Evidence in vivo that the DEAD-box RNA helicase RhlB facilitates the degradation of ribosome-free mRNA by RNase E

Abstract

The RNA degradosome of Escherichia coli is a ribonucleolytic multienzyme complex containing RNase E, polynucleotide phosphorylase, RhlB, and enolase. Previous in vitro and in vivo work has shown that RhlB facilitates the exonucleolytic degradation of structured mRNA decay intermediates by polynucleotide phosphorylase in an ATPase-dependent reaction. Here, we show that deleting the gene encoding RhlB stabilizes a lacZ mRNA transcribed by bacteriophage T7 RNA polymerase. Deleting the gene encoding enolase has little if any effect. Other messages transcribed by T7 polymerase are also stabilized by ΔrhlB. The effect of point mutations inactivating RhlB is comparable with the effect of deleting the gene. Primer extension analysis of the lacZ message indicates that RhlB facilitates endoribonucleolytic cleavage by RNase E, demonstrating a functional interaction between the RNA helicase and the endoribonuclease. The possible physiological role of an RhlB-RNase E pathway and the mechanisms by which RhlB could facilitate RNase E cleavage are discussed.

Instability is an intrinsic property of mRNA, permitting rapid response to changes in transcription and, in some situations, posttranscriptional control of gene expression. In Escherichia coli, the overall pathway for mRNA degradation is well established although much remains to be learned about controlling the stability of specific transcripts (1, 2). Messenger RNA is initially inactivated and fragmented by endoribonuclease cleavage. The fragments are then digested by exoribonucleases in a tightly coupled process. Intermediates in mRNA turnover are rapidly degraded and are thus generally difficult to detect except in mutant strains where a gene encoding a component of the degradation machinery has been disrupted. The RNA degradosome contains RNase E, polynucleotide phosphorylase (PNPase), RhlB, and enolase (3–5). RNase E is the principal endoribonuclease in mRNA degradation. It is a large multidomain protein with an N-terminal catalytic region and a C-terminal noncatalytic region containing RNA binding sites and the protein scaffold for degradosome assembly (6, 7). PNPase is one of the enzymes that degrade mRNA fragments produced by RNase E. The function of enolase, a glycolytic enzyme, in the RNA degradosome is not established although recent work suggests a role in the posttranscriptional regulation of the major glucose transporter, encoded by ptsG, involving a small regulatory RNA (8, 9). RhlB is a DEAD-box RNA helicase (10) that facilitates the degradation of structured mRNA decay intermediates by PNPase (5, 11, 12). This activity, requiring ATP hydrolysis, is believed to involve the local unwinding of structures that block PNPase (5, 13). Ribonucleolytic multienzyme complexes in other organisms have also been shown to interact with DEAD-box or related proteins (13, 14).

The DEAD-box proteins are a family of putative RNA helicases that generally have RNA-dependent ATPase activity in vitro. They contain a characteristic set of conserved sequence motifs that were discovered in a comparison between the eukaryotic translation initiation factor eIF4A and the E. coli protein SrmB (15). eIF4A is involved in loading the eukaryotic small ribosomal subunit onto mRNA (16). SrmB is a chaperone in the assembly of the large ribosomal subunit of E. coli (17). The DEAD-box proteins are now known to be part of the ubiquitous DExD/H-box family of helicases that participate in many RNA unwinding and remodeling reactions (18). The DExD/H-box proteins together with other RNA and DNA unwinding enzymes constitute a super family containing a structurally conserved ATPase domain with RecA-like architecture (19). This domain is likely to be a generic motor for the unwinding and remodeling of nucleic acids. RhlB and eIF4a, which are among the smallest members of the DExD/H-box family, contain little more than this conserved catalytic core. The DExD/H-box domain often associates with additional domains, either covalently in the same polypeptide or noncovalently within a multienzyme complex. RhlB and eIF4A are DExD/H-box helicases that require a protein partner for biological activity (6, 20).

