Cold shock induction of RNase R and its role in the maturation of the quality control mediator SsrA/tmRNA

Authors


Summary

In this paper we show that RNase R is a cold shock protein that is induced seven- to eightfold by cold shock and that its expression is tightly regulated by temperature. Transcriptional studies reveal that the rnr gene is co-transcribed with flanking genes as an operon induced under cold shock. The induction of RNase R levels is mainly a result of the stabilization of the rnr transcripts. The transient stability of the rnr transcripts is shown to be regulated by PNPase at the end of the acclimation phase. Studies with an rnr mutant revealed a cold-shock phenotype showing that RNase R contributes to growth at low temperatures. We have shown that RNase R can be involved in the maturation of SsrA/tmRNA, an important small stable RNA involved in protein tagging and ribosome rescue. The wide biological significance of RNase R regarding adaptation to cold shock and its involvement in RNA surveillance, protein quality control and pathogenesis is discussed.

Introduction

Most organisms respond to a sudden change in temperature with a specific cascade of adaptive reactions in order to survive and resume growth at the new environmental conditions (Eriksson et al., 2002). In E. coli there is growth arrest after cold shock. During this period of adaptation (acclimation phase) cold shock proteins (CSP) are synthesized to overcome the deleterious effects of cold shock. Subsequently the synthesis of most proteins is resumed in such a way that 4 h after temperature downshift, the response is completed (Thieringer et al., 1998). Many CSP have been shown to facilitate translation by adapting ribosomes to the lower temperature and preventing the formation of RNA secondary structures (Jiang et al., 1997; Phadtare et al., 2002).

RNA-binding proteins and factors affecting mRNA stability are the major components involved in the expression of cold shock proteins (Brandi et al., 1996; Goldenberg et al., 1996; Yamanaka and Inouye, 2001). Polynucleotide phosphorylase (PNPase) is the only exoribonuclease so far shown to be inducible under cold shock (Jones et al., 1987; Zangrossi et al., 2000; Beran and Simons, 2001; Mathy et al., 2001). Polynucleotide phosphorylase is essential for adaptive growth resumption of cold shock-treated cells at the end of the acclimation phase (Yamanaka and Inouye 2001). Polynucleotide phosphorylase and RNase II have been considered as the two main exonucleases in E. coli (reviewed in Régnier and Arraiano, 2000; Arraiano and Maquat, 2003). Different factors affect the expression of RNase II and the levels of RNase II and PNPase have been shown to be interregulated (Zilhão et al., 1995; 1996a; Cairrão et al., 2001). RNase II together with RNase R belong to the RNR superfamily of 3′ to 5′ processing exoribonucleases (Zuo and Deutscher, 2001). RNR family members are widely distributed among bacteria and eukaryotes and in many organisms the RNR family is only represented by RNase R homologues (Mian, 1997; Zuo and Deutscher, 2001). Structurally the E. coli RNR family is characterized by three main domains (SMART http://smart.embl-heidelberg.de): S1-like/cold shock domain at the N-terminal (Weber et al., 2002), a central RNase II-like domain (Mian, 1997) and a C-terminal RNA binding S1 domain (Bycroft et al., 1997). Double mutants lacking both PNPase and RNase R are unviable indicating that RNase R has an essential function in RNA metabolism, that cannot be substituted by any other cellular exoribonuclease (Cheng et al., 1998). RNase R is mainly responsible for the residual hydrolytic exonucleolytic activity observed in extracts of E. coli strains that are deficient in RNase II (Cheng et al., 1998). Recent studies have shown that RNase R is more effective than RNase II against structured RNAs such as tRNAs (Cheng and Deutscher, 2002) and it is involved in the quality control of ribosomal RNA (Cheng and Deutscher, 2003) indicating a more specific role for RNase R in RNA metabolism.

RNase R has been shown to co-purify with the SsrA/SmpB system that is involved in protein quality control and ribosome recycling in a process called trans-translation (Karzai and Sauer, 2001). SsrA is a small stable RNA present in all bacterial genomes and in some chloroplasts. SsrA, also known as transfer-messenger RNA (tmRNA), or 10Sa RNA, is an unique molecule that functions both as a tRNA and a messenger RNA (Chauhan and Apirion, 1989; Komine et al., 1994). SsrA/tmRNA together with SmpB and other cofactors, releases stalled ribosomes and tags truncated proteins directing them to degradation by cellular proteases (Withey and Friedman, 2002;2003). In E. coli the primary ssrA transcript is 457 nt and processing of this transcript is critical for generating the mature 363 nt SsrA tRNA-like structure with a 3′-terminal CCA required for aminoacylation (Karzai et al., 2000). Maturation of SsrA/tmRNA has been reported to result from endonucleolytic processing by RNase P, RNase III and RNase E followed by exoribonucleolytic trimming (Li et al., 1998; Lin-Chao et al., 1999; Karzai et al., 2000; Withey and Friedman, 2003). The association of RNase R with the SsrA/SmpB system and the fact that the absence of a functional RNase R leads to an aberrant tagging of proteins (Karzai and Sauer, 2001) suggests an important role for RNase R in this system. RNase R is involved in pathogenesis, being required for the expression of virulence genes in Shigella and enteroinvasive strains of E. coli. However, the exact role of RNase R in establishing virulence still remains to be elucidated.

