The SmpB-tmRNA-mediated trans-translation system has two well-established activities: rescuing ribosomes stalled on aberrant mRNAs and marking the associated protein fragments for proteolysis. Although the causative non-stop mRNAs are known to be degraded, little is known about the enabling mechanism or the RNases involved in their disposal. We report that Escherichia coli has an enabling mechanism that requires RNase R activity and is dependent on the presence of SmpB protein and tmRNA, suggesting a requirement for active trans-translation in facilitating RNase R engagement and promoting non-stop mRNA decay. Interestingly, this selective transcript degradation by RNase R targets aberrant (non-stop and multiple-rare-codon containing) mRNAs and does not affect the decay of related messages containing in-frame stop codons. Most surprisingly, RNase II and PNPase do not play a significant role in tmRNA-facilitated disposal of aberrant mRNAs. These findings demonstrate that RNase R is a crucial component of the trans-translation-mediated non-stop mRNA decay process, thus providing a requisite activity well suited to complement the ribosome rescue and protein tagging functions of this unique quality control system.
In prokaryotes, transcription and translation are spatio-temporally coupled. Each message is co-transcriptionally recognized and decoded by multiple ribosomes. Therefore, a quality control mechanism is essential for preventing translation of defective messages and locking of multiple ribosomes in futile translation cycles. The eubacterial trans-translation system has evolved to recognize ribosomes stalled on non-stop mRNAs and help resolve the obstruction (Karzai et al., 2000; Withey and Friedman, 2003). SmpB protein and tmRNA, the key required components of trans-translation, form a complex that recognizes stalled ribosomes and acts to rescue individual 70S ribosomes (Keiler et al., 1996; Karzai et al., 1999; 2000; Barends et al., 2001; Sundermeier et al., 2005; Dulebohn et al., 2006). Once bound in the A-site of a stalled ribosome, tmRNA initially acts as a tRNA to transfer the incomplete polypeptide to itself. It then acts as a surrogate mRNA to direct the translation (trans-translation) of a open reading frame encoded within tmRNA, thus appending an 11-amino-acid degradation tag to the end of the incomplete peptide. The appended tag marks the potentially toxic protein fragment for proteolysis by C-terminal specific proteases (Gottesman et al., 1998; Herman et al., 1998). The tmRNA peptide-reading frame also encodes an authentic stop codon to promote normal translation termination and ribosome recycling. The SmpB-tmRNA system thus solves two critical problems associated with non-stop or defective mRNAs. A third, and scarcely explored, problem associated with defective mRNAs is the persistence of the causative aberrant mRNA, which if not degraded could engage the ribosomal machinery in futile cycles of translation. Studies have shown a link between the identification of stalled ribosomes by the trans-translation system and degradation of aberrant mRNAs (Yamamoto et al., 2003). However, little is known about how these defective mRNAs are recognized and removed or which RNases are involved in their degradation. Additionally, not much is known about the mechanism by which the putative RNases and the trans-translation machinery are coupled.
Levels of gene expression are determined via a fine balance between transcriptional efficiency and post-transcriptional events, namely, message stability and translatability. The signals that govern the rate of transcription and translation are well understood. Messenger RNAs have widely varying degrees of stability; however, the signals that govern transcript stability are poorly understood. Initial indications are that non-stop mRNA decay is carried out primarily by 3′-to-5′ exonucleases (Tu et al., 1995). Out of the eight exoribonucleases identified in Escherichia coli, three (RNase R, RNase II and PNPase) have been implicated in 3′-to-5′ mRNA turnover (Deutscher, 2003; 2006). These three exoribonucleases show significant functional overlap in vivo, though some data suggest a level of specificity for each enzyme (Deutscher, 2006). None of these three RNases are individually essential for E. coli, though double mutants of PNPase and either RNase II or RNase R are lethal (Deutscher, 2006), suggesting some level of functional overlap. RNase R and RNase II belong to the RNR family of RNases and exhibit similarities in domain architecture and sequence (Zuo and Deutscher, 2001). Both enzymes work by a hydrolytic mechanism and release nucleoside monophosphates. PNPase on the other hand is a phosphorolytic nuclease that generates nucleoside 5′-diphosphates. RNase R, encoded by the vacB or rnr gene, was originally identified as a residual 3′-to- 5′ exoribonuclease activity in an RNase II-deficient E. coli strain (Gupta et al., 1977; Kasai et al., 1977). RNase R is highly processive and degrades RNAs containing significant secondary structure, including ribosomal RNAs, and mRNAs containing stable secondary structure elements (Cheng et al., 1998; Cheng and Deutscher, 2005). Interestingly, RNase R was found associated with a ribonucleoprotein complex containing SmpB protein and tmRNA (Karzai and Sauer, 2001). Despite this biochemical association, the exact role of RNase R in trans-translation, processing of tmRNA, or degradation aberrant mRNAs has not been well defined.
