Although SsrA(tmRNA)-mediated trans-translation is thought to maintain the translation capacity of bacterial cells by rescuing ribosomes stalled on messenger RNA lacking an in-frame stop codon, single disruption of ssrA does not crucially hamper growth of Escherichia coli. Here, we identified YhdL (renamed ArfA for alternative ribosome-rescue factor) as a factor essential for the viability of E. coli in the absence of SsrA. The ssrA–arfA synthetic lethality was alleviated by SsrADD, an SsrA variant that adds a proteolysis-refractory tag through trans-translation, indicating that ArfA-deficient cells require continued translation, rather than subsequent proteolysis of the truncated polypeptide. In accordance with this notion, depletion of SsrA in the ΔarfA background led to reduced translation of a model protein without affecting transcription, and puromycin, a codon-independent mimic of aminoacyl-tRNA, rescued the bacterial growth under such conditions. That ArfA takes over the role of SsrA was suggested by the observation that its overexpression enabled detection of the polypeptide encoded by a model non-stop mRNA, which was otherwise SsrA-tagged and degraded. In vitro, purified ArfA acted on a ribosome-nascent chain complex to resolve the peptidyl-tRNA. These results indicate that ArfA rescues the ribosome stalled at the 3′ end of a non-stop mRNA without involving trans-translation.
Trans-translation occurs characteristically in eubacteria and is mediated by SsrA (tmRNA), which enters the A-site-vacant ribosome and allows continued translation by acting both as an alanyl-tRNA and a messenger RNA. Its impact on cellular physiology appears twofold. First, trans-translation releases the ribosome stalled at the end of non-stop mRNA as its coding sequence ends with a normal stop codon. Second, SsrA(tmRNA)-mediated trans-translation leads the premature peptides encoded by truncated mRNA to proteolysis by adding the proteolysis-provoking amino acid sequence to the C-terminus (for reviews, see Moore and Sauer, 2007; Keiler, 2008; Hayes and Keiler, 2010). In some bacterial species, SsrA deficiency is associated with appreciable phenotypes, which are complemented largely, if not completely, by SsrADD (Huang et al., 2000; Moore and Sauer, 2007; Keiler, 2008), an SsrA variant that mediates trans-translation to attach a proteolysis-resistant tag sequence (Keiler et al., 1996). Therefore, it is believed that the ribosome-rescuing function of SsrA is physiologically more important than its function to destabilize the product polypeptides (Moore and Sauer, 2007; Keiler, 2008). In other words, ribosome stalling at the end of non-stop mRNA is deleterious to the cell and needs to be rescued.
Interestingly, essentiality of SsrA differs among bacterial species. It is essential in Neisseria gonorrhoeae (Huang et al., 2000), but dispensable in Escherichia coli (Komine et al., 1994). It is conceivable that E. coli and other bacteria, which do not require SsrA to survive, possess an alternative mechanism for rescuing the stalled ribosome (Moore and Sauer, 2007; Hayes and Keiler, 2010). However, exact fates of ribosomes stalled at 3′ ends of mRNAs in the absence of SsrA have not been investigated to any considerable extent. Although peptidyl-tRNA hydrolase (Pth) is known to be conserved throughout biological kingdoms, its role has been shown only in the extra-ribosomal hydrolysis of short peptidyl-tRNAs generated by the ‘drop off’ mechanisms (Cruz-Vera et al., 2004; Singh and Varshney, 2004; Das and Varshney, 2006). In the present work, we attempted to identify an alternative ribosomal rescue factor of E. coli by undertaking genetic screening for mutants that require SsrA to survive. Our synthetic lethality screening revealed that E. coli cannot withstand the simultaneous absence of SsrA and YhdL, a previously uncharacterized protein (now renamed ArfA). We suggest that ribosomes stalled at the end of non-stop mRNAs are rescued either through the SsrA-mediated trans-translation pathway or by the ArfA-mediated alternative pathway.
YhdL is required for growth in ssrA-deficient E. coli cells
Taking advantage of the conditional instability of a par-defective Ampr mini-F plasmid pFZY1 (Koop et al., 1987; Bernhardt and de Boer, 2004), we previously screened transposon-mutagenized E. coli for mutants that require SsrA for growth (Ono et al., 2009). Using the same strategy, except for the use of N-methyl-N′-nitro-N-nitrosoguanidine (MNNG) as a means of mutagenesis, we obtained a mutant that exhibited a strict SsrA requirement for growth at a temperature range, 30–42°C, examined. We then selected a plasmid clone from the E. coli genomic library that complemented the growth defect of the mutant. Such a plasmid proved to contain the yhdL gene, a 219 bp open reading frame (ORF) located at 90.4 min on the E. coli chromosome. Genomic sequencing revealed that the mutant strain had a transition mutation (adenine to guanine) at the 52nd nucleotide within the yhdL ORF, causing a substitution of Ala18 to Thr (A18T) in the YhdL protein (Fig. 1A). As discussed below, we rename this gene arfA (alternative ribosome-rescue factor).
