The RNA degradosome and poly(A) polymerase of Escherichia coli are required in vivo for the degradation of small mRNA decay intermediates containing REP-stabilizers

Authors

  • Vanessa Khemici,

    1. Laboratoire de Microbiologie et Génétique Moléculaire (CNRS, UMR 5100) and Paul Sabatier Université, 118 Route de Narbonne, 31062 Toulouse, France.
    Search for more papers by this author
  • Agamemnon J. Carpousis

    Corresponding author
    1. Laboratoire de Microbiologie et Génétique Moléculaire (CNRS, UMR 5100) and Paul Sabatier Université, 118 Route de Narbonne, 31062 Toulouse, France.
    Search for more papers by this author

Summary

In Escherichia coli, REP-stabilizers are structural elements in polycistronic messages that protect 5′-proximal cistrons from 3′→5′ exonucleolytic degradation. The stabilization of a protected cistron can be an important determinant in the level of gene expression. Our results suggest that RNase E, an endoribonuclease, initiates the degradation of REP-stabilized mRNA. However, subsequent degradation of mRNA fragments containing a REP-stabilizer poses a special challenge to the mRNA degradation machinery. Two enzymes, the DEAD-box RNA helicase, RhlB and poly(A) polymerase (PAP) are required to facilitate the degradation of REP-stabilizers by polynucleotide phosphorylase (PNPase). This is the first in vivo evidence that these enzymes are required for the degradation of REP-stabilizers. Furthermore, our results show that REP degradation by RhlB and PNPase requires their association with RNase E as components of the RNA degradosome, thus providing the first in vivo evidence that this ribonucleolytic multienzyme complex is involved in the degradation of structured mRNA fragments.

Introduction

Messenger RNA stability is an important determinant in the control of gene expression. Over the past decade, our understanding of the mechanism of mRNA degradation in Escherichia coli has advanced significantly (Coburn and Mackie, 1999; Grunberg-Manago, 1999; Regnier and Arraiano, 2000; Steege, 2000). It is now generally accepted that a single-strand-specific ribonuclease, RNase E, initiates mRNA decay (Kushner, 2002). RNase E works together with two 3′→5′ exonucleases, RNase II and PNPase, that degrade mRNA fragments to nucleotides (Donovan and Kushner, 1986). In E. coli and other bacteria, there is neither a 5′ mRNA cap structure nor any known 5′→3′ exoribonucleases. RNase E and the 3′→5′ exonucleases act cooperatively as mRNA decay intermediates are not experimentally detectable except in mutant strains where the exonucleolytic pathway has been inactivated. Thus, cleavage by RNase E triggers the rapid degradation of the message to nucleotides. However, this view is oversimplified as many prokaryotic transcripts are polycistronic and the individual cistrons can have significantly different rates of translation and decay. For large polycistronic messages, the decay of 5′ proximal cistrons often occurs before the distal cistrons are transcribed. Indeed, now classical work by Yanofsky and colleagues showed that mRNA levels of cistrons within the trp operon vary considerably, as does the synthesis of the corresponding enzymes, even though the operon is transcribed as a single polycistronic message (Morse et al., 1969; Rose et al., 1970). Thus, RNA decay in E. coli appears to operate at the level of the cistron, not the full-length message.

These considerations suggest that polycistronic messages have barriers preventing the spread of degradation to adjacent cistrons. Several barriers against degradation by the 3′→5 exonucleases are well known. RNase II and PNPase both require free single-strand 3′ends for binding to their substrates. Thus, nascent transcripts are protected from the exonucleases as their 3′ ends are sequestered in a ternary elongation complex containing the DNA template and RNA polymerase. A second barrier is the RNA structure formed by rho-independent transcription termination. The short oligo(U) stretch at the 3′ end of these messages is insufficient for exonuclease binding and the adjacent RNA stem-loop, which is often very GC-rich, locks the 3′ end into a stable, double-stranded structure. The protection from the exonucleases of nascent messages and messages with rho-independent terminators appears to largely explain the requirement for an endonuclease, such as RNase E, to initiate degradation. However, a third type of barrier against the exonucleases involves repeated extragenic palindrome (REP) sequences. Escherichia coli has nearly six hundred REPs located in intercistronic regions in single copy or more often in multiple copies (Bachellier et al., 1999). When transcribed, they fold into stable RNA structures between cistrons within a polycistronic message. Previous work with the malEFG operon has shown that a REP-stabilizer protects the upstream malE cistron from exonucleolytic degradation (Newbury et al., 1987a, b). A long-lived malE message containing the 3′ REP-stabilizer arises by processing of the polycistronic malEFG message. The stabilization of the malE message is required for normal, high-level synthesis of the malE gene product. In vitro work has shown that the malE REP-stabilizer slows the progression of exonucleases such as RNase II and PNPase leading to the production of stable intermediates in exonucleolytic decay (McLaren et al., 1991; Py et al., 1996; Blum et al., 1999; Coburn et al., 1999).

The in vivo experiments reported here show that maturation of the malE message requires RNase E in a pathway that involves endonucleolytic cleavage of the malEFG primary transcript and exonucleolytic trimming to the REP-stabilizer. The degradation of the mature malE message is also initiated by an endonuclease, which is likely to be RNase E. The subsequent degradation of mRNA fragments containing the 3′ REP-stabilizer poses a special challenge to the RNA degradation machinery. Our results show that two enzymes, RNA helicase B (RhlB) and poly(A) polymerase (PAP), are required to facilitate the exonucleolytic degradation of mRNA fragments containing REP-stabilizers. RhlB, a DEAD-box RNA helicase (Kalman et al., 1991; Schmid and Linder, 1992), is part of the RNA degradosome, which is a ribonucleolytic complex containing RNase E, RhlB, enolase and polynucleotide phosphorylase (PNPase) as major components (Carpousis et al., 1994; Py et al., 1994; 1996; Miczak et al., 1996). Poly(A) polymerase has previously been implicated in mRNA decay (Hajnsdorf et al., 1995; O’Hara et al., 1995; Haugel-Nielsen et al., 1996; Carpousis et al., 1999). Our results have been extended to the analysis of other messages containing REP-stabilizers and in each case, in the absence of RhlB and PAP, mRNA decay fragments containing REP-stabilizers accumulate to high levels. This work is the first in vivo evidence that RhlB and PAP act in the degradation of REP-stabilized mRNA. Furthermore, although the REP-stabilizer at the 3′ end of the mature malE message provides effective protection against the exonucleases in vivo, decay intermediates containing REP-stabilizers, produced by endonucleolytic cleavage, are rapidly degraded. These results suggest an ordered pathway in which the initial endonucleolytic cleavage triggers exonucleolytic degradation of the REP-stabilizer that is facilitated by RhlB and PAP.

