Ribosome rescue: tmRNA tagging activity and capacity in Escherichia coli


  • Sean D. Moore,

    Corresponding author
    1. Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139, USA.
      E-mail smoore@mit.edu; Tel. (+1) 617 258 0791; Fax (+1) 617 258 0673.
    Search for more papers by this author
  • Robert T. Sauer

    1. Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139, USA.
    Search for more papers by this author

E-mail smoore@mit.edu; Tel. (+1) 617 258 0791; Fax (+1) 617 258 0673.


When protein synthesis stalls in bacteria, tmRNA acts first as a surrogate tRNA and then as an mRNA in a series of reactions that append a peptide tag to the nascent polypeptide and ‘rescue’ the ribosome. The peptide tag encoded by wild-type tmRNA promotes rapid degradation of rescued proteins. Using a mutant tmRNA that encodes a tag that does not lead to degradation, we demonstrate that the synthesis of approximately 0.4% of all proteins terminates with tagging and ribosome rescue during normal exponential growth of Escherichia coli. The frequency of tagging was not significantly increased in cells expressing very high levels of tmRNA and its binding protein SmpB, suggesting that recognition of ‘stalled’ ribosomes does not involve competition between tmRNA and other translation factors for A-sites that are unoccupied transiently during protein synthesis. When the demand for ribosome rescue was increased artificially by overproduction of a non-stop mRNA, tmRNA levels did not increase but tmRNA-mediated tagging increased substantially. Thus, the ribosome-rescue system usually operates well below capacity.


In bacteria, translating ribosomes ‘stall’ when the A-site is empty because the end of the mRNA has been reached without encountering a stop codon, levels of the cognate tRNA required to decode the A-site codon are low, or translation termination is inefficient (Keiler et al., 1996; Roche and Sauer, 1999; 2001; Hayes et al., 2002; Ueda et al., 2002). In each case, the stalled ribosome can be rescued through the actions of tmRNA, the product of the ssrA gene, and SmpB, a dedicated tmRNA-binding protein (Keiler et al., 1996; Karzai et al., 1999; Withey and Friedman, 2003; Hallier et al., 2004). During ribosome rescue, the tRNA-like portion of tmRNA, charged with alanine, acts as a surrogate tRNA to accept the nascent polypeptide. Then the ribosome switches templates from the stalled mRNA to a small open-reading frame (ORF) in tmRNA. Translation resumes on this ORF and continues to a stop codon, allowing translation termination and recycling of the ribosome. The peptide tag appended to the rescued polypeptide is a signal for proteolytic degradation, ensuring that proteins whose normal synthesis cannot be completed are destroyed (Keiler et al., 1996).

The tmRNA/SmpB system is universally conserved in bacteria (Keiler et al., 2000; Withey and Friedman, 2003). Although tmRNA is not essential in Escherichia coli, it increases the growth rate and resistance to several antibiotics that affect protein synthesis (de la Cruz and Vioque, 2001; Abo et al., 2002). In other bacteria, the importance of the tmRNA system is more pronounced. It is essential for the growth of Neisseria gonorrhoeae (Huang et al., 2000) and Mycoplasma sp. (Hutchison et al., 1999), allows the growth of Bacillus subtilis under harsh stress conditions (Muto et al., 2000), and is required for proper developmental timing in Caulobacter crescentus (Keiler and Shapiro, 2003). In terms of cell viability, the ribosome-rescue function of tmRNA appears to be more important than its function in protein-quality control (Huang et al., 2000; Withey and Friedman, 2003).

Ribosome rescue and protein tagging by the tmRNA system provides a window into translational events that have gone awry, but surprisingly little is known about the frequency of these events in E. coli relative to normal levels of translation. The E. coli tmRNA ORF encodes the sequence ANDENYALAA, which can be re-engineered to make tagged proteins refractory to degradation within the cell. For example, changing the six C-terminal residues to HHHHHH (His6) in the tmRNA-H6 mutant was used to slow the degradation of tagged proteins for proteomic studies, revealing the existence of a very large number of tagged species (Roche and Sauer, 2001). Here, we use tmRNA-H6 as a tool to estimate the frequency of ribosome rescue and tagging in E. coli and to establish the capacity of this system. We show that 0.4% or more of all protein synthesis reactions terminate via the tmRNA-mediated rescue pathway, with an overall capacity at least three- to fourfold higher than this level.

Two mechanisms for the early stages of ribosome rescue can be envisioned. In one model, tmRNA initiates ribosome rescue simply by competing with charged tRNAs or release factors for A-site binding. In this ‘kinetic’ model, the length of time that an A-site remains unoccupied determines the probability that tmRNA will initiate ribosome rescue, and the level of tagging should therefore be proportional to the cellular concentration of tmRNA. In an alternative model, a stalled ribosome acquires some unique structural feature that allows specific recruitment of tmRNA. For example, ribosome stalling in vivo can lead to mRNA cleavage at or near the A-site codon (Hayes and Sauer, 2003; Sunohara et al., 2004), and tmRNA mediates efficient tagging of nascent chains on ribosomes stalled on truncated mRNAs in vitro (Ivanova et al., 2004). We show that 20-fold overproduction of tmRNA-H6 causes no significant increase in tagging or slowing of cellular growth and conclude that tmRNA must recognize stalled ribosomes by some direct mechanism.