Like ssrA, arfA itself is dispensable for E. coli and its disruption strain, JW3253 (BW25113 yhdL(arfA)::FRT-Kmr-FRT), is available among the Keio collection of NBRP: E. coli (NIG, Japan) (Baba et al., 2006). The yhdL(arfA)::FRT-Kmr-FRT segment of the JW3253 chromosome was successfully transduced by P1 bacteriophage into the wild-type strain, W3110, but not into an isogenic ssrA mutant, TA331. To substantiate that this failure was attributable to the absence of SsrA, we introduced a plasmid pBAD24-ssrA, which expressed SsrA from the arabinose-inducible promoter, into the recipient TA331 before using it as a recipient in P1 transduction. Indeed, TA331 harbouring pBAD24-ssrA produced Kmr transductants only when the selective plate was supplemented with arabinose (data not shown). From these results, we conclude that the ssrA disruption and the arfA disruption result in a synthetically lethal phenotype.
To follow bacterial growth under the conditions of ArfA-SsrA double depletion, cells of strain CH115 (ΔarfAΔssrA) harbouring pBAD24-ssrA were grown first in LB medium containing arabinose and then in the same medium containing glucose instead of arabinose. Northern analysis revealed that SsrA decreased as early as 15 min after medium change, and almost disappeared after 1 h (Fig. S1). After the medium change and consequent shutting off of the ara promoter-directed expression of ssrA, growth of CH115/pBAD24-ssrA started to slow down and stopped after ∼1 h at 37°C (Fig. 1B, solid squares). Also, the onset of SsrA depletion was accompanied by a loss of viability as shown by the declined numbers of colony forming unit (cfu) after the medium change (Fig. 1B, open squares). This growth impairment was complemented by ArfA supplied from a compatible plasmid but not by a frameshifted version of ArfA or ArfA-A18T (data not shown). These results indicate that the ArfA protein is required for growth and viability in the absence of SsrA and that the A18T substitution of ArfA results in a loss of its function.
Complementation of the synthetic lethality of the ssrA–arfA double disruption by SsrADD
To address which of the ribosome-rescuing function and the proteolysis-enhancing function of SsrA is important for cellular survival in the absence of ArfA, we examined whether SsrADD, an SsrA variant encoding a proteolysis-refractory tag sequence (Keiler et al., 1996), can complement the synthetic growth defect of the ΔssrAΔarfA double mutant. The yhdL(arfA)::FRT-Kmr-FRT marker was successfully transferred by P1 transduction into TA371 (W3110 ssrADD), indicating that the chromosomally encoded SsrADD can support growth of E. coli lacking arfA (Fig. S2). Thus, it is the role of ribosome rescue, but not degradation of the tagged nascent polypeptide, that SsrA must fulfil in the absence of ArfA.
Effects of depletion of SsrA and ArfA on transcription and translation
The ability of SsrADD to restore the growth of the ΔssrAΔarfA double mutant is consistent with a notion that ArfA as well has a function related to translation. We studied effects of the SsrA–ArfA double depletion on transcription and translation of an IPTG-inducible recombinant model protein, N-terminally His6-tagged Salmonella FlgA without the signal peptide (referred to as FlgA hereafter). We introduced pCH300, encoding FlgA under the IPTG-inducible T5 promoter control, into CH115 (ΔssrAΔarfA) harbouring pBAD24-ssrA. Depletion of SsrA was achieved by a medium change from LB-arabinose to LB-glucose, followed by incubation for an hour. Then IPTG was added to induce the expression of FlgA and incubation was continued further before harvesting cells and examining the contents of the flgA mRNA and the FlgA protein (Fig. 2, lanes 1–6). The FlgA protein content was reduced markedly in the SsrA-depleted cells (Fig. 2, lanes 4–6) as compared with cells growing in the presence of arabinose. Northern analysis showed that the abundance of flgA mRNA in the SsrA-depleted cells was no less than that in the SsrA-sufficient cells; in fact it rather increased upon SsrA depletion (Fig. 2, lanes 4–6). Accumulation of flgA mRNAs with higher mobility seen in Fig. 2, lane 6 may be due to the protection of truncated mRNA by stalled ribosome. These results indicate that the double depletion impairs primarily translation instead of transcription.
The apparent increase in the flgA message observed above might be explained in terms of enhanced transcription of rRNA induced by the translation inhibition, leading to a titration of RNases, stabilization of mRNA and accumulation of premature rRNA (Lopez et al., 1998). This was supported by the observation that incorporation of 14C-labelled Leucine into CH115 harbouring pBAD24-ssrA was decreased upon SsrA depletion (Table S1). These observations appear to suggest that translation was generally inhibited in the simultaneous absence of ArfA and SsrA. Inhibitory effect of SsrA depletion on flgA translation was not observed when TA331, having arfA+, was used as a host instead of CH115 (Fig. 2, lanes 7–12). Taken together, these results are in accordance with the notion that ArfA functions to rescue ribosome stalled at the end of mRNA as SsrA does and that this role of ArfA becomes highlighted in the absence of SsrA. In the absence of this functionally redundant set of ribosome-rescue factors, increasing number of ribosomes will accumulate in the cell in mRNA-associated non-functional states, resulting in the inhibition of translation and eventual cell death.