Results

Disruption of the gene encoding RhlB

We have disrupted the gene encoding RhlB (Experimental procedures). Briefly, because rhlB is part of a bicistronic operon, we used a strategy that was designed to disrupt RhlB expression, but avoid interfering with expression of the downstream gppA gene. The mutant strain contains an in frame deletion of 405 codons yielding a mini rhlB gene encoding the N-terminal and C-terminal eight amino acids of RhlB, and an internal eight amino acid insertion that served as a ‘tag’ for identifying strains with the deletion. The loss of sequences hybridizing to a probe from within the coding sequence was verified by Southern blotting and the loss of a 50 kDa protein that reacts with antibodies against RhlB was verified by Western blotting (data not shown). Characterization of SVK1, the strain harbouring the rhlB deletion, showed that it grows normally at temperatures between 30 and 42°C. The analysis of the decay of bulk mRNA as well as specific mRNAs such as those encoded by the rpsO and rpsT genes did not reveal any effect on their stability (data not shown).

Processing and degradation of the malEFG message

Because previous in vitro work has implicated RhlB and PAP in the degradation of the malE REP-stabilizer (Py et al., 1996; Blum et al., 1999; Coburn et al., 1999), we wanted to analyse their effect on the malEFG message in more detail. We thus constructed a strain in which we combined rhlB with pcnB (PAP mutant). Figure 1 shows a schematic representation of the malEFG operon and its transcripts, and the mRNA from the intercistronic malE-malF region folded into secondary structure. The ‘BIME′ is a DNA element containing repeated palindromic sequences (Bachellier et al., 1999). In preliminary experiments with Northern blots of agarose gels hybridized with a probe against malE, we detected no effect of the rhlBpcnB double mutant on the stability of the full-length malEFG or the mature malE message (data not shown). However, we detected a smear of small RNAs at the bottom of the gel that were not present in wild type. Because we suspected that these small RNAs could be intermediates in mRNA decay, we decided to first examine the role of RNase E in the degradation of the malE message. Figure 2 shows a Northern blot of an agarose gels hybridized with a probe against malE, in which RNA was isolated from a temperature-sensitive (ts) mutant at the permissive and non-permissive temperatures. In agreement with previous work, the malE message is much more abundant than the full-length malEFG message. The high levels of the malE message have previously been attributed to its long lifetime relative to the short-lived malEFG message (Newbury et al., 1987a). The nearly complete absence of the malE transcript in lane 4 shows that the maturation of malE depends on RNase E. The presence of more mRNA in lane 2 is probably due to partial inactivation of RNase E even at the permissive temperature, which is a well-known characteristic of the mutant allele employed here. These results show that RNase E initiates the maturation of the malE message, probably by cleaving the polycistronic transcript in malF or malG to create a free 3′ end that can then be trimmed by the 3′→5′exonucleases (see Discussion). The mature malE message could be degraded by either of two possible pathways: a ‘pure’ exonucleolytic degradation that does not require RNase E cleavage of the mature malE message or a ‘typical’ exonucleolytic degradation that is initiated by an RNase E cleavage in the body of the malE message. If the latter possibility is correct, endonucleolytic cleavage by RNase E would be involved in both the maturation and degradation of the malE message.

Figure 1.

The malEFG operon.
A. Schematic diagram of the malEFG operon and its transcription. At the top, the thick lines represent the coding sequences, the thin lines the intergenic regions, and the box, a bacterial interspersed mosaic element (BIME) containing three repeated extragenic palindrome (REP) sequences. Below (wavy lines), the primary transcript (malEFG) and the mature malE transcript, with a 3′ REP-stabilizer, that arises by processing of the primary transcript.
B. RNA structure of the malE REP-stabilizer. Y1, Z2 and Y2 indicate the REP sequences in the malE-malF intergenic space. The RNA structure is based on previous biochemical work (McLaren et al., 1991). Although Y1 and Z2 can form RNA stem–loops by themselves, in this structure they pair to form an extended region that is mostly double-stranded (c. 70% of the nucleotides are base-paired). The sequence shown here is from the E. coli K12 genomic data ( Blattner et al., 1997). REP-stabilized RNA (RSR) indicates the 3′ ends of the mature malE message, which extends 3–9 nt from the base of the stem-loop formed by Y1 and Z2. The stop codon of malE and the start of malF are boxed. SD is the Shine–Dalgarno sequence for translation initiation of malF.

Figure 2.

The role of RNase E in the processing and degradation of the malEFG message. Northern blot of RNA separated by agarose gel electrophoresis, hybridized with a probe against the malE gene. The RNA was isolated from MC1061 (wild type) and SVK39, containing the rne1 allele that encodes a temperature-sensitive (ts) RNase E. The RNA was isolated at the permissive and non-permissive temperature as described (Experimental procedures). Lanes 1–4, respectively: wild type (+) and rne1 (ts) at 30°C; wild type (+) and rne1 (ts) at 42°C. The position of the ribosomal RNA (16S and 23S) is indicated to the left of the panel.

Characterization of mRNA fragments from the malE message

To further characterize the small RNAs detected with the malE probe in the rhlBpcnB background, the RNAs were separated on polyacrylamide gels, blotted and hybridized with three DNA probes corresponding to the 3′ end of the malE coding sequence, the malEF intercistronic region and the 5′ end of the malF coding sequence. In Northern blots, the small RNAs were only detected with the intercistronic probe (data not shown). These results suggest that the small RNAs correspond to mRNA fragments containing the REP-stabilizer. Figure 3A is an example of the results obtained with the intercistronic probe. RNA species from 80 to 370 nt are detected. In these experiments, the cells were grown in the presence of maltose to induce the expression of the malEFG message. Because REP sequences in different operons are closely related to each other, we wanted to test if the RNAs detected in Fig. 3A depend on maltose for synthesis. Figure 3B shows an experiment with the rhlBpcnB strain in which the cells were grown in the presence of glycerol, glucose or maltose. The synthesis of the major RNAs at 370, 290, 124 and the lower half of the 80–90 nt region depends on maltose whereas the minor RNAs at 140 nt and the upper half of the 80–90 nt region are maltose-independent. These results show that the major species detected in Fig. 3 depend on induction by maltose.