Questions that still need to be answered include the expression of the rnr gene, the regulation of its expression, the factors involved in its regulation and more importantly the physiological role of RNase R in the cell. Our results indicate that RNase R is a new member of the cold shock inducible proteins and that this induction is mainly a result of its regulation at post-transcriptional level through regulation of mRNA stability. A cold-shock phenotype is also observed in a rnr mutant revealing the importance of RNase R in the recovery of cells from cold shock conditions. We have also shown that RNase R is involved in the maturation of SsrA/tmRNA under cold shock. The consequences of these discoveries in quality control mechanisms and pathogenesis are discussed.

Results

Induction of RNase R under cold shock

Some of our previous studies have focused on the expression of RNase II (Zilhão et al., 1995; 1996a,b; Régnier and Arraiano, 2000; Cairrão et al., 2001). The structural similarities among the RNR family of exoribonucleases lead us to investigate the role and regulation of RNase R, the other E. coli member of this family. A histidine tagged version of RNase R (H6-RNR) was constructed and purified to raise polyclonal antibodies against RNase R. These antibodies were shown to be highly specific to RNase R (a 91 kDa protein) and did not show cross-hybridization with other proteins (Fig. 1). To determine the importance of RNase R in RNA metabolism a mutant strain (rnr) was constructed by inserting a kanamycin cassette into the chromosomal rnr coding sequence (see Experimental procedures, Fig. 2B). Western analysis confirmed that the rnr strain did not express RNase R (Fig. 1A and C).

Figure 1.

Effect of temperature in the protein levels of RNase E, RNase R, RNase II and PNPase.
A. Western blot analysis of cell extracts from cold-shocked cells (10°C) and upon reversion to 37°C (see Experimental procedures) using antibodies against RNase R and RNase E; the period of time under cold shock is indicated in hours. wt- wild-type strain MG1693; rnrrnr::kan (HM104).
B. Quantification of Fig. 1A showing the relative fold of RNase R induction in the wild-type strain at 10°C, normalized to basal levels at 37°C. Quantification of protein levels was estimated by scanning densitometry.
C. Western blot analysis for RNase R, RNase II and PNPase for the same experimental conditions as in A. Δrnbrnb::tet (CMA201).
D. Quantification of RNase R induction under cold shock; average and standard deviations (vertical bars) were estimated from several independent experiments. *– non-specific cross hybridization with RNase E or RNase II antibodies.

Figure 2.

Characterization of the rnr operon transcripts.
A. Northern blot analysis of steady state of rnr mRNA levels at 37°C and in cold shocked cells. wt- wild type (MG1693); rnr- rnr::kan (HM104). A, B and C correspond to the transcripts detailed in Fig. 1B. RNA size markers are indicated in the left. Period of time under cold shock is indicated in hours. The rnr probe used (a) is indicated in Fig. 2.
B. Schematic map of the E. coli rnr region. Coding sequences are indicated by shadowed boxes with the first and last aa of the corresponding proteins. The transcripts starting from the 5′-end mapped in Fig. 2C are indicated by A, B and C. The insertion of the kan cassette in site StuI of the rnr coding sequence is indicated by an open box. The possible promoter regions (−35 and −10) of the rnr operon are underlined; the initiation start (+1) is indicated by an arrow. Starting and termination codons for the yjeB gene and rnr gene are represented as shadow boxes; SD- Shine Dalgarno of the rnr gene. Solid lines (a, b, c, d) indicate probes used in Northern analysis. 1 and 2 indicate the 5′ end labelled primers used for Primer extension analysis.
C. Primer extension analysis of the rnr operon using primer 1. A, C, G and T- M13 DNA sequence. 1-control reaction with tRNA; 2-wt RNA at 37°C; 3-wt RNA 30 min at 10°C; 4-wt RNA 1 h at 10°C, 5-wt RNA 3 h at 10°C.

Among the E. coli ribonucleases only PNPase has been reported to be regulated by cold shock (Jones et al., 1987; Zangrossi et al., 2000; Beran and Simons, 2001; Mathy et al., 2001). Cold shock conditions were tested to evaluate whether the levels of RNase II and RNase R were subjected to some kind of regulation under cold-shock conditions in a similar way to PNPase. Western analysis revealed that RNase R is induced upon cold shock conditions (Fig. 1). The RNase R levels increased by an average of seven- to eightfold after a downshift from 37°C to 10°C (Fig. 1B and D). A mechanism of regulation also seemed to act at 37°C as reversion of cells from 3 h at 10°C back to 37°C lead to a decrease of the protein levels in the wild-type strain (Fig. 1A and B). These results show that RNase R levels are regulated by temperature.