Here, we have studied in vivo the role of RhlB and enolase in the degradation of a lacZ mRNA transcribed by bacteriophage T7 RNA polymerase (T7-lacZ mRNA). Because the phage polymerase is 8-fold faster than its E. coli counterpart, transcription outpaces translation, producing long stretches of ribosome-free mRNA. This message is exceptionally sensitive to inactivation and degradation by RNase E (21). The integrity of the noncatalytic region of RNase E, including the sites that bind RhlB and enolase, is necessary for this sensitivity (22, 23). Deleting the enolase gene has little if any effect on the T7-lacZ mRNA. Deleting the RhlB gene stabilizes the T7-lacZ mRNA, leading to a significant increase in β-galactosidase synthesis. We present evidence that RhlB is part of a specialized pathway involved in the degradation of ribosome-free mRNA by RNase E.

Materials and Methods

The construction of ENS134-2, containing the rne131 mutation, and ENS134-3, containing the pnp::Tn5 mutation, has been described (22). SVK16, constructed here as a control, is isogenic to ENS134-3. SVK4 and SVK5 are derivatives of ENS134 and ENS134-2, respectively, in which the rhlB gene was disrupted (ΔrhlB) essentially as described (12). SVK62 and SVK63, containing the rhlB^DKAD and rhlB^AAA alleles, respectively, were constructed by the same replacement strategy. SVK17 is a derivative of SVK4 containing the pnp::Tn5 mutation. TAZ2 is a derivative of ENS134 in which the enolase gene was disrupted by using the same strategy as for rhlB. The strains used in β-galactosidase assays with pEZ201 and pEZ206 (24) are derivatives of MC1061 that have been described (12). Liquid cultures were grown at 30°C in Mops medium (25) containing 0.4% glycerol, 0.2% casamino acids, and 0.1 mM isopropyl β-d-thiogalactoside (IPTG). For krox20 and engrailed expression, cells were grown to mid-log phase in LB medium containing 50 μg/ml ampicillin, induced with 1 mM IPTG and then incubated for 2 h. Preparation of protein extracts, immunoprecipitation, and Western blotting were as described (6, 26). RNA extractions and Northern blotting were as described (12). Northern blots were hybridized by using the HincII-KpnI fragment of pT7lacZ-arg5 (27). For primer extension, the cells were cultured at 30°C without IPTG, then shifted to 42°C for 40 min. IPTG (0.1 mM) was added 10 min after the shift. RNA (4 μg) was mixed with ³³P-5′ end-labeled attcgcgtctggccttcctgt, dried, and suspended in 4 μl of annealing buffer (20 mM Tris·HCl, pH 7.5/80 mM KCl), heated to 85°C for 5 min, quenched on ice, and then incubated at 30°C for 30 min. Four microliters of RT buffer (75 mM Tris·HCl, pH 8.3/20 mM MgCl₂/4 mM DTT/0.8 mM each dNTP), containing 4 units of AMV reverse transcriptase (Promega), was added. Transcription was for 30 min at 42°C. A lacZ sequence was generated by using pT7lacZ-arg5, the same primer, and a Thermosequenase kit (Amersham Pharmacia).