Figure 3.

Detection of small mRNA decay intermediates with a probe against the malE REP-stabilizer. Northern blot of RNA separated by polyacrylamide gel electrophoresis, hybridized with a probe against the malE REP-stabilizer.
A. RNA was isolated from wild type, rhlB, pcnB and rhlBpcnB strains (lanes 1–4, respectively). The markers (M) are a mixture of HaeIII and TaqI digested pBR322 (Appligene) end-labelled with γ-[33P]-ATP.
B. RNA was isolated from the double mutant strain (rhlBpcnB) grown with glycerol, glucose or maltose (lanes 1–3, respectively). To the right of each panel (dots and bar) is the size of the RNA species detected in the mutant strains.

In addition to testing the double mutant, Fig. 3A also shows the wild- type control and the rhlB and pcnB single mutants. In the wild-type control, little or no small RNA was detected. The signal at the top of the blot corresponds to non-specific hybridization to high molecular weight nucleic acid as well as specific hybridization to the 1.3 and 3.9 kb mal messages. In the single mutants, traces of the 370 and 290 nt species can be detected, but they are less abundant than in the double mutant. The smaller maltose-induced species at 124 nt and the top half of the 80–90 nt region can be detected in pcnB alone, but we have reproducibly detected more of these species in the double mutant. In these and subsequent experiments, equal amounts of total RNA were separated by electrophoresis and in some experiments we verified the quality of the blots with a probe against the 5S ribosomal RNA (data not shown). In a further characterization of the mRNA decay intermediates detected in Fig. 3, we inhibited transcription by rifampicin and RNA levels were measured by Northern blotting (data not shown). In the double mutant, the 370 nt species disappeared with a half-life of about 2 min whereas all the other species were virtually stable in a 20 minute time course (greater than 80% remaining). These results cannot be interpreted in terms of RNA stability because rifampicin blocks transcription, not the conversion of mRNA to fragments. Nevertheless, the apparent stability in a 20 minute time course suggests that the turnover of these mRNA fragments is markedly inhibited in the rhlBpcnB strain. The detection of small mRNA fragments containing the malE 3′ REP-stabilizer is incompatible with a purely exonucleolytic pathway for malE mRNA degradation. An endonuclease, which is likely to be RNase E, is required for their production.

Degradation of REP-stabilizers in other messages

To test if the rhlB and pcnB mutations affect other REP-containing messages, we examined six intercistronic regions containing REP sequences. Figure 4 shows a panel of Northern blots hybridized with REP-specific probes against nrdA-nrdB, cca-bacA, ppc-argE, sdhB-sucA, and sucB-sucC(Table 1). Indicated to the right of each lane, is the position of prominent RNA species that are present in the double mutant, but not detected in the wild-type control. Each probe reveals a distinctive set of small RNAs. Thus, the accumulation of REP-stabilized mRNA decay intermediates, when rhlB and pcnB are disrupted, appears to be a general effect. The result with the sucB probe (Fig. 4E) was particularly striking as a predominant species of 210 nt accumulates. In other experiments, we also detected the accumulation of mRNA decay intermediates from the lamB-malM intercistronic region (data not shown). Examination of the Northern blot with the sucB probe (Fig. 4E), shows that the 210 nt species is not detected in wild type, it is present as a minor species in rhlB, clearly visible in pcnB, and very abundant in the double mutant. Although the patterns are more complex, comparable results are seen with the other REP probes (Fig. 4A–D). The hierarchy of effects was always rhlB < pcnB < rhlBpcnB. These results show that RhlB and PAP both have a role in the degradation of the REP-containing mRNA. The cumulative effect seen in the double mutant suggests that RhlB and PAP act independently of each other. The hierarchy suggests that PAP by itself is more effective than RhlB although the results clearly show that both enzymes are required for the normal rapid degradation of mRNA decay intermediates containing REP-stabilizers.

Figure 4.

Detection of small mRNA decay intermediates with a selection of probes against different REP-stabilizers. Northern blot of RNA separated by polyacrylamide gel electrophoresis, hybridized with probes against a selection of REP-stabilizers (Table 1). In each panel, RNA was isolated from the wild type, rhlB, pcnB and double mutant strain (rhlBpcnB): lanes 1–4, respectively. The probes were against nrdA-nrdB, cca-bacA, ppc-argE, sdhB-sucA and sucB-sucC; A–E respectively. Prominent small RNAs that were detected in the double mutant, but not in the wild type, are indicated to the right of each panel by dots (discrete bands) and bars (multiple bands and smears). The size range indicated to the left of A is based on markers as described (Fig. 3). In E, the major small RNA that accumulates has a size of 210 nt as determined by comparison to molecular weight markers.

Table 1. . BIMEs studied in this work a.
Intercistronic regionbp (total/BIME)Structure
  • a

    . BIMEs are bacterial interspersed mosaic elements. The data in this table was taken from a web site at the Pasteur Institute constructed by Sophie Bachellier (http://www.pasteur.fr/recherche/unites/pmtg/repet/tableauBIMEcoli.html). Y, Z1 and Z2 are symbols representing different classes of REP (repeated extragenic palindrome) sequences. S, s and r represent different classes of sequences located between the REPs. For further details see (Bachellier et al., 1999).

malE-malF 153/130YSZ2sY
nrdA-nrdB 233/205YSZ2sYSZ2Y
cca-bacA 180/129YSZ2sY
lamB-malM 242/129YsZ2SY
ppc-argE 597/348Z2rY > 20 < Z2rY > 20 < Z2rY
sdhB-sucA 300/122YsZ2SZ1
sucB-sucC 274/169YsZ2SYsZ2

