Inhibitory Effect of Heterologous Ribosome Recycling Factor on Growth of Escherichia coli

ABSTRACT Ribosome recycling factor (RRF) of Thermotoga maritimawas expressed in Escherichia coli from the cloned T. maritima RRF gene and purified. Expression of T. maritima RRF inhibited growth of the E. coli host in a dose-dependent manner, an effect counteracted by the overexpression of E. coli RRF. T. maritima RRF also inhibited the E. coli RRF reaction in vitro. Genes encoding RRFs fromStreptococcus pneumoniae and Helicobacter pylori have been cloned, and they also impair growth of E. coli, although the inhibitory effect of these RRFs was less pronounced than that of T. maritima RRF. The amino acid sequence at positions 57 to 62, 74 to 78, 118 to 122, 154 to 160, and 172 to 176 in T. maritima RRF differed totally from that ofE. coli RRF. This suggests that these regions are important for the inhibitory effect of heterologous RRF. We further suggest that bending and stretching of the RRF molecule at the hinge between two domains may be critical for RRF activity and therefore responsible forT. maritima RRF inhibition of the E. coli RRF reaction.

Protein synthesis consists of three steps: initiation, peptide chain elongation, and termination with the release of the completed peptide chain. The step that follows these three steps, disassembly of the posttermination complex, is less well known but is essential for the next round of protein synthesis (for reviews, see references 15, 18, 20, and 21). Two factors, EF-G and ribosome recycling factor (RRF; formerly called ribosome releasing factor [19]), catalyze this step (19). In the absence of RRF, ribosomes not only remain on the mRNA but also resume unscheduled translation downstream of the termination codon (17,33).
RRF, discovered in 1970 (7) and confirmed independently in 1973 (40), is encoded by the frr gene, which was mapped to near 4 min, close to the gene encoding EF-Ts in Escherichia coli (14). The gene encoding RRF is found among all living organisms so far examined except for Archaea (see reviews listed above). Approximately 30 frr genes have so far been sequenced, partly in connection with bacterial genome sequencing projects (for example, see reference 2), and some RRFs have been characterized (22,29,32,44). In vitro protein synthesis is stimulated four-to eightfold by the addition of RRF (27,31,34). The GTP requirement (8,23), the fate of ribosomes at the termination complex in the presence of RRF (8,23), and the possible role of IF3 in certain situations (4,23,28) have been studied. In addition to its role in ribosome recycling, RRF maintains translational fidelity (18). Although the exact mechanism of RRF action remains elusive, RRF has been proposed to bind to the ribosomal A site (20) because it reduces translational error and because its action is inhibited by the antibiotics which interfere with the A site (10). RRF competes for ribosomal binding with peptide release factors that are assumed to bind to the A site (5). RRF has a nearly perfect structural similarity to tRNA (36). We therefore proposed that RRF is translocated from the A site to the P site similarly to tRNA during disassembly of the post-termination complex (36). This paper describes the cloning, expression, and purification of RRF from Thermotoga maritima, which has been used for determination of the crystal structure (36). In addition, genes coding for RRF of Streptococcus pneumoniae and Helicobacter pylori were isolated and studied. In contrast to Pseudomonas aeruginosa frr, which functions in E. coli (29), we found that these other heterologous RRFs were toxic to E. coli. We suggest that freezing at the hinge region of the T. maritima RRF structure may be partly responsible for the toxic effects. BL21(DE3)pLysS is a lysogen of lambda phage derivative DE3, which carries the gene for T7 RNA polymerase under the control of the inducible lacUV5 promoter in the chromosome. T7 RNA polymerase induced by the addition of IPTG (isopropyl-␤-D-thiogalactopyranoside) drove the expression of various frrs. BL21(DE3)pLysS contains plasmid pLysS expressing T7 lysozyme. This enzyme is a natural inhibitor of T7 RNA polymerase and reduces its ability to transcribe target genes in noninduced cells.

Strains and plasmids.