Results

Deletion of the rhlB Gene Stabilizes the T7-lacZ mRNA. The ENS134 strain contains a chromosomally encoded lacZ gene that is transcribed by a chromosomally encoded bacteriophage T7 RNA polymerase. This system is under the control of the lac repressor and requires IPTG for induction. The level of synthesis of β-galactosidase in this strain can be used as an indicator of the stability of the lacZ message (21, 22). We will refer to this message as the T7-lacZ mRNA although it differs from authentic lacZ mRNA only due to its synthesis by T7 RNA polymerase. When the rhlB gene was deleted, 4.2-fold more β-galactosidase was produced (Table 1). To further examine this effect, we analyzed the amount of T7-lacZ mRNA by Northern blotting using a 5′-end probe (Fig. 1a). On this blot, the same quantities of RNA were loaded, and the quality of the transfer was checked with a probe against 5S rRNA (data not shown). The 4.3-kb mRNA is the full-length transcript containing lacZ, lacY′ (a fragment of lacY), and a tRNA (Fig. 1c). RNase E cleaves the 4.3-kb mRNA between lacZ and lacY′ to produce the 3.2-kb monocistronic lacZ transcript. The prominent small RNA in the lower half of the blot (Fig. 1a, asterisk) corresponds to the 5′-end fragments of the lacZ mRNA (27). They were not detected with an internal probe (data not shown). Lanes 1 and 2 in Fig. 1a show that the signal depends on IPTG. Comparison of lanes 3 and 4 shows that there is more T7-lacZ mRNA in the ΔrhlB strain. The effect on mRNA levels could be due to either increased transcription or increased mRNA stability. Because the tRNA processed from the T7-lacZ transcript is stable, it can be used as a reporter for transcription (21, 22, 27). The filter in Fig. 1a was stripped and rehybridized with a probe against the tRNA (Fig. 1b). Quantification showed a small, 1.3-fold increase in the tRNA in the ΔrhlB strain. Quantification of the 3.2-kb monocistronic mRNA and normalization to the tRNA level showed a 3.6-fold increase. Although the signal is too close to the background to accurately quantify, the 4.3-kb transcript also seems to increase in the ΔrhlB strain based on visual inspection of the image. Thus, the principal effect of deleting rhlB is the stabilization of the T7-lacZ mRNA.

Table 1. β-Galactosidase activity.

Strain	Genotype	Activity (±SD, %)	Relative activity
ENS134	Wild type	1,670 (±9.4)	1.0
SVK4	ΔrhlB	7,090 (±9.4)	4.2
SVK16	pnp-	5,430 (±7.1)	3.3
SVK17	ΔrhlB pnp-	11,700 (±19)	7.0
ENS134-2	rne131	19,400 (±12)	11.6
SVK62	rhlBDKAD	6,780 (±6.6)	4.1
SVK63	rhlBAAA	6,710 (±9.6)	4.0

The results are the average of at least three independent measurements. Activities were measured as described (23). The SD as a percentage is in parentheses. Relative activity for each strain is normalized to wild type.

Fig. 1.

Northern blot. (a) Total RNA from ENS134 (wild type) and SVK4 (ΔrhlB) was separated on a 1.5% agarose formaldehyde gel, transferred, and hybridized with a 5′ lacZ probe. In lanes 1 and 2, T7-lacZ was not induced. In lanes 3 and 4, the strains were cultivated continuously in the presence of IPTG. The position of the 4.3-kb primary transcript, the 3.2-kb monocistronic lacZ transcript, and a 5′ end fragment (asterisk) is indicated to the right of the panel. (b) The filter was stripped and hybridized with a probe against the arg5 tRNA. (c) Diagram of the 4.3-kb T7-lacZ transcript indicating the intercistronic RNase E processing site involved in the formation of the 3.2-kb monocistronic lacZ transcript (not drawn to scale).

Deletion of the eno Gene Does Not Stabilize the T7-lacZ mRNA. Enolase, like RhlB, is a major component of the RNA degradosome, and the purified degradosome has enolase activity (4, 5). To test its role in expression of the T7-lacZ mRNA, the eno gene was disrupted. The media used for the measurements in Table 2 contained glycerol (0.4%) and lactate (0.2%) because the Δeno strain requires both for growth (28). As a control, the ΔrhlB strain grown under the same conditions has 4.1-fold more β-galactosidase activity, which is comparable with the result in Table 1. There is a small but significant reduction of β-galactosidase in the Δeno strain (84% of wild type). This difference could be due to the disruption of glycolysis in the Δeno strain, which grows slower than wild type. From this experiment, we conclude that Δeno does not stabilize the T7-lacZ mRNA.