PNPase, but not RNase II, is involved in the degradation of REP-stabilizers

To test if RhlB and PAP are acting with PNPase or RNase II, we have constructed strains in which the pcnB and rhlB mutations were combined with mutations in the genes encoding PNPase or RNase II. Figure 5A shows a Northern blot with RNA from strains in which PNPase was inactivated. The blots were hybridized with the probe against the sucB REP. For comparisons, the isogenic wild-type controls are shown in Fig. 5B. In pnp (Fig. 5A, lane 1) there are visible amounts of the 210 nt species as well as a second slightly larger species. These RNAs are not visible in the wild-type control (Fig. 5B, lane 1). Combining rhlB with pnp had no additional effect whereas combining pcnB with pnp leads to complete stabilization of the 210 nt RNA. The pnppcnBrhlB triple mutant has no additional effect when compared to the pnppcnB double mutant. These data implicate PNPase in the degradation of the REP-containing mRNA decay intermediates and show that the effect of the pnp mutation is comparable to that of the rhlB mutation. This result suggests that PNPase and RhlB are in the same pathway. However, these results also show that, in the pnp background, a poly(A)-dependent ribonuclease other than PNPase, which does not require RhlB activity, participates in the degradation of the REP-containing mRNA as there is a large difference in the accumulation of the intermediates between the pcnB+ and pcnB backgrounds (Fig. 5A, lanes 1 and 2 versus lanes 3 and 4). Figure 5C and D show, respectively, Northern blots with RNA from strains in which RNase II has been inactivated and the isogenic wild-type controls. The blots are essentially identical showing that RNase II does not have a role in the degradation of the REP-containing RNAs in a pnp+ background. Results comparable to those shown in Fig. 5 have been obtained in an analysis of the malE REP-stabilizer (data not shown).

Figure 5.

The role of PNPase and RNase II in the degradation of the sucB REP stabilizer. Northern blots of RNA separated by polyacrylamide gel electrophoresis, hybridized with a probe against the sucB REP-stabilizer. The RNA was isolated from strains with pnp (PNPase) or rnb (RNase II) mutations (A and C, respectively). In each panel, RNA was isolated from a strain containing the single mutant (lane 1) or the single mutant combined with rhlB, pcnB or rhlBpcnB (lanes 2–4, respectively). Also shown for comparison are the isogenic, wild-type controls (B and D, respectively). The 210 nt RNA is the predominant species that accumulates when the degradation of the sucB REP-stabilizer is impaired.

PNPase and RhlB act as components of the RNA degradosome

Because RhlB and PNPase both associate with RNase E in the E. coli RNA degradosome, we wanted to test if this interaction is necessary for their function in vivo. Figure 6A shows a cartoon of the RNA degradosome. The C-terminal half of RNase E is an extended non-catalytic region that contains the sites of interaction with RhlB, enolase and PNPase. The rneΔ10 and rneΔ18 mutations encode deleted forms of RNase E that are incapable, respectively, of forming a complex with PNPase, or RhlB and enolase (Vanzo et al., 1998; Leroy et al., 2002). The full-length RNase E polypeptide has 1061 amino acids. The RneΔ18 and RneΔ10 polypeptides contain, respectively, deletion of residues 728–845 and 844–1045. It should be noted that strains harbouring these mutant alleles are viable and that the mutant RNase E has ribonuclease activity (Vanzo et al., 1998; Leroy et al., 2002). Figure 6B shows a Northern blot with the sucBC probe using RNA prepared from strains with the rneΔ18 mutation. Figure 6D shows the isogenic wild-type controls. The 210 nt species is clearly visible with the rneΔ18 mutation. Thus, rneΔ18 alone has an effect on the turnover of the REP-containing mRNA. The rneΔ18 rhlB double mutant shows a noticeable increase in the 210 nt species, which has been observed reproducibly. In the pcnB background, the abundance of the small RNAs with the rneΔ18 pcnB double mutant (Fig. 6B, lane 3) is essentially equivalent to the rhlBpcnB double mutant (Fig. 6D, lane 4) and there is no additional effect with the rneΔ18 pcnBrhlB triple mutant (Fig. 6B, lane 4). These data show that the rneΔ18 mutation, which disrupts the interaction with RhlB and enolase, has an effect on the degradation of the REP-containing mRNA decay intermediates. In the pcnB background, the effect of rneΔ18 is comparable to rhlB, thus suggesting that disrupting the interaction between RhlB and RNase E is equivalent to disrupting RhlB. The difference in the levels of the 210 nt intermediate detected in the pcnB+ background (lanes 1 versus 2) suggests that RhlB, in the absence of an interaction with RNase E, has partial activity when PAP is present. Alternatively, enolase could have a role in the degradation of the REP-containing mRNA decay intermediates although we know of no other experimental support for this hypothesis.

Figure 6.

The role of the RNA degradosome in the degradation of the sucB REP stabilizer.
A. Cartoon of the RNA degradosome. RNase E contains a C-terminal non-catalytic region that serves as the scaffold for the interaction with other proteins. The rneΔ18 and rneΔ10 alleles encode, respectively, RNase E mutants in which the region binding RhlB and enolase, or PNPase have been deleted (Vanzo et al., 1998; Leroy et al., 2002). Previous work has demonstrated the disruption of the protein–protein interactions by a variety of experiments including immunoprecipitation. The full-length RNase E polypeptide has 1061 amino acids. The RneΔ18 and RneΔ10 polypeptides contain, respectively, deletion of residues 728–845 and 844–1045.
B, C and D. Northern blots of RNA separated by polyacrylamide gel electrophoresis, hybridized with a probe against the sucB REP-stabilizer. The RNA was isolated from strains with the rneΔ18 and rneΔ10 mutations (B and C, respectively), and their isogenic wild type controls (D). In each panel, RNA was isolated from a strain containing the rne mutants or wild type (lane 1), or the rne mutants or wild type combined with rhlB, pcnB or rhlBpcnB (lanes 2–4, respectively).

