Article

• The EMBO Journal (1998) 17, 1141 - 1151
• doi:10.1093/emboj/17.4.1141

Evidence for in vivo ribosome recycling, the fourth step in protein biosynthesis

Laszlo Janosi¹, Salim Mottagui-Tabar², Leif A. Isaksson², Yasuhiko Sekine³, Eiichi Ohtsubo³, Shijie Zhang¹, Scarlett Goon¹, Sarah Nelken¹, Masahiro Shuda¹ and Akira Kaji¹

1. Department of Microbiology, School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA
2. Department of Microbiology, Stockholm University, S-106 91 Stockholm, Sweden
3. Institute of Molecular and Cellular Biosciences, University of Tokyo, Tokyo, Japan

Correspondence to:

Akira Kaji, E-mail: kaji@mail.med.upenn.edu

Received 8 October 1997; Accepted 13 November 1997; Revised 12 November 1997

• Keywords:

  • ribosomal frameshift,
  • ribosome recycling,
  • temperature-sensitive RRF,
  • termination complex,
  • translational re-initiation

Introduction

Protein biosynthesis consists of four steps: initiation, elongation, termination, and disassembly of the ribosome–mRNA–tRNA post termination complex (hereafter referred to as the termination complex). A survey of the current textbooks of biological sciences, however, shows that most authors complete the description of protein synthesis with termination, as if the fourth step of protein synthesis did not exist. Neglecting the fourth step of protein synthesis is a gross oversight. A recent review addresses this issue (Nakamura et al., 1996).

Ribosome recycling factor (RRF)—formerly called ribosome releasing factor (see recent review by Janosi et al., 1996a)—was identified as an in vitro protein synthesis factor as early as 1970 (Hirashima and Kaji, 1970). Since then, RRF, together with elongation factor G (EF-G), have been shown to catalyze the fourth step of protein synthesis in vitro (Hirashima and Kaji, 1972a,b, 1973; Ogawa and Kaji, 1975). The absence of RRF in vitro causes the terminating ribosomes to remain on mRNA and initiate protein synthesis at the codon immediately 3' to the termination codon. In addition, RRF stimulates in vitro protein synthesis as much as 8-fold (Ryoji et al., 1981b), suggesting that recycling the components of the termination complex increases the efficiency of protein synthesis in vitro. Our in vitro studies on RRF using natural mRNA (Ogawa and Kaji, 1975; Ryoji et al., 1981b) were mostly confirmed recently by elegant in vitro studies with short synthetic mRNA (Pavlov et al., 1997). Recent studies in our laboratory indicate that it may prevent errors in vitro during peptide elongation (Janosi et al., 1996b).

Although there is no direct in vivo evidence available for ribosome recycling by RRF, there exists significant evidence that RRF is biologically important. The crucial biological role of RRF became evident when it was found that the gene encoding RRF is widely distributed in prokaryotes (Himmelreich et al., 1996; Vizcaino et al., 1996) including Mycoplasma genitalium (Fraser et al., 1995). (For a complete list of other known RRF sequences, see Janosi et al., 1996b). Mycoplasma genitalium is the smallest known free-living organism with only 500 genes; this organism has dispensed with other protein synthesis factors such as RF2 and RF3 which are involved in termination but has retained RRF. In Escherichia coli, RRF has been shown to be an essential factor for growth (Janosi et al., 1994).

In this communication, we demonstrate in vivo evidence for RRF function using E.coli temperature-sensitive (Ts) mutants with altered RRF genes (frr). In agreement with previous in vitro results (Ryoji et al., 1981a), translational re-initiation occurs downstream from the termination codon in vivo upon inactivation of RRF. The downstream re-initiation is at multiple random sites up to 45 nucleotides 3' to the termination codon and does not require a canonical initiation codon sequence. This suggests that, in the absence of RRF, the ribosome of the termination complex remains on the mRNA, scans a short distance downstream, and re-initiates translation in vivo. Inactivation of RRF during the transition from the stationary to the growing phase had a bactericidal effect, while it had a bacteriostatic effect during the growing phase.

Isolation of 12 independent alleles of frr coding for temperature-sensitive RRF

To prepare Ts RRF mutants, targeted random mutagenesis was performed on the wild-type RRF DNA (frr) by the error-prone PCR method. The mutagenized DNA was subcloned in a plasmid encoding CM^r (chloramphenicol resistance). Since this is a mixture of mutagenized frr DNA cloned into a plasmid, it was called pMIX. The plasmids, pMIX, were then introduced into recA^-E.coli LJ3(pPEN1560) which carries the functionally null mutant (frame shifted) frr1(Fs) in its chromosome (Janosi et al., 1994). This E.coli harbors a resident plasmid, pPEN1560, containing wild-type frr [frr(Wt)] and the sacB-Neo^r cassette (conferring sucrose sensitivity and kanamycin resistance) (Blomfield et al., 1991; see Table I). The resident plasmid (pPEN1560) and pMIX are incompatible because they have the same replicon. Growth at 32°C in the presence of sucrose and CM selects for segregants which lack the resident plasmid but harbor the plasmid with a mutagenized frr insert. Among 4548 transformants selected this way, 18 (0.4%) showed both the temperature- and the kanamycin-sensitive (because of loss of the resident plasmid) phenotypes. These transformants were considered to harbor a plasmid with a temperaturesensitive mutation in the frr gene.

Table 1: Escherichia coli strains and plasmids

View full table

To establish that the PCR-mutagenized frr is indeed responsible for the temperature sensitivity of the host, we isolated this portion from the plasmid of the 18 isolates and ligated it to plasmid pMW119 giving ampicillin resistance (AP^r). The E.coli strain LJ3 [recA^-frr1(Fs)] harboring pPEN907 (incompatible with pMW119) which carries CM^r and wild-type frr was transformed with pMW119 carrying the frr (Ts) allele and Ap^r. From the AP^r transformants, CM-sensitive segregants which had lost their pPEN907 plasmid due to plasmid incompatibility were isolated. All of the isolated segregants were temperature- and CM-sensitive, establishing that the Ts phenotype is associated with the insert portion of the plasmids from our 18 Ts isolates.

To confirm that the Ts growth phenotype is associated with the open reading frame (ORF) of frr, the 0.9 kb KpnI–HindIII fragment containing the promoter, the mutated frr, and the transcription terminator were isolated and sequenced. As shown in Table II, all mutational changes (sometimes more than a single change) fell within the coding sequence of frr. This suggests that the Ts growth phenotype is due to the mutational change in the coding sequence of frr carried by the plasmid.