Table 2. β-Galactosidase activity.

Strain	Genotype	Activity (±SD, %)	Relative activity
ENS134	Wild type	2,242 (±2.3)	1.0
TAZ2	Δeno	1,877 (±7.1)	0.84
SVK4	ΔrhlB	9,238 (±2.6)	4.1

The results are the average of three independent measurements. Activities were measured as described (23). The SD as a percentage is in parentheses. Relative activity is normalized to wild type.

RhlB Acts Independently of PNPase in the Degradation of the T7-lacZ mRNA. Previous in vitro and in vivo work has shown that RhlB facilitates the exonucleolytic degradation of structured mRNA decay fragments by PNPase (5, 11, 12). Although the major pathway for the turnover of the T7-lacZ mRNA is RNase E-dependent, a secondary PNPase-dependent pathway has been described (22). This pathway could involve exonucleolytic degradation of the 3.2-kb monocistronic transcript (Fig. 1c). The overproduction of β-galactosidase in the pnp^– strain is 3.3-fold, under the conditions used here (Table 1). When pnp^– was combined with ΔrhlB, the β-galactosidase level increased 7.0-fold. This result is strikingly different from previous in vivo work on the degradation of mRNA decay intermediates, in which the disruption of the pnp gene was epistatic to ΔrhlB (12). Thus, the effect of ΔrhlB on the turnover of the T7-lacZ mRNA is independent of PNPase, suggesting that RhlB has another function in mRNA degradation that has not previously been characterized. For comparison, the effect of the previously described rne131 mutation (22, 23), in which the entire noncatalytic region of RNase E is missing, is an 11.6-fold increase in β-galactosidase (Table 1). Taken together, these results suggest that there are at least three pathways involved in the inactivation of the T7-lacZ mRNA: a PNPase pathway that does not require RhlB, an RNase E pathway that does not require RhlB, and an RNase E pathway that requires RhlB.

Deletion of the rhlB Gene Impairs Cleavage of the T7-lacZ mRNA by RNase E. We examined RNase E cleavages in the 5′ end of the T7-lacZ mRNA by primer extension (Fig. 2). In this experiment, strains were grown at 30°C and then shifted to 42°C. T7-lacZ transcription was induced after the shift by the addition of IPTG. Lanes 2 and 3 in Fig. 2 show, respectively, the result with either the wild-type control or the rne1 temperature-sensitive strain. Lanes 1 (wild-type) and 4 (rne1) are background controls in which IPTG was not added. The 5′ ends marked by the asterisks were detected in wild type, but absent in the rne1 strain (i.e., no signal above the background control in lane 4), showing that their production is RNase E-dependent. The RNase E-independent 5′ ends (x) could correspond to an RNase III cleavage that was previously mapped in vivo (29). The same experiment was performed with the ΔrhlB strain, with or without IPTG (Fig. 2, lane 5 and 6, respectively). The RNase E-dependent 5′ ends are detectable in ΔrhlB (lane 5), but the signal is weaker than in wild type (lane 2) and close to the background control (lane 6). In contrast, the 5′ end of the primary transcript (+1) is more abundant in the ΔrhlB strain (lane 5) than in wild type (lane 2). These results show that RNase E cleavages in the 5′ end of the T7-lacZ mRNA are inhibited in the ΔrhlB strain. The decreased RNase E cleavage in the 5′ end of the T7-lacZ mRNA correlates with the increase in functional full-length T7-lacZ mRNA (Table 1 and Fig. 1), suggesting that these cleavages are involved in the inactivation of the T7-lacZ mRNA.

Fig. 2.