Comparative Western analysis between PNPase, RNase II, RNase R and RNase E (Fig. 1A and C) revealed that only RNase R and PNPase were regulated by cold shock conditions. Consistent with other reports (Zangrossi et al., 2000; Mathy et al., 2001) the levels of PNPase showed a two- to threefold increase 3 h after temperature downshift. RNase R showed an unexpected seven- to eightfold increase, a higher increase then PNPase which has been described as a cold shock protein of class I (Thieringer et al., 1998). RNase II levels had been shown to change under other stress conditions (Cairrão et al., 2001) but its levels did not change after temperature downshift to 10°C (Fig. 1C). Likewise, RNase E levels did not show any significant variations through the range of temperature tested. RNase II and RNase E levels remained the same in both the wild type and rnr mutant, indicating that under these conditions RNase R does not seem to regulate the protein levels of these ribonucleases.

Characterization of the rnr operon under cold shock

Northern analysis was performed in order to determine whether the cold-shock induction of RNase R could be due to a change in mRNA levels. The results showed the presence of several rnr transcripts (A, B and C, Fig. 2A and Fig. 2B) with a molecular weight higher than the expected 2.4 kb corresponding to a monocistronic rnr transcript. In the first hour of cold shock the steady state levels of the rnr transcripts increase by at least 100-fold (Fig. 2A). However this increase was transient, returning to lower levels throughout the acclimation phase (Fig. 2A). The rnr gene is flanked by open reading frames coded by the yjeB gene (upstream) and the yjf H and yjf I genes (downstream) (Fig. 2B). The yjeB gene has 94% homology with LT2, a Salmonella thypimurium transcriptional repressor, yjf H (also known as rlmB) is the E .coli GM2251 rRNA methyltransferase (Lövgren and Wikström, 2001), whereas the possible function of yjf I is still unknown. Northern analysis was performed with probes specific for the yjeB and yjf H genes to clarify the co-transcription mechanism. The larger transcripts A and B were detected with both probes (data not shown). The transcription map emerging from this analysis is reported in Fig. 2B.

In order to identify the transcriptional start of the rnr operon, primer extension analysis was performed using primers complementary to the coding sequence of the rnr and yjeB genes (Fig. 2B and Experimental procedures). The results showed that the transcription of the rnr operon began at a guanine residue located 147 nucleotides upstream of the start codon of the yjeB gene (Fig. 2B and C). This long leader (5′ UTR) is a characteristic feature of many cold-inducible genes in E. coli and other bacteria. The − 10 promoter region is extended with an upstream 5′-TGTGN-3′ motif (Fig. 2B). These extended − 10 promoters are believed to contribute to promoter strength in absence of a strong − 35 hexamer and to the formation of the open-complex formation at the lower temperature (Burns et al., 1996).

Increased stability of rnr transcripts in the cold acclimation phase

It has been demonstrated that cold shock proteins can be post-transcriptionally regulated at the level of mRNA stability (Brandi et al., 1996; Yamanaka, 1999; Yamanaka and Inouye, 2001; Giangrossi et al., 2002). To determine whether the cold shock-induced increase in mRNA levels could be due to a change in mRNA stability, we analysed the decay of the rnr mRNAs before cold shock (37°C), after one hour in the acclimation phase and after reversion of cells from 10°C to 37°C (Fig. 3A and B). At 37°C (0 min before cold shock) the transcripts were hardly visible and disappeared completely 5 min after blocking transcription with rifampicin (in average half-life ≤ 2.7 min). The RNase R levels present at 37°C (Fig. 1) are probably a result of the translation of the yjeB +rnr transcript mRNA as this is the only transcript detected at 37°C (Fig. 3A). One hour after cold shock there was a strong stabilization of the yjeB +rnr transcript (half-life ≥93 min) and co-transcripts (Fig. 3B). After acclimation phase, the transcripts became more labile. These results show that there is a strong increase in the mRNA stability of the rnr during early cold adaptation, which contributes to the accumulation of those transcripts in the first hour of cold shock. However, upon reversion of temperature from 10°C to 37°C a mechanism of degradation was induced and the yjeB +rnr transcripts showed a faster decay (half-life ≤ 11 min), disappearing completely 30 min after reversion. The shift to 37°C therefore induced a rapid decay of the stabilized transcripts. These results demonstrate that the stability of the transcripts from the rnr operon is tightly regulated by temperature.

Figure 3.

Stability of rnr transcripts under cold shock conditions.
A. Northern blot analysis of the stability of rnr transcripts from wild-type cells grown at 37°C, under cold shock (wild type 1 h) and after reversion to 37°C (wild type-1 h at 10°C followed by shift to 37°C). A representative Northern blot analysis of the expression of rnr at 37°C was overexposed for better visualization (Fig. 3A, left panel). Transcription was blocked by addition of rifampicin (0 min) and aliquots harvested at the indicated times for RNA extraction. RNA size markers are indicated in the left. Levels of the yjeB +rnr transcript were detected by hybridization of the Northern blots with the rnr probe (b) as indicated in Fig. 2.
B. Graphical representation of the decay rates of the yjeB +rnr mRNA at 37°C(•), after 1 h at 10°C (▪) and upon reversion to 37°C (□). Quantification of the rnr transcripts was done by scanning densitometry and values of the amount of RNA at zero minutes considered as 100%. The percentage of transcript remaining after rifampicin addition was plotted as a function of time. The half-lives were determined by linear regression analysis.