Expression and efficiency of tmRNA-H6

We replaced the wild-type tmRNA gene in the chromosome of E. coli strain X90 with a gene encoding tmRNA-H6. Northern hybridization studies showed that log-phase cultures [37°C, Luria–Bertani (LB) broth] contained tmRNA-H6 at levels similar to wild-type tmRNA in the parental strain (Fig. 1, lanes 2 and 3). Quantification showed that the level of mature tmRNA-H6 was approximately 95% of that of wild-type tmRNA. The numbers of wild-type tmRNA or tmRNA-H6 molecules per cell (650–700) were roughly 5% of the total number of ribosomes as estimated from the ratio of tmRNA to 5S ribosomal RNA (data not shown), which is generally consistent with a previous report (Lee et al., 1978).

Figure 1.

Cellular levels of tmRNA-H6. Northern blot of total cellular RNA probed with a 32P-labelled oligonucleotide complementary to the 3′ end of mature tmRNA. The ‘control’ RNA (516 nucleotides, including sequences from tmRNA and the E. coli 5S, 16S and 23S ribosomal RNAs) was added to each sample to allow corrections for losses during sample handling.

In an effort to overexpress tmRNA-H6, we introduced a multicopy plasmid containing the tmRNA-H6 gene (Roche and Sauer, 2001) into a strain lacking the chromosomal tmRNA gene. Interestingly, we found that this strain expressed tmRNA-H6 at levels no higher than strains containing the chromosomal gene (Fig. 1, lanes 3 and 4). When we inserted the smpB gene in its normal position upstream of plasmid-encoded tmRNA-H6, however, we observed accumulation of tmRNA-H6 to levels about 20 times higher than in the chromosomal construct (Fig. 1, lane 5). Cells overexpressing plasmid-borne SmpB/tmRNA-H6 doubled only slightly more slowly (37 min) than cells expressing chromosomal tmRNA-H6 (32 min). This result suggests that having tmRNA at levels roughly equal to the total number of ribosomes does not interfere with normal translation to any substantial degree. Processing of a precursor transcript is required to generate mature tmRNA (Komine et al., 1994; Li et al., 1998). Only a small amount of the precursor was observed in cells overproducing tmRNA-H6 (Fig. 1, lane 5), indicating that processing still occurred efficiently under these conditions.

To determine the efficiency of tmRNA-H6 in ribosome rescue and tagging, we expressed a non-stop mRNA (no in frame stop codons) that stalled ribosomes at its 3′ end and determined the fraction of the encoded ‘reporter’ protein that was tagged. In cells lacking tmRNA, the nascent protein accumulated as an untagged ‘fall-off’ product as expected (Fig. 2A, lane 1; Keiler et al., 1996). When wild-type tmRNA was present, no significant level of the ‘fall-off’ product was observed, indicating that stalled ribosomes were efficiently rescued with subsequent degradation of the tagged reporter protein (Fig. 2A, lane 2). In cells expressing tmRNA-H6 from the chromosome, approximately 70% of the recovered reporter protein was tagged and 30% was untagged (Fig. 2A, lane 3). The relative amounts of tagged and untagged forms of the reporter protein may not quantitatively reflect the efficiency of ribosome rescue because the stabilities and nickel-binding properties of the two forms may be different. Nonetheless, this result is consistent with studies that suggest tmRNA-H6 is less active than wild-type tmRNA (Roche and Sauer, 2001). Plasmid overexpression of SmpB/tmRNA-H6 resulted in complete tagging of the reporter protein (Fig. 2A, lane 4), demonstrating that increased levels compensate for lower activity. Studies presented below show that the demand for ribosome rescue is significantly higher than normal in cells expressing the non-stop mRNA. Thus, tagging by the tmRNA-H6 system provides a reasonable estimate of total ribosome-rescue activity in the cell.

Figure 2.

Assays employing a non-stop mRNA that stalls ribosomes.
A. Cartoon of the gene for the non-stop mRNA encoded by pPW500; trpAt is a strong transcriptional terminator from the trpA attenuator (trpAt) (Keiler et al., 1996). SDS-PAGE of proteins purified by Ni-NTA affinity chromatography from cells harbouring pPW500 that lack tmRNA (lane 1), contain the wild-type chromosomal tmRNA gene (lane 2), contain the chromosomal tmRNA-H6 gene (lane 3), or contain a plasmid overexpressing SmpB and tmRNA-H6 (lane 4). Cells were induced with IPTG for 1 h prior to harvest. The gel was stained with Coomassie blue.
B. Denaturing PAGE of total RNA isolated from cells containing pPW500. Cultures were split and IPTG was added to one aliquot for 1 h prior to harvest. The control RNA described in Fig. 1 was added to each lysate at ∼500 copies per cell prior to RNA purification. The top gel was stained with Sybr-Green II dye. In the bottom panel, RNA from a duplicate gel was transferred to a membrane and probed using 32P-labelled oligonucleotides complementary to the control RNA and to the non-stop RNA.