Oligonucleotide primers for PCR. All oligonucleotide primers for PCR were synthesized using Beckman Oligo 1000M. The sense and antisense primers for cloning T. maritima frr were 5Ј-AGG GGA TAC ATA TGG TTA ATC CGT TCA-3Ј and 5Ј-GCA ACG TGC TGT GGG ATC CTC AAA ATT-3Ј, respectively. The sense and antisense primers for cloning S. pneumoniae frr were 5Ј-GGA ATA AGA AAG CAT ATG GCT AAC GCA-3Ј and 5Ј-GAG TTT TTC TGT GGA TCC TTA GAC TTC-3Ј, respectively. The sense and antisense primers for cloning H. pylori frr were 5Ј-AAA AAG GAT GAA AAC ATA TGT TAC AGG-3Ј and 5Ј-CCT TAA TAT CGA ATT CTT AGA CCT TTA-3Ј. These primers were designed according to their published sequences (42) and those given in The Institute for Genomic Research (TIGR) microbial database on the World Wide Web (http://www.tigr.org/tdb/mdb/mdb.html). The NdeI, BamHI, and EcoRI sites are underlined, while the initiation codons are in boldface italics.
DNA amplification, preparation of plasmids, cloning, and sequence analysis. Genomic DNAs of T. maritima were prepared as described previously (12). Mikhail Shchepetov, University of Pennsylvania, kindly provided the genomic DNAs of S. pneumoniae. A genomic DNA fragment of H. pylori was provided by TIGR through the American Type Culture Collection. PCR amplifications of the frr DNAs of T. maritima, S. pneumoniae, and H. pylori were carried out with Taq DNA polymerase (Stratagene) using the primers described in the preceding section. The frr DNA fragments thus obtained were treated with NdeI and BamHI (T. maritima and S. pneumoniae) or NdeI and EcoRI (H. pylori).
Each frr DNA was excised at the introduced restriction enzyme sites and was ligated into pET-11a (T. maritima) or pET-24a (ϩ) (S. pneumoniae or H. pylori) to obtain pKA1 or pKA5 and pKA6, respectively. The inserts of T. maritima and S. pneumoniae frr DNA in the pET vector were sequenced by our DNA Sequencing Facility (Department of Genetics, University of Pennsylvania) with the vector-specific oligonucleotide primers (corresponding to the T7 promoter and terminator).
Expression of various heterologous frrs in E. coli. Plasmid pKA1, pKA5, pKA6, pRR3, pHSG299, or pRR2 was placed in BL21(DE3)pLysS (host), depending on the experiment. The host harboring the plasmid was grown at 37°C overnight in Luria-Bertani (LB) medium containing the appropriate antibiotic to select against the emergence of the strain without the plasmid. These antibiotics were ampicillin (50 g/ml for pKA1, pRR2, and pUC19), kanamycin (50 g/ml for pKA5, pKA6, pRR3, and pHSG299), and chloramphenicol (25 g/ml for pLysS). The cultures were diluted 100-fold with fresh LB medium supplemented with the same antibiotics and shaken at 37°C to early mid-log phase (optical density at 600 nm [OD 600 ] Ͼ 0.5).
Cells from a 1-ml culture were collected by centrifugation and suspended in 50 l of water-50 l of 2ϫ loading buffer for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and heated in a boiling water bath. The lysates were applied to a 15% acrylamide gel, which was subjected to electrophoresis at 30 mA as described previously (35), followed by staining with Coomassie brilliant blue R.
Purification and antigenic properties of T. maritima RRF. E. coli BL21 (DE3)pLysS harboring pKA1 carrying T. maritima frr was grown in 4 liters of LB medium supplemented with appropriate antibiotics. Cells were induced with 1 mM IPTG for 4 h and collected, and the crude extract (136.6 mg) was obtained as described previously (9). The heat-treated extract (1.9 mg/ml; total, 16 ml; purity, 66.5%) was applied on a Sephadex G-100 column (15 g of Sephadex G-100 superfine [Pharmacia]; 2.6 by 40 cm). The column had been equilibrated with buffer J (10 mM Tris-HCl [pH 7.6], 50 mM NH 4 Cl, 10 mM MgSO 4 , 0.5 mM dithiothreitol), and RRF was eluted with the same buffer at 4 drops/min at 4°C.