Alleviation of the synthetic growth defect by puromycin
The results presented so far are consistent with our hypothesis that ArfA facilitates release of the ribosome stalled at the 3′ end of a non-stop mRNA and this function becomes crucial when the stalled ribosome cannot be rescued by trans-translation. In this case, the lethality of the ΔssrAΔarfA double mutation should be brought about by sequestration of increasing number of ribosomes in a sate stalled on the aberrant mRNA. Puromycin can enter the A-site of the ribosome without instruction by a codon and accept peptidyl transferase centre (PTC)-catalysed transpeptidation from peptidyl-tRNA located in P-site (Wilson, 2004; Muto et al., 2006), thus dissociating the translation complex. It is expected then that this antibiotic will force the ribosome stalled at the 3′ end of mRNA dissociate from the mRNA and, consequently, to function again. We examined whether puromycin could restore the bacterial growth in the absence of SsrA and ArfA. Strikingly, growth rate of CH115 harbouring pBAD24-ssrA after medium exchange from LB-arabinose (Fig. 3A, solid circles) to LB-glucose (Fig, 3A, open circles) was improved by addition of puromycin (Fig. 3A, triangles, diamonds and squares). The growth-enhancing effect of puromycin was dependent on its concentration up to 50 µg ml−1 (Fig. 3A, triangles for 10 µg ml−1 and diamonds for 50 µg ml−1) and then appeared to be counterbalanced by its general inhibitory action (Fig. 3A, squares for 100 µg ml−1). The growth-enhancing effect of puromycin was specific to the cells lacking both SsrA and ArfA because puromycin did not enhance the growth rate if TA331 or CH101 was used instead of CH115 (Fig. 3B and C).
Association of ArfA with the large subunit of the ribosome
Involvement of ArfA in translation and ribosomal rescue raises a possibility that it interacts with the ribosome or some translation factor. We constructed a plasmid pCH201, which expresses His6-tagged ArfA upon IPTG induction. We used an ArfA variant lacking C-terminal 12 amino acid residues because His6-tagged full-length ArfA was expressed only poorly while the His6-tagged and C-truncated ArfA (referred to simply as His6-ArfA hereafter) was expressed at much higher level. His6-ArfA could support the growth of ΔssrAΔarfA double mutant, indicating that it is functional in vivo (data not shown). We prepared His6-ArfA from cell extract of CH101 harbouring pCH201 by means of Ni-resin affinity isolation under non-denaturing conditions. SDS-PAGE and silver-staining showed that a number of protein species were co-isolated with His6-ArfA, which were not detected in a similarly prepared sample from cell extract of CH101 harbouring empty vector, pQE80L (Fig. S3). We then determined the identity of the ArfA-associated proteins by excising the protein bands, digesting them with trypsin, and subjecting the digested materials to LC-MS/MS analysis. Proteins detected in this analysis are summarized in Table S2. Most of them were identified as ribosome proteins (Fig. S3), strongly suggesting that ArfA is associated with the ribosome.
We then fractionated a lysate of CH101 cells harbouring pCH201 by centrifugation into S100 and P100 fractions. Anti-His6 immunoblotting showed that most of His6-ArfA was fractionated into the P100 fraction, in which the ribosomes should be recovered (Fig. 4A). The A18T-mutant form of His6-ArfA (His6-YhdL-A18T) showed essentially the same fractionation profile, indicating that A18T alteration does not affect binding of ArfA to the ribosome (Fig. 4A). Further analysis of the P100 materials by sucrose gradient centrifugation demonstrated that His6-ArfA co-sedimented mostly with the 50S subunit of the ribosome (Fig. 4B). From these results, we conclude that ArfA is stably associated with the ribosomal large subunit.
ArfA-dependent rescue of the ribosome stalled at the end of a model non-stop mRNA
To examine the effects of SsrA–ArfA double depletion on translation of mRNA lacking an in-frame stop codon, we engineered the cloned crp gene. First, the His6 sequence was attached to the N-terminus of CRP. Secondly, the native stop codon of crp was removed. Finally, the trpA transcription terminator sequence was inserted into the 3′ region of the crp ORF. The resulting plasmid was named pCH333 whereas pCH323 served as a control having a stop codon upstream to trpA terminator (Fig. 5A). If transcription is terminated at the trpA terminator, mRNA from pCH333 has no stop codon until it is truncated at the trpA site. If transcription continues beyond the terminator, the crp ORF on pCH333 will encounter an in-frame stop codon after the terminator and produce a C-terminally extended His6-CRP protein. pCH323 will produce the normally terminated His6-CRP protein irrespective of the occurrence of transcription termination. We designated the His6-tagged CRP polypeptide encoded by the trpAt-terminated non-stop mRNA CRP-NST and the normally terminated His6-CRP protein CRP-ST. Note that the length of mRNAs for CRP-NST and CRP-ST was the same. Using these plasmids, we analysed the expression profiles of CRP-ST and CRP-NST as well as their mRNAs in TA341 (W3110 Δcrp), TA501 (TA341 ΔssrA) or CH231 (TA341 ΔarfA) (Fig. 5B, lanes 1–6). Expression of CRP-NST in TA481 (TA341 ssrADD) or in CH281 (TA481 ΔarfA) was also investigated (Fig. 5B, lanes 7 and 8).