Analysis of the 5′ end of the T7-lacZ message by primer extension. Lanes 1, 4, and 6 show, respectively, wild type, rne1, and ΔrhlB without IPTG; lanes 2, 3, and 5 are the same but with IPTG. The lacZ DNA sequence, run in parallel, is shown to the left. Signals corresponding to mRNA fragments whose production is RNase E dependent or independent are marked by * or x, respectively. From bottom to top, the sequences of the RNase E cleavage sites are as follows: gaau/ccga (+398), acgg/agaa (+393), acgg/ucaa (+369), and gacc/uauc (+356). The fifth site could not be precisely located. The coordinates in parentheses indicate the 5′ end of the cleavage product (underlined). The numbering takes +1 as the 5′ nucleotide of the T7-lacZ message (5′-ggaauu...).

The Enzymatic Activity of RhlB Is Required for Its Role in T7-lacZ mRNA Degradation. If the absence of RhlB affects RNase E activity, this effect could be due to either allosteric control of RNase E activity by the binding of RhlB or a direct role for the enzymatic activity of RhlB. In the former case, the enzymatic activity of RhlB would not be necessary. In the conserved sequences of the DEAD-box proteins, the DEAD motif is involved in ATP hydrolysis and the SAT motif is involved in coupling ATPase and helicase activity (18). We constructed two mutants, RhlB^DEAD→DKAD and RhlB^SAT→AAA. These mutants were introduced into the ENS134 strain by replacing the wild-type rhlB allele. We have shown previously that the RhlB^DKAD is inactive in an in vitro ATPase assay (6). Far Western blots showed that the RhlB mutants bind to RNase E in vitro (data not shown). Their interaction in vivo was tested by immunoprecipitation with RNase E antibody. The degradosome components in the immunoprecipitates were analyzed by Western blotting (Fig. 3). Fig. 3, lane 1 is a control with preimmune serum. Lanes 2–5 show the results with wild type, ΔrhlB, rhlB^DKAD, and rhlB^AAA. The mutant RhlB proteins are present as components of the RNA degradosome (lanes 4 and 5). In all of these strains, we detected enolase and PNPase, showing that the absence of RhlB or the presence of the mutants did not interfere with the association of the other degradosomal proteins. The relative levels of β-galactosidase are 4.1- and 4.0-fold for rhlB^DKAD and rhlB^AAA, respectively (Table 1). Thus, the extent of overproduction of β-galactosidase with these mutants is the same as in the ΔrhlB strain. These results show that the effect on the T7-lacZ mRNA of point mutations that inactivate RhlB is the same as the deletion of the gene. We conclude that the enzymatic activity of RhlB is required for its role in T7-lacZ mRNA degradation.

Fig. 3.

Immunoprecipitation using antibodies raised against RNase E. Extracts from wild-type (lane 2), ΔrhlB (lane 3), rhlB^DKAD (lane 4), and rhlB^AAA (lane 5) strains were analyzed by Western blotting with the antibodies against the proteins noted to the right. Lane 1 is a control in which the wild-type extract was immunoprecipitated by using preimmune serum.