Figure 6C shows an experiment comparable to that in Fig. 6B, but in the rneΔ10 background. Very little 210 nt intermediate is detected with the rneΔ10 mutation. The rneΔ10 rhlB double mutant has low levels of the 210 nt species whereas its abundance with rneΔ10 pcnB double mutant (lane 3) is essentially equivalent to the rhlBpcnB double mutant (Fig. 6D, lane 4) and there is no additional effect with the rneΔ10 pcnBrhlB triple mutant (Fig. 6C, lane 4). Thus, disrupting the interaction between PNPase and RNase E affects the RhlB-PNPase pathway in the pcnB background whereas there is little if any effect in the pcnB+ background. Results essentially comparable to those shown in Fig. 6B and C have been obtained in an analysis of the malE REP-stabilizer (data not shown). Altogether, these results show that in the pcnB background, disrupting the interaction between RNase E and RhlB or PNPase is comparable to inactivating RhlB or PNPase. In the pcnB+ background, these effects are masked suggesting that PAP can bypass the requirement for RhlB and PNPase as a component of the RNA degradosome. This is consistent with the results in the previous section suggesting that a poly(A)-dependent ribonuclease other than PNPase, which does not require RhlB activity, participates in the degradation of the REP-containing mRNA.

The degradation of an mRNA fragment from the end of the rpsT message

Nearly complete stabilization of an mRNA fragment from the 3′ end of the rpsT mRNA, in a pcnB strain, was observed previously (Coburn and Mackie, 1998). We wanted to test if the rhlB mutation affected the degradation of this fragment. Figure 7A shows the structure of the rpsT decay intermediate. Its 5′ end is formed by an RNase E cleavage; its 3′ end results by rho-independent transcription termination. Note the short 3′ oligo(U) stretch and the GC-rich stem–loop. In addition, GCAA forms a ‘tetra loop’ that further increases the stability of this RNA structure. This stable stem–loop is critical for the resistance of the rpsT message to exonucleolytic degradation. Figure 7B shows a Northern blot of a polyacrylamide gel hybridized with a probe against the entire rpsT mRNA. The decay intermediate that migrates at 143 nt (lanes 4 and 6) is the stable mRNA fragment that was identified in the pcnB background previously. The slightly larger 147 nt intermediate in the pcnB+ background corresponds to a species that was identified in a pnp mutant (Mackie, 1989). Because the previous work suggests that the 5′ ends of the 143 and 147 nt species are identical, we believe that the small difference in size detected here is due a short oligo(A) tail in the pcnB+ background. That the 143 and 147 nt species in Fig. 7B are derived from the 3′ end of the rpsT mRNA was confirmed using a 3′ end-specific probe (data not shown). The 110/104 doublet (lanes 4 and 6) likely corresponds to minor cleavages by RNase E. The abundance of the 143 nt intermediate in the pcnB strain shows that polyadenylation has a major role in its degradation. However, unlike the degradation of the REP-containing mRNA, the abundance of the rpsT 3′ fragment in the pcnBrhlB double mutant is equivalent to the pcnB single mutant (lane 4). Thus, RhlB contributes very little to the decay of this intermediate. The low level of the mRNA decay intermediate in the rhlB single mutant supports this conclusion. In the pnp single mutant and pnprhlB double mutant, the accumulation of the intermediate is weak in the pcnB+ background (lanes 3 and 5). Furthermore, the rneD10 and rneD18 mutations have little effect on the accumulation of the mRNA decay intermediate (data not shown). It should be noted that the rhlB, pnp and pcnB alleles employed here are complete disruptions. Thus, consistent with the results in the previous two sections, a ribonuclease other than PNPase must participate in the degradation of the polyadenylated mRNA decay intermediate. It seems unlikely that this degradation is due to RNase II as previous in vivo work showed only a small increase in the 147 nt intermediate when a pnprnb double mutant was compared to a pnp single mutant (Mackie, 1989). Furthermore, in vitro experiments have shown that the intermediate is resistant to degradation by RNase II even with polyadenylation (Coburn and Mackie, 1996; 1998). Our results suggest that RhlB and PNPase have a minor role in the degradation of the rpsT decay intermediate in the pcnB+ background. The apparent complete stabilization of the 143 nt intermediate in the pcnB background shows that there is a strict requirement for polyadenylation in its degradation.

Figure 7.

The role of RhlB and PNPase in the degradation of an mRNA fragment from the 3′ end of the rpsT message.
A. Structure of the 3′ mRNA decay fragment from the rpsT message (Coburn and Mackie, 1998). The 5′-monophosphate end is formed by RNase E cleavage; the 3′ end by rho-independent transcription termination.
B. Northern blot of RNA separated by polyacrylamide gel electrophoresis, hybridized with a probe against the rpsT gene. Lanes 1–6, respectively: RNA isolated from wild type, rhlB, pnp, pcnB, pnprhlB, and pcnBrhlB. The markers (left) are as described (Fig. 3). To the right of the panel, P1 and P2 are primary transcripts that arise from different promoters, but terminate at the same site. The 143 and 147 nt species are 3′ mRNA fragments that arise by RNase E cleavage. The small difference in size is likely due to a short oligo(A) extension in the pcnB+ background. The 110 and 104 nt fragments probably arise by RNase E cleavage at secondary sites.

Discussion

RhlB and poly(A) polymerase are required for degrading REP-stabilizers

REP-stabilizers are known to be a critical element in E. coli mRNA stability. The study of the malEFG operon has contributed significantly to our understanding of these mRNA stability elements. In vivo, disruption of the malE REP-stabilizer destabilizes the message and the malE gene product is underexpressed (Newbury et al., 1987a). In vitro, the malE REP-stabilizer decreases the processivity of RNase II and PNPase suggesting that mRNA protection is a result of the inhibition of 3′→5′ degradation (McLaren et al., 1991; Py et al., 1996). However, subsequent in vitro work showed that two enzymes, RhlB and PAP, can overcome this protection.