PNPase is involved in the regulation of the rnr mRNA levels

PNPase has been shown to be involved in the degradation of some cold-shock mRNAs at the end of the acclimation phase (Brandi et al., 1996; Goldenberg et al., 1996; Yamanaka and Inouye, 2001). In order to establish whether PNPase is one of the factors involved in the cold shock regulation of rnr mRNA levels, Northern analysis was performed with the pnp7 mutant of PNPase. At 37°C and after reversion of cells from 10°C to 37°C there were no significant differences in terms of stability of the rnr transcripts in the wild type versus the pnp7 mutant (Fig. 4A). Similarly, 1 h after temperature downshift there was no difference regarding the steady state levels of these mRNAs in the two strains (Fig. 4B). However, 3 h and 6 h after cold shock there was a strong accumulation of the rnr transcripts in the pnp7 strain when compared to the wild-type strain (Fig. 4B). A threefold accumulation of yjeB +rnr transcripts was observed in the pnp7 strain relative to the wild-type strain and this accumulation was maintained even after the acclimation phase (6 h) (Fig. 4C). Likewise there was a 4.5-fold increase of the levels of the yjeB +rnr+yjfH transcript in the pnp7 mutant. These results indicate that PNPase is an important factor involved in the regulation of rnr transcripts at the end of the acclimation phase.

Figure 4.

Effect of PNPase in the expression of the rnr operon.
A. Northern blot analysis of the effect of a pnp7 mutation in the stability of the rnr operon at 37°C and after temperature reversion (wild type 1 h at 10°C followed by shift to 37°C).
B. Northern blot analysis showing the post-transcriptional regulation of the rnr operon under cold shock (1 h and 3 h) and after the acclimation phase (6 h) in the wild-type and the pnp7 strains. The rnr probe (b) described in Fig. 2B was used for detection of the levels of the rnr transcripts. wt- wild-type strain MG1693; pnp7- SK5691. RNA size markers are indicated in the left.
C. Graphical representation of relative steady state levels of rnr transcripts B and C in the pnp7strain normalized to the wt strain (pnp7/wt). Identified bands were quantified by densitometry (imagequant; Molecular Dynamics) and plotted versus time of extraction.

Phenotype of the rnr mutant under cold shock

The increase of expression levels of RNase R under cold shock conditions can be important for cell physiology under such stress conditions. We have compared the cellular growth at 10°C and 37°C in the wild-type and the rnr mutant strains. An arabinose inducible expression plasmid (pBAD24) harbouring the rnr gene was constructed (pRNR – see Experimental procedures) in order to complement the rnr mutant strain with a functional RNase R. The other genes of the rnr operon were not present in the pRNR plasmid. At 37°C the colonies were slightly smaller in the rnr strain than in the wild type (data not shown). To evaluate the effect of cold shock in the phenotype of the wild-type and rnr strains, bacteria were submitted to 3 h of cold shock and plated in rich media, followed by incubation at 10°C (Fig. 5). Comparative growth revealed a phenotype for the rnr mutant, a growth defect characterized by the formation of colonies that were considerably smaller than the wild-type strain (Fig. 5). This effect was even more pronounced with prolonged cold shock conditions (data not shown). Complementation of the rnr mutant in trans with a functional RNase R expressed from pRNR, eliminated the cold phenotype of the mutant strain (Fig. 5). These results indicate that RNR can have an important role in the adaptation of cells to cold shock. The rnr mutant strain only recovers the normal phenotype when complemented with the pRNR plasmid in the presence of arabinose, showing that the cell ability to withstand cold conditions correlates with the content of functional RNase R in the cell.

Figure 5.

Effect of cold shock in colony growth. Bacterial cells were grown under cold shock (10°C) for 3 h. Equal aliquots of the different strains were platted in Luria agar. The size of colonies was compared after incubation at 10°C. wt- MG1693; rnrrnr::kan (HM104); rnr+ pRNR –rnr strain complemented with plasmid pRNR (pHM103). (+Ara) denotes conditions of induced expression of RNase R from pRNR plasmid by addition of 0.2% of arabinose. The phenotype of uninduced cultures (control) was equally compared under the same conditions (data not shown).