Deletion of the rhlB Gene Does Not Generally Inhibit RNase E Activity. The decreased RNase E cleavage of the T7-lacZ mRNA could be due to a global inhibition of RNase E activity in the ΔrhlB strain. We recently showed that small mRNA decay fragments accumulate in ΔrhlB and pcnB^– strains and that accumulation was the strongest in a ΔrhlB pcnB^– double mutant (12). The pcnB gene encodes poly(A) polymerase, which is involved in the exonucleolytic degradation of mRNA fragments. The accumulation of intermediates in mRNA decay could interfere with RNase E activity by competing with full-length mRNA. We tested this possibility by using an rne-lacZ fusion that is an indicator of RNase E activity in the cell. Briefly, the message that encodes RNase E is exceptionally sensitive to inactivation by RNase E due to elements in its 5′ UTR. This sensitivity permits the self-regulation of RNase E expression (30, 31). Previous work has shown that β-galactosidase expression from the ez1 gene, which is a fusion between the 5′ regulatory elements of the rne gene and the coding sequence of lacZ, makes lacZ expression sensitive to the level of RNase E activity in the cell (24). The ez6 fusion, in which the regulatory elements in the 5′ UTR have been disrupted, is a control for effects on transcription or plasmid copy number. In Table 3, β-galactosidase expression was measured in strains harboring ez1 or ez6. The differences between the β-galactosidase levels of wild type harboring the ez6 control and the mutant strains are small and not significant. Derepression was calculated as the ratio of β-galactosidase activity of ez1 to ez6 normalized to this ratio in wild type. The 2.6-fold derepression in the rne131 strain (AC27) agrees well with previous measurements by Western blotting showing that the mutant RNase E protein is overexpressed 2.8-fold (23). In the ΔrhlB and pcnB^– strains, there is no detectable derepression. In agreement with this result, we were unable to detect differences by Western blotting (data not shown). However, the 1.7-fold derepression in the ΔrhlB pcnB^– strain suggests that the very high levels of mRNA decay intermediates in this strain could interfere with RNase E activity. Because ΔrhlB by itself does not derepress expression of the ez1 fusion, it seems unlikely that the stabilization of the T7-lacZ message can be explained by general inhibition of RNase E activity. This conclusion is consistent with previous characterizations of ΔrhlB showing that growth is not impaired and that there is no obvious defect in the RNase E-mediated maturation of tRNA and rRNA, or in the degradation of mRNAs synthesized by E. coli RNA polymerase (V.K., unpublished results, and refs. 12 and 32).

Table 3. β-Galactosidase activity.

		Activity (±SD, %)		ez1/ez6 Derepression
Strain	Genotype	ez1	ez6
MC1061	Wild type	1,220 (±6.3)	1,660 (±8.6)	1.0
AC27	rne131	3,320 (±8.7)	1,730 (±2.3)	2.6
SVK1	ΔrhlB	1,470 (±4.9)	1,940 (±9.9)	1.0
SVK45	pcnB-	1,050 (±5.6)	1,480 (±4.7)	1.0
SVK46	ΔrhlB pcnB-	1,970 (±4.4)	1,580 (±5.3)	1.7

The results are the average of at least three independent measurements. Activities were measured as described (23). The SD as a percentage is in parentheses. Derepression is the ratio of the activity with the ez1 fusion to the activity with the ez6 fusion normalized to this ratio in wild type.

Deletion of the rhlB Gene Stabilizes Other T7 Messages. To test whether the effect of the ΔrhlB mutant is specific to the lacZ transcript or due to transcription by the T7 RNA polymerase, we examined the expression of two recombinant mouse proteins, krox20 and engrailed, expressed from plasmids in which the gene is under the control of a T7 promoter. Previous work showed that the expression of these recombinant proteins is significantly improved by the rne131 mutation (22). A third protein, the HTLV protease, was also tested, but we were unable to detect a specific signal by Western blotting, apparently due to loss of activity of the antiserum. Plasmids harboring the genes encoding the recombinant proteins were transformed into ENS134 and into the isogenic SVK4 (ΔrhlB) and ENS134-2 (rne131) strains. Expression was induced by IPTG and analyzed by Western blotting (Fig. 4). The total protein loaded in each lane is equivalent (SDS/PAGE). In each panel of Fig. 4, lane 1 is a background control showing the signal from BL21(DE3) transformed with pET11a. BL21(DE3) is the isogenic parent of ENS134. As expected, the expression in the rne131 mutant (lane 3) is significantly higher than in the wild-type control (lane 2). The expression also increases significantly in the ΔrhlB strain (lane 4). The effect of ΔrhlB is equal to or slightly stronger than the rne131 mutation, suggesting an important role for RhlB in the inactivation of these messages. The β-galactosidase expression from the chromosomal T7-lacZ gene is visible in the SDS/PAGE (asterisk, clearly resolved on the 12% gel used to separate krox20). Consistent with the enzyme assays showing ≈3-fold more β-galactosidase activity in rne131 than in ΔrhlB (Table 1), β-galactosidase expression in rne131 (lane 3) is higher than in ΔrhlB (lane 4). Thus, in contrast to the RNase E-dependent inactivation of the T7-lacZ mRNA, which is partially RhlB-dependent, the RNase E-dependent inactivation of the krox20 and engrailed messages is fully RhlB-dependent.