Northern analysis showed that accumulation of CRP-NST (non-stop) mRNA was negatively affected by SsrA such that ΔssrA cells contained much higher level of it (Fig. 5B, middle panel, lane 5) than the ssrA+ cells (lanes 4 and 6). Negative effect of SsrA on CRP-ST mRNA (with normal stop codon) was not so strong as that on CRP-NST mRNA (Fig. 5B, middle panel, lanes 1–3). Accumulation of truncated mRNA was observed for CRP-ST mRNA in ssrA background. They may represent the products of stalled ribosome mediated mRNA cleavage. The 3′ end of such mRNAs would be protected by the ribosomes which could not be rescued because SsrA was absent. Similar observations were reported and discussed previously in terms of the mRNA quality control function of trans-translation (Yamamoto et al., 2003).
The CRP-ST protein accumulated similarly when expressed in the three strains (Fig. 5B, upper panel, lanes 1–3), with some smaller products in the ΔssrA mutant (Fig. 5B, upper panel, lane 2). Probably, the latter observation was due to ribosomal stalling within the crp ORF, which became apparent in the absence of SsrA. CRP-NST, the translation product of crp non-stop mRNA, was observed in cells lacking SsrA (Fig. 5B, lane 5). Interestingly, the CRP-NST band diminished in the ΔarfA mutant (Fig. 5B, lanes 6 and 8). This can be explained if ArfA competes with SsrA for rescuing the ribosome stalled at 3′ end of non-stop crp mRNA. Although most of the stalled ribosome-CRP-NST complexes would be dissociated through trans-translation, in which CRP-NST will be tagged and degraded, a subset of ribosomes might be released by ArfA, protecting CRP-NST from degradation. In the absence of ArfA, all the stalled ribosome would be released through trans-translation, resulting in diminished CRP-NST level.
The longer polypeptide (indicated by arrow with ‘readthrough and stop’ in Fig. 5B) should correspond to the polypeptide whose translation had continued until the stop codon located 170 nt downstream of the terminator on the readthrough mRNA transcribed beyond the terminator (Fig. 5A). This was confirmed by using a series of plasmids having a stop codon at different positions downstream of the trpA terminator; the ‘readthrough and stop’ band was shifted in accordance with the positions of the stop codon (Fig. S4). We also introduced pCH333 into the SsrADD mutant cells (TA481 and CH281) and found that a major fraction of the protein products migrated at the position of tagged CRP-NST (Fig. 5B, lanes 7 and 8, ‘tagged’). Also, the ΔarfA mutation was found again to reduce the non-tagged CRP-NST (Fig. 5B, compare lanes 7 and 8 for ‘non-stop’).
We also examined the effect of overproduction of His6-ArfA on the expression profile of CRP-ST and CRP-NST. We introduced pQE80L (vector), pCH201 (His6-ArfA), or pCH221 (His6-ArfA-A18T) into TA341 harbouring a compatible plasmid, pCH102 (vector), pCH323 (CRP-ST) or pCH333 (CRP-NST). Overexpression of His6-ArfA did not exert any effect on the expression of CRP-ST (Fig. 5C, lane 5). In contrast, His6-ArfA overproduction resulted in a marked increase in the intensity of the CRP-NST band (indicated by arrow with ‘non-stop’; Fig. 5C, lane 8). We interpret this to mean that overproduced ArfA deals with a higher fraction of the ribosomes stalled at the 3′ end of non-stop mRNA in a manner competitive with SsrA and rescues them from a CRP-NST-tRNA-bearing state. In support of this interpretation, overproduction of the non-functional variant, ArfA-A18T, was ineffective (Fig. 5C, lane 9).
ArfA-dependent resolution of peptidyl-tRNA in vitro
The ribosome stalled at the end of non-stop mRNA should bear nascent peptidyl-tRNA in the A-site (Kuroha et al., 2009). We addressed whether the ArfA protein has an ability to resolve the accumulated peptidyl-tRNA. We prepared an S30 lysate from CH101 cells and treated it with the anti-SsrA oligonucleotide to inhibit the SsrA activity (Hanes and Plückthun, 1997). Non-stop crp mRNA was prepared by in vitro transcription using a DNA template that was truncated after the 250th codon of crp. These materials were used for in vitro translation of the crp non-stop ORF. The translation reaction mixture was then treated with the purified preparation of C-terminally truncated form of His6-ArfA, that of His6-ArfA-A18T, puromycin or RNase A. Samples were separated by Bis-Tris SDS-polyacrylamide gel electrophoresis under the neutral conditions and proteins were visualized by anti-His6 immunoblotting. As shown in Fig. 6, CRP-NST band was seen in all the reaction analysed in Fig. 6. We speculate this is the product of spontaneous hydrolysis. Without any treatment, a slowly migrating band was clearly visible (Fig. 6, lane 1, ‘peptidyl-tRNA’). This band should have represented the CRP-NST peptidyl-tRNA because treatment of the sample with puromycin or RNase A abolished it (Fig. 6, lanes 4–8). Strikingly, the CRP-NST peptidyl-tRNA disappeared when the translation products were incubated with His6-ArfA (Fig. 6, lane 2). The non-functional variant with the A18T alteration did not affect the intensity of the peptidyl-tRNA band (Fig. 6, lane 3). The in vitro translation samples corresponding to the samples shown in Fig. 6, lanes 1 and 2 were further analysed by sucrose gradient centrifugation. In the absence of ArfA, CRP-NST peptidyl-tRNA and CRP-NST mRNA were present in the ribosome fraction. Upon ArfA addition, CRP-NST peptidyl-tRNA was disappeared and the level of CRP-NST mRNA in the ribosome fraction was greatly reduced (Fig. 7). This indicates that ribosome complex containing CRP-NST peptidyl-tRNA is formed at the 3′ end of CRP-NST mRNA and that this complex is somehow dissolved and CRP-NST peptidyl-tRNA was hydrolysed when the His6-ArfA was added to the reaction. His6-ArfA added to the reaction was distributed to both ribosome fraction and ribosome-free fraction (Fig. 7), in accordance with its distribution shown in Fig. 4A. Taken together with the finding that ArfA associates firmly with the ribosomal large subunit, we conclude that ArfA induces hydrolysis of ribosome-bound peptidyl-tRNA by binding to the large subunit of the ribosome.