RNase R is involved in the maturation of SsrA/tmRNA

It has been reported that bacterial cells lacking a functional SsrA/tmRNA exhibit slower growth (Gillet and Felden, 2001). RNase R was shown to be associated with the SsrA/SmpB system (Karzai and Sauer, 2001). In order to elucidate if RNase R could have a role in the maturation of SsrA/tmRNA under cold shock conditions, Northern blot analysis was performed. The levels of the SsrA/tmRNA were evaluated in the wild-type strain, rnr mutant and an rnr strain transformed with an arabinose inducible plasmid encoding only RNase R (pRNR). Comparative analysis showed that under cold shock conditions in the rnr mutant there was a stabilization of the half-life of precursor SsrA/tmRNA (Fig. 6A–C). This stabilization may account for a threefold increase in the levels of precursor SsrA/tmRNA in the rnr mutant strain (Fig. 6B and C). This stabilization decreases once the rnr strain is complemented with RNase R induced from the pRNR. These results indicate a role of RNase R in the tmRNA maturation process under cold shock. Because of the great stability of SsrA/tmRNA RNA (Komine et al., 1994), the amounts of the mature form of SsrA/tmRNA probably represent molecules processed before the temperature downshift. This was also observed when RNase E was shown to be required for the maturation of SsrA/tmRNA (Lin-Chao et al., 1999). At 37°C the role of RNase R in the maturation of SsrA/tmRNA was less pronounced (Fig. 6A). RNase R seems to specifically process the precursor form of SsrA/tmRNA at low temperature. The results are consistent with the accumulation of RNase R at low temperature. Northern analysis was performed to evaluate the role of other exonucleases in the processing of the precursor form of SsrA/tmRNA. We have compared the wild-type and the rnr mutant strain with a mutant in PNPase (the other cold shock exonuclease) and a mutant in RNase II (the other member of the RNR family that was not induced by cold shock). In the rnr mutant strain the steady state levels of the SsrA/tmRNA percursor increased threefold comparative to the wild-type strain, whereas for the mutants in PNPase and RNase II no significant differences were found from the wild type (Fig. 6D). The comparative analysis has shown that under cold shock RNase R has a more important role in the maturation of SsrA/tmRNA than PNPase or RNase II.

Figure 6.

RNase R and tmRNA maturation.
A. Northern blot analysis of tmRNA maturation at 37°C and under cold shock (3 h) in the wild-type and rnr mutant strains.
B. Northern blot analysis of tmRNA maturation under cold shock (3 h). wt- MG1693; rnr– MG1693 rnr::kan; rnr+ pRNR (pHM103) –rnr strain complemented with plasmid pRNR (pHM103). Pre-ssrA precursor form of SsrA/tmRNA; SsrA/tmRNA-mature form of SsrA/tmRNA. A ssrA PCR generated DNA probe was used to detect ssrA transcripts (see Experimental procedures).
C. Graphical representation of the decay rates of the pressrA transcript plotted (logarithmic scale) as the percentage of pressrA remaining after rifampicin addition (time 0′) in function of time (min). Values of half-lives were determined by linear regression. shows a representative Northern blot analysis from three independent experiments.
D. Northern analysis of SsrA/tmRNA maturation in ribonucleases deficient strains rnr, pnp7, Δrnb, under cold shock conditions. Total RNA was extracted from cells submitted to 3 h under cold shock and steady state levels of ssrA transcripts were detected using the same ssrA probe as described above. M – Labelled markers are indicated in the left of Fig. 6A and B.

Discussion

Bacteria are ubiquitous colonizers and they are often subjected to drastic temperature alterations. The adaptive responses to temperature changes involve a remodelling of bacterial gene expression, aimed at adjusting cell physiology to the new environmental demands (Eriksson et al., 2002). In this report we show that the E. coli exoribonuclease RNase R is a new cold shock protein. Expression of RNase R levels is tightly regulated by temperature mainly due to the stabilization of rnr transcripts. In the first hour upon cold shock there is a dramatic increase in the steady state levels of the rnr transcripts that can account for the increase in RNase R levels. Under these conditions the rnr gene is co-transcribed with other flanking genes [yjeB – upstream, yjf H (rlmB) and yjf I – downstream]. These studies open new perspectives regarding the study and characterization of these genes of the rnr operon. The results from Figs 2 and 3 show that the transcripts from the rnr operon are considerably stabilized at the low temperature. A large extent of stabilization and a pronounced variation of the stability in the course of the acclimation are rather typical of cold shock mRNAs. We can not completely exclude a role of transcription but up to now there is no cold shock protein that has been shown to be increased based on transcriptional effect. Under cold shock some mRNAs become much more stabilized than others (Giangrossi et al., 2002; Polissi et al., 2003). Moreover, if the effect of the cold was simply a stabilization of all the RNAs then we would not observe a differential stability of the same RNA in different moments of the cold adaptation (early versus late). This indicates that in-creased stability in cold shock is not simply a matter of temperature but rather some kind of regulation. This stabilization is transient since after 1 h of induction under cold shock, steady state levels of rnr transcripts start to decline and messages are not so stable (same had been observed for pnp transcripts and other mRNAs from cold shock genes). This transient stability suggests two sequential mechanisms of regulation. In the early acclimation phase there is a protection of the rnr operon transcripts from degradation, either by increased stability of secondary structures formed under cold shock or by the action of some cold shock proteins. Subsequently, in the next hours of the acclimation phase, there is a reduction of that protection and the decay of most transcripts is faster. Consistent with this mechanism of regulation, PNPase, a cold shock induced RNase regulates the rnr mRNA levels at the end of the acclimation phase (Fig. 4). Other examples of a similar post-transcriptional regulation by PNPase have been described (Yamanaka and Inouye, 2001). A shift to 37°C induces a rapid decay of the stabilized transcripts (Fig. 3) showing a temperature dependent regulation of the stability of the rnr transcripts. There is a concomitant decrease in protein levels once the temperature is reversed to 37°C (Fig. 1). Therefore the expression of RNase R seems to be tightly regulated by temperature.