Fig. 4.

SDS/PAGE and Western blots from E. coli strains expressing recombinant krox20 (a) or engrailed (b) transcribed by T7 RNA polymerase. In each panel, lanes 1–4 are, respectively, as follows: BL21(DE3)/pET11a (–), ENS134/krox20 or engrailed (wt), ENS134-2/krox20 or engrailed (rne131), SVK4/krox20 or engrailed (ΔrhlB). In a, 12% SDS gel was stained with Sypro Orange and visualized by fluorimaging. In b, same as a, except that the proteins were separated on a 15% gel. The Western blots were developed by enhanced chemiluminescence (ECL) detection. The krox20 and engrailed proteins are not visible on the SYPRO Orange-stained SDS/PAGE. Their approximate migration, based on the Western blots, is indicated by the arrow. The asterisk to the right of the SDS/PAGE indicates the position of the β-galactosidase expressed from the chromosomal T7-lacZ gene.

Discussion

The principal result of this work is that the DEAD-box RNA helicase, RhlB, facilitates the inactivation by RNase E of messages synthesized by bacteriophage T7 RNA polymerase. This conclusion is supported by the primer extension analysis (Fig. 2), which shows that the ΔrhlB mutation impairs RNase E cleavages in the 5′ end of the T7-lacZ mRNA. This decrease in RNase E cleavage correlates with an increase in full-length lacZ mRNA (Fig. 1) and an increase in the level of β-galactosidase activity (Table 1). These results provide evidence for a functional interaction between RhlB and RNase E. Furthermore, until now, the RNA helicases identified in mRNA degradation have been characterized as part of exonucleolytic degradation pathways involving enzymes such as the prokaryotic PNPase or the eukaryotic exosome (13, 14). Our analysis of the self-regulation of RNase E expression in the ΔrhlB strain (Table 3), which is normal, argues against a general inhibition of RNase E activity. The strains used in Table 3 are the same strains in which we previously demonstrated the accumulation of structured mRNA decay intermediates due to the disruption of the RhlB-PNPase pathway (12). Thus, an inhibition of RNase E activity by the mRNA decay intermediates that accumulate in the ΔrhlB strain is unlikely. The lack of a general effect on RNase E is consistent with recent work suggesting that the ΔrhlB mutation has at most a small effect on the lifetime of messages synthesized by E. coli RNA polymerase (12, 32). Furthermore, the results in the Northern blot (Fig. 1) suggest that the intercistronic RNase E cleavage between lacZ and lacY′ is not affected because the abundance of the processed 3.2-kb species increases in the ΔrhlB strain. These results suggest that RhlB is not generally required for RNase E activity in vivo. It should be noted that this is similar to the RhlB-PNPase pathway because RhlB is not required for PNPase activity except under particular circumstances involving the degradation of highly structured mRNA decay intermediates.

The increased expression of krox20 and engrailed (Fig. 4) suggests that the effect of ΔrhlB is general for messages synthesized by T7 RNA polymerase. Previous work has shown that the effect of the rne131 mutation, in which the entire noncatalytic region of RNase E is disrupted, on the T7-lacZ message is equivalent to the effect of an rne thermosensitive mutation, in which the endoribonuclease activity is abolished at the nonpermissive temperature (21, 22). Thus, the exceptional sensitivity of the T7-lacZ message to RNase E is mediated by elements in the noncatalytic region (22, 23). The effect of ΔrhlB on the expression of krox20 and engrailed is comparable with the effect of rne131 whereas its effect on the T7-lacZ message is weaker than rne131. Regardless of this difference, these results show that RhlB is important in the inactivation and degradation of the T7 messages. There are several differences between T7-lacZ, and krox20 and engrailed that could explain the differential effect of ΔrhlB, including the length and sequence of the messages, the signals for translation (lac vs. T7), and the number of copies of the gene. It is unlikely that the effect of ΔrhlB on krox20 and engrailed is due to a trivial effect on plasmid copy number because the ΔrhlB mutation does not affect pBR322 replication (V.K., unpublished results). The stringent requirement for RhlB in the inactivation of krox20 and engrailed could, for example, be due to high-level transcription from multicopy genes or differences in the translation signals.