We are interested in gaining a better mechanistic understanding of the trans-translation process, particularly in deciphering the role of tmRNA in facilitating non-stop mRNA decay. In a related study, we have identified tmRNA determinants required for non-stop mRNA decay (J. Richards, P. Mehta and W. Karzai, in preparation). The work herein was undertaken to identify RNases involved in degradation of aberrant mRNAs that promote ribosome stalling. We present evidence to demonstrate that RNase R is the key 3′-to-5′ exoribonuclease involved in the tmRNA-mediated degradation of non-stop mRNAs, in a process that requires active trans-translation of the defective mRNA.
RNases potentially involved in tmRNA-dependent non-stop mRNA decay
We sought to determine which E. coli RNases were involved in degradation of defective transcripts that trigger ribosome stalling. Toward this end, we utilized a well-characterized non-stop mRNA expression construct (pλ-cI-N-trpAt) that produces a transcript, which lacks an in-frame stop codon and thus promotes ribosomes to stall at its 3′ end. We transformed wild-type (WT) and tmRNA-deficient (ssrA–) strains with the λ-cI-N-trpAt non-stop reporter mRNA expression plasmid, and either a plasmid encoding tmRNA or the control vector (pKW1). We isolated total cellular RNA from cells expressing the λ-cI-N-trpAt non-stop mRNA and used Northern blot analysis to measure the steady-state level of the reporter transcript. Evaluation of the fate of the λ-cI-N-trpAt non-stop mRNA showed high levels of reporter transcript accumulation in cells lacking tmRNA (Fig. 1, lane 3). In contrast, in the presence of tmRNA only a fraction of the non-stop transcript accumulated in these cells (Fig. 1, lane 1). The degradation of the non-stop mRNA required the presence of tmRNA, irrespective of whether it was present on the chromosome or supplied on a plasmid (Fig. 1, lanes 1 and 2). These results demonstrated that E. coli cells were capable of efficiently degrading non-stop mRNAs, even if expressed at high levels. These data also showed that the degradation process was dependent on the presence of tmRNA, suggesting that tmRNA is capable of facilitating the selective degradation of non-stop mRNAs. These findings also provided us with a facile assay to search for ribonucleases required for the disposal of aberrant mRNAs.
Using this mRNA decay assay, we evaluated the effect of disruption of specific RNase genes on the stability of the λ-cI-N-trpAt non-stop mRNA. We constructed E. coli strains where the chromosomal copy of ssrA was disrupted in combination with one or more genes that are known to encode for enzymes involved in RNA metabolism (Table 1). The first strain we tested for the level of tmRNA-dependent mRNA degradation was a multi-RNase-deficient strain with disruptions in genes encoding ribonucleases I, II, D and BN (Fig. S1). Northern blot analysis showed this strain to be fully proficient in degradation of non-stop mRNAs, as no greater than wild-type accumulation of λ-cI-N-trpAt non-stop mRNA was observed in this background. These results indicated that RNases I, II, D and BN were not involved in tmRNA-dependent decay of defective mRNAs. Of particular interest was the finding that RNase II, a key enzyme implicated in general mRNA degradation, was not involved in tmRNA-mediated non-stop mRNA decay. Next, we tested a mutant strain deficient in the ribonucleases T and PH functions. Previous studies have shown that although RNase T and PH are important for maturation of the 3′-end of tmRNA, RNase BN, and a number of other exoribonucleases, can effectively participate in the maturation reaction in vivo (Li et al., 1998). The RNases T and PH-deficient strain was fully proficient in tmRNA-dependent non-stop mRNA decay (Fig. S1), showing wild-type steady-state levels of λ-cI-N-trpAt non-stop mRNA. These results indicated that ribonucleases T and PH, like RNase I, II, D and BN, were not involved in tmRNA-dependent non-stop mRNA decay.
Table 1. Strains used in this study.