We have identified YhdL (ArfA) as a novel factor involved in the maintenance of translation capacity of E. coli cells. The importance of ArfA becomes apparent in the absence of SsrA, an essential component of the trans-translation system with its proposed role to rescue the ribosome stalled at the 3′ end of a non-stop mRNA (Moore and Sauer, 2007; Keiler, 2008; Hayes and Keiler, 2010). We have revealed the following properties of ArfA and cells deficient in it. (i) E. coli cells lacking either ArfA or SsrA alone are viable but those lacking both of them are non-viable. (ii) Growth-supporting role of SsrA in the absence of ArfA is fulfilled by SsrADD, indicating that the growth-essential functions of ArfA/SsrA are related to translation, but not to proteolysis of the translation products. (iii) Depletion of SsrA and ArfA impairs translation but not transcription of an inducible model gene flgA. (iv) Puromycin effectively alleviates the growth defect of the double disruptant. (v) ArfA binds stably to the large subunit of the ribosome. (vi) Cellular abundance of ArfA positively correlates with that of the polypeptide product (CRP-NST) of a model non-stop mRNA, which is ordinarily eliminated by trans-translation and subsequent proteolysis. (vii) Addition of ArfA to in vitro translation system caused the decrease of peptidyl-tRNA that is associated with the nascent polypeptide–ribosome complex stalled on a non-stop mRNA. These results collectively suggest that ArfA rescues the stalled ribosome by somehow inducing hydrolysis of the ribosome-bound peptidyl-tRNA.
Our results revealed that SsrA and ArfA independently deal with aberrant stall of the ribosome. While SsrA forces the stalled ribosome to continue translation on its own message, ArfA seems to force the ribosome to dissociate from the non-stop mRNA. Either one of these mechanisms seems to be sufficient, indicating that the E. coli cell is equipped with functionally redundant but mechanistically distinct systems for the quality control of the translational apparatus. Thus, ArfA should be considered to represent one of the alternative ribosome recycling systems that have been assumed to exist in E. coli (Moore and Sauer, 2005; 2007).
Kuroha et al. (2009) showed that a nascent, non-stop mRNA-directed polypeptide was released from the ribosome even in the absence of SsrA and functional Pth. Also, RRF and RF3, involved in the peptidyl-tRNA drop-off pathway, did not accelerate the ribosome rescue from non-stop mRNA (Janssen and Hayes, 2009). It seems likely that ArfA was responsible for the SsrA-independent release of nascent polypeptides reported by these authors. It should be noted here, however, that Kuroha et al. (2009) did not exclude the possibility that remaining Pth activity in their system was responsible to the reduction of peptidyl-tRNA. So, it is not clear if the mechanism of ArfA to reduce peptidyl-tRNA is distinct from that of Pth. Szaflarski et al. (2008) showed that ribosome could be rescued from non-stop mRNA in the absence of SsrA in their in vitro poly-U translating system. We do not know if the ribosome preparation they used in their experiment contained ArfA or not. It may be possible that Pth, ArfA or other factors were associated with the ribosome preparation, but non-enzymatic rescue of ribosome may not be excluded. Even in in vivo situation, non-enzymatic ribosome rescue may occur. Its efficiency, however, may not high enough to maintain cellular condition in the absence of both trans-translation system and ArfA system.
Database analysis revealed that ArfA is conserved among enterobacteriaceae and that Ala18, an amino acid residue, which we have shown to be required for the ArfA function, is invariant among enterobacterial ArfA proteins (Fig. 1A). The proteins which show similarity to E. coli ArfA were also found in N. gonorrhoeae, whose growth requires the SsrA trans-translation system (Huang et al., 2000), and other bacteria, but they do not have similarity in the region around Ala18. We suggest that the strict requirement of N. gonorrhoeae for SsrA is due to the lack of the ArfA system to handle the stalled ribosome. It may be necessary to see if there are any relationship between the SsrA dependency and the existence of the proteins which show significant homology to E. coli ArfA in other bacteria.