RhlB was identified by sequence similarity to other DEAD-box RNA helicases (Kalman et al., 1991; Schmid and Linder, 1992). The biochemical characterization of RNase E showed that RhlB was part of a complex, now known as the RNA degradosome, that also contains RNase E, enolase and PNPase (Carpousis et al., 1994; Py et al., 1994; 1996; Miczak et al., 1996). Previous in vitro work has shown that RhlB in the RNA degradosome can facilitate the 3′→5′ exonucleolytic degradation of the malE REP-stabilizer (Py et al., 1996; Coburn et al., 1999). The involvement of PAP in the degradation of this REP-stabilizer was also suggested by in vitro experiments (Blum et al., 1999). RNA substrates were synthesized in which the 3′ end corresponds to the GGCA sequence at the base of the large double-stranded structure formed by the Y(1) and Z2 REPs (Fig. 1B). This substrate, which contains a double-stranded 3′ end, is resistant to attack by free PNPase or PNPase in the RNA degradosome, even with ATP, suggesting that RhlB cannot unwind 3′ double-stranded ends. Nevertheless, the addition of a short oligo(A) extension is sufficient to overcome this barrier.

Altogether the in vitro work with RhlB and PAP raises a paradox: how can the REP-stabilizer at the end of the malE message act as a barrier against 3′→5′ exonucleolytic degradation if RhlB and PAP can destabilize this structure? Indeed, over the past few years, questions of this sort have arisen repeatedly in efforts to better understand the control of exoribonucleolytic activity in E. coli mRNA degradation. Our results suggest an explanation. The decay of REP-stabilized mRNA appears to follow an orderly pathway in which an endonucleolytic cleavage by RNase E is required for exonucleolytic degradation facilitated by RhlB and PAP. Thus, a REP-stabilizer that is an effective barrier in an intact mRNA, can be efficiently degraded once the message has been attacked by RNase E.

Processing and degradation of the malEFG message

Our results show that RNase E is required for the maturation of the malE message (Fig. 2). However, it seems unlikely that the 3′ end is formed directly by endonucleolytic cleavage as there is no evidence that RNase E can cleave within the malE-malF intercistronic region. The 3′ end of the mature malE message is probably produced by RNase E cleavage downstream of malE and exonucleolytic trimming to the REP-stabilizer. The accumulation of small mRNA decay intermediates containing the REP-stabilizer in the rhlBpcnB strain shows that the degradation of the malE message is initiated by endonucleolytic cleavage. This conclusion is reinforced by the fact that the mature malE message is not stabilized in the rhlBpcnB strain. The simplest interpretation of the stabilization of the full-length malEFG mRNA in the RNase E mutant strain is that RNase E is involved in both the maturation and degradation of the malE message. Altogether, these considerations suggest that the stability of the mature malE message is a result of efficient protection by the REP-stabilizer and relatively slow attack by RNase E. Thus, the REP-stabilizer alone does not assure stability.

Our work complements other results suggesting that poly(A)-dependent mRNA degradation is triggered by RNase E cleavage (Cohen, 1995; Xu and Cohen, 1995; Coburn and Mackie, 1998; Hajnsdorf and Regnier, 1999; Dreyfus and Regnier, 2002). Indeed, recent in vitro results with RNA I suggest that it could be PAP itself that discriminates between the full-length message and mRNA decay fragments produced by RNase E (Feng and Cohen, 2000) although RNase E could help to target PAP as a physical interaction has been detected between these enzymes in vitro (Raynal and Carpousis, 1999). It is noteworthy that in the pcnB background, the pathway by which RhlB facilitates the degradation of REP-stabilizers also appears be activated by endonuclease cleavage. Thus, RhlB, like PAP, appears to distinguish between the full-length message and decay intermediates.

The role of the RNA degradosome in degrading mRNA decay intermediates

In the pcnB background, the mutants that disrupt RhlB or PNPase, abolished degradation of mRNA fragments containing the REP-stabilizers, suggesting that, in the absence of poly(A) polymerase, only the RhlB-PNPase pathway is capable of degrading the REPs. Furthermore, the results in Fig. 6 show that the RNA degradosome is required for the activity of RhlB and PNPase in the degradation of REP-stabilizers, in the pcnB background. Mutations in RNase E that disrupt the interaction with either RhlB or PNPase interfere with degradation. This effect is equivalent to the disruption of the genes encoding RhlB (Fig. 4) and PNPase (Fig. 5). Thus, in the pcnB background, the free enzymes are unable to facilitate the degradation of REP-stabilizers. This is the first in vivo evidence that RhlB and PNPase act together as part of the degradosome. Previous in vitro work suggested that an intact RNA degradosome would be necessary for RhlB activity. Its association with RNase E strongly activates ATP hydrolysis and RNA unwinding (Vanzo et al., 1998). Furthermore, the reconstitution of a mini-degradosome from RNase E, RhlB and PNPase showed that a complex containing all three components is required to degrade the malE REP-stabilizer (Coburn et al., 1999). Nevertheless, recent work has demonstrated a weak direct interaction between RhlB and PNPase (Liou et al., 2002) and our results suggest that free RhlB might have partial activity when PAP is present (Fig. 6B). Altogether, these results show that the physical interactions between RhlB, PNPase and RNase E are important for controlling the activity of the exonucleolytic pathway in the degradation of the REP-stabilizers.

The eukaryotic exosome is a 3′→5′ mRNA degrading complex with a multisubunit catalytic core that is related to bacterial PNPase (Aloy et al., 2002; Raijmakers et al., 2002; Symmons et al., 2002). Interestingly, the exosome has associated proteins that are putative RNA helicases. Whether these proteins have a role similar to that of RhlB is not known.

The role of RhlB and poly(A) polymerase in controlling exonucleolytic activity

Our in vivo results with REP-stabilized mRNA have revealed a major role for PAP in facilitating degradation although it is noteworthy that some degradation via RhlB occurs in the pcnB mutant. This is particularly evident in the degradation of a 210 nt mRNA fragment containing the sucB REP (Fig. 4B). Previous work has suggested that the 3′ end of the mature malE message is heterogeneous with single-stranded 3′ extension ranging from three to nine nucleotides (McLaren et al., 1991). This heterogeneity arises because exonucleolytic trimming by RNase II and PNPase does not end at a precise point. Our in vivo results suggest that a proportion of the REP-stabilized mRNAs, perhaps those with the longest 3′ single-stranded extensions, can be degraded without polyadenylation. Another possibility is that spontaneous unwinding of the 3′ end of the REP-stabilizer is sufficient to permit some exonucleolytic degradation. Although our results suggest that RhlB and PAP can act independently of each other, we believe it is likely that, in wild type, these enzymes work together.