Table 2: Mutational changes rendering the E.coli ribosome recycling factor temperature sensitive

View full table

Replacement of the chromosomal wild-type frr with its frr(Ts) allele and in vitro thermolability of the mutationally altered RRF

The chromosomal wild-type frr allele of the MC1061 strain was replaced with mutant alleles frr13, frr14, frr15 and frr16 via homologous recombination as described previously (Janosi et al., 1994) resulting in strains LJ13 to LJ16 respectively. Strains LJ13 to 16 were all temperature-sensitive; furthermore, complementation of the Ts phenotype was observed with all of the strains by a plasmid carrying wild-type frr (data not shown). LJ14 was the most clear-cut Ts mutant.

To test the thermolability of RRF in vitro, E.coli LJ14 harboring pPEN2399 [pUC19 with frr14(Ts) insert], was grown at 32°C, and Ts RRF was purified to homogeneity. This preparation was then examined for its ability to convert polysomes into monosomes in the presence of EF-G and GTP (Hirashima and Kaji, 1970) after exposure to 45°C for varying periods. Figure 1 shows that the RRF coded for by frr14 lost its ribosome-releasing activity upon exposure to the elevated temperature. Even with prolonged heat treatment, there was a residual (20%) RRF activity. It should be noted that the specific activity of the purified Ts RRF was 10% of that of the wild-type RRF even before the heat treatment (data not shown).

Figure 1.

Ribosome recycling factor coded for by frr14 is temperature-sensitive. Wild-type RRF () and mutant RRF coded for by frr14 () were purified to homogeneity (Hirashima and Kaji, 1972b). In vitro ribosome release assay (Hirashima and Kaji, 1972b) was carried out after exposing the isolated RRF to 45°C for varying periods as shown. Percentages of the activity remaining are shown.

View full figure (47 KB)

Physiological characterization of the temperature-sensitive RRF mutants

Table III shows that the mutants fall into four major classes with regard to temperature sensitivity. The allele frr8, belonging to the first category, is moderately temperature-sensitive with only 90% reduction in the number of colonies at 43°C. With the most severely Ts mutants (frr10, 14–16), the colony-forming ability was reduced both at 37 and 43°C to 10^-2 to 10^-5 of that of wild-type.

Table 3: Diversity of the temperature-sensitive growth phenotype determined by the mutant frr alleles

View full table

Table III shows also that moving the mutant allele to the chromosome does not change the overall characteristics of the conferred Ts phenotype. The temperature sensitivity, however, increases when the mutant allele is moved to the chromosome. Since the oligocopy plasmid used in this test is present in approximately six copies (Hashimoto-Gotoh and Sekiguchi, 1977), this may be due to a dosage effect of the residual activity of Ts RRF, as indicated in Figure 1.

When the growing cells were exposed to the non-permissive temperature, the growth was reduced within 20 min (Figure 2, circles) but did not stop completely for a prolonged period, probably due to residual activity of Ts RRF at the non-permissive temperature. As shown in this figure (lower panel), when LJ14 [frr14(Ts)] cells were grown overnight at 32°C, diluted into fresh medium, and exposed immediately to the non-permissive temperature, they started losing viability after 4 h at the high temperature (triangles). The loss of viability is dependent on the addition of fresh media (Figure 2, triangles versus closed diamonds). This suggests that the loss of RRF has a severe deleterious effect when cells switch from the stationary to the growing phase. Similar results were obtained with LJ13, LJ15 and LJ16 (data not shown).

Figure 2.

Temperature-sensitive growth of E.coli carrying the mutant frr14 allele in the chromosome. E.coli LJ14 [frr14(Ts)] (panels A and C) and its parent (wild-type frr) (panels B and D) were grown in Luvia–Bertani broth (LB) medium overnight at 32°C and divided into two parts. The first part was diluted with fresh LB to OD₅₄₀ of 0.01 and divided into three aliquots. One aliquot of this diluted cell suspension was immediately exposed to 43°C and grown at that temperature (). The second aliquot was grown at 32°C until OD₅₄₀ reached 0.45 when it was switched to 43°C (). The third aliquot was grown at 32°C throughout the experiment (). The second part from the overnight culture was further incubated at 32°C () or at 43°C () without dilution with fresh medium. The cell density was measured at various time points for the different culturing conditions and the results are plotted in panels A and B. In addition, aliquots from the different culturing conditions were removed at various times, plated, and grown at a permissive temperature. The number of viable colonies per ml is plotted in panels C and D.

View full figure (80 KB)

In support of the notion that Ts RRF has the residual activity at the non-permissive temperature, we observed that an excess amount of the residual activity of Ts RRF expressed from multicopy pUC19 eliminated the Ts phenotype. Figure 3 shows that in such cells the effect of the non-permissive temperature was considerably delayed and diminished, compared with the control (panel B) which had only the chromosomal copy of the frr14(Ts) allele, suggesting a gene dosage effect. These results suggest that an excess Ts RRF does not exert toxic effects.

Figure 3.

Suppression of the temperature-sensitive growth phenotype by high gene dosage. E.coli LJ14 [(frr14(Ts)] harboring pPEN2399 [pUC19 plasmid containing frr14(Ts)] (A) and LJ14 harboring pUC19 (B) were grown at 27°C in LB supplemented with AP. The cultures were then diluted into the same medium to give 0.05 OD₅₄₀ and grown at 27°C. At the time indicated by the arrows, the cultures were divided into two aliquots. One aliquot was kept at 27°C (), and the other was shifted to 43°C ().

View full figure (45 KB)

Amino acid alterations of RRF giving a temperature-sensitive phenotype

As shown in Table II, the 18 Ts isolates were represented by 12 independent genotypes in the 558 nucleotides encoding RRF (185 amino acids). In the N-terminal portion, we observed three mutations each involving the 1st, 9th and 21st codons respectively. In the frr8 allele, the GUG initiation codon was changed to GCG (Ala) codon. A possible new translation initiation point at the Met13 codon probably functions in this case. At 5 nt upstream from Met13, there is an AAG which may function as a Shine–Dalgarno sequence for the 13th codon AUG. If this is the case, allele frr8 has the most significant change in the primary sequence of RRF (loss of 12 N-terminal amino acids), yet it has only moderate thermolability (see Table III). Allele frr3 is a double point mutant, with Leu182Pro and Ala9Val. This mutant is only slightly more temperature-sensitive than frr17 with a single Leu182Pro mutation, indicating that the mutation Ala9Val has very little influence on thermostability. In contrast, the Lys21Arg mutation (frr15) gives temperature sensitivity, even at 37°C. These three mutants therefore, suggest that at least 12 N-terminal amino acids are not important for RRF thermostability, but the maintenance of the sequence from somewhere beyond the 12th amino acid is important for thermostability, as indicated by the temperature sensitivity of frr15 with the mutation at the 21st position.