We have shown that purified ArfA has an in vitro activity to lead the ribosome-bound peptidyl-tRNA to hydrolysis. ArfA does not share any homology to known translation factors, suggesting that it has a unique function in completion of the aborted translation event. ArfA is a small protein. It is intriguing to see if it has a hydrolase enzymatic activity by itself. If ArfA does not have an enzymatic activity by itself, there must be another factor that functions as a peptidyl-tRNA hydrolase in conjunction with ArfA. We speculate that binding of ArfA to the ribosome somehow enhances the hydrolase activity intrinsically associated with the ribosomal PTC. Alternatively, it may enhance the activity of hydrolases such as Pth or activate an unidentified ribosome-associated hydrolase, such as recently reported ICT1 of the mitochondrial ribosome (Richter et al., 2010). Also, the possibility that ArfA only accelerates the spontaneous dissociation of peptidyl-tRNA from ribosome complex formed at the 3′ end of non-stop mRNA cannot be excluded. The molecular mechanism, by which ArfA acts against ribosome-nascent chain complex to release the ribosome from the non-stop mRNA, is an important and interesting question to be addressed in future studies.
Why do E. coli and other bacteria have the ArfA system in addition to trans-translation? It is possible that these bacteria often experience conditions where ribosome stalling takes place at the level beyond the capacity of trans-translation in natural environments. Another possibility is that the ArfA system and trans-translation have different roles in ribosome rescue. For example, when ribosome stalls upon amino acid starvation, trans-translation has a greater advantage because accompanied protein degradation will supply amino acids (Garza-Sanchez et al., 2008; Li et al., 2008). Bacteria may utilize these two ribosome rescue systems depending upon different circumstances. Our analysis of CRP-NST and CRP-ST indicates that ArfA affects specifically the translation complex in the process of translation of non-stop mRNA. Elucidation of molecular mechanisms of ArfA-mediated ribosome rescue may answer the important question of how ArfA senses ribosome stalling.
The A18T mutation in ArfA was found to impair the in vitro activity to induce hydrolysis of peptidyl-tRNA and to have the lethal consequence in ssrA-deficient cell. However, this mutation does not affect the ArfA's ability to bind to the ribosome, pointing to the importance of Ala18 in the essential function of this protein that is executed in the ribosome-associated state. We speculate Ala18 contribute to the local conformation around the essential residue and the bulky side chain of Thr residue in the A18T mutant inhibits the catalysis and/or the intermolecular interaction directly or via local conformational distortion. Systematic substitution-scanning analysis of ArfA, structural determination of ArfA and its complex with the ribosome, and identification of a factor that directly interacts with ArfA will give us a new standpoint of understanding the translation process, especially quality control of the translation complex that allows for effective recycling of its components.
E. coli strains, phages, plasmids and primers
Escherichia coli strains used in this study are listed in Table 1. Phage P1-mediated transduction was used to introduce mutations in yhdL(arfA) (JW3253) or lacZ (EJ2845) to appropriate strains. Removal of the FRT-Kmr-FRT cassette from the chromosome of the transductant was performed using pCP20 as described in Datsenko and Wanner (2000). Construction of ssrA arfA double mutant was performed in the presence of plasmid-borne SsrA.
Plasmids and primers used in this study are listed in Tables 2 and 3 respectively. pFZY1-ssrA was reported previously (Ono et al., 2009). pSTV29 (TAKARA) was used as a vector to construct an E. coli genome library, from which a clone suppressing the SsrA requirement of the mutant primarily obtained by MNNG mutagenesis was screened. pQE80L (Qiagen) was used to construct plasmids carrying an inducible arfA gene. DNA fragment containing arfA was PCR-amplified from W3110 genome using primers LF01_KpnI and LR01_PstI, digested with KpnI and PstI, and cloned into the corresponding site of pSTV28 (TAKARA) to construct pSTV-yhdL. pSTV-yhdL-A18T was constructed similarly except that the SsrA-dependent mutant derived from TA461/pFZY1-ssrA was used as a source of genomic DNA. DNA fragments PCR-amplified from pSTV-yhdL and pSTV-yhdL-A18T using primers LF02_BamHI and yhdL_dC12PstIRV were digested with BamHI and PstI and cloned into the corresponding site of pQE80L to construct pCH201 and pCH221 respectively. DNA fragment containing multiple cloning site and flanking region of pQE80L was PCR-amplified using pQE-MCS_fw and pQE-MCS_rv and directly cloned into FspI–NaeI region of pSTV28. The resulting plasmid was digested with Tth111I and blunt-ended using Klenow enzyme, and ligated with a 1.5 kb ScaI–FspI fragment of pQE80L containing the lacIq gene to construct pCH101. pCH101 was digested completely with ScaI and partially with AccI, blunt-ended using Klenow enzyme, and the 3.2 kb fragment containing p15A replication origin was isolated and ligated with the blunt-ended 1.2 kb PstI fragment of pUC4K (Amersham) containing the Kmr gene to construct pCH102. PCR fragment amplified from pTN601 (Nambu and Kutsukake, 2000) using primers FF01_BamHI and flgA trpAT BamHI RV was digested with BamHI and cloned into BamHI site of pCH101 to construct pCH300. PCR fragment containing crp was amplified from pHA7 (Aiba et al., 1982) using primers CF01_BamHI and CR01_SphI, digested with BamHI and SphI, and cloned into the corresponding site of pCH102 to construct pCH320. PCR fragment containing trpA terminator was amplified from pTN601 using primer pairs FF01_BamHI and flg trpA BamHI RV or FF01_BamHI and flg non-ST trpA BamHI RV, digested with SspI, and cloned into SmaI site of pCH320 to obtain pCH321 or pCH331 respectively. pCH321 and pCH331 were SphI-digested, blunt-ended, and self-ligated to construct pCH322 and pCH332 respectively. pCH322 and pCH332 were AccI- and NheI-digested, blunt-ended and self-ligated to construct pCH323 and pCH333 respectively. pCH353 was constructed as follows. A 708 bp EcoRI–SalI fragment of pCH320 was cloned into the corresponding site of pFLAG-CTC (Sigma) to construct pCH340. Using pCH340 as a template, DNA fragment encoding CRP-FLAG was PCR-amplified using primers CF01_BamHI and FLAG-PstI_rv. The amplified DNA was then digested with BamHI and PstI and cloned into the corresponding site of pQE80L to obtain pCH341. DNA fragment containing the trpA terminator was PCR-amplified using mutually annealing primer pair Ttrp-fw-nst and Ttrp_rv, digested with PstI and HindIII, and cloned into the corresponding site of pCH341 to construct pCH352. A 773 bp EcoRI–HindIII fragment of pCH352 was cloned into the corresponding site of pSPT18 to construct pCH353.