The results from Western analysis show that RNase R is a cold shock protein (Fig. 1). RNase R and PNPase are the only cold shock ribonucleases so far described. The results from Fig. 1C showed that the levels of RNase E and RNase II are not regulated by a decrease of temperature. After a shift to low temperature RNase R shows a higher increase in protein levels (seven- to eightfold induction) than PNPase (two- to threefold), which has been described as a cold shock protein of class I (Thieringer et al., 1998). The increase in RNase R levels under cold shock conditions raises the question about the physiological role of this exoribonuclease at low temperature. It is known that rnr has an essential function in the cell because a double mutant, pnp rnr, is not viable (Cheng et al., 1998). Both enzymes are induced upon cold shock reinforcing the notion that they have some overlapping function that cannot be substituted by any other ribonuclease in the cell.

Taking into account our results and the present knowledge on RNase R we propose a model for the involvement of RNase R in the network of adaptative responses to changes in environmental conditions, namely cold shock (Fig. 7). Previous studies demonstrate that RNase R is particularly more effective than other ribonucleases on the processing/degradation of RNA molecules containing extensive secondary structures such as tRNAs, 16S and 23S RNA (Cheng and Deutscher, 2002;2003). These roles may be especially important under stress conditions such as cold shock, when the secondary structures of RNAs are increased. Our results show that RNase R is involved in the maturation of SsrA/tmRNA, a small stable RNA. This fact broadens the importance of RNase R in the maturation of structured RNAs (Fig. 7). RNase R was shown to co-precipitate with the SsrA/SmpB system that is involved in the trans-translation process (Karzai and Sauer, 2001). Our results clearly demonstrate that RNase R has a role on the processing of the precursor form of SsrA/tmRNA under cold shock. It is surprising that inactivation of an exoribonuclease delays the maturation of a primary transcript that has been demonstrated to be processed by endonucleases. One would rather expect that exonucleases are involved in the final step of processing. In this case inhibition of exoribonucleases should cause accumulation of intermediates, not stabilization of a percursor. The data here presented imply that RNase R catalyses (a) rate limiting cleavage(s). In the rnr mutant strain the steady state levels of the SsrA/tmRNA percursor increased threefold when compared to the wild-type strain. The stabilization of the half-life of the precursor form of SsrA/tmRNA in the rnr mutant is decreased once the mutant rnr is complemented with a functional RNase R (Fig. 6). The other proteins of the rnr operon do not seem to be necessary for this process. This is the first report showing that RNase R is involved in the processing/maturation of small, stable RNAs.

Figure 7.

Model for the involvement of RNase R in several biological processes. RNase R is implicated in RNA quality control. The role of RNase R in the maturation of the small stable SsrA/tmRNA (present work) also link RNase R with the assurance mechanisms that contribute for protein quality control. The role of RNase R in maturation of structured RNAs (rRNAs, tRNAs, tmRNA) has indirect consequences for ribosome rescue, ribosome biogenesis and consequently translation. RNase R can be particularly important in stress conditions such as cold shock. Changes in temperature regulate virulence factors and permanence in cold can increase pathogenicity of Enterobacteriaceae. RNase R and SsrA/tmRNA are required for expression of virulence genes. The results from this work suggest a broader role for RNase R in pathogenesis.

The retardation in the maturation of SsrA/tmRNA is only clearly observed in a cold shock dependent manner, a fact that is consistent with the accumulation level of RNase R. The fact that PNPase and RNase II did not seem to participate significantly in the maturation of precursor form of SsrA/tmRNA, showed that RNase R is the main exonuclease involved in this process.

RNase R was implicated in the degradation of defective structured tRNAs (Li et al., 2002) and recently it has been shown to be involved in the quality control of ribosomal RNA (Cheng and Deutscher, 2003) showing that RNase R is involved in RNA quality control (Fig. 7). The eukaryotic exosome, a multiprotein complex involved in 3′ to 5′ exonucleolytic degradation, has been shown to be involved in RNA surveillance processes (Hilleren et al., 2001; Torchet et al., 2002). It is interesting that one of the subunits of the exosome is Rrp44p has high homology to RNase R (Mitchell et al., 1997). The absence of a functional RNase R leads to an aberrant tagging of proteins by the tmRNA system (Karzai and Sauer, 2001), demonstrating that RNase R is also important in the quality control of proteins (Fig. 7). We hypothesize that the accumulation of unprocessed forms of tmRNA in the rnr strain may be one of the factors responsible for the observed impairment in the trans-translation system.