It is interesting to compare the degradation of the REP-stabilizers to the degradation of the 3′ decay fragment of the rpsT message. Whereas the pcnB mutation has only a partial effect on the degradation of the REP-containing mRNA fragments (Figs 3 and 4), it completely blocks the degradation of the 3′rpsT fragment (Fig. 7). The strict requirement of the rpsT fragment for polyadenylation is probably due to complete protection by the 3′ end by the rho-independent termination structure (Fig. 7). Besides being too short to bind to the exonucleases, in vitro experiments have shown that oligo(U) tails are poor substrates for initiating an attack by the RNA degradosome (Blum et al., 1999). In contrast to the REP-stabilized mRNA fragments, the effect of the rhlB mutation is negligible on the 3′rpsT fragment suggesting that RNA helicase activity is not necessary to degrade the short stem–loops formed by rho-dependent termination. A poly(A)-dependent pathway that does not require RhlB has been elucidated (Coburn and Mackie, 1998; 1999). In the presence of purified PNPase, PAP and ATP, continuous polyadenylation permits multiple attacks by PNPase in a reaction that ultimately degrades the 3′ stem–loop of the rpsT mRNA. The same reaction could explain the partial degradation of REP-stabilizers that we observe in the absence of RhlB.

Our work differs in an important way from previous studies with RNA I, rpsT and rpsO (Xu and Cohen, 1995; Coburn and Mackie, 1998; Hajnsdorf and Regnier, 1999). It has been suggested that either free PNPase or PNPase in the RNA degradosome is the principal ribonuclease involved in poly(A)-dependent degradation. Our results show that disrupting PNPase leads to the accumulation of REP and rpsT mRNA decay intermediates (Figs 5 and 7). However, based on the level of accumulation of mRNA decay intermediates, the effect of disrupting PAP is much stronger than disrupting PNPase. Thus another ribonuclease, which is unlikely to be RNase II, appears to have an important role in poly(A)-dependent degradation. This conclusion is reinforced by our results with the RNase E mutants (Fig. 6) showing that the requirement for the RNA degradosome is masked in the pcnB+ background. A poly(A)-dependent degradation activity has been observed in vivo under conditions where RNase E, RNase II and PNPase were inactivated, in what appears to be a secondary pathway in the degradation of the full-length rpsO message (Hajnsdorf et al., 1995). Possible candidates for this poly(A)-dependent degradation activity include RNase PH, an enzyme related to PNPase, and RNase R, a paralogue of RNase II (Kelly and Deutscher, 1992; Deutscher and Li, 2001; Cheng and Deutscher, 2002). An important objective of future work will be to identify this enzyme, as it appears to be capable of degrading structured RNA with an activity nearly comparable to that of RhlB-PNPase in the RNA degradosome.

Experimental procedures

Strain construction

The E. coli strains used in this study are described in Table 2. SVK1 is a derivative of MC1061 in which the rhlB gene was replaced by a mini rhlB gene encoding the first eight amino acids of the gene, an eight amino acids tag derived from the gene32 of bacteriophage T4, and the last eight amino acids of rhlB. The replacement of rhlB was made using pVK19, which is a derivative of pLN135 (Cornet et al., 1996). pVK19 was constructed in two steps. The chromosomal segments flanking rhlB were amplified by PCR using ORhlB1/ORhlB2 and ORhlB3/ORhlB4 respectively (Table 3). The oligonucleotides ORhlB2 and ORhlB3 contain overlapping sequences that encode the gene32 tag; ORhlB1 and ORhlB4 contain BamHI restriction sites. The initial PCR products were mixed and amplified using the outer oligonucleotides. The cross-over PCR product was purified, digested with BamHI and ligated to pLN135 digested by BglII and treated with phosphatase. The rhlB deletion was introduced into MC1061 as described (Cornet et al., 1996). Candidates were screened for the presence of the gene 32 sequence and then examined by Southern and Western blot analysis.

Table 2. . Strains used in this work. a
StrainGenotypeReference
  • a

    . The background throughout is MC1061.

  • b

    . SVK39, SVK37 and SVK72, which are isogenic to the previously described AC23, AC26 and AC24 strains (Carpousis et al., 1994; Leroy et al., 2002), were reconstructed here as controls.

MC1061 araB139 Δ(ara-leu)7696 galE15 galK16 ΔlacX74 rpsL(StrR) hsd2(rK mK) mcrA mcrB1New England Biolabs catalogue (1994); Casadaban and Cohen (1980)
SVK1 ΔrhlB this work
SVK39 zce-726::Tn10 rne1this workb
SVK29 pnp::Tn5this work
SVK30 ΔrhlB pnp::Tn5this work
SVK45 Δpcnb::Tn10KanRthis work
SVK46 ΔrhlB Δpcnb::Tn10KanRthis work
SVK27 Δrnb::Tn10this work
SVK28 Δrnb::Tn10 ΔrhlBthis work
SVK57 Δrnb::Tn10 Δpcnb::Tn10KanRthis work
SVK58 Δrnb::Tn10 ΔrhlB Δpcnb::Tn10KanRthis work
SVK37 zce-726::Tn10 rneΔ18this workb
SVK38 zce-726::Tn10 rneΔ18 ΔrhlBthis work
SVK65 zce-726::Tn10 rneΔ18 ΔpcnB::Tn10this work
SVK66 zce-726::Tn10 rneΔ18 ΔrhlB Δpcnb::Tn10KanRthis work
SVK72 zce-726::Tn10 rneΔ10this workb
SVK73 zce-726::Tn10 rneΔ10 ΔrhlBthis work
SVK76 zce-726::Tn10 rneΔ10 Δpcnb::Tn10KanRthis work
SVK77 zce-726::Tn10 rneΔ10 Δpcnb::Tn10KanRΔrhlBthis work
SVK78 zhe-6::Tn10this work
SVK79 zhe-6::Tn10 pnp::Tn5this work
SVK80 zhe-6::Tn10 ΔrhlBthis work
SVK81 zhe-6::Tn10 pnp::Tn5 ΔrhlBthis work
SVK82 zhe-6::Tn10 Δpcnb::Tn10KanRthis work
SVK83 zhe-6::Tn10 pnp::Tn5 Δpcnb::Tn10KanRthis work
SVK84 zhe-6::Tn10 ΔrhlB, Δpcnb::Tn10KanRthis work
SVK85 zhe-6::Tn10 pnp::Tn5 ΔrhlB, Δpcnb::Tn10KanRthis work
Table 3. . DNA oligonucleotides used in this work.
NameSequence 5′→3′Target
  • a

    . The first 17 nucleotides, in italic, correspond to the bacteriophage T7 promoter although we have not used in vitro transcription in the work reported here.