E. coli genome project, University of Wisconsin, Madison
This work. CA244 II– R– parental strain, ssrA::cat transductant
X90 II– tmRNA–
This work. MG1655 II– parental strain, rnb::kan transductant
X90 R– tmRNA–
This work. X90 R– (rnr::kan) parental strain, ssrA::cat transductant
X90 PNP– tmRNA–
This work. X90 PNP– parental strain, ssrA::cat transductant
PNPase and RNase II do not participate in tmRNA-facilitated decay of non-stop mRNA
In E. coli, PNPase, RNase II and RNase R form a group of 3′-to-5′ exoribonucleases that are known to be responsible for degradation of mRNA fragments (Mohanty and Kushner, 2000; 2003; Zuo and Deutscher, 2001; Deutscher, 2006). These three enzymes are known to have significant functional overlap in vitro. First, we examined a double RNase-deficient strain, with combined disruptions in genes encoding the RNase II and RNase R exoribonucleases. This analysis revealed a substantial accumulation of the λ-cI-N-trpAt non-stop mRNA in rnb–rnr– cells, irrespective of the presence or absence of tmRNA (Fig. S2), indicating that tmRNA-dependant degradation of defective mRNAs was not functional in this background. Next, we investigated the possible involvement of PNPase in the degradation of the λ-cI-N-trpAt reporter mRNA. In a PNPase-deficient strain, degradation of non-stop mRNA proceeded as in wild-type cells, indicating PNPase was not involved in tmRNA-dependent mRNA degradation (Fig. S3). We observed an identical non-stop mRNA degradation phenotype in an independently derived pnp disruption strain, suggesting that the PNPase phenotype was due to pnp gene disruption and not other unrelated mutations (Table 1, and data not shown).
Given these observations, we endeavoured to dissect the individual contributions of RNase R and RNase II to non-stop mRNA decay. To this end, we assessed the stability of λ-cI-N-trpAt reporter transcript in strains lacking either RNase II or RNase R function. We first examined an RNase II-deficient strain (Fig. 2A). Northern blot analysis of the reporter transcript showed the RNase II-deficient strain to be fully proficient in tmRNA-dependent non-stop mRNA decay (Fig. 4A and B), showing wild-type steady-state levels of λ-cI-N-trpAt non-stop mRNA. In contrast, the RNase R-deficient strain exhibited a pronounced defect in non-stop mRNA decay, accumulating substantial amount of the λ-cI-N-trpAt reporter transcript. These results indicated that RNase R was required for the tmRNA-dependant degradation of non-stop mRNA (Fig. 2C and D). Additionally, these data demonstrated that expression of the λ-cI-N-trpAt non-stop mRNA was not affected by the presence or absence of RNase R or tmRNA, only its stability was subject to tmRNA-dependent degradation by RNase R. In the absence of tmRNA (Fig. 2C, lane 2), despite the fact that all RNases (including RNase R) were present, the λ-cI-N-trpAt transcript was expressed and accumulated in the cell, consistent with a requirement for tmRNA in non-stop mRNA decay process. Likewise, in the absence of RNase R (Fig. 2C, lanes 3–4), regardless of the presence or absence of tmRNA, the λ-cI-N-trpAt transcript was expressed and accumulated in these cells, consistent with a requirement for RNase R in decay process. Taken together, these data support the conclusion that RNase R degrades non-stop mRNAs in a tmRNA-dependent manner. Therefore, we conclude that of the three E. coli 3′-to-5′ exoribonucleases involved in general mRNA degradation, tmRNA-dependent non-stop mRNA decay is a function unique to RNase R.
RNase R is involved in the selective degradation of non-stop mRNA
The increased steady-state level of the reporter λ-cI-N-trpAt transcript in the rnr– strain strongly indicated that RNase R was involved in the selective degradation of non-stop mRNAs. To determine the relative half-life of the reporter non-stop mRNA, we performed rifampicin chase experiments in an rnr– mutant and its isogenic parental strain. Cells were grown to mid-log phase and expression of the reporter mRNA was induced for 15 min by the addition of IPTG, at which point rifampicin was added to prevent further rounds of transcription initiation. Total cellular mRNA was isolated at various time points post rifampicin addition and subjected to Northern blot analysis (Fig. 3). This analysis showed that in the wild-type background the λ-cI-N-trpAt transcript had a half-life of approximately 2.5 min (Fig. 3A). In contrast, in the rnr– strain the non-stop λ-cI-N-trpAt transcript was substantially stabilized with a half-life of approximately 12 min (Fig. 3B). These data demonstrate that RNase R has a major effect on the stability of non-stop mRNA, as its absence in cells increased the half-life of the λ-cI-N-trpAt non-stop mRNA by four- to fivefold.