Fractions (4.7 ml/tube) were collected. The eluates were concentrated by Centriprep 10 (Amicon) and examined by SDS-PAGE. The protein concentration was measured with the DC protein assay kit (Bio-Rad). Western blotting was performed as described previously (35).
RRF assay. For disassembly of the posttermination complex, a model substrate was prepared from polyribosomes isolated from growing E. coli in the presence of tetracycline, which inhibits binding of aminoacyl tRNA to the ribosomal A site. The polysomes have nascent growing polypeptides, which are removed by puromycin. The resulting complex consists of ribosomes with deacylated tRNA on the P site, E site, and an empty A site. Each ribosome in this polysome can be regarded as a model for the posttermination ribosomal complex. The only difference between this model and the actual posttermination complex is that the real posttermination complex has a termination codon at the A site while the ribosome isolated from the polysome can have any codon at this site. Since disassembly of the posttermination complex does not depend on a specific codon, this model substrate is adequate for examining the RRF reaction (9).
Disassembly (release of ribosomes from mRNA) of the model complex was measured by the conversion of polysomes to monosomes. The conversion is detected by monitoring the sedimentation profile of the ribosomes in sucrose gradient centrifugation. Polysomes and purified E. coli RRF were prepared as described previously (9). EF-G was prepared using the purification method previously described (11,25). The reaction mixture was incubated at 30°C for 15 min, and the conversion of polysomes to monosomes was examined as follows. The mixture was placed on top of 5 ml of sucrose gradient (15 to 30%) and centrifuged for 90 min at 40,000 rpm at 4°C in a Beckman SW 50 rotor. The gradient was analyzed by measuring UV absorption at 254 nm using the ISCO gradient analyzer. The release of ribosomes from mRNA was monitored by measuring the increase in the amount of 70S ribosomes and the decrease of polyribosomes. Depending on the preparation of polysomes, some 70S monosomes were already present in the preparation before the RRF reaction. The amount of preexisting monosomes was subtracted from the amount of monosomes present after the RRF reaction to calculate the conversion of polyribosomes to monosomes due to the RRF reaction. Details of the assay procedure have been described previously (9).

RESULTS
Purification of T. maritima RRF and the antigenic properties of purified T. maritima RRF. E. coli BL21(DE3)pLysS cells harboring pKA1 carrying T. maritima frr were treated as described in Materials and Methods, and crude T. maritima RRF with a purity of 23.5% was obtained. For further purification, the crude T. maritima RRF was incubated at 75°C for 30 min to remove the bulk of the proteins from the host cells as shown in Fig. 1A. By this procedure, extracts (30.5 mg of protein of which 66.5% was RRF) were obtained. A similar treatment was equally effective for the purification of other T. maritima proteins expressed in E. coli (30,39,43). The heat-treated extract was then applied to a Sephadex G-100 column. The representative fractions were examined for purity as shown in Fig. 1B. The purified T. maritima RRF (5.5 mg/ml and 92.0% purity) was thus obtained with 25.0% recovery. Total recovery over the entire purification procedure was 15.8%.
The cross-reactivity of purified T. maritima RRF with an anti-E. coli RRF polyclonal antibody was examined as shown in Fig. 2. Although T. maritima RRF cross-reacted with the anti-E. coli RRF antibody, the reactivity was at least 300-fold less than that of E. coli RRF. T. maritima RRF is therefore only remotely related to E. coli RRF immunologically. It is noted that T. maritima RRF migrated more slowly than E. coli RRF (see also Fig. 6A), indicating that the band observed in lane 1  (Fig. 3). The pET-11a vector carried T. maritima frr with the ampicillin resistance gene (Ap r ). The expression of T. maritima frr was induced by the addition of IPTG. Induction of T. maritima frr inhibited the growth of E. coli (Fig. 3A and D). The maximum killing effect was such that 90% of the viable counts were lost; the data represent typical results that were reproduced several times. The inhibitory effect depended on the concentration of the inducer, suggesting a dose-dependent relationship in the inhibition by T. maritima frr. The inhibitory effect appears to affect the viability of the host more than the increase in OD. Similarly, inactivation of temperature-sensitive (ts) RRF had a greater pronounced effect on viability than on the ODs of the cell cultures (17).
