by radiometry as described above. At these concentrations of release factors, the rate-limiting step was ester bond hydrolysis. Accordingly, the time evolution of peptide release can be described by Equation 10.
Here, P t and P max are the amounts of released peptide at time t and at infinite time, respectively. The OriginPro fitting software was used to obtain the best fit to Equation 10 by varying parameters p and b.

Determination of the Recycling Rate (q d ) of Release Factors without and with RF3
The pretermination complexes prepared as described above with a final active concentration of 100 nM were mixed with 10 nM RF1 or RF2 depending on the stop codon used. For the reactions with RF3 the reaction mix also contained 200 nM RF3 along with 1 mM GTP, 2 mM phosphoenolpyruvate, 1 mM ATP, and 0.05 g/l pyruvate kinase in 1ϫ polymix buffer. After different times of incubation, aliquots were withdrawn, the reaction was quenched with 600 l of ice-cold 5% trichloroacetic acid, and the released tetrapeptide was quantified from the supernatant by radiometry. The amounts of spontaneously released tetrapeptide in the absence of release factors were comparatively small and were subtracted as background. Under these conditions, the release factors had to cycle many times to remove all tetrapeptides from the pretermination complexes. The rates of recycling of the factors were estimated from the slopes of the straight lines in plots of the pmol (amount) of MFTI released/pmol of release factor versus time.

Estimation of the Accuracy of Translation in Dipeptide Formation Assays
Initiated ribosomes, containing [ 3 H]fMet-tRNA fMet at the P site and a UUU codon at the A site, were prepared in an initiation mix, containing WT-L11, Cter-L11, or

Selective Effects of L11 Alterations on the Termination Efficiencies of RF1
and RF2-We have compared the kinetic efficiencies (k cat /K m ) of the release factors RF1 and RF2 in translation termination at UAA, UAG, and UGA codons in the A site of pretermination ribosomes, containing fulllength L11 (WT-L11), NTD truncated L11 (Cter-L11), and no L11 (⌬L11). Pretermination complexes with the same stop codon but containing different variants of L11 were mixed in pairs and allowed to compete for RF1 ( Fig.  1) or RF2 (Fig. 2) at varying concentrations. RF1 was used for complexes carrying either the UAA or UAG codon, whereas RF2 was used for complexes with the UAA or UGA codon. The reaction was stopped after 20 s, a time long enough for single round tetrapeptide release, but short enough to  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 prevent multiple round peptide release. To differentiate between the tetrapeptides released from the two competing complexes, we have used 14 Table 2). Our experiments show that deletion of the N-terminal domain of L11 or of the whole L11 protein leads to a reduction of the RF1 termination efficiency at both UAA and UAG codons by a factor of 4 or 6, respectively (Fig. 1, B and D, and Table 2). When the Cter-L11 and ⌬L11 complexes were competed for RF1, the ratio of k cat /K m values was 1.4 (Fig. 1, E and F,   either the UAA or UGA stop codon, the termination efficiency was reduced much less, only about 20 or 30% because of the N-terminal truncation and L11 deletion, respectively ( Fig. 2 and Table 2).

Determination of the Intrinsic Rate Constant for Ester Bond
Hydrolysis-The reduction of termination efficiency (k cat /K m ) caused by N-terminal truncation of L11 or deletion of the whole protein (Table  2) could a priori be the result of (see "Experimental Procedures") (i) a reduced intrinsic rate constant for ester bond hydrolysis (k c ); (ii) a reduced rate constant for factor association (k a ); or (iii) an increased rate constant for factor dissociation from the pretermination ribosome complex (k d ). To clarify this question, we determined k c for cognate termination by RF1 or RF2 at a high concentration (1-3 M) for all combinations of pretermination complexes (200 nM) and stop codons ( Table 3). For all of the pretermination complexes the k c values for RF1and RF2-mediated peptide release were estimated as ϳ0.55 s Ϫ1 and 2.5 s Ϫ1 , respectively (Table 3), in line with previous observations regarding wild type ribosomes (33). The data showing the fraction of tetrapeptide released with time by RF1 and RF2 (3 M) from the pretermination complexes (200 nM) with UAG and UGA stop codons, respectively, are presented in Fig. 3, A and C. The natural logarithms of the fraction of tetrapeptide released against time result in linear fits as shown in Fig. 3, B and D. These results demonstrate that neither N-terminal truncation nor complete deletion of protein L11 affects the intrinsic rate constant for RF-induced ester bond hydrolysis during termination of translation.