To determine whether tmRNA levels increased in response to higher demand for ribosome rescue, we purified total RNA from cells induced or uninduced for expression of the non-stop mRNA and used PAGE and staining to assay the levels of non-stop mRNA and tmRNA (Fig. 2B). These experiments showed that the level of tmRNA did not change in response to overexpression of the non-stop mRNA in wild-type cells, in cells containing chromosomal tmRNA-H6, or in cells containing the SmpB/tmRNA-H6 plasmid. Non-stop mRNA induction also failed to increase transcription from a fusion of the natural promoter for tmRNA to lacZ (data not shown). The non-stop mRNA, which runs above tmRNA on the gel, was present at higher levels in the strain lacking tmRNA than in the strains containing wild-type tmRNA or tmRNA-H6. This finding agrees with reports that ribosome rescue accelerates the degradation of stalled mRNAs (Hayes and Sauer, 2003; Yamamoto et al., 2003). Among the strains containing tmRNA, the induced level of the non-stop mRNA was highest with chromosomal tmRNA-H6, next highest with the SmpB/tmRNA-H6 plasmid, and lowest in the strain with wild-type tmRNA. These results are consistent with lower activity of chromosomal tmRNA-H6 relative to wild-type tmRNA.

How common is ribosome rescue?

To assay normal levels of ribosome rescue in cells containing tmRNA-H6, we developed an indirect competitive elisa. Following cell disruption under non-denaturing conditions, His6 epitopes in the lysate were determined by comparison with a standard curve generated using a purified protein with the same C-terminal peptide (AANDEH HHHHH) added during ribosome rescue by tmRNA-H6. Log-phase cultures with chromosomal tmRNA-H6 contained approximately 7000 His6-tagged proteins per cell (Fig. 3A); tagging levels were similar for stationary-phase cultures (not shown). Tagging was elevated only slightly in cells overproducing SmpB/tmRNA-H6 from a plasmid (Fig. 3A). Thus, an approximate 20-fold increase in tmRNA-H6 concentration in the cell did not increase the levels of ribosome rescue and tagging appreciably. We conclude that ribosome rescue is unlikely to occur by an A-site competition mechanism (see Discussion). In addition, the level and activity of chromosomal tmRNA-H6 appears to be sufficient to satisfy normal demands for ribosome rescue.

Figure 3.

Proteins tagged by tmRNA-H6 ribosome rescue.
A. An indirect competitive elisa was used to quantify the amount of His6 epitope in cell lysates, which were matched for total protein content. The error bars represent standard deviations of four independent measurements.
B. Western blot probed with an anti-His6 antibody of proteins from strains containing the chromosomal tmRNA gene (lane 1), the chromosomal tmRNA-H6 gene (lane 2), or the chromosomal tmRNA-H6 gene with null alleles of the clpP, clpX and lon genes (lanes 3–5). The samples in lanes 1–3 were prepared from equal numbers of starting cells. Lane 4 has half as much protein as lane 3. Lane 5 has half as much protein as lane 4.

A steady-state concentration of about 7000 tmRNA-tagged proteins per cell establishes a lower limit for the number of rescue events that would need to occur during each cell generation. However, some rescue reactions probably produce non-native polypeptides that are degraded irrespective of the nature of the attached tag sequence. Such degradation would lower steady-state levels and result in an underestimate of the rate of ribosome rescue. To test this possibility, we constructed a strain containing chromosomal tmRNA-H6 with clpP, clpX and lon mutations to inactivate the ClpAP, ClpXP and Lon proteases (Kanemori et al., 1997). The level of His6-tagged molecules in this strain increased almost twofold (Fig. 3A), with at least 13 000 protein-synthesis reactions being terminated by tmRNA-mediated ribosome rescue during each cell generation. This steady-state level of tagged proteins represents about 0.4% of the steady-state level of all cellular proteins. However, because some tmRNA-tagged proteins may still be degraded faster than bulk proteins, the number of protein synthesis reactions that terminate in tmRNA tagging and ribosome rescue could be higher than 0.4% (see Discussion and Experimental procedures).

To characterize tmRNA-tagged proteins, Western blotting experiments with anti-His6 antibodies were used to analyse total cell lysates from the protease proficient and deficient strains, allowing comparisons of the relative sizes and abundance of specific tagged species (Fig. 3B). In accord with the elisa data (Fig. 3A), the protease-deficient lysate had roughly twice the immunoreactivity of the protease-proficient lysate. Interestingly, although the same His6-tagged species appeared to be present in both lysates, certain bands were clearly more prominent in the clpP, clpX , lon lysate, suggesting that ClpAP, ClpXP, and/or Lon, degrade these proteins in the parental strain.