In a further downstream region, between Lys21 and Val117 codons, mutations did not occur, suggesting that slight alterations in this region of the protein may be lethal or have no effect on thermostability. These possibilities are strengthened by the fact that there was more than one independent isolate with identical mutational sites in other regions.

In the portion further downstream from this large silent region, Ts mutations occurred approximately every 15 codons at positions 128, 148, 163 (two more nearby) and 179 (five more nearby at the C-terminal end). Mutations in the proximal and distal portion of the region stretching from amino acid 117 to 163 render the molecule labile, even at 37°C. Thus, frr14 (Val117Asp), frr2 (Val128Gly), frr4 (Leu163Pro) and frr10 (Leu163 Pro and Asp159Asn) are severely temperature-sensitive (Table III). Mutations occurring in the middle of this region, as in frr6 (Ile148Thr) and frr7 (Val160Glu), render the molecule inactive only at 43°C. These data suggest that the ends of this stretch of 46 amino acids play a critical role in the thermal stabilization of RRF.

The two alleles, frr13 and frr16, code for a C-terminal truncation, five and four amino acids long respectively. Allele frr16 is a G(535) deletion mutant. It causes a frameshift, effecting codons 179 to 181 giving termination at codon 182. The frr17 and frr3 carry a common point mutation (Leu182Pro) and exhibit approximately the same temperature-sensitive phenotype (inactivated at 43°C but not at 37°C) as frr13 (loss of C-terminal amino acids 182 to 185). This suggests a critical role of Leu182 for stability. These C-terminal mutations also suggest that extensive alterations in the C-terminal region are not fatal. This region, however, must be important for the thermal stability of RRF because C-terminal mutants are all sensitive at 43°C and RRF encoded for by frr16 is temperature-sensitive, even at 37°C. Consistent with this concept is the fact that the C-terminal end of RRF is well-preserved among RRF of various species (Janosi et al., 1996a).

RRF functions to disassemble the termination complex in vivo

Earlier in vitro experiments (Ryoji et al., 1981a) showed that translation in the absence of RRF results in re-initiation and translation downstream from the termination codon without depending on the initiation signals because ribosomes are not released from the termination complex in the absence of RRF (Ogawa and Kaji, 1975). To demonstrate that RRF releases ribosomes from the termination complex in vivo, we constructed two plasmids with a reporter gene coding for -galactosidase to detect unscheduled translation downstream from the termination codon. In these constructs, the lac promoter is followed by the lacZ ribosome binding site (RBS) and a short ORF coding for 13 amino acids followed by a stop codon. The stop codon is followed by the lacZ coding sequence (from the 10th authentic lacZ) either in zero frame (pPEN2363) or in -2 frame (pPEN2369). The reporter lacZ lacks its own initiation codon.

The above-mentioned plasmids, pPEN2363 and pPEN2369 were placed separately into LJ14 [frr14(Ts)]. If ribosomes are not released due to the inactivation of RRF at 39°C, and they re-initiate translation downstream from the termination codon without depending on the initiation signals as we observed in vitro, one would expect the lacZ gene to be expressed. The controls were the wild-type parent strain (MC1061) or LJ14 harboring pPEN907, the plasmid encoding wild-type RRF.

Figure 4 shows that a temperature increase induced synthesis of -galactosidase in LJ14 (Figure 4A, upper left panel) with the in-frame lacZ'-'lacZ reporter plasmid. The induction was minimal at 31°C, where most of the Ts RRF should be active. The induced -galactosidase synthesis is mostly attributable to RRF depletion at 39°C because the MC1061 strain with wild-type RRF (Figure 4A, upper middle panel) and LJ14 harboring pPEN907 carrying wild-type frr (upper right panel) showed only a negligible -galactosidase activity.

Figure 4.

Temperature-dependent re-initiation of translation downstream from the termination codon in LJ14 (Ts frr). Plasmid pPEN2363 (the -galactosidase coding sequence is in-frame with the upstream termination codon) (A) or plasmid pPEN2369 (the -galactosidase coding sequence is out-of-frame with the upstream termination codon) (B) was placed in E.coli LJ14 (left figures), or in its parent MC1061 (middle figures) or in LJ14 harboring pPEN907 carrying wild-type frr (right figures). The strains were grown in LB (supplemented with kanamycin) at 31°C. Then cultivation continued either at 31°C () or at 39°C () for an additional 4 h. -Galactosidase activity (top row figures in both panels) and OD₆₀₀ [bottom row figures in (A) and (B)] were measured as indicated. Data for at various time periods are overlapping with in some of the figures of (B) and shown as .

View full figure (73 KB)

As shown in Figure 4B, when RRF was depleted, -galactosidase was also synthesized in the cells transformed with pPEN2369, the out-of-frame (-2 frame) reporter plasmid (Figure 4B, upper left panel). If RRF was not inactivated, or wild-type RRF was present, no -galactosidase synthesis was observed. As shown in Figure 5, the optimum re-initiation for the in-frame construct was at 35°C, while it was at 39°C with the out-of-frame construct. The observed difference suggests that ribosomes remaining at the termination codon in the post-termination complex may undergo thermal agitation to change the reading frame in vivo.

Figure 5.

Temperature optimum for out-of-frame re-initiation is higher than for in-frame re-initiation. E.coli LJ14 harboring either pPEN2363 (in-frame -galactosidase reporter plasmid) () or pPEN2369 (out-of-frame -galactosidase reporter plasmid) () was grown at 31°C and the cultures were shifted to the indicated temperatures and grown for another 4 h. The induced -galactosidase activity was then measured and plotted against the temperature.

View full figure (62 KB)

As controls, similar constructs without the upstream termination codons were prepared; pPEN2345 (with the lacZ gene in-frame with the initiation codon) and pPEN2367 (with the lacZ gene in -2 frame with the initiation codon). As expected, pPEN2345 constitutively expressed -galactosidase in all situations, while pPEN2367 did not express -galactosidase in either the wild-type E.coli or in the frr(Ts)LJ14 mutant, regardless of the temperature (data not shown).

Since pPEN2369 used in the experiment shown in Figure 4 has the reporter in -2 frame, we wished to establish that -1 out of frame re-initiation of translation can also occur due to inactivation of RRF. We used plasmid pSEK336C which contains an upstream ORF (orfA of insertion sequence IS3) followed by the orfB'-lacZ fusion in -1 frame with respect to the termination codon of orfA (Sekine et al., 1994). The initiator AUG codon of orfB was eliminated. In this construct, -galactosidase is expected to be synthesized only by ribosomes that have finished the translation of orfA and re-initiate downstream out-of-frame from the orfA termination codon. Plasmid pSEK336C was placed in LJ14 [frr (Ts)], and the culture temperature was raised to 39°C. As shown in Figure 6, in vivo inactivation of RRF significantly increased -galactosidase synthesis. At the permissive temperature for LJ14, or at both temperatures in the case of MC1061 (wild-type frr), such an increase was not observed.