To examine whether the observed defect in non-stop mRNA decay was due solely to RNase R function, we took three independent but complementary approaches. First, we evaluated λ-cI-N-trpAt mRNA stability in an independent rnr gene disruption strain (rnr::cm), where the rnr gene was disrupted with the chloroamphenicol-resistance gene marker rather than the original kanamycin-resistance gene marker (rnr::kan). We obtained identical non-stop mRNA decay defects with the rnr::cm strain as with the rnr::kan strain (data not shown). Second, we complemented the non-stop mRNA decay defect associated with the rnr– strains by supplying RNase R function in trans on a plasmid (pRNR). To this end, we constructed a plasmid harbouring the rnr gene under the control of its native promoter (see Experimental procedures). An rnr– strain (rnr::kan) harbouring this plasmid showed significantly greater degradation of the non-stop λ-cI-N-trpAt transcript than the deletion strain alone. Using this construct, the non-stop mRNA degradation phenotype was restored to near wild-type levels (Fig. 4), indicating that the observed defects in non-stop mRNA decay were due to loss of RNase R function and not other unrelated mutations. Third, using the rnr complementation plasmid, we made strategic inactivating mutations in the catalytic domain of RNase R. In contrast to the wild-type complementing plasmid, vectors expressing two independent catalytically inactive RNase R variants did not complement the non-stop mRNA decay defect of the rnr– strain (data not shown). Taken together, these results strongly support the conclusion that RNase R plays a direct role in tmRNA-dependant non-stop mRNA decay.
tmRNA-facilitated mRNA decay by RNase R is restricted to aberrant mRNAs
To examine whether the mRNA decay phenotype attributed to tmRNA and RNase R was specific to defective messages, we analysed the degradation of a related functional mRNA carrying an authentic in-frame stop codon. The construct we used for this assay was identical to the plasmid encoding λ-cI-N-trpAt, apart from the introduction of an in-frame stop codon at the 3′ end of the λ-cI-N coding sequence. This construct, λ-cI-N-Stop-trpAt, was designed to express a stable λ-cI N-terminal domain, and by virtue of having a normal translation termination signal, it should not have triggered ribosome stalling. Therefore, it should not have been a substrate for tmRNA-facilitated degradation by RNase R. We used Northern blot analysis to evaluate the steady-state intracellular levels of this λ-cI-N-Stop-trpAt reporter transcript in the presence or absence of tmRNA and RNase R. This analysis revealed that the λ-cI-N-Stop-trpAt reporter transcript accumulated in wild-type cells, irrespective of the presence or absence of tmRNA (Fig. 5, lanes 1–2). Similarly, this analysis revealed that the intracellular levels of the λ-cI-N-Stop-trpAt mRNA were not substantially affected by the presence or absence of RNase R (Fig. 5, lanes 3–4). We observed a slight increase in the steady-state levels of the reporter transcript in the rnr– cells, as expected if RNase R were involved in general mRNA homeostasis (Fig. 5B). These results suggest that only aberrant mRNAs that promote ribosome stalling are targeted by tmRNA for degradation by RNase R.
Next, we asked whether other defective mRNAs that are known to cause ribosome stalling are also a substrate for tmRNA-mediated degradation by RNase R. Previous work has shown that mRNAs harbouring multiple consecutive AGA or AGG rare codons cause ribosome stalling and are substrates for trans-translation (Roche and Sauer, 1999; Hayes and Sauer, 2003). It is thought that ribosomes translating these defective transcripts stall when they reach the consecutive rare codons, primarily due to cognate tRNA scarcity. Ribosome stalling then activates an as-of-yet unidentified endonucleolytic activity that cleaves the faulty transcript in the decoding centre of the ribosomal A-site, generating a non-stop mRNA (Hayes and Sauer, 2003; Sunohara et al., 2004a; Li et al., 2006). The non-stop mRNA-bound stalled ribosomes thus become substrates for the SmpB-tmRNA-mediated trans-translation process. To assess the role of RNase R in degradation of rare codon-containing mRNAs, we constructed a variant of the λ-cI-N-Stop mRNA, used by Roche and Sauer (1999), by introducing four consecutive AGG rare codons at the end of the λ-cI-N coding sequence but prior to the stop codon (Fig. 6A). As before, we induced the expression of the λ-cI-N-4-AGG-Stop mRNA and used Northern blot analysis to evaluate the role of tmRNA and RNase R in the degradation of this reporter mRNA (Fig. 6). Our analysis showed that in rnr+ssrA– background, cells lacking only tmRNA function, the rare codon-containing mRNA was expressed and co-translationally cleaved, as evidenced by the appearance of shorter λ-cI-N products below the full-length λ-cI-N-4-AGG-Stop transcript (Fig. 6B). However, the cleaved products accumulated in this RNase-proficient background, indicating tmRNA function was required for removal of the cleaved products (Fig. 6B, lane 1).