Ribosome-rescue capacity

Do the tagging levels determined in the previous section represent the maximal capacity of tmRNA-mediated ribosome rescue? Two plasmids constructs were used to address this question. One (pPWSM) produced an IPTG-inducible non-stop mRNA for a reporter protein with no encoded His6 epitope; the other (pPW-stops) was isogenic except for tandem in frame termination codons. After induction of expression from pPWSM in a strain with chromosomal tmRNA-H6, the tagged reporter protein was the principal His6-tagged protein in the cell (Fig. 4A, lane 4) and the total number of His6-tagged proteins per cell increased to approximately 34 000 (Fig. 4B). This increase depended on production of the non-stop mRNA, as induction of the normal mRNA from pPW-stops resulted in the expected steady-state level of roughly 7000 His6-tagged proteins per cell (Fig. 4B). Hence, the capacity of chromosomal SmpB/tmRNA-H6 is substantially higher than needed to deal with the normal cellular demand for ribosome rescue. Overexpression of SmpB/tmRNA-H6 increased the level of His6-tagged proteins in the pPWSM strain modestly to roughly 42 000 per cell (Fig. 4B). This result is consistent with our finding that tmRNA-H6 expressed from the chromosomal locus rescues most but not all of the ‘stalled’ ribosomes under high demand conditions in the cell.

Figure 4.

Levels of tmRNA-H6 ribosome rescue in cells with or without non-stop RNA.
A. Left– Cartoons showing reporter genes in pPWSM, which expresses a non-stop mRNA with no encoded His6 sequence, and pPW-stops, which expresses an mRNA that is isogeneic except for tandem in frame stop codons (UAAUAA) before the transcriptional terminator. Right– SDS-PAGE of cellular proteins purified by Ni-NTA affinity chromatography and stained with Coomassie blue. Lane 1, molecular-weight standards; lane 2, sample from strain with wild-type chromosomal tmRNA gene; lane 3, uninduced sample from strain with chromosomal tmRNA-H6 gene and pPWSM; and lane 4, IPTG-induced sample from strain with chromosomal tmRNA-H6 gene and pPWSM.
B. An indirect competitive elisa was used to quantify His6 epitopes in cell lysates from strains containing chromosomal tmRNA-H6 (1×) or overexpressed tmRNA-H6 (20×) and pPW-stops or pPWSM. The three lysates shown on the right were prepared from cultures induced with IPTG for 60 min prior to harvest. Error bars were calculated as described in Fig. 2.

As discussed above, if tmRNA-H6-tagged proteins are degraded faster than bulk proteins, then steady-state levels of tagged proteins in the cell underestimate the fraction of translation reactions that terminate with ribosome rescue. To determine the intracellular stabilities of tmRNA-H6-tagged proteins, cells containing chromosomal tmRNA-H6 and pPWSM were induced with IPTG to express the reporter protein for 60 min and then treated with spectinomycin to stop further protein synthesis. Time points were taken after stopping translation, and His6-tagged proteins were separated by SDS-PAGE and detected by Western analysis (Fig. 5A, lanes d–h). Even though the tagged N-domain was the most abundant tagged protein at ‘time zero’ (Fig. 5A), it appeared less abundant than in stained gels (Fig. 4A). Indeed, we found that the tagged N-domain reproducibly generated a relatively weak signal on Western blots. To generate a rough standard curve, the ‘time zero’ sample was subjected to twofold serial dilutions into a control lysate that did not contain His6-tagged proteins (Fig. 5A, lanes a–c). The tagged N-domain protein had a half-life of approximately 38 min. Other major bands had half-lives of roughly 240 min (49 kDa), 150 min (6 kDa) and 100 min (30 kDa). The prominence of these tagged ‘background’ proteins therefore results both from a reasonably high rate of synthesis/rescue and from relatively slow rates of degradation.

Figure 5.

Degradation/accumulation of proteins tagged by tmRNA-H6 ribosome rescue.
A. Cells containing chromosomal tmRNA-H6 and pPWSM were induced with IPTG for 60 min, 100 µg ml−1 of spectinomycin was added to halt protein synthesis, and aliquots were taken at different times after spectinomycin addition. Samples were lysed and proteins were resolved by SDS-PAGE and detected by Western blotting using an anti-His6 antibody. Lanes d–h show samples taken from 0 to 120 min following spectinomycin addition. The zero-time sample (lane d) was serially diluted with a His6–free control lysate to generate a standard curve (lanes a–c; triangle). The half-life of the tagged N-domain protein is approximately 38 min.
B. Growth curves of strains containing chromosomal tmRNA-H6 and pPWSM. Cells were grown to early log phase and one culture was induced to express the non-stop mRNA.
C. Following induction of the pPWSM-encoded non-stop mRNA with IPTG, samples were taken at different times from the cultures shown in B, and the concentration of His6 epitope in each lysate was determined by elisa. The solid line is a single-exponential fit of the data (see Experimental procedures).

To obtain independent estimates of the rates of ribosome rescue and degradation of the reporter protein, we induced its expression in cells containing chromosomal tmRNA-H6 and pPWSM and used the elisa to monitor the kinetics of His6-protein accumulation (Fig. 5C). In this experiment, the increase in steady-state His6-protein levels following induction results from tagging of the N-domain reporter. The observed increase in tagged proteins fit well to a single-exponential function with a rate constant of 0.033 min−1. This value represents the sum of rate constants for the dilution of tagged protein from cell division and growth (kgrow) and degradation (kdeg) of the reporter protein (see Experimental procedures). Based on a cell-doubling time of 110 min following induction (Fig. 5B), we calculate a degradation half-life of 26 min for the tagged reporter protein. This value is probably more accurate than the 38 min half-life in the spectinomycin experiment, but the important point is that both experiments give generally similar results. From the new steady-state plateau in the Fig. 5C experiment, a ribosome-rescue rate of 840 min−1 can be calculated for the reporter protein (see Experimental procedures).