Figure 6.

Out-of-frame (-1) re-initiation of translation downstream from the termination codon of orfA of IS3. E.coli LJ14 [frr14(Ts)] (circles) and its parent MC1061[frr(Wt)] (squares) were transformed with pSEK336C which contains the reporter gene in the -1 frame with respect to the upstream termination codon. Transformants were grown in LB overnight at 32°C, diluted with the same medium, grown to 0.4 OD₅₄₀, and then one aliquot was further grown at 32°C (,), and the other was shifted to 39°C (,). The data representing and are overlapping and shown with the . The -galactosidase activity (in Miller units) is plotted against the periods after the temperature shift.

View full figure (43 KB)

The efficiency and the site of re-initiation caused by RRF inactivation in vivo

In the system described in the preceding section, only the re-initiated proteins with the reporter -galactosidase, but not the translation of the upstream ORF, could be measured. It was therefore not possible to estimate the fraction of ribosomes that reach the upstream termination codon and engage in re-initiation as a result of RRF inactivation. To estimate this fraction, we used a modified version of the 3A' reporter system (Björnsson et al., 1996) in which we can quantify both the proteins that correspond to the upstream cistron and to the unscheduled re-initiated translation downstream from the termination codon. This system consists of an engineered protein A of Staphylococcus aureus [called A' (Björnsson et al., 1996)]. We made a construct so that the upstream cistron is a dimer of A' (M_r =12 000; 2A') and the downstream ORF is a trimer of protein A' (M_r = 18 000; 3A'). The ATG initiation signal for 3A' is preceded by a weak Shine–Dalgarno sequence giving only a minimal background initiation. Unscheduled re-initiated protein should be detected as proteins slightly larger than 3A' due to the addition of amino acids at the N-terminal end coded by codons downstream from the 2A' termination signal and extended translation into the 3A' cistron.

Plasmid pSM–T272 (the downstream 3A' sequence is in -1 frame with respect to the upstream 2A' termination codon) and pSM–T275 (in zero frame) were placed separately in LJ14. As shown in Figure 7, all the test strains (lanes 2–9) synthesized 2A' and 3A' proteins regardless of the presence of RRF, as expected. Upon inactivation of RRF in vivo by raising the culture temperature to 39°C, proteins with slightly higher molecular weight than 3A' were observed (bands 3A'R in lanes 5 and 7). In accordance with the in vivo re-initiation observed in Figures 4 and 6, 3A'R was observed with pSM–T272 (out-of-frame) as well as with pSM–275 (in-frame). No product smaller than 3A' was observed as a result of RRF inactivation, suggesting that the ribosomes remaining on the mRNA cannot re-initiate downstream from the canonical initiation site of 3A'. This observation also suggests that non-specific proteolytic degradation of the re-initiated products is not significant in LJ14. Because 3A'R bands did not appear either in MC1061 at the elevated temperature (Figure 7, lanes 3 and 9) or in LJ14 at low temperature (lanes 4 and 6), we conclude that their appearance is due to inactivation of RRF by the elevated temperature.

Figure 7.

Re-initiation of translation upon in vivo inactivation of RRF in E.coli LJ14 with the 2A'–3A' system. E.coli MC1061 (wild-type frr) and its temperature-sensitive derivative E.coli LJ14 [frr14(Ts)] were transformed either with pSM–T272 (2A' followed by 3A' out-of-frame), or with pSM–T275 (2A' followed by 3A' in-frame). The transformants were grown overnight at 32°C in LB, diluted with fresh LB and grown at 32°C until OD₅₄₀ reached 0.65. One aliquot was then grown at 32°C, and the other at 39°C for 1 h. IPTG (0.4 mM) was then added to induce expression of 2A' and 3A' for 1 h. The cultures were then treated as described (Björnsson et al., 1996). Lane 1, 3A' size control, MC1061/pAB93 with no stop codon insert (Mottagui-Tabar et al., 1994) grown at 32°C; lane 2, MC1061/pSM–T272 grown at 32°C; lane 3, MC1061/pSM–T272 at 39°C; lane 4, LJ14/pSM–T272 at 32°C; lane 5, LJ14/pSM–T272 at 39°C; lane 6, LJ14/pSM–T275 at 32°C; lane 7, LJ14/pSM–T275 at 39°C; lane 8, MC1061/pSM–T275 at 32°C; lane 9, MC1061/pSM–T275 at 39°C. Arrows on the left indicate the 2A' and 3A' proteins. The arrow on the right marked as 3A'R indicates re-initiated proteins (in lanes 5 and 7).

View full figure (32 KB)

The results suggest that ribosomes which are not released from the termination codon of 2A' re-initiate upstream from the initiation codon of 3A'. This re-initiation must occur at several points, because the 3A'R band is not discrete in the gels. The results in Table IV show that the mean molar amount relative to that of 2A' for the out-of-frame and the in-frame 3A'R products were 0.41 and 0.19, respectively. It should be pointed out that the system provides measurement only for the frame which gives 3A' and 3A'R but not for the other two frames, which yield proteins that are eliminated in the purification of 2A', 3A' and 3A'R. The actual efficiency of re-initiation should therefore be higher than the values shown in Table IV. This then suggests that most, if not all, of the ribosomes terminating 2A' may remain on the mRNA and begin the unscheduled translation in the absence of RRF.

Table 4: Majority of ribosomes at the termination complex engage in translation downstream from the termination codon

View full table

Re-initiation in the absence of RRF occurs at several points and does not require canonical initiation codon

The results using the 3A' system suggest that the proteins synthesized via re-initiation upon inactivation of RRF consist of a mixture of proteins having various N-termini depending on the site of re-initiation. As shown in Table V, this notion was confirmed by N-terminal analysis of -galactosidase synthesized in the absence of RRF. In this experiment, the induced reporter -galactosidase proteins from the experiments described in Figures 4 and 6 were purified and Edman analysis was performed on the mixture of the induced proteins. The results were compared with the DNA sequences of the plasmid constructs. Several different major peptides were detected as shown in Table V. The control analysis of -galactosidase coded for by pPEN2345 ('lacZ coding sequence aligned in-frame with the initiation codon of the lacZ' sequence without the upstream termination codon) gave the single N-terminal amino acid sequence encoded by lacZ' and pUC19 (Fowler and Zabin, 1977; Yanisch-Perron et al., 1985) (Table V; pPEN2345).