  • b

    . The primers were chosen to amplify the entire intergenic region. Bold letters show the stop or start codon of the upstream and downstream genes.

  • c

    . The sequence in italic is the BamHI restriction site.

  • d

    . The first 24 nucleotides, in italic, correspond to the tag derived from gene32 of bacteriophage T4.

rpsT 1ttggctaatatcaaatcagctaagaag rpsT coding sequence
rpsT T7a gtaatacctcactataggagccagtttgttgatctgtgcagtc 
malE 1gcacgcatcctcgcattatcc malE coding sequence (with malEF T7)
malEF 1gacgcgcagactcgtatcaccaag malE-malF b
malEF T7a gtaatactcactataggacatccatggggttcttcctcat 
sucB-sucC 1ggacgtgtagtagtttaagtttcacc sucB-sucC
sucB-sucC 2ttcatgtaagttcatgtgttctgtcc 
nrdA-nrdB 1gtaagatctgatattgagatgccgg nrdA-nrdB
nrdA-nrdB 2gtggtatatgccatgagtgtgtcc 
sdhB-sucA 1cgtaaaccgtaggcctgataagacg sdhB-sucA
sdhB-sucA 2gttctgcatcgtgatcccttaagc 
lamB-malM 1atctggtggtaatagcaaaacctgg lamB-malM
lamB-malM 2cgatgagacttttattcattttcattgt 
ppc-argE 1ggaatattgttcgttcatattacccc ppc-argE
ppc-argE 2cattaaaacgtaggccggataaggc 
cca-bacA 1gcccaaagcctgaatgagtattgg cca-bacA
cca-bacA 2tgtcgtgttcttttaattgaggaatag 
c ORhlB 1cgcggatcccttacgccaggttagcgtcgagregion upstream of rhlB
d ORhlB 2 agccatttgtgcagcgagttcagcttctgttaaatgtgttttgctcatagtg 
d ORhlB 3 gctgaactcgctgcacaaatggctccgcgtaatcgtaatcgtcgtcgttcaggtregion downstream of rhlB
c ORhlB 4cgcggatcctcttcaccgctgatcacctgt 

Other strains were constructed by serial P1 transduction using the following alleles: Δpcnb::Tn10KanR from strain SK7988 (Liu and Parkinson, 1989; O’Hara et al., 1995); pnp::Tn5 from strain LM160 (McMurry and Levy, 1987); Δrnb::Tn10 from strain CMA200 (Zilhao et al., 1996); rneΔ18 and rneΔ10 from strains ENS134-18 and ENS134-10 respectively (Leroy et al., 2002). The Δpcnb::Tn10KanR, pnp::Tn5 and Δrnb::Tn10 alleles were shown to abolish activity when they were isolated. In the strains constructed here, the complete absence of PAP, PNPase or RNase II was confirmed by Western blotting (data not shown). In Fig. 5A the strains used are SVK79, SVK81, SVK83 and SVK84. SVK79 and SVK81 differ from SVK29 and SVK30 only by the insertion of the zhe-6::Tn10 from the CAG12153 strain (Singer et al., 1989). The insertion of this Tn10 into SVK29 permitted the transduction of pnp::Tn5 allele using tetracycline selection instead of kanamycin, thus allowing the construction of the double and triple mutant strains SVK83 and SVK85 respectively. The isogenic wild-type controls in Fig. 5B correspond to strains SVK78, SVK80, SVK82 and SVK84.

Northern blot

Standard techniques were as described (Miller, 1972; Sambrook and Russell, 2001). For malEFG and rpsT, cells were cultivated in MOPS growth medium (Neidhardt et al., 1974) supplemented with maltose (0.4%) and casamino acids (0.2%). For other mRNAs, MOPS was supplemented with glycerol (0.4%), amino acids (50 µg ml−1 V, A, L, I, H, M, S, P, G, N, D, C, E, K) and thiamine (50 µg ml−1). Overnight cultures were diluted into fresh medium and grown at 30°C to an OD600 = 0.3. One volume of culture was mixed with 1/10 volume of 10× lysis buffer (0.5 M Tris-HCl, pH 6.8, 50 mM EDTA, 10% SDS), immediately frozen in liquid nitrogen, then stored at − 20°C until extraction. RNA was prepared by heating at 95°C for 3 min, phenol extraction and ethanol precipitation. RNA pellets were dissolved in water treated with DEPC and concentrations were determined by UV absorption at 260 nm.

For the analysis of mRNA decay intermediates, 5–10 µg total RNA was separated on denaturing 6% polyacrylamide gels (29 : 1, 0.5 × TBE, 8 M urea) and electrotransferred to Hybond N (Amersham Pharmacia) in 0.5× TBE. For full-length mRNA, RNA (10 µg) was separated on 1.5% agarose gels (1 × MOPS electrophoresis buffer) containing formaldehyde (2.2 M) and transferred by capillary blotting. RNA was cross-linked to the membrane by UV irradiation. Hybridization and washing were as recommended (Amersham Pharmacia). Radioactive probes were prepared by either direct PCR labelling with α-[33P]–ATP or, after purification of an unlabelled PCR product, by random primer labelling using a Megaprime Kit (Amersham Pharmacia). The PCR primers are described in Table 3.

Acknowledgements

We thank C. Arraiano, I. Iost and P. Régnier for providing strains; L. Poljak for help with Northern blotting; J.-Y. Bouet, L. Poljak and B. Py for suggestions to improve the manuscript. This research was supported by the CNRS (UMR 5100) with aid from the technical platform of the IEFG (IFR 109). Additional funding was from the Cancer Research Association (ARC) and the Fundamental Microbiology Program of the Ministry of Education (MENRT). V.K. is a predoctoral student with support from the MENRT and the ARC.

Ancillary

Advertisement