The distinctive features of T7 transcription are the synthesis of long stretches of ribosome-free mRNA due to rapid elongation and the relative insensitivity of the RNA polymerase to transcription termination by Rho, a factor that inhibits the synthesis of untranslated mRNA by E. coli RNA polymerase (33). If the RhlB-RNase E pathway in E. coli were involved in the degradation of ribosome-free mRNA, this function would be complementary to transcription termination by Rho. Messages that are under the control of a translational repressor are a source of ribosome-free mRNA that might escape transcription termination by Rho. Indeed, it has already been shown that the repressed threonine synthetase mRNA is exceptionally sensitive to degradation by RNase E and that this sensitivity is abolished by the rne131 mutation (34). That is, full-length translationally repressed messages accumulate in the rne131 strain whereas in wild type they are rapidly degraded. Coupling mRNA degradation to translational control makes repression irreversible. The RhlB-RNase E degradation pathway could also be involved in the posttranscriptional regulation of gene expression by small antisense RNAs such as RhyB because RhyB-mediated mRNA degradation, which is RNase E-dependent, is slower in the rne131 strain (35, 36). In contrast to the situation in uninfected E. coli, bacteriophage T7 messages are unusually long-lived (20-min half-life), and it has been suggested that this stability is due to inhibition of the host mRNA degradation machinery during an infection (37). Bacteriophage T7 synthesizes a protein kinase that phosphorylates numerous host proteins. Expression of this kinase in the ENS134 strain leads to the phosphorylation of RNase E and RhlB, and the stabilization of the T7-lacZ message (38). Thus, messages synthesized by the T7 RNA polymerase during an infection might be protected from inactivation by the RhlB-RNase E pathway due to phosphorylation of these enzymes by the protein kinase.

Our results show that point mutations inactivating RhlB have the same effect as deleting the rhlB gene (Table 1). Furthermore, the inactive RhlB protein is associated with RNase E as a component of the RNA degradosome (Fig. 3). Thus, it is unlikely that the physical interaction between RhlB and RNase E is involved in an allosteric regulation of RNase activity. The requirement for enzymatic activity suggests a role for RNA unwinding or remodeling. Because RNase E is single-strand specific, RhlB could counteract the collapse of RNase E cleavage sites into double-stranded structures. Experiments in vitro using the purified RNA degradosome show no effect of ATP on the maturation by RNase E of 5S rRNA from its 9S precursor (A.J.C. and L.P., unpublished results). This result is not unexpected because the RNase E sites in the 9S precursor are in single-stranded regions that are unlikely to collapse into double-stranded structures (39). Similarly, we have been unable to detect an effect of ATP on the cleavage of an mRNA fragment from the 5′ end of lacZ. However, these experiments need to be extended to an analysis of the full-length lacZ message as well as the krox20 and engrailed messages. Another possibility is that RNA binding proteins are involved. Ribosome-free mRNA is unlikely to be protein-free in vivo. Because NPH-II, a DExD/H protein from vaccinia virus, has been shown to displace RNA binding proteins (40), RhlB could be involved in the disruption or remodeling of ribonucleoprotein to facilitate RNase E cleavage. It should be interesting in future experiments to elucidate the mechanism by which RhlB facilitates the degradation of ribosome-free mRNA.