Table 5: N-terminal amino acid sequence of re-initiation proteins

View full table

We could not detect any -galactosidase which was initiated upstream from the termination codon. With pPEN2363 and pPEN2369, re-initiation can occur as far downstream as 22 and 23 nt from the termination codon respectively. With pSEK336C, re-initiation occurred as far as 45 nt downstream. None of the re-initiation points corresponds to the codon immediately downstream from the termination codon as was observed in the in vitro system (Ryoji et al., 1981a). Although the possibility exists that these proteins are degradation products of a re-initiated protein, this is unlikely because we did not observe any re-initiated product smaller than 3A' in the experiment described in Figure 7. If significant non-specific protein degradation occurs, one would expect smaller products than 3A'.

On the basis of the following considerations, we propose that RRF does indeed catalyze the fourth step of protein biosynthesis, disassembly of the termination complex in vivo. First, in vitro data have established that RRF releases ribosomes at the termination codon on natural mRNA (Ogawa and Kaji, 1975) and on synthetic mRNA (AUG XXX UAA) (G.Grentzmann, P.J.Kelly, S.Laalami, M.Shuda, M.A.Firpo, Y.Cenatiempo and A.Kaji, manuscript in preparation). Recent in vitro experiments with short synthetic mRNA containing the ribosome-binding sequence suggest that released ribosomes may immediately bind to the mRNA again when the Shine–Dalgarno sequence is very close to the termination codon (Pavlov et al., 1997). This would give the appearance of ribosomes not released from mRNA and slide back to the Shine–Dalgarno sequence on the mRNA. Second, in the absence of RRF, the ribosomes remaining at the termination codon on mRNA re-initiated translation starting at the codon immediately following the termination codon in vitro (Ryoji et al., 1981a). Third, as shown in this paper, inactivation of RRF induced most, if not all, of the ribosomes to re-initiate unscheduled protein synthesis downstream from the termination codon in three separate in vivo systems.

Some features of in vivo translational re-initiation in the absence of RRF have been delineated by the current study. First, the re-initiation of protein synthesis in vivo occurred at multiple sites as far as 45 nt downstream from the termination codon (Table V). We suspect that, in the absence of RRF, the ribosome of the termination complex scans downstream from the termination codon without synthesizing protein. Eventually, ribosomes re-initiate protein synthesis without depending on a canonical start codon or a Shine–Dalgarno sequence. Second, frame shifting occurs during re-initiation; hence, the reporter protein was synthesized in the absence of RRF regardless of whether the reporter gene was in-frame (Figures 4 and 7), -2 nucleotides out-of-frame (Figure 4), or -1 nucleotide out-of-frame (Figures 6 and 7). This frame shifting was stimulated by elevated temperatures (Figure 5). To the best of our knowledge, RRF is the first reported soluble protein factor involved in preventing ribosomal frame shifting besides EF-G which functions to translocate ribosomes precisely three codons without causing frame shifting (Dahlfors and Kurland, 1990). Third, the translation re-initiation following inactivation of RRF occurs only downstream from the termination codon; therefore, no reporter protein which initiated upstream from the termination codon was detected (Table V). It should be noted however, that if re-initiation occurred upstream from the termination codon, only the out-of-frame reporter protein would be detected since the in-frame reporter protein would have an in-frame termination codon interrupting the sequence. Fourth, ribosomes which are scanning downstream from the termination site do not scan beyond an authentic initiation signal with a ribosome-binding sequence. Thus, in the 3A' experiment described in Figure 7, we did not detect any re-initiation products which were smaller than 3A'.

Inactivation of RRF was bacteriostatic in the growing phase. Like RRF, in vivo inactivation of any one of the currently known 10 essential protein synthesis factors causes bacteriostasis. The bacteriostatic effects of in vivo inactivation of these factors were demonstrated either with the use of temperature-sensitive mutants such as IF3 (Grunberg-Manago, 1996), EF-G (Hou et al., 1994), RF1 (Ryden et al., 1986), RF2 (Kawakami et al., 1988) and RRF (this paper), or with conditional synthesis of factors such as IF1 (Cummings and Hershey, 1994), IF2 (Cole et al., 1987), RRF (Janosi et al., 1994) and IF3 (Olsson et al., 1996). RRF inactivation and faulty ribosomal recycling may cause bacteriostasis by many possible mechanisms. Since RRF inactivation causes random re-initiation short distances downstream from a termination codon, inter-cistronic and post-cistronic regions of poly-cistronic mRNA will be translated. These new proteins may be harmful to the cell. Unplanned post-cistronic translation may also be harmful in other ways. On poly-cistronic messages, the re-initiating ribosomes may interfere with translation of downstream cistrons. Alternatively, ribosomes translating post-cistronically will eventually reach the 3' end of an mRNA. This can occur under normal conditions when damaged mRNA (truncated within the cistron) is inadvertently used for protein synthesis. For the latter case, the cell has evolved a mechanism to eliminate the truncated protein and rescue the ribosome through the use of 10 Sa RNA which functions as tRNA as well as mRNA (Keiler et al., 1996). In the absence of RRF however, this machinery cannot function to rescue ribosomes since ribosomes will simply re-initiate downstream from the 10 Sa termination codon and eventually reach the 3' physical end. Ribosomes may then be pushed off from either mRNA or 10 Sa RNA in a form difficult for them to engage in the next round of translation.

Unlike the bacteriostasis observed when other protein synthesis factors are inactivated, RRF inactivation is bactericidal during the transition state from the stationary to the growing phase (Figure 2). As discussed above, in vivo inactivation of any one of the protein synthesis factors in E.coli does not cause a bactericidal effect. Likewise, most protein synthesis inhibitors, such as tetracycline and erythromycin, stop bacterial growth but do not kill bacteria. An exception reported is the bactericidal effect of in vivo inactivation of EF-G in Bacillus subtilis (Aharonowitz and Ron, 1972). It is tempting to speculate that this mutation in the B.subtilis EF-G gene is such that it specifically interferes with the ribosome recycling step in which EF-G participates (Hirashima and Kaji, 1973). Though the exact mechanism remains obscure, the bactericidal effect of RRF inactivation at the stationary phase is reminiscent of a similar effect of loss of ribosome modulation factor (RMF). Wada et al. showed that ribosomes at the stationary phase are maintained as 100S dimers by RMF (Wada et al., 1995). If RMF is absent, stationary phase cells die but growing cells do not (Yamagishi et al., 1993). It is possible that the lethal effect of inactivation of RRF and RMF at the stationary phase may be related.