Active recognition of stalled ribosomes

Detecting ribosome-rescue events in vivo requires a method of preventing or slowing proteolysis of tmRNA-tagged proteins. Replacing the last six residues of the degradation tag encoded by wild-type tmRNA with His6 allows detection of tagged proteins (Roche and Sauer, 2001), but we found that high-demand rescue by tmRNA-H6 was mildly compromised unless this variant was overproduced. The reason for the reduced activity of the tmRNA-H6 is unknown. In other studies, we have assayed the activities of approximately 10 tmRNAs with variant ORFs encoding unrelated amino-acid sequences including a commonly used mutant in which the last two codons encode aspartate instead of alanine. In every case, these mutants were also less active than wild-type tmRNA (data not shown), suggesting that some feature in or near the wild-type ORF participates in ribosome rescue and/or in stabilizing tmRNA. The tmRNA-H6 variant was the most active tmRNA mutant tested.

Interestingly, plasmid-mediated tmRNA-H6 overproduction required the presence of the smpB gene at its natural upstream position. This effect could result from additional transcription from the smpB promoter or from SmpB protection of tmRNA-H6 from ribonuclease cleavage. We favour the protection model for several reasons: (i) SmpB is known to increase the intracellular levels and lifetimes of wild-type tmRNA (Karzai et al., 1999; Keiler and Shapiro, 2003; Hallier et al., 2004); (ii) With smpB present on the plasmid, tmRNA-H6 levels were 20-fold higher than normal, but neither an smpB-tmRNA transcript nor a fragment expected for cleavage of the smpB portion of such a transcript was observed; and (iii) Supplying SmpB in trans from a separate plasmid also increased tmRNA-H6 levels (S.D. Moore, unpubl. obs.), suggesting that SmpB protein and not smpB transcription was responsible for tmRNA overexpression.

Although overproduced tmRNA-H6 was very abundant and present at levels similar to ribosomal RNAs, cells grew only 15% more slowly despite the additional burden of RNA synthesis. Hence, dramatic over accumulation of SmpB/tmRNA-H6 does not appear to interfere significantly with normal translation. Moreover, overproduction of tmRNA-H6 and SmpB increased the level of tagging of cellular proteins only slightly. It might be argued that a factor other than SmpB limits tmRNA activity in the cell, but these two molecules alone are sufficient to bind the A-site of purified ribosomes (Valle et al., 2003) and to mediate tagging and rescue in highly purified translation systems (Ivanova et al., 2004). Collectively, these results suggest that SmpB/tmRNA recognizes stalled ribosomes actively and does not compete for A-sites that are unoccupied transiently during normal translation. Recognition of stalled ribosomes may depend on having fewer than 15 nucleotides of mRNA-3′ to the A-site (Ivanova et al., 2004). This situation would arise if a ribosome reached the end of a non-stop message or if the mRNA was cleaved in the A-site or at the 3′ boundary of the ribosome-mRNA complex during translational stalling (Keiler et al., 1996; Hayes and Sauer, 2003; Sunohara et al., 2004). SmpB associates with ribosomes in the absence of tmRNA (Hallier et al., 2004) and interacts with both the ribosome and tmRNA during the initial stages of rescue (Valle et al., 2003), suggesting that it may recognize ‘stalled’ ribosomes and recruit tmRNA subsequently.

Normal levels of ribosome rescue

An E. coli cell contains ≈ 1.5 × 10−13 g of protein in a volume of 10−15 l (Chohji et al., 1976), corresponding to ≈ 3 million proteins with an average molecular weight of 30 kDa. Most of these proteins have lifetimes substantially longer than the cell generation time (Nath and Koch, 1970; Mosteller et al., 1980). In an otherwise wild-type strain containing tmRNA-H6, we measured a steady-state level of roughly 7000 proteins per cell generated by ribosome rescue. The steady-state level of rescued proteins increased to approximately 13 000 per cell when the activities of the ClpAP, ClpXP and Lon proteases were eliminated. To generate this level of tagged protein, 0.4% of all translation reactions would need to terminate in tmRNA rescue, assuming that rescued proteins and bulk proteins have comparable half-lives in this strain. If rescued proteins in the clpPclpXlon strain were degraded faster than bulk protein, then the rate or ribosome rescue would have to be even higher.