GAG GAT CCA AAA AAG CCC GCT CAT TAG GCG GGC TGC TTA TAG GTT AAT AAG AAT ATT CCC AT
flgA nonST trpAT BamHI RV
GAG GAT CCA AAA AAG CCC GCT CAT TAG GCG GGC TGC TAA TAG GTT AAT AAG AAT ATT CCC AT
AAG GAT CCG TGC TTG GCA AAC CG
ATG CAT GCG TGC CGT AAA CGA C
ACC TCG AGA AAT CAT AAA
CGC CAA GCT AGC TTG GA
GAC TGC AGC TTG TCG TCA TCG T
AAC TGC AGA AAG CAG CCC GCC TAA TGA
CCA AGC TTA AAA AAG CCC GCT CAT TAG
CCG AAT TCT ATT GAC CAG TTC CTC ACC G
CCC AAG CTT AAT GGG CCT AAA AGG TTC GG
Bacterial growth and screening
Escherichia coli cells were grown in LB at 37°C unless otherwise noted. LB containing 1.5% (w/v) agar was used to prepare culture plate. Arabinose or glucose was added to the media to the final concentration of 0.2% (w/v) or 0.4% (w/v) respectively. Appropriate antibiotics were added to the media. Bacterial growth in liquid medium was monitored by measuring OD660.
Mutagenesis of TA461/pFZY1-ssrA was performed as described (Miller, 1992). Briefly, cells were treated with 50 µg ml−1 MNNG for 10 min at 37°C, washed with saline, and plated out onto LB agar plate containing X-Gal. After incubation at 37°C, deep blue colonies were picked up as candidates which require SsrA as described previously (Ono et al., 2009). Out of several candidates, one mutant which showed a severe SsrA-dependent phenotype was further analysed.
Identification of protein factors that interacted with ArfA in vivo
CH101/pCH201 was grown to a mid-log phase and His6-ArfA was overexpressed in the presence of 0.5 mM IPTG for 1.5 h. Protein factors interacting with ArfA in vivo were co-purified with His6-ArfA using Ni-chelating resin (Ni-NTA agarose, Qiagen) from the cell extract of CH101/pCH201. As a control, the same procedure was done with CH101/pQE80L. Proteins were separated by SDS-PAGE and visualized by silver staining. Protein bands were excised, digested with trypsin and analysed by LC-MS/MS (LTQ ion trap mass spectrometer, Thermofisher Scientific, San Jose, USA) as described by Ozawa et al. (2009). The obtained data were analysed with a Beadworks software (ver. 3.3) using the E. coli K12 protein database (Ecogene, http://ecogene.org/, Rudd, 2000).
Fractionation of cellular components
Crude lysate of cells and subcellular fractions were prepared as follows. CH101/pCH201, CH101/pCH221 or CH101/pQE80L was grown in LB. At a mid-log phase, IPTG was added to the final concentration of 0.5 mM and cells were incubated for 1.5 h to induce protein expression. Cells were harvested, washed with ice-cold STE (0.1 M NaCl, 10 mM Tris-HCl pH 8.0, 1 mM EDTA), and then suspended in TMN buffer (25 mM Tris-HCl pH 7.5, 20 mM MgCl2, 100 mM NH4Cl, 2 mM DTT, 1 mM PMSF; 1 ml for 100 mg cell). Lysozyme was added to the final concentration of 250 µg ml−1 and samples were incubated for 30 min on ice. Then the cells were disrupted by sonication. Debris was sedimented by centrifugation (15 krpm, 10 min, 4°C) and the supernatant was recovered as crude lysate, which was fractionated into S30 and P30 fractions by ultracentrifugation (25 krpm, 30 min, 4°C using Beckman Optima-TM TLM Ultracentrifuge and TLA100.3). S30 fraction was further fractionated into precipitate and supernatant fractions by ultracentrifugation (50 krpm, 100 min, 4°C). The precipitates were rinsed with TMN buffer, suspended in TMN buffer whose volume was equal to that of supernatant, and centrifuged again at 30 000 g to remove insoluble components. The resulting supernatant was recovered as P100 fraction.