The in vivo results detailed in this communication, in conjunction with previously published in vitro results, establish ribosomal recycling as the essential fourth step in prokaryotic protein synthesis. No direct information is currently available regarding an analogous step in eukaryotes; however, there are significant pieces of indirect evidence suggesting that prokaryotes and eukaryotes may complete the fourth step of protein synthesis in a different manner (discussed in Janosi et al., 1996a). First, unlike prokaryotes, eukaryotic ribosomal subunits are believed to remain on mRNA when ribosomes pass through ORFs upstream from a cistron (Kozak, 1987). Unlike prokaryotes, initial binding of mRNA by ribosomes in eukaryotes is mostly limited to the 5' cap region, and this region is frequently followed by an ORF before the cistron (for example, Lindahl and Hinnebusch, 1992). This makes it difficult to translate cistrons if ribosomes are released at the end of each ORF as observed in prokaryotes (Hirashima and Kaji, 1972a,b; Ogawa and Kaji, 1975).

Second, though homologs of frr exists in eukaryotic genomes (Janosi et al., 1996b), eukaryotic RRF is found in only organelles such as mitochondria and chloroplasts (to be published elsewhere) which are believed to have originated from symbiotic bacteria during phylogenetic development. The notion that these eukaryotic RRF do not participate in cytoplasmic protein synthesis in eukaryotes is supported by a recent finding that Methanococcus jannaschii does not carry the frr equivalent (Bult et al., 1996). This is the only reported bacterium which has eukaryotic protein synthesis machinery and carries neither the frr gene equivalent nor mitochondria, suggesting that the eukaryotic frr gene product does not participate in the cytoplasmic protein synthesis.

Given the likely difference between eukaryotic and prokaryotic ribosomal recycling described above, the fourth step of protein synthesis is an enticing new target for antimicrobial agents. RRF is especially suited as a target in view of the present observation that its inhibition could lead to bacterial death. Unlike other protein synthesis inhibitors such as tetracycline and erythromycin, the one which targets RRF and the fourth step of protein synthesis may be bactericidal. It should be noted that each of the first three steps of protein synthesis, namely initiation, elongation and termination, is sensitive to one or more antibiotics. One antibiotic, fusidic acid, is already known to inhibit the fourth step of protein synthesis in vitro because of its effect on EF-G (Hirashima and Kaji, 1973). As a first step towards developing antimicrobial agents directed at RRF, we are pursuing the three-dimensional structure of RRF. Preliminary nuclear magnetic resonance spectra data from large quantities of purified RRF (Hirashima and Kaji, 1972b; Ichikawa and Kaji, 1989) suggest that the three-dimensional structure is attainable by this approach. Preliminary attempts to obtain crystals of RRF have been successful. Once this novel protein's structure becomes available, antimicrobial drugs targeting it may be designed in a rational manner.

Bacterial growth

Bacterial strains were grown as described (Janosi et al., 1994). Media were supplemented with antibiotics such as tetracycline (TC, 12.5 g/ml), chloramphenicol (CM, 20 g/ml), kanamycin (KM, 50 g/ml), ampicillin (AP, 100 g/ml), 6% sucrose, 0.4 mM isopropyl--D-thiogalactoside (IPTG) and 40 g/ml 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-gal) when needed.

Bacterial strains and plasmids

Strains and plasmids were constructed with standard methods (Miller, 1972; Sambrook et al., 1989). The E.coli strains and plasmids used in this study are shown in Table I. Plasmid pPEN728 contains the wild-type frr (gene coding for RRF) from pRR2 (Shimizu and Kaji, 1991). For pPEN874, the temperature-resistant replicon of pSC101 was inserted in place of the Ts replicon of pMAK705. Plasmid pPEN907 was constructed from pPEN874 by inserting the wild-type frr with its promoter (Shimizu and Kaji, 1991) and transcriptional termination signal (Ichikawa and Kaji, 1989). Plasmid pPEN1560 was constructed from pPEN728 by adding the sacB-Neo^r (sucrose sensitivity and KM resistance) cassette from pIB209 (Blomfield et al., 1991) and deleting the 117 bp ScaI–XmnI fragment inactivating the AP^r gene.

Isolation of temperature-sensitive frr alleles

The PCR reaction with error-prone Taq polymerase was carried out on pPEN907 with the M13/pUC19 universal sequencing primers. The product was digested with KpnI and HindIII, ligated to pPEN874 resulting in a plasmid mixture carrying mutated frr and CM^r, (pMIX). E.coli LJ3 cells (frr1 Fs, Fs stands for frame shift) harboring pPEN1560 (contains wild-type frr) were transformed with pMIX followed by selection for CM^r and sucrose-resistant transformants at 32°C and screened for temperature-sensitive growth at 43°C by replica plating. The chromosomal frr of MC1061 was replaced with Ts frr alleles on cloned plasmid as described earlier (Hamilton et al., 1989; Janosi et al., 1994) to obtain LJ13, LJ14, LJ15 and LJ16.

Semiquantitative characterization of the temperature-sensitive growth phenotype

E.coli strain LJ3 (frr1Fs) carrying the Ts frr alleles in pMW119 or LJ13, LJ14, LJ15 and LJ16 harboring Ts frr allele in the bacterial chromosome were grown overnight in Luria–Bertani broth (LB) medium (Janosi et al., 1994) at 32°C. The cultures were diluted serially 10-fold and 15 l aliquots were spotted onto Luria agar (LA; Janosi et al., 1994) plates, incubated overnight at 32°C, 37°C and 43°C and the dilutions giving rise to 10–20 colonies were recorded.

RRF and other assays

Wild-type RRF from E.coli DH5 harboring pRR1 (carrying wild-type frr; Ichikawa and Kaji, 1989) and Ts mutant RRF from E.coli LJ14 harboring pPEN2399 (carrying Tsfrr) were purified and assayed as described (Hirashima and Kaji, 1972b). Purified EF-G was a gift from Dr Barend Kraal (Faculty of Mathematics and Natural Sciences, Leiden Institute of Chemistry, Leiden, The Netherlands). Protein concentrations and -galactosidase activity were measured by the DC Protein Assay kit (Bio Rad) and the Miller method (Miller, 1972), respectively.

lacZ'-'lacZ gene fusion and re-initiated products

Plasmids pPEN2345, pPEN2367, pPEN2363 and pPEN2369 were constructed from pLKC481 and pLKC480 (American Type Culture Collection) and pUC19 (Yanisch-Perron et al., 1985). They have the following junction sequence: pPEN2345, Plac (promoter for lac)-RBS (ribosome binding sequence) -ATG-36 nucleotides-ACT-39 nucleotides-GTC (10th codon of lacZ)---; pPEN2367, Plac-RBS-ATG-36 nucleotides-ACT-37 nucleotides-GTC (10th codon of lacZ)---; pPEN2363, Plac-RBS-ATG-36 nucleotides-TAG-39 nucleotides-GTC (10th codon of lacZ)---; pPEN2369, Plac-RBS-ATG-36 nucleotides-TAG-37 nucleotides-GTC (10th codon of lac Z)---. The initiation codon ATG of pPEN2345 is in-frame with -galactosidase ('lacZ) while that of pPEN2367 is in the -2 frame. The termination codon is at the 14th triplet of these plasmids (pPEN2363 and pPEN2369) and is in-frame with the initiation codon ATG. The termination codon TAG of pPEN2363 is in-frame with the downstream -galactosidase ('lacZ) while that of pPEN2369 is in the -2 frame. The Edman analysis (six cycles) (Matsudaira, 1993) was carried out on 50–200 pmol of affinity-purified (Protosorb lacZ, Promega) -galactosidase induced upon inactivation of RRF.