In contrast, in rnr+ssrA– cells supplemented with tmRNA function, the rare codon-containing mRNA was cleaved and the cleavage products were no longer detectable, suggesting that the selective removal of the cleaved mRNA products by RNase R was dependent on tmRNA (Fig. 6B, lane 2). In rnr– cells the λ-cI-N-4-AGG-Stop mRNA was cleaved but the cleavage products were stable, irrespective of the presence or absence or tmRNA (Fig. 6B, lanes 3–4). These results clearly demonstrated that degradation of rare codon-containing mRNAs was dependent on RNase R activity. Intriguingly, in cells used in these experiments only the rnr gene was disrupted, while all other RNase genes were intact (including genes encoding RNase II and PNPase functions), yet these cells still accumulated the cleavage products of the rare codon-containing mRNA. We also generated a similar rare-codon construct containing four consecutive AGA codons. Although we observed similar patterns of mRNA cleavage, and dependence on RNase R for disposal of the cleaved products, the 4-AGA construct was much less efficient in inducing ribosome stalling and mRNA cleavage (data not shown). In agreement with our earlier conclusions, these results indicated that RNase R was the key RNase involved in tmRNA-mediated degradation of non-stop mRNAs and that RNase II and PNPase did not make any substantial contribution to this process. Consistent with a previous report (Hayes and Sauer, 2003), we observed a slight decrease in the level of the full-length λ-cI-N-4-AGG-Stop mRNA in the presence of tmRNA (Fig. 6B, compare lane 3 and lane 4), supporting the conclusion that although tmRNA was not required for the endolytic activity, its presence made the cleavage process more efficient. Furthermore, these results showed that RNase R was not involved in the initial endonucleolytic activity that produced the non-stop mRNA. Based on these findings we conclude that RNase R plays a crucial role in tmRNA-dependent disposal of non-stop mRNAs, irrespective of how these defective mRNAs are generated.
SmpB protein is required for degradation of non-stop mRNAs
Thus far, we have shown that degradation of non-stop mRNA is performed by RNase R in a process that is dependent on the presence of tmRNA. To gain further insights, we wished to investigate whether the non-stop mRNA decay was simply a tmRNA-dependent process or if it also required other components of the trans-translation machinery. The SmpB protein is a key component of the trans-translation process, as in its absence trans-translation does not ensue. Specific and high-affinity binding of SmpB to tmRNA is essential for recognition and proper accommodation of tmRNA into stalled ribosomes (Karzai et al., 1999; Sundermeier et al., 2005; Dulebohn et al., 2006). A requirement for SmpB in enabling degradation of defective mRNAs would indicate that an active trans-translation process is needed for disposal of non-stop mRNAs. Therefore, we assessed the requirement for SmpB protein in disposal of non-stop mRNAs. We transformed an smpB deletion strain with the λ-cI-N-trpAt expression construct and examined the fate of the non-stop mRNA transcript in the presence and absence of tmRNA (Fig. 7A and B). The λ-cI-N-trpAt non-stop transcript accumulated to high levels in the smpB– background, irrespective of the presence or absence of tmRNA, suggesting SmpB function was required for tmRNA-facilitated non-stop mRNA degradation by RNase R. These data also suggested that the selective decay of defective mRNA by RNase R was dependent on active trans-translation and required both SmpB and tmRNA functions.
We have presented evidence to demonstrate that RNase R activity is essential for degradation of aberrant mRNAs in E. coli. Our data show that this targeted RNase R activity is dependent on the presence of SmpB protein and tmRNA, suggesting that active trans-translation is required to engage RNase R with its intended targets. Moreover, the tmRNA-facilitated mRNA decay appears to be selective, preferentially degrading aberrant mRNAs that promote ribosome stalling, while leaving related normal stop codon-containing transcripts unaffected. Of the three known processive 3′-to-5′ exoribonucleases, only RNase R is capable of this targeted mRNA decay. PNPase and RNase II do not appear to make a substantial contribution to tmRNA-mediated non-stop mRNA decay. Thus, by identifying RNase R as the enzyme charged with the disposal of aberrant mRNAs these findings answer a long-standing question about the SmpB-tmRNA-mediated trans-translation process.