Because E. coli strains lacking tmRNA exhibit only modest growth defects, the importance of ribosome rescue could be questioned. However, if cellular mechanisms to free stalled ribosomes did not exist, then the number of operational ribosomes in the cell would diminish rapidly. For example, the steady-state level of tmRNA-tagged proteins that we observed in the protease-deficient strain was roughly comparable to the total number of ribosomes. Hence, in the absence of any rescue mechanism, most ribosomes would become inoperative within a single cell generation. Rescue by tmRNA cannot be the only mechanism to liberate stalled ribosomes in E. coli, because proteins encoded by non-stop mRNAs do accumulate in the absence of tmRNA (Keiler et al., 1996; Sunohara et al., 2004). Indeed, recent studies have shown that overexpression of tmRNA reduces the cellular requirement for peptidyl-tRNA hydrolase and that strains with mutations in the gene for this enzyme (pth) display more severe temperature sensitivity when tmRNA is absent (Singh and Varshney, 2004). These results suggest that ‘drop off’ of the peptidyl-tRNA molecule from the ribosome provides an alternative way to free stalled ribosomes, with peptidyl-tRNA hydrolase being needed to cleave the protein-RNA linkage to regenerate functional tRNA. Ribosome rescue by tmRNA is clearly more efficient than this alternative competing process, which is only observed at significant levels in tmRNA-defective strains.

In bacteria containing wild-type tmRNA, the protein products of ribosome rescue are marked for degradation by the ssrA tag (Keiler et al., 1996). ClpXP is largely responsible for degradation of these ssrA-tagged proteins in the cytoplasm of E. coli (Gottesman et al., 1998; Farrell et al., 2005). Recent studies show that cellular ClpXP levels are sufficient to degrade approximately 100 000 copies of a model ssrA-tagged protein per generation (Farrell et al., 2005). Thus, cytoplasmic degradation of ssrA-tagged proteins produced by ribosome rescue under normal conditions would be well within the proteolytic capacity of the cell.

Capacity of the ribosome-rescue system

By overexpressing a non-stop mRNA encoding the N-domain of λ repressor, we increased the cellular demand for ribosome rescue and directed tmRNA tagging mainly to one protein substrate (Figs 2A and 4A). Cells lacking tmRNA did not form colonies when plated on 1 mM IPTG to induce expression of the non-stop mRNA (data not shown), suggesting that the pth system alone is not capable of enough ribosome rescue to allow efficient synthesis of housekeeping proteins. Induction of the non-stop message in cells encoding chromosomal tmRNA-H6 was accompanied by an increase in the steady-state level of tmRNA-tagged proteins to roughly 34 000 per cell. This result indicates that E. coli has significant reserve capacity for ribosome rescue, although growth of these cells was slowed substantially upon induction of the non-stop mRNA. The capacity of the tmRNA/SmpB system appears to be at least three- to fourfold greater than needed to deal with normal cellular demands for ribosome rescue. Mutants of B. subtilis lacking polynucleotide phosphorylase, a-3′-to-5′ exoribonuclease, accumulate truncated mRNAs and show a threefold increase in tmRNA tagging (Oussenko et al., 2005). Interestingly, overexpression of tmRNA improves the growth of these strains, suggesting that the capacity of the wild-type B. subtilis tmRNA system has been exceeded under these conditions.

Experimental procedures

Strains and plasmids

Strains were derivatives of E. coli X90 (araΔ (lac-pro) nalA argEam rif thi-1/F′lacIqlac+pro+) (Parsell and Sauer, 1989) or X90 ssrA::cat (Keiler et al., 1996). To introduce tmRNA-H6 into the chromosome, we constructed a plasmid containing E. coli yfjG, smpB, ssrA and intA with a kanR marker inserted between ssrA and intA. The ORF of ssrA was then mutated to encode tmRNA-H6 (ANDEHHHHHH) as described (Roche and Sauer, 2001). This plasmid was used to generate a linear DNA fragment that was recombined into the chromosome of strain W3110 using λ-RED (Datsenko and Wanner, 2000) to generate SM682. P1 transduction was used to introduce kanR-linked tmRNA-H6 into X90 to generate SM694; this construct was verified by diagnostic polymerase chain reaction (PCR) and sequencing of the smpB and ssrA genes. A strain with multiple protease deficiencies was generated by P1 transduction of the ΔclpP-lon::cat marker from strain KY2263 (Kanemori et al., 1997) into SM694.

A PCR fragment containing the smpB promoter and gene was cloned upstream of the tmRNA-H6 gene in pKW-24 (Roche and Sauer, 2001) to generate pSM680. This p15A origin plasmid contained 111 nucleotides upstream of the smpB ORF and the wild-type smpBssrA intergenic region. Plasmid pPW500 encodes the N-domain of λ repressor followed by an M2 (FLAG) epitope, a His6 tag and the trpAt terminator (Keiler et al., 1996). Plasmid pPWSM was generated by PCR removal of the M2 and His6 coding sequences from pPW500; pPW-stops was generated from pPWSM by PCR insertion of tandem UAA codons immediately after the λ repressor N-domain coding sequence.

Northern analysis

An RNA containing nucleotide sequences of immature tmRNA (−20 through +47), 5S rRNA (13–36), 16S rRNA (1516–1542) and 23S rRNA (2872–2904) was transcribed from a T7 promoter in vitro. This RNA was precipitated with isopropanol, washed with 75% ethanol, and suspended in 100 µl DNase buffer [10 mM TrisCl, 2 mM magnesium acetate, 10 U ml−1 RNase-free DNase I (Roche)]. After 30 min at room temperature, samples were extracted with acidic phenol/chloroform (Ambion, Austin, TX), precipitated with isopropanol, washed with ethanol, resuspended in 10 mM TrisCl (pH 7.3), and stored at −80°C. The concentration of this ‘control’ RNA, which contained ∼95% full-length product in denaturing PAGE, was determined using an extinction coefficient of 5.08 × 106 M−1cm−1 at 260 nm.