The 3A' reporter system

The 3A' plasmids pSM–T272 and pSM–T275 were constructed from plasmid pSM11 (Mottagui-Tabar et al., 1994), pAB7 (Björnsson and Isaksson, 1993) and necessary synthetic nucleotides. They have the following junction sequence between the upstream 2A' and the downstream 3A' cistrons; pSM–T272; 2A'-TAA-8 nucleotides-GGT-9 nucleotides-ATG-3A'. pSM–T275; 2A'-TAA-9 nucleotides-GGT-9 nucleotides-ATG-3A'. Bold TAA is the termination codon for the upstream 2A' segment and bold ATG is the initiation codon for the downstream 3A' cistron. Procedures for growth of bacterial cultures harboring either pSM–T272 or pSM–T275, purification of protein-A' species, separation on SDS–PAGE and determination of band densities were as described earlier (Björnsson et al., 1996).

Plasmid pSEK336C derived from IS3

Plasmid pSEK336C was prepared from pSEK336 (Sekine et al., 1994) by deletion of a HindIII fragment within the cI857 gene to inactivate the repressor. Plasmid pSEK336C carries an IS3 (insertion sequence 3) segment where the 3' portion of orfA ('orfA) and the 5' portion of orfB (orfB') are allocated in a tandem array. The lacZ reporter gene is in-frame with orfB' which is in -1 frame with respect to 'orfA. The IS3 segment has the following sequence: ATG-16 nucleotides-AACAG-29 nucleotides-CTGAA---. C represents the mutation changing the initiation codon ATG of orfB' to CTG. The bold ATG and TGA are the initiation codon and the termination codon for orfA' respectively. To avoid confusion with ribosomal frameshifting which occurs from orfA to orfB in the natural IS3 sequence, the sequence causing the programmed frame shift has been eliminated by a mutation (the bold C above) (Sekine et al., 1994). The cells harboring pSEK336C were grown at 39°C and the cultures were kept in the logarithmic growth phase (OD₆₀₀ between 0.03 and 0.6). Purification of the induced -galactosidase and amino-terminal sequencing was performed as described (Sekine et al., 1994).

The authors thank Mr G.Hirokawa of the University of Penna and Dr E.Kaji of University of Chicago for critically reading this manuscript.

Aharonowitz Y and Ron EZ (1972) A temperature sensitive mutant in Bacillus subtilis with an altered elongation factor G. Mol Gen Genet, 119, 131–138. | Article | PubMed | ISI | ChemPort |

Björnsson A and Isaksson LA (1993) UGA codon context which spans three codons: reversal by ms²i⁶A37 in tRNA, mutation in rpsD(S4) or streptomycin. J Mol Biol, 232, 1017–1029. | Article | PubMed | ISI

Björnsson A, Mottagui-Tabar S and Isaksson LA (1996) Structure of the C-terminal end of the nascent peptide influences translation termination. EMBO J, 15, 1696–1704. | PubMed | ISI |

Blomfield IC, Vaughn V, Rest RF and Eisenstein BI (1991) Allelic exchange in Escherichia coli using the Bacillus subtilis sacB gene and a temperature-sensitive pSC101 replicon. Mol Microbiol, 5, 1447–1457. | Article | PubMed | ISI | ChemPort |

Bult CJ et al. (1996) Complete genome sequence of the methanogenic archaeon, Methanococcus jannaschii. Science, 273, 1058–1073. | Article | PubMed | ISI | ChemPort |

Cole JR, Olsson CL, Hershey JWB, Grunberg-Manago M and Nomura M (1987) Feedback regulation of rRNA synthesis in Escherichia coli. Requirement for initiation factor IF2. J Mol Biol, 198, 383–392. | Article | PubMed | ISI | ChemPort |

Cummings H and Hershey J (1994) Translation initiation factor IF1 is essential for cell viability in Escherichia coli. J Bacteriol, 176, 198–205. | PubMed | ISI | ChemPort |

Dahlfors AAR and Kurland CG (1990) Novel mutants of elongation factor G. J Mol Biol, 215, 549–557. | Article | PubMed | ISI

Fowler AV and Zabin I (1977) The amino acid sequence of -galactosidase of Escherichia coli. Proc Natl Acad Sci USA, 74, 1507–1510. | Article | PubMed | ChemPort |

Fraser CM et al. (1995) The minimal gene complement of Mycoplasma genitalium. Science, 270, 397–403. | Article | PubMed | ISI | ChemPort |

Grunberg-Manago M (1996) Regulation of the expression of aminoacyl-tRNA synthetases and translation factors. In Neidhardt,F.C. et al. (eds), Escherichia coli and Salmonella typhimurium Cellular and Molecular Biology. ASM Press, Washington DC, pp. 1432–1457.

Hamilton CM, Aldea M, Washburn BK, Babitzke P and Kushner SR (1989) New method for generating deletions and gene replacements in Escherichia coli. J Bacteriol, 171, 4617–4622. | PubMed | ISI | ChemPort |

Hashimoto-Gotoh T and Sekiguchi M (1977) Mutations to temperature sensitivity in R plasmid pSC101. J Bacteriol, 131, 405–412. | PubMed | ChemPort |

Himmelreich R, Hilbert H, Plagens H, Pirkl E, Bi-hen L and Herrmann R (1996) Complete sequence analysis of the genome of the bacterium Mycoplasma pneumoniae. Nucleic Acids Res, 24, 4420–4449. | Article | PubMed | ISI | ChemPort |

Hirashima A and Kaji A (1970) Factor dependent breakdown of polysomes. Biochem Biophys Res Commun, 41, 877–883. | Article | PubMed | ISI | ChemPort |

Hirashima A and Kaji A (1972a) Factor-dependent release of ribosomes from messenger RNA. Requirement for two heat-stable factors. J Mol Biol, 65, 43–58. | Article | PubMed | ISI | ChemPort |

Hirashima A and Kaji A (1972b) Purification and properties of ribosome-releasing factor. Biochemistry, 11, 4037–4044. | Article | PubMed | ISI | ChemPort |

Hirashima A and Kaji A (1973) Role of elongation factor G and a protein factor on the release of ribosomes from messenger ribonucleic acid. J Biol Chem, 248, 7580–7587. | PubMed | ISI | ChemPort |