Termination of Protein Synthesis by Release Factors in Recycling
Mode-To study the rate of recycling of RF1 and RF2 during termination at the various pretermination ribosome complexes, either one of these was mixed in excess with a small amount of release factor. The recycling rate was then obtained from the slope of the straight line when the amount of released tetrapeptide per amount of active release factor was plotted as a function of time ( Fig. 4 and Table 4). Under these conditions, the slow dissociation of release factor from the post-termination ribosome was rate-limiting and, accordingly, the recycling rate represented the rate constant q d for dissociation of the release factor from the post-termination ribosome (Reaction 1). As a control, we also estimated the rate of spontaneous tetrapeptide release in the absence of any release factor from the pretermination complexes in an otherwise identical experimental setup, which was found negligible. Fig. 4 illustrates the recycling rates of RF1 (Fig. 4A) and RF2 (Fig. 4B) on pretermination complexes with a UAA codon in the A site. In the absence of RF3, RF1 recycled with a rate close to 3 ϫ 10 Ϫ3 s Ϫ1 for all the three ribosomal complexes (Fig. 4A and Table 4). There was essentially no difference when a UAG codon was present instead of a UAA codon (data not shown). For WT-L11 complexes programmed with a UAA RF2 recycled with a larger rate close to 9 ϫ   FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 10 Ϫ3 s Ϫ1 (Fig. 4B and Table 4). For the Cter-L11 and ⌬L11 complexes, RF2 recycled with significantly smaller rates close to 6 ϫ 10 Ϫ3 and 3.5 ϫ 10 Ϫ3 s Ϫ1 , respectively, for UAA ( Fig. 4B and Table 4) as well as for UGA codons (data not shown). When the recycling of RF1 and RF2 occurred in the presence of an excess amount of RF3 (see "Experimental Procedures"), both factors recycled much more rapidly, but the recycling rates responded to the L11 alterations in a way similar to that in the absence of RF3 (Table 4). Accuracy of mRNA Translation-Studies of termination efficiency in living cells based on nonsense suppression (6,18) are ambiguous in the sense that enhanced nonsense suppression may either be interpreted as reduced termination efficiency of a release factor or enhanced readthrough efficiency of the stop codon by a tRNA. Therefore, we checked whether a complete or an N-terminal domain deletion of the L11 protein affects the accuracy of codon reading (i.e. the missense error level) by aminoacyl-tRNA. For this, we compared the effective association rate constants k cat /K m , for a near cognate Leu-tRNA GAG Leu in ternary complex with EF-Tu and GTP when interacting with wild type or L11altered ribosome complexes programmed with UUU in the A site. The k cat /K m values were 143 Ϯ 7.0 for the WT-L11, 82 Ϯ 5.5 for the Cter-L11, and 78 Ϯ 8.6 M Ϫ1 s Ϫ1 for the ⌬L11 ribosome variant. At the same time, k cat /K m for UUU codon reading by Phe-tRNA GAA Phe was unaffected by the L11 alterations (see Fig. 5, A and B), suggesting about a 2-fold reduction in the missense error level by either one of the L11 changes.