In biological systems in which SsrA/tmRNA activity is required for growth and/or viability, the data show that the primary function of tmRNA is the release of stalled ribosomes from RNA (Withey and Friedman, 2002) –Fig. 7. This ribosome release mediated by trans-translation becomes more critical for organisms under stress conditions (Karzai et al., 2000; Gillet and Felden, 2001). Phenotypes associated with bacterial cells lacking a functional SsrA RNA include growth defects under stress conditions (Gillet and Felden, 2001). In this work we report a similar phenotype associated with the rnr mutant under cold shock, characterized by smaller colonies compared to the wild-type strain. The complementation of this mutant with a functional RNase R (without the other proteins of the rnr operon) is enough to rescue the cold-shock phenotype. The various roles of RNase R in the physiology of the cell (Fig. 7) through RNA maturation, RNA quality control (RNA surveillance), and its role in trans-translation affecting protein quality control and ribosome rescue, may contribute to the overall growth deficiency observed in the rnr mutant.

The fact that PNPase regulates RNase II and RNase R, the two major hydrolytic exoribonucleases at the level of mRNA stability (Zilhão et al., 1996b and this work), strengthens the idea of a coordination between the levels of exonucleases. Moreover, the inviability of double mutants between the major exoribonucleases (pnp rnr and pnp rnb mutants) indicates distinct overlapping functions between RNase II and RNase R. The fact that a double mutant rnb rnr is viable (our unpub. results), whereas pnp rnr and pnp rnb mutants are not, suggests that PNPase has at least two essential functions: one that can only be compensated by RNase II (probably its role in the degradation of mRNAs) and the other by RNase R (probably its role in quality control).

RNase R was described as a gene required for full expression of the virulence phenotype of Shigella spp. and enteroinvasive E. coli (Tobe et al., 1992; Cheng et al., 1998). PNPase, another cold shock RNase, was also involved in virulence (Clements et al., 2002). Many pathogens can grow in cold conditions and survival of enterohaemorrhagic E. coli O157:H7 is greatest at 8°C (Wang and Doyle, 1998). Although the exact role of RNase R in establishing virulence still has to be determined, the fact that RNase R is a cold shock protein may help to clarify some mechanisms of pathogenesis associated with low temperature conditions (Fig. 7).

Considering that the role of RNase R has been underestimated, and in view of the widespread presence of RNase R among Prokaryotes and Eukaryotes, it is important to pursue the investigation of this exoribonuclease. We have shown that RNase R is involved in the maturation of SsrA/tmRNA and perhaps that it may also be involved in the processing of other small stable RNAs. This would be quite interesting taking into account that recent studies indicate that small, non-coding RNAs are far more abundant and important than initially imagined (Storz, 2002). The finding that RNase R is a cold shock protein opens new perspectives in the understanding of the general mechanisms of adaptation to cold shock conditions and pathogenesis.

Experimental procedures

Bacterial strains and growth conditions

All strains used in this work are isogenic derivatives of the wild-type strain MG1693 (thyA715) and were previously described in Zilhão et al. (1996a): SK5691(thyA715, pnp7), CMA201(thyA715, Δrnb), HM104 (thyA715, rnr::kanR this work). For cold shock experiments bacterial cultures were grown at 37°C in Luria broth up to middle exponential phase (OD600 = 0.5–0.6) and then transferred to a water bath at 10°C. Samples were removed before and after cold shock at times indicated on figures. Ampicillin (100 µg l−1), kanamycin (50 µg ml−1) and thymine (50 µg l−1) were supplemented when required. For comparative growth of the different strains, equal amounts of cells were plated on Luria broth, followed by incubation either at 37°C or at 10°C.

For construction of the rnr::kanR strain (HM104), a 5.3 kb EcoRI fragment of the 95 min region from the E. coli genome (Kihara et al., 1996) was cloned into pUC119 (Biolabs®Inc.) resulting in plasmid pHM102. A kanamycin marker derived from pUCAK (Amersham Biosciences) was inserted into the unique StuI site of the coding region of rnr, in pHM102. A 7 kb ApaLI fragment containing the insertion mutation was electroporated into strain FS1576 (Stahl et al., 1986) and subsequently transferred into strain MG1693 by P1 transduction, resulting in strain HM104. The rnr mutation in HM104 was confirmed by polymerase chain reaction (PCR), restriction enzyme digestion patterns, and Western analysis. Plasmid pHM103 contains the coding sequence of the rnr gene under the arabinose inducible promoter of pBAD24 (Invitrogen®) (a 3.7 kb KpnI-SphI fragment from pHM102 fused with a PCR generated NcoI–KpnI N-terminal fragment of the rnr sequence). Induction of the RNase R was done by addition of 0.2% of l-arabinose to growth cultures.