Consistent with these findings, we had previously reported that RNase R co-purified with a ribonucleoprotein complex that contained SmpB protein and tmRNA (Karzai and Sauer, 2001). Based on biochemical evidence and phenotypic characterization of an rnr– strain, we had proposed a role for RNase R in degradation of defective mRNAs (Karzai and Sauer, 2001). However, a direct link between RNase R and trans-translation had not been established. Intriguingly, a number of indirect links do exist between RNase R and components of the trans-translation machinery. For instance, the rnr gene is located immediately upstream of the smpB gene in numerous bacteria. The functional coupling algorithm available on the STRING site (http://string.embl.de/), which is based on quantitative integration of interaction data from a large number of sources and organisms, assigns a high degree of confidence to colocalization of rnr and smpB genes. Additionally, rnr, smpB and ssrA have been shown to be virulence genes in several bacteria (Cheng et al., 1998; Julio et al., 2000; Okan et al., 2006), and rnr is one of three upregulated genes in Salmonella typhimurium ssrA mutants (Julio et al., 2000). All of these findings point to a functional link between RNase R and trans-translation. The significance of the work described herein lies in the fact that it firmly establishes a direct link between RNase R and trans-translation, by demonstrating an explicit role for this RNase in selective removal of non-stop mRNAs in a trans-translation-dependent manner.
Recent reports have documented a major role for RNase R in disposal of rRNA, defective tRNAs, and mRNAs containing extensive stable secondary structures (Deutscher, 2006). Additionally, a recent study suggested a role for RNase R in maturation of tmRNA during cold shock (Cairrao et al., 2003). Furthermore, in Caulobacter crescentus, RNase R has been shown to be involved in regulated degradation of the two-piece tmRNA in a cell cycle-dependent manner (Hong et al., 2005). Interestingly, in Caulobacter crescentus RNase R was reported not to be involved in tmRNA processing under cold shock, as well as a number of other altered growth conditions. Consistent with the latter report, although RNase R did co-purify with an SmpB–tmRNA complex, cells lacking RNase R function possessed efficient peptide tagging and ribosome rescue activities, suggesting that RNase R was not involved in tmRNA stability or processing (Karzai and Sauer, 2001). Indeed, RNase R has not been implicated in maturation of any RNA, and its processive nature makes such a processing role unlikely (Cheng and Deutscher, 2003; 2005). In a recent study, Chen and Deutscher demonstrated substantial elevation of RNase R level and activity in E. coli in response to a variety of adverse environmental conditions, including growth in cold shock, minimal media, and carbon, nitrogen, or phosphorus starvation (Chen and Deutscher, 2005). Although the details of the regulatory mechanisms that resulted in elevated RNase R activity were not determined, it was clear from the study that RNase R was subject to extensive physiological control both at the level of expression and activity. Considering this information, and the fact that most adverse environmental conditions could result in reduced availability of the building blocks for nucleic acid and protein synthesis, we postulate that trans-translation might become essential under these growth-limiting conditions. Indeed, trans-translation has been implicated in survival under adverse conditions and successful transition to new developmental stages (Julio et al., 2000; Hong et al., 2005; Okan et al., 2006).
Based on the new information presented here, RNase R appears to be an important component of the trans-translation machinery. Of particular interest is the finding that the tmRNA-facilitated degradation of transcripts by RNase R is restricted to defective mRNAs and does not affect related normal mRNAs. We do not wish to suggest that RNase R is not involved in general mRNA homeostasis. There is substantial evidence to show that RNase R plays a key role in cellular RNA metabolism (Deutscher, 2003; Cheng and Deutscher, 2005). Our findings support these conclusions, and further suggest that of the three key 3′-to-5′ RNases in E. coli (RNase II, RNase R and PNPase), RNase R has the unique distinction of being involved in tmRNA-mediated removal of defective mRNAs. However, in a recent report Yamamoto et al. have indicated (data not shown) that RNase R was not involved in tmRNA-mediated decay of faulty mRNAs (Yamamoto et al., 2003). As that report did not present any relevant data on RNase R, or any other RNases, we cannot evaluate further the reasons for the apparent inconsistency. Strain difference or acquired spurious mutations could provide possible explanations. Our findings clearly show that RNase R plays a critical role in tmRNA-mediated decay of both non-stop and rare codon-containing mRNAs, while it does not affect the decay of a related normal mRNA (Figs 5 and 6).