The presence of multicopy E. coli frr reversed the inhibitory effect of T. maritima frr (Fig. 3B and E). These cells were identical to those used in Fig. 3A but contained plasmid pRR3, which carried E. coli frr and kanamycin resistance. These figures show that a functional extrachromosomal E. coli frr overcame in large part the toxic effect of T. maritima RRF. The empty vectors used in these experiments, as well as the presence of pRR3, had no deleterious effect on the host cells, as shown in Fig. 3C and F. Figures 4 and 5 show similar but less-pronounced effects by S. pneumoniae and H. pylori frr, respectively. In both cases, the addition of 1 mM IPTG exerted a deleterious effect, whereas, in the absence of IPTG, bacterial growth was identical to that for the controls (Fig. 4A and C and 5A and C). The effect of H. pylori frr, however, was so mild that it did not reduce the viable count but only retarded the increase (Fig. 5A and C). These toxic effects were significantly reduced when plasmids carrying E. coli frr were present simultaneously ( Fig. 4B and D and 5B and D). Since the expression of E. coli frr carried by pRR2 did not depend on adding inducer (Fig. 6A, lane 3) (13,38), the constant presence of excess RRF is not toxic.
Evidence for the expression of heterologous RRF in E. coli. The preceding results can be interpreted by assuming that each of the plasmids carrying non-E. coli frr expresses the corresponding RRF. Indeed, plasmids pKA1 (pET-11a carrying T. maritima frr) and pRR3 (pHSG299 carrying E. coli frr) expressed the respective RRFs (Fig. 6A, lane 2). The relative amounts of these RRFs were in the ratio of 10 (E. coli)/8 (T. maritima) (average of the densities of the bands on three different gels). Because E. coli RRF migrated slightly faster than that of T. maritima, one can detect the presence of both. That the slower-moving band corresponds to T. maritima RRF is clear from lanes 3 to 5. We conclude from this experiment that expression of the E. coli frr gene neutralizes the deleterious effect of T. maritima RRF.
In a similar manner, under the conditions where the plasmid carrying E. coli frr reduced the deleterious effect of S. pneumoniae frr, the expression of both RRFs indeed took place. The relative amounts of these RRFs in this cell were 7 (S. pneumoniae) to 10 (E. coli) (data not shown).
Although H. pylori RRF shows the weakest inhibition of E. coli growth, this is not due to a weak expression of H. pylori RRF, as shown in lane 3 in Fig. 6B. We conclude that the relative amount of H. pylori RRF expressed was similar to that of T. maritima RRF (compare Fig. 6B and A).

Inhibition of the E. coli RRF reaction by T. maritima RRF.
We examined the effect of T. maritima RRF on the in vitro RRF assay using naturally occurring polysomes (9). In this system, polyribosomes isolated from growing E. coli cells were treated with puromycin to remove the nascent peptide. We regard the resulting complex of ribosome, tRNA, and mRNA as a model substrate for the posttermination ribosomal complex. Disassembly of this model posttermination complex converts the polysomes into monosomes (8). Table 2 shows that the strongest inhibition (65%) of E. coli RRF by T. maritima RRF was observed when the largest amount of T. maritima RRF (10-fold larger than the amount of E. coli RRF) was added. Equal amounts of T. maritima RRF and E. coli RRF in the reaction mixture yielded only a slight (7%) inhibition. This result suggests that the affinity of T. maritima RRF for the E. coli polysomes must be less than 10% that of E. coli RRF. T. maritima RRF did not disassemble the posttermination complex of E. coli (Table 2).