Northern blot analysis

Northern blot analysis was performed according to Sambrook et al. (1986). Total RNA was extracted as previously described (Zilhão et al., 1996a). For decay experiments, transcription was blocked by addition of rifampicin (500 µg ml−1) and nalidixic acid (20 µg ml−1) at time 0 min. Aliquots of cell cultures were withdrawn at times indicated in figures for RNA extraction. For analysis of the rnr operon equal amounts of total RNA (15 µg) were electrophoresed on a 1.5% agarose MOPS/formaldehyde gel and blots hybridized with DNA probes labelled with 32P. The probes used are indicated in Fig. 2B: a 1.2 kb [PstI-PstI](a) or 2.8 kb [EcoRI-NdeI](b) DNA fragment of the rnr gene; a 0.75 kb [DraI-DraI] probe encoding the downstream gene yjfH (d) and a PCR generated DNA fragment encoding the yjeB gene (c). Northern analysis for detection of SsrA/tmRNA was done in denaturing 6% polyacrylamide gels: for each strain equal amounts of total RNA (7.5 µg) were electrophoresed and blots hybridized with a ssrA generated PCR DNA probe. Quantification of mRNA transcripts was done by scanning densitometry imagequant program (Molecular Dynamics). The half-lives of mRNA in each strain were determined by linear regression using the logarithmic of the percentage of mRNA remaining versus time, considering the amount of RNA at zero minutes as 100%. The oligos used for each probe were the following: yjeB-F [5′-GGATCGTACTGAAACCATGATTCTGC-3′], yjeB-R [5′-CCGTAATCAGTGAAACTCGTTAACTGC-3′]; ssrA-F [5′-CAGGCTACATGGGTGCTAAATC-3′] and ssrA-R [5′-GAC TTCGCGGGACAAATTGAGG-3′]. The markers used were either a RNA ladder (Sigma®) or a labelled DNA sized standard derived from pBR322 digested with HaeIII and MaeI.

Primer extension analysis

To detect the transcription start of the rnr operon, primer extension analysis was performed with oligos labelled at the 5′ end with [γ-32P]-ATP: primer 1(5′-CCGTAATCAGTGAAAC TCGTTAACTGC-3′) and primer 2(5′-CGCGTATTTTTCAGCT TCGCGTTCC-3′). Total RNA (30 µg) was ethanol precipitated with the labelled primers and primer extension reactions performed according to Sambrook et al. (1986). The reaction products were analysed on a 6% sequencing gel. Sequencing reactions were prepared with the T7 Sequenase kit vs2.0 (Amersham Biosciences) using primer1 (yjeB-R) and as template plasmid pyjeB (pWSK29 containing a blunt PCR amplified DNA fragment of yjeB) or the M13 sequence.

Purification of a His(6)-RNR and antibody production

To create a pET28a(RNR) the rnr gene was amplified by PCR, using as template a chromosomal DNA preparation from strain MG1655 (E. coli Genetic Stock Center). Oligonucleotides primers RNR1 [5′-GCAGGAAGCCATATGTCACAA GATCCTTTC-3′] and rRNR2 [5′-CGCATTTTGTCAGGATC CTAACCC-3′] were designed in order to generate a NdeI site in frame with the histidine tag of pET28a (Novagen®). Cloning strategy included an intermediate cloning step into pSTBlue-1 (Invitrogen®) and posterior insertion of a NdeI-EcoRI fragment into the NdeI and EcoRI sites of pET28a. Overproduction of His(6)-RNR was carried out in strain BL21(DE3) (Novagen®). Cells were grown at 30°C in Luria broth supplemented with 1 mM PMSF. Induction was imposed at an OD(600 nm) of 0.4 by adding IPTG at a final concentration of 1 mM. Induced cells (2 h after IPTG addition) were collected by centrifugation (10 min at 7000 r.p.m., 4°C) and washed once in cold PBS before storage of pellets at − 70°C for posterior protein purification. After sonication, clarified lysates in buffer A (50 mM NaHPO4, 300 mM NaCl, 0,5% Triton X-100, 20 mM Imidazol, 1 mM PMSF, 1 mg ml−1 lysozyme) were applied to a column of Ni2+-NTA agarose (QIAGEN®) pre-equilibrated with 10 volumes of buffer A without lysozyme. After washing of the column with six volumes of buffer A, the elution was conducted with a gradient increase of imidazol in buffer B (50 mM NaHPO4, 300 mM NaCl, 0.5% Triton X-100, 250 mM Imidazol). The different elutions were analysed by SDS-PAGE to confirm purity of the His(6)-RNR. Quantified amounts of the purified His(6)-RNR were used for antibodies production according to procedures of EUROGENTEC®.

Western blot analysis

For Western blotting 40 µg of total protein were separated on SDS-PAGE gels and the blotting performed as described (Cairrão et al., 2001). Equivalence of protein loading between lanes was judged by Ponceau-S staining of the membrane. For the detection of the different RNases the membranes were incubated with diluted primary antibodies (1:10000- RNase R, RNase II; 1:100000 PNPase; 1:25000 RNase E). After exposure of the blots to Biomax-MR film (Kodak), quantification of the intensity of bands was estimated by scanning densitometry imagequant program (Molecular Dynamics).

Acknowledgements

We would like to thank A.J. Carpousis for the gift of antibodies against RNase E and PNPase, Federica Briani and Gianni Dehò for discussion and communication of unpublished results, and Claudio Gualerzi for helpful suggestions. F.C. was recipient of a doctoral grant from F.C.T. This work was in part supported by grants from Fundação para Ciência e Tecnologia, Portugal, to C.M.A.

Ancillary