Of particular interest is the finding that other faulty transcripts that are known to promote ribosome stalling (Roche and Sauer, 1999), such as transcripts containing multiple rare codons, are also ultimately degraded by RNase R in a tmRNA-dependent manner (Fig. 6). Such rare codon-containing transcripts are known to be cleaved in a translation-dependent reaction by an as-of-yet unidentified endonucleolytic activity (Hayes and Sauer, 2003; Sunohara et al., 2004a,b; Li et al., 2006). Ribosomes stalled on these messages become a substrate for SmpB and tmRNA only after transcript cleavage has occurred. Significantly, our results demonstrate that RNase R is responsible for removal of the cleaved rare codon-containing transcripts, and that RNase R engagement is dependent on the presence of tmRNA (Fig. 6). Equally important, our data (Figs 3, 5–7) show that the expression of the reporter mRNA is not affected by the absence of tmRNA, RNase R or SmpB, supporting the conclusion that the reporter transcript is expressed normally in these cells and is degraded only if all three components (SmpB, tmRNA and RNase R) of the complete trans-translation process are present in the cell.
Interestingly, in a related study (P. Mehta, J. Richards and W. Karzai, in preparation) we have identified tmRNA determinants that, when mutated, affect only the RNase R-mediated defective mRNA decay activity without affecting the peptide tagging and ribosome rescue functions of tmRNA. Consistent with these findings, we have observed that in at least two previous studies non-stop mRNAs were found to accumulate in cells that either lacked tmRNA function or carried a tmRNA peptide-reading frame variant (Hayes and Sauer, 2003; Moore and Sauer, 2005). Although the exact timing of defective mRNA release from the rescued 70S ribosome has not been determined, it has been postulated to be released from the rescued ribosome prior to engagement of the tmRNA peptide-reading frame and resumption of translation with tmRNA as the surrogate template (Williams et al., 1999; Ivanova et al., 2005). Thus, RNase R must be engaged with the defective transcript at an early stage of the trans-translation process, most likely prior to the discharge of the faulty mRNA from the rescued ribosome. The fact that SmpB protein and tmRNA are both required, combined with the finding that certain mutations in tmRNA that adversely affect non-stop mRNA decay but do not affect peptide tagging or ribosome rescue (P. Mehta, J. Richards and W. Karzai, in preparation), suggests that RNase R engagement must be an active process. There are two possible mechanisms for an SmpB-tmRNA-facilitated engagement of RNase R with its target non-stop mRNA. First, RNase R could already be present on ribosomes and become activated only after the SmpB-tmRNA complex engages stalled ribosomes. Alternatively, SmpB protein or tmRNA might deliver RNase R directly to the defective mRNA prior to its release from the rescued 70S ribosome. Previous reports support the latter model (Karzai et al., 1999; Hong et al., 2005). However, direct and specific interaction between E. coli RNase R and SmpB protein or tmRNA have not been examined in detail, but are actively being pursued in our laboratory. Clearly, more work is required to elucidate the exact mechanism of how RNase R is loaded selectively on non-stop mRNAs and how the SmpB-tmRNA system facilitates this process. From the new findings presented here, it is distinctly evident that RNase R plays a major role in SmpB-tmRNA-mediated degradation of non-stop mRNAs.
In conclusion, this work represents a major step forward towards a better understanding of this fascinating biological quality assurance system. It provides clear evidence to demonstrate that RNase R is a crucial component of a complete trans-translation process, a process that in addition to rescuing stalled ribosomes and marking the associated protein fragments for proteolysis, also degrades the causative defective mRNAs in an SmpB and tmRNA-dependent fashion.
Strains and plasmids
The E. coli strains used in this work (Table 1) were grown in Luria–Bertani (LB) broth (10 g tryptone, 5 g yeast extract and 5 g NaCl per litre) or on LB plates (containing 1.5% agar). Ampicillin was used at 100 μg ml−1, kanamycin at 50 μg ml−1, tetracycline at 24 μg ml−1 and chloroamphenicol at 30 μg ml−1 where required. All mutations were generated in the E. coli X90 ssrA::cat background (hereafter referred to as ssrA–) (Keiler et al., 1996), using standard microbiological techniques. The pKW11 tmRNA expression plasmid, encoding ssrAWT, and the pKW1 control plasmid have been described previously (Karzai et al., 1999; Roche and Sauer, 1999). The pPW500 plasmid encoding the λ-cI-N-Flag-trpAt non-stop mRNA under the control of an IPTG-inducible promoter has been described previously (Keiler et al., 1996). The pRNR plasmid was constructed by insertion of an NdeI-XhoI digested PCR fragment containing the rnr gene, with its intrinsic promoter, into similarly cut pACYC184 vector. The absence of unintended mutations in the rnr gene was confirmed by DNA sequencing.