Sucrose gradient fractionation was performed as described by Dijk and Littlechild (1972). Cells were disrupted by glass beads (0.1 mm) using Beadbeater (Wakenyaku-Kogyo, Japan), and fractionated into S100 and P100 fractions. P100 was suspended in TN buffer (25 mM Tris-HCl pH 7.5, 100 mM NH4Cl, 2 mM DTT) and dialysed twice in the same buffer. Seven hundred microlitres of P100 fraction was applied onto a 5–30% sucrose gradient containing TN buffer prepared in Thinwall Polyallomer Centrifuge tube (13 × 51 mm, Beckman) and fractionated by ultracentrifugation (Beckman Optima-TM L-80, SW50.1-Ti, 25 krpm, 8 h, 4°C). After centrifugation, each fraction (approximately 130 µl) was collected. Ribosome and ArfA in each fraction were monitored by A260 and anti-His6 Western blotting respectively. To analyse S30 in vitro translation sample by sucrose gradient fractionation experiment, 200 µl of sample treated with 100 mM magnesium acetate was applied onto a 10–40% sucrose gradient containing 25 mM Tris-HCl pH 7.4, 100 mM NH4Cl, 10 mM MgCl2, 1 mM DTT prepared in Thinwall Polyallomer Centrifuge tube (13 × 51 mm, Beckman) and fractionated by ultracentrifugation (Beckman Optima-TM L-80, SW50.1-Ti, 33 krpm, 2.7 h, 4°C). After centrifugation, each fraction (approximately 100 µl) was collected.
The in vitro translation reaction was performed as described (Takai et al., 1996; Kanda et al., 2000) with some modifications. S30 extract of CH101 was prepared according to Pratt (1984). Translation of in vitro prepared non-stop mRNA was performed in the mixture containing 62.5 mM HEPES-KOH pH 7.4, 1.7 mM DTT, 174 µg ml−1E. coli tRNA, 50 mM Cr-phosphate, 2% PEG8000, 20 µg ml−1 folinic acid, 1 mM each amino acid, 1.25 mM ATP, 0.8 mM GTP, 0.63 mM cyclic-AMP, 30 mM ammonium acetate, 175 mM potassium glutamate, 1.5 mM spermidine, 80 µg ml−1 Cr-kinase, 100 µg ml−1 mRNA, 2.5 mM magnesium acetate, 50 ng µl−1 anti-SsrA probe (5′-TTAAGCTGCTAAAGCGTAGTTTTCGTCGTTTGCGACTA-3′; Hanes and Plückthun, 1997), 30% volume of S30 extract for 30 min at 37°C. After translation reaction, 200 nM of purified ArfA, 10 µM (5.4 µg ml−1) of puromycin, or 10 µg ml−1, 100 µg ml−1 or 1000 µg ml−1 of RNase A was added. After 10 min, translation was stopped by adding equal volume of SDS sample buffer (100 mM Tris-HCl pH 6.8, 10% 2-mercaptoethanol, 4% SDS, 0.02% Bromophenol blue, 20% Glycerol).
Non-stop mRNA used as a template for cell-free translation was prepared by transcribing HindIII-digested pCH353 using ScriptMAX™Thermo T7 Transcription Kit (TOYOBO). The transcript, which contained crp ORF truncated at the 250th codon, was then purified using Microspin G-50 column (GE healthcare). His6-ArfA used in the cell-free reaction system was purified from the extract of BL21/pCH201 cells which were grown for 3 h in the presence of 0.5 mM IPTG using Ni-chelating resin (Ni-NTA agarose) according to the supplier's instruction.
Bis-Tris SDS-PAGE analysis
After addition of SDS sample buffer, samples were incubated for 2 h at 37°C and analysed by Western blotting after 12% Bis-Tris NuPAGE (Invitrogen) of which running buffer system contained MOPS.
Preparation of the probes used in Northern blotting
DNA probe hybridizing to the crp gene was prepared by PCR using a primer pair CF01_BamHI and CR01_SphI and pHA7 (Aiba et al., 1982) as a template. DNA probe hybridizing to flgA gene was prepared by PCR using a primer pair FF01_BamHI and flgA trpAT BamHI RV and pTN601 (Nambu and Kutsukake, 2000) as a template. DNA probe hybridizing to SsrA RNA was prepared by PCR using a primer pair ssrA-F and ssrA-R and pSsrA-AA (Abo et al., 2000) as a template. They were labelled with DIG using DIG DNA labelling kit (Roche). DIG-labelled RNA probe hybridizing to 5S rRNA was prepared by transcribing pRT18 digested with EcoRI using DIG RNA labelling kit (Roche).
We thank Dr Hiroji Aiba for providing pHA7 and anti-CRP antibody, Dr Hideji Yoshida for providing anti-L2 antibody, and NBRP:E.coli (NIG, Japan) for providing JW3253. We also thank Dr Hiroji Aiba, Dr Joshua Sakon, Dr Nobuo Shimamoto and Dr Koreaki Ito for critical reading of this article.