DISCUSSION
L11 has been implicated in class I release factor-mediated translation termination since the 1970s, when scientists tried to understand the function of individual ribosomal proteins by neutralizing them with specific antibodies (34) or removing them from the ribosome with salt wash followed by reconstitution (3,4). Removal of L11 from the ribosome appeared to reduce the activity of RF1 and, curiously, enhance the activity of RF2, and this scenario was reversed by readdition of L11 to the ribosomal core (4). Similar observations were made with ribosomes from three different mutant strains of E. coli lacking L11 (14,35). When, furthermore, antibodies raised against the NTD of L11 (amino acids 1-64) were allowed to bind to wild type ribosomes, termination by RF1, but not RF2, was inhibited, suggesting that the NTD of L11 interacts directly with RF1, but not RF2 (14). In line with these observations, more recent in vivo studies demonstrated UAG suppression, growth deficiency, and temperature sensitivity for an L11 chromosomal knock-out strain (6). Overexpression of the L11-CTD from a plasmid partially reversed the growth defects but failed to reverse completely the UAG   suppression (18). Despite these findings, the mechanistic role of L11 in termination of protein synthesis and its structural basis have remained obscure.
In the present study, we have obtained precise estimates of how the k cat /K m values (efficiencies) for the interaction between the pretermination ribosome and class I release factors are affected by removal of the N-terminal domain from L11 as well as by depletion of the whole L11 protein from the ribosome. For RF1, we have found 4-fold or 6-fold reductions in the k cat /K m values by L11 truncation or removal, respectively. For RF2, there were also reductions in the k cat /K m values by these L11 alterations but much smaller and not exceeding 30% (Table 2). There was, in other words, a large reduction in the activity of RF1-dependent termination and a small but significant reduction in the activity of RF2-dependent termination. Another release factor-selective effect of these L11 alterations was an about 2-fold decrease in the rate constant for spontaneous release of RF2, but not of RF1, from the post-termination ribosome. Recent structural data on RFs in complex with pre-as well as post-termination ribosomes may have a bearing on our kinetic data.
To set the perspective for a discussion about selective effects of L11 on the kinetics of RF1 and RF2, we note that L11 is located at the GTPase-associated center of the ribosome at the base of the ribosomal stalk (36,37) quite distant from the decoding and peptidyltransferase centers, believed to be the major points of action of class I release factors (38). However, domain I of the release factors is seen in close proximity to L11 and the L11BR in recent cryo-EM reconstructions. Domain I is important for RF interaction with RF3 3 (13), and when it was removed or swapped between the RFs, this resulted in a general loss of guanine nucleotide exchange on RF3 (29) and in selective effects on the efficiencies of RF1 and RF2; although the efficiency of RF1 remained unaltered when domain I was removed, the efficiency of RF2 decreased considerably (29). These data may suggest that the interaction of domain I with the GTPase-associated center is different for the two factors, in line with recent cryo-EM observations; that is, despite an overall similarity in the trilobed cryo-EM density of RF1 and RF2 on the ribosome, there is an arc-like extra density, seen close to domain I of RF1, which is absent in the cryo-EM structure of RF2 3 (13), and attributed to the NTD of L11. 3 In contrast, domain I of RF2 appears to interact with the CTD of L11 and L11BR 3 (13). These differences in the way the two release factors interact with L11 as well as L11BR may explain their different kinetic responses to L11 truncation or deletion. However, establishment of firmer links between structural data on RFs in ribosomal complexes and the present kinetic results will require refined modeling approaches based on structures at much higher resolution than are available today.
The results of our study are compatible with in vivo data showing increased read-through of the RF1-specific UAG codon in an L11 knock-out (6) and an L11 truncated (18) E. coli strain, compared with wild type. Our results seem, at the same time, to be incompatible with in vivo data, demonstrating decreased read-through of the RF2-specific UGA codon by the same L11 alterations (6,18). However, the outcome of such in vivo experiments depends on the competition between an RF and a tRNA for stop codon reading. Therefore, increased read-through of a stop codon may either be explained by decreased efficiency of an RF or increased ability of a tRNA to read the stop codon. Decreased read-through may, according to the same logic, be explained as enhanced efficiency of a release factor or decreased ability of a tRNA to read the stop codon. We have found a 2-fold decrease in the efficiency by which Leu-tRNA GAG Leu reads the near cognate UUU, suggesting that the efficiency of near cognate stop codon reading by tRNAs was reduced by about the same factor in these read-through experiments. If correct, this would predict increased read-through of UAG and decreased read-through of UGA stop codons by the L11 alterations because the 2-fold decrease in the efficiency of near cognate tRNA reading is smaller than the 4 -6-fold decrease in termination efficiency by RF1, but larger than the 30% decrease in termination efficiency by RF2. The results of our study are also compatible with in vitro experiments suggesting that RF1-dependent termination is inhibited by removal of L11 from the ribosome, but in apparent contradiction to the further observation that RF2-dependent termination is stimulated by L11 depletion (4). It may, however, be borne in mind that these early experiments were performed with ribosome complexes bound to separate AUG and UGA triplets, which we now know is a poor model system for pretermination ribosomes (31,39).