Cultures were grown in LB broth at 37°C to an OD600 of 1, and 1 ml (∼109 cfu) was taken chilled on ice for 10 min, harvested at 4°C, and suspended in 400 µl of Trizol reagent (Invitrogen). Control RNA (80 mol cfu−1), linear polyacrylamide (10 µg) and 80 µl of chloroform were added. Samples were vortex mixed and centrifuged at 20 000 g for 15 min at room temperature. RNA in the top phase was purified largely as described for the control RNA, and resuspended in 50 µl formamide loading dye buffered with 50 mM bis-TrisCl (pH 6.5). One tenth of the sample was resolved by denaturing TBE-urea PAGE, electro-transferred to a nylon membrane (Nytran SuperCharge, Schleicher and Schuell), and cross-linked with UV light. The membrane was washed for 60 min with 0.1 SSC, 0.1% SDS, and prehybridized with 6× SSC, 10× Denhardt’s, 0.1% SDS for 1 h at 65°C. 32P-end-labelled oligonucleotide (40 pmol) was added to the hybridization buffer and allowed to hybridize overnight at 55°C. The membrane was washed three times with 6× SSC, 0.1% SDS and exposed to a phosphorimager plate (Molecular Dynamics). Image analysis and integration of band intensities was performed using ImageQuant software (Molecular Dynamics).

Reporter protein purification

Cells harbouring pPW500 were induced for 1 h with 1 mM IPTG, chilled on ice for 10 min, and roughly 109 cells were harvested at 4°C, and suspended in 150 µl of DB buffer (6 M GuHCl, 25 mM potassium-HEPES, 0.1% Tween, 2 mM imidazole, pH 7.5). After vigorous mixing, insoluble material was removed by centrifugation (15 min at 20 000 g), the supernatant was transferred to a new tube with 10 µl of Ni-NTA agarose resin (Qiagen, prewashed in DB buffer), and the sample was rocked for 30 min at room temperature. Resin was collected by centrifugation (2000 g for 1 min), washed twice with 500 µl of 8.3 M urea, 50 mM sodium phosphate, 10 mM TrisCl, 10 mM imidazole, pH 8.0, washed twice with 500 µl of 10 mM TrisCl (pH 8.0), and dried under vacuum. Dried resin was rehydrated in 30 µl 8.3 M urea, 100 mM acetic acid to elute the bound proteins. The masses of the ‘fall-off’ and ‘tagged’ pPW500 translation products in the purified sample were within 1 Da of the expected masses as determined by MALDI-TOF mass spectrometry (M.I.T. Biopolymers facility). For SDS-PAGE, an aliquot of the sample was neutralized with 100 mM Tris base before addition of SDS sample buffer.

Total RNA staining

RNA was purified from 1 ml of log-phase cultures (OD600 ≈ 1) as described above, except for the addition of more control RNA (∼ 500 molecules cfu−1). One tenth of the sample was electrophoresed on denaturing urea-PAGE and stained for 15 min with SybrGreen II (Molecular Probes) according to the manufacturer's instructions. The gel was rinsed with H2O and imaged using a fluorimager (Molecular Dynamics, no filter, 488 nm excitation). In control experiments, tmRNA and the control RNA responded linearly to dose in gels stained with SybrGreen II at concentrations at least 10-fold higher and lower than those shown in the Fig. 2B gel.

Native lysate preparation

Cell cultures were chilled on ice for 10 min, harvested by centrifugation at 4°C, resuspended (∼20 OD600 ml−1 in 15% glycerol, 100 mM NaCl, 10 mM TrisCl, pH 8.0), and stored at −80°C. Samples were thawed on ice, reharvested, and resuspended at ∼33 OD600 ml−1 in lysis buffer (B-Per II (Pierce) plus 1 mM EDTA, 0.01 mg ml−1 egg-white lysozyme (EM Science; fresh 10 mg ml−1 stock), and 1/10th volume protease inhibitor cocktail [‘Complete, EDTA-free’ (Roche); one tablet per millilitre of water)]. After lysis at room temperature (about 3 min), lysates were diluted to ∼25 OD600 ml−1 with nuclease solution [B-Per II augmented with 10 mM magnesium acetate and 0.25 U ml−1 Benzonase (Novagen)]. After incubation for 1 min, lysates were cleared by centrifugation at 20 000 g for 15 min and supernatants were transferred to new tubes. Protein concentrations were determined by Bradford assay (Bio-Rad). Absorbance values determined in triplicate were averaged and compared with a standard curve generated using serial dilutions of BSA (Sigma). Lysates generated from 2.5 × 1010 cells had 1.7–1.8 mg total soluble protein.