DISCUSSION
It is known that P. aeruginosa RRF (29) and the L-lactate dehydrogenase gene from T. maritima (30) function in E. coli. We therefore expected that heterologous RRFs would function in E. coli because their sequences are very similar to that of E. coli RRF (Fig. 7). Contrary to this expectation, expression of the genes coding for these RRFs was deleterious to E. coli. We suggest that this toxic effect is due to an inhibitory action of a heterologous RRF on the reaction catalyzed by E. coli RRF for the following reasons.
First, the toxic effects of the heterologous frr depend on the extent of induction of these heterologous RRFs by IPTG. Second, analysis of the E. coli extract harboring the heterologous frr showed the presence of the corresponding heterologous RRF. Third, pure heterologous RRF represented by T. maritima RRF inhibited the in vitro E. coli RRF reaction. Fourth, the heterologous RRF, represented by T. maritima RRF, was close enough to E. coli RRF immunologically to be cross-reactive. Although cross-reactivity does not necessarily indicate a similar structure, it suggests that the heterologous RRF can compete with E. coli RRF because of the structural similarities of these two proteins. Fifth, and most importantly, the simultaneous expression of an excess amount of E. coli RRF overcame the toxic effect of the heterologous RRF. In addition, T. maritima frr did not complement LJ15 (17), an E. coli mutant carrying ts RRF at 47°C (the nonpermissive   Figure 7 shows the amino acid sequences of RRF from E. coli (13), P. aeruginosa (29), T. maritima (36; see Addendum in Proof), S. pneumoniae, and H. pylori (42). In the five sequence segments (a to e) there are many amino acids of E. coli RRF identical to those of P. aeruginosa RRF but the RRFs of the other species showed only a few identical amino acids. This suggests that these segments may be responsible for the inhibitory effect of the heterologous RRFs.
We have recently resolved the crystal structure of T. maritima RRF prepared as described in this paper; it consists of two domains, domain I being a long three-helix bundle and domain II being a three-layer ␤-␣-␤ sandwich (36). Recent work on E. coli RRF (26) confirmed this structure, which essentially agrees with the computer-predicted secondary structure (3) of RRF (16). These two domains are connected through the hinge region. Out of the five segments marked in Fig. 7, segments a and b are in domain II while the remaining three are in domain I.
A heterologous RRF may act on E. coli RRF directly and inactivate it by forming an inactive complex. Our preliminary nuclear magnetic resonance studies on RRF (24), however, indicate that RRF tends to stay as a monomer. In addition, it takes at least a 10-fold-higher molar concentration of heterologous RRF for a 50% inhibition of E. coli RRF (Table 2). Since RRF is a nearly perfect mimic of tRNA, we postulated that it behaves like tRNA on the ribosome (36). Our mutation data (17) support this hypothesis. It is therefore more likely that the heterologous RRF competes with E. coli RRF for the ribosomal A site (the site at which aminoacyl tRNA binds to the ribosome).  Why then does the A site-bound T. maritima RRF not work for E. coli? We may speculate that RRF must bend at the hinge region during its action. It is possible that T. maritima RRF can bend at 80°C but not at 37°C because molecular flexibility increases with temperature. This makes it difficult for T. maritima RRF bound at the A site to function at 37°C. The following observations lead to this speculation. First, crystallization of E. coli RRF depends on decyl-␤-D-maltopyranoside, which fits into the pocket of the hinge region of RRF (26), probably fixing the molecule in one form. Second, no such agent is required for crystallization of T. maritima RRF at room temperature (37), probably because this RRF keeps itself in one form due to its nonflexibility at room temperature. This is conceivable because T. maritima RRF is probably designed to be bendable at the hinge region at 80°C but not at 37°C. Third, the hinge region represents the high-mobility region of this molecule (estimated from the crystal structure) despite the fact that it is not near the N or C terminal. It should be noted that less pronounced inhibitory effects by S. pneumoniae and H. pylori RRFs probably do not involve this mechanism because their RRFs are designed to function at 37°C. For inhibition by these RRFs and T. maritima RRF, the a-e regions may also play important roles.