The λ-cI-N-Flag-Stop-trpAt mutation was introduced into the λ-cI-N-Flag-trpAt coding sequence of the pPW500 plasmid using appropriate mutagenic primers and the Quick-Change site-directed mutagenesis kit, in accordance with the manufacturer's recommendations. The multiple rare codon-containing reporter construct, λ-cI-N-4-AGG-Flag-stop, carries four consecutive AGG rare codons at the end of the λ-cI-N gene sequence, followed by a Flag-tag epitope and an in-frame stop codon. The rare-codon mutations were introduced by site-directed mutagenesis into the pER118 plasmid (Roche and Sauer, 1999), which harbours the coding sequence for the λ-cI-N-Flag-stop reporter construct. Individual clones were analysed by sequencing to confirm the presence of the intended mutations and verify that the remaining gene sequence was unaffected by the procedure.
Induction of λ-cI-N-trpAt non-stop mRNA
Escherichia coli X90-ssrA– cells harbouring pPW500 and a complementing plasmid borne copy of ssrA were grown overnight at 30°C with ampicillin and tetracycline selection. Fresh subcultures of ssrA– cells harbouring pPW500 and a complementing plasmid borne copy of ssrA were grown at 37°C in LB media to an OD600 of 0.5–0.6. The cultures were induced with IPTG at 1 mM final concentration. The control strain, lacking ssrA, contained the pPW500 and the pKW1 control plasmids. Cells were grown for 90 min post induction and 3 ml of cells harvested by centrifugation. E. coli X90 cells are rifampicin resistant therefore experiments aimed at analysing the decay rate of the reporter mRNA were performed using MG1655-rnr– and the isogenic wild-type cells harbouring pPW500, the λ-cI-N non-stop mRNA expression plasmid. In these experiments the induction period with IPTG was 15 min. At this point, rifampicin was added to the culture and 3 ml samples of the culture were taken at 0, 2, 4, 8, 16 and 32 min after the addition of rifampicin. Whole cell pellets were resuspended in equal volumes of sterile water, and OD600 measured to enumerate cell numbers. Total RNA from equal numbers of cells (corresponding to equal OD600) was then extracted using Tri-Reagent (Molecular Research Center). The quantity of recovered RNA was determined by absorbance at 260 nm.
Northern blot analysis
Equal amounts of RNA for each sample were resolved by electrophoresis on a denaturing 1.5% (v/v) formaldehyde agarose gel. Fifteen microlitres of RNA samples was mixed with 10 μl of denaturing solution (6.5 μl formamide, 2 μl formaldehyde, 1.5 μl 10× MOPS buffer), heated at 65°C for 10 min and 5 μl of sample dye added (2% bromophenol blue, 10 mM EDTA, 50% glycerol) to each sample before loading on the gel. Resolved RNAs were transferred to Hybond N+ membranes (Amersham Biosciences), by overnight capillary transfer with 20× SSC buffer (175.3 g l−1 sodium chloride, 88.2 g l−1 sodium citrate). Membranes were cross-linked by UV and pre-hybridized at 45°C in hybridization solution (50% formamide, 5× SSC, 2.5× Denhardts solution, 1% SDS, 50 μg ml−1 salmon sperm DNA). Biotin-labelled probes specific for λ-cI-N reporter transcript or 16S rRNA were denatured at 65°C for 10 min and then added to the hybridization solution. Hybridization was carried out overnight at 65°C. The biotin-labelled DNA probes were generated from purified PCR fragments of 200–400 nucleotides in length using EZ-Link Psoralen-Biotin reagent (Pierce). The relative level of Psoralen-Biotin incorporation into each specific probe varied from experiment to experiment, as reflected in the relative intensities of the detected λ-cI-N or 16S rRNA targets. Unbound probe was washed from Northern blots with three low stringency washes (0.2× SSC/0.1% SDS at room temperature) followed by three high-stringency washes (0.1× SSC/0.1% SDS at 65°C). Blots were developed using alkaline phosphatase-conjugated streptavidin and chemiluminescent substrate (Biotin Luminescent Detection Kit, Roche). Bands were quantified using Image-J software (NIH, http://rsb.info.nih.gov/ij/).
We thank Dr Robert Sauer and Tom Sundermeier for critical reading and insightful comments on the manuscript. We thank Dr Murray Deutscher for generously sharing strains and reagents. We also thank members of the Karzai lab for helpful discussions and suggestions. We are grateful to Dr Jorge Benach and members of The Center for Infectious Diseases for their continued support. This research was supported in part by grants (to A.W.K) from the National Institutes of Health, and the Pew Scholars Program.