Western analysis

Proteins were separated by SDS-PAGE and transferred to PVDF membrane (Millipore) in 10 mM CAPS, 20% methanol (pH 11) in a semidry transfer apparatus (Owl) according to the manufacturer's instructions. After transfer, the membrane was shaken for 1 h in THBST (150 mM NaCl, 50 mM TrisCl, 10 mM sodium-HEPES, 0.1% Tween, pH 7.4) supplemented with 5 mg ml−1 BSA. The membrane was washed once with THBST and incubated with HRP-conjugated anti-His6 polyclonal antibody (Santa Cruz) at 1/15 000 dilution in THBST for 1 h. After three 5 min washes, the membrane was treated with an HRP-reactive chemiluminescent reagent (ECL-Plus, Pierce) according to the manufacturer's instructions and exposed to film.


The concentration of His6 epitopes in native lysates was determined using an indirect, competitive elisa and two purified AANDEH6-tagged proteins. The first protein (λN-AANDEH6) was expressed from a derivative of pPW500, modified to encode a C-terminal AANDEH6 immediately following the λ repressor N-domain and M2 (FLAG) epitope. The second protein (GFP-AANDEH6) was expressed from a derivative of the λN-AANDEH6 plasmid, in which the gene for GFP mutant ‘3b’ (Cormack et al., 1996) was modified by PCR to encode a C-terminal AANDEH6 and then cloned under control of the Ptrc promoter in place of the λN-AANDEH6 gene. Both AANDEH6 tagged proteins were purified by a combination of Ni-NTA affinity chromatography (Qiagen) and anion-exchange chromatography (Hi-Trap Q, Amersham). Purified protein from the final column was concentrated (10 000 MWCO Centricon filter, Millipore) and exchanged into storage buffer (20 mM sodium-HEPES, 100 mM NaCl, 1 mM EDTA, pH 7.5). Concentrations were determined using extinction coefficients of 6400 M−1cm−1 (λN-AANDEH6; 280 nm; 6 M GuHCl) or 29 330 M−1cm−1 (GFP-AANDEH6; 292 nm; 0.1 N NaOH).

Individual wells of a 96-well plate (MaxiSorp, Nunc) were coated with 50 µl of λN-AANDEH6 (50 nM in 100 mM sodium carbonate, pH 9.5) for 1 h at 37°C. Unbound protein was aspirated from the wells, which were then washed with 50 µl, 2 × 100 µl, and then 200 µl of SuperBlock (Pierce). For each experiment, a lysate was prepared from strain X90 (no H6 epitopes) to use as a control and for dilution of lysates containing AANDEH6 epitopes. A series of GFP-AANDEH6 dilutions (typically from 0.5 nM to 25 µM) was prepared in SuperBlock, and 25 µl of each dilution was added to an empty well of the coated plate followed by 25 µl of control lysate. For each plate, samples for duplicate standard curves were included. Experimental samples were prepared in triplicate; 25 µl of SuperBlock was added to each well followed by 25 µl of the lysate. In most cases, samples were also included in which experimental lysates were diluted into the control lysate. Monoclonal HRP-conjugated anti-His6 antibody (R&D Systems) was diluted 1/3000 in THBST and 50 µl was added to each well. The plate was incubated at room temperature for 2 h with gentle shaking, wells were cleared by aspiration and washed four times with THBST, a TMB solution (100 µl well−1; 1-step Slow-TMB, Pierce) was added, and the plate was incubated until blue colour developed in the standard-curve wells (about 1 h). Reactions were stopped by adding 100 µl of 1.5 M H2SO4, and the absorbance at 450 nm was determined for each well using a plate reader (SpectraMax). Wells that had not been coated with λN-AANDEH6 were used to calculate a blank value. Absorbance values for the standard curved were averaged for each concentration, plotted against log[GFP-AANDEH6] using Kaleidagraph (Synergy Software), and fitted in this program to the function D + (A − D)/1 + 10^([GFP-AANDEH6] −B•log (C), where D was the minimum plateau, A was the maximum plateau, C was the concentration at the mid-point of the transition, and B was the slope of the transition (Oellerich, 1984). AANDEH6 epitope concentrations in experimental samples were determined by calculation from the fitted standard function and varied by approximately 10% when independent samples prepared from aliquots of the same frozen cultures were analysed in separate experiments.

Steady-state concentrations of rescued proteins

With the assumption that protein degradation can be approximated as a first-order process, the rate of change in the concentration for any given protein tagged during ribosome rescue (TP) is given by:


where kgrow is the rate constant for cell doubling (ln(2)/tdouble). At steady state, krescue = (kdeg + kgrow)[TP ]steady-state. For a set of n rescued proteins with different rate constants for rescue and degradation,


In the absence of significant degradation, the total rescue rate is kgrow[TPtotal]stedy-sate; this expression underestimates the rescue rate as the degradation of tagged proteins increases. The approach to equilibrium in Fig. 5C was fit to the equation:



This work was supported by Grant NIH AI-16892. S.D.M. was supported by a postdoctoral fellowship from the Damon Runyon Cancer Research Foundation. We thank C. Farrell, J. Kenniston, A. Martin and K. McGinness for their helpful discussions.