Hou Y, Lin Y-P, Sharer JD and March PE (1994) In vivo selection of conditional-lethal mutations in the gene encoding elongation factor G of Escherichia coli. J Bacteriol, 176, 123–129. | PubMed | ISI | ChemPort |

Ichikawa S and Kaji A (1989) Molecular cloning and expression of ribosome releasing factor. J Biol Chem, 264, 20054–20059. | PubMed | ISI | ChemPort |

Janosi L, Shimizu I and Kaji A (1994) Ribosome recycling factor (ribosome releasing factor) is essential for bacterial growth. Proc Natl Acad Sci USA, 91, 4249–4253. | Article | PubMed | ChemPort |

Janosi L, Hara H, Zhang S and Kaji A (1996a) Ribosome recycling by ribosome recycling factor (RRF) - an important but overlooked step of protein biosynthesis. Adv Biophys, 32, 121–201. | Article | PubMed | ISI | ChemPort |

Janosi L, Ricker R and Kaji A (1996b) Dual functions of ribosome recycling factor in protein biosynthesis: disassembling the termination complex and preventing translational errors. Biochimie, 78, 959–969. | Article | PubMed | ISI | ChemPort |

Kawakami K, Inada T and Nakamura Y (1988) Conditionally lethal and recessive UGA-suppressor mutations in the prfB gene encoding peptide chain release factor 2 of Escherichia coli. J Bacteriol, 170, 5378–5381. | PubMed | ISI | ChemPort |

Keiler KC, Waller PRH and Sauer RT (1996) Role of a peptide tagging system in degradation of proteins synthesized from damaged messenger RNA. Science, 271, 990–993. | Article | PubMed | ISI | ChemPort |

Kozak M (1987) Effects of intercistronic length on the efficiency of re-initiation by eucaryotic ribosomes. Mol Cell Biol, 7, 3438–3445. | PubMed | ISI | ChemPort |

Lindahl L and Hinnebusch A (1992) Diversity of mechanisms in the regulation of translation in prokaryotes and lower eukaryotes. Curr Opin Genet Dev, 2, 720–726. | Article | PubMed | ChemPort |

Matsudaira P (1993) Introduction. In Matsudaira,P. (ed.), A Practical Guide to Protein and Peptide Purification for Microsequencing. Academic Press, New York, pp. 1–14.

Miller JH (1972) Experiments in Molecular Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Mottagui-Tabar S, Björnsson A and Isaksson LA (1994) The second to last amino acid in the nascent peptide as a codon context determinant. EMBO J, 13, 249–257. | PubMed | ChemPort |

Nakamura Y, Ito K and Isaksson LA (1996) Emerging understanding of translation termination. Cell, 87, 147–150. | Article | PubMed | ISI | ChemPort |

Ogawa K and Kaji A (1975) Requirement for ribosome-releasing factor for the release of ribosomes at the termination codon. Eur J Biochem, 58, 411–419. | Article | PubMed | ISI | ChemPort |

Olsson CL, Graffe M, Springer M and Hershey JWB (1996) Physiological effects of translation initiation factor IF3 and ribosomal protein L20 limitation in Escherichia coli. Mol Gen Genet, 250, 705–714. | Article | PubMed | ISI | ChemPort |

Pavlov MYu, Freistroffer D, MacDougall J, Buckingham RH and Ehrenberg M (1997) Fast recycling of Escherichia coli ribosomes requires both ribosome recycling factor (RRF) and release factor RF3. EMBO J, 16, 4134–4141. | Article | PubMed | ISI | ChemPort |

Ryden M, Murphy J, Martin R, Isaksson L and Gallant J (1986) Mapping and complementation studies of the gene for release factor 1. J Bacteriol, 168, 1066–1069. | PubMed | ISI | ChemPort |

Ryoji M, Berland R and Kaji A (1981a) Re-initiation of translation from the triplet next to the amber termination codon in the absence of ribosome-releasing factor. Proc Natl Acad Sci USA, 78, 5973–5977. | Article | PubMed | ChemPort |

Ryoji M, Karpen JW and Kaji A (1981b) Further characterization of ribosome releasing factor and evidence that it prevents ribosomes from reading through a termination codon. J Biol Chem, 256, 5798–5801. | ISI | ChemPort |

Sambrook J, Fritsch EF and Maniatis T (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Sekine Y, Eisaki N and Ohtsubo E (1994) Translational control in production of transposase and in transposition of insertion sequence IS3. J Mol Biol, 235, 1406–1420. | Article | PubMed | ISI | ChemPort |

Shimizu I and Kaji A (1991) Identification of the promoter region of the ribosome-releasing factor cistron (frr). J Bacteriol, 173, 5181–5187. | PubMed | ISI | ChemPort |

Tiedeman AA and Smith JM (1988) lac ZY gene fusion cassettes with Kan^R resistance. Nucleic Acids Res, 16, 3587. | Article | PubMed | ISI | ChemPort |

Vizcaino N, Cloeckaert A, Dubray G and Zygmunt M (1996) Cloning, nucleotide sequence, and expression of the gene coding for a ribosome releasing factor-homologous protein of Brucella melitensis. Infect Immun, 64, 4834–4837. | PubMed | ISI | ChemPort |

Wada A, Igarashi K, Yoshimura S, Aimoto S and Ishihama A (1995) Ribosome modulation factor: Stationary growth phase-specific inhibitor of ribosome functions from Escherichia coli. Biochem Biophys Res Commun, 214, 410–417. | Article | PubMed | ISI | ChemPort |

Yamagishi M, Matsushima H, Wada A, Sakagami M, Fujita N and Ishihama A (1993) Regulation of the Escherichia coli rmf gene encoding the ribosome modulation factor: growth phase- and growth rate-dependent control. EMBO J, 12, 625–630. | PubMed | ISI | ChemPort |

Yanisch-Perron C, Vieira J and Messing J (1985) Improved M13 phage cloning vectors and host strains: nucleotide sequences of the M13mp18 and pUC19 vectors. Gene, 33, 103–119. | Article | PubMed | ChemPort |

Main navigation

• Search EMBO Journal

  Search

• Advanced search

• The EMBO Journal

  • Home
  • Aims and scope
  • Current issue
  • Advance online publication
  • Web focuses
  • Archive:
  • Browse by issue
  • Browse by subject
  • Browse by category
  • Free online sample issue
  • Press releases

• Authors and Referees

  • Editorial process
  • Guide for authors
  • Submit an article
  • Guide for referees
  • Editors and Editorial Board
  • Contact editorial office

• Customer Service

  • Subscribe
  • Order sample copy
  • Purchase articles
  • Reprints and permissions
  • Contact NPG
  • Advertising