Escherichia coli rng gene (previously called cafA) encodes a novel RNase, named RNase G, which is involved in the 5′ end-processing of 16S rRNA. In rng mutant cells, a precursor form of 16S rRNA, 16.3S rRNA, is accumulated. Here we report a role of RNase G in the in vivo mRNA metabolism.
We found that rng::cat mutant strains overproduced a protein of about 100 kDa. N-terminal amino acid sequencing of this protein showed that it was identical to the fermentative alcohol dehydrogenase, the product of the adhE gene located at 28 min on the E. coli genetic map. The level of adhE mRNA was significantly higher in the rng::cat mutant strain than that in its parental strain, while such differences were not seen in other genes we examined. A rifampicin-chase experiment revealed that the half-life of adhE mRNA was 2.5-fold longer in the rng::cat disruptant than in the wild-type.
These results indicate that, in addition to rRNA processing, RNase G is involved in in vivo mRNA degradation in E. coli.
The cafA gene, an old name for the rng gene which encodes Escherichia coli RNase G, was originally found as a putative open reading frame (orf) of unknown function during analysis of mecillinam-resistant shape-determination mutants (Wachi et al. 1991). Extensive overproduction of CafA protein caused the formation of chained cells and minicells in E. coli. We observed long axial filament bundles, termed cytoplasmic axial filament, running through the centre of their cytoplasms (Okada et al. 1994). McDowall et al. (1993) reported a significant sequence similarity between the CafA protein and the N-terminal half of RNase E, which is encoded by the rne gene at 24 min on the E. coli genetic map. RNase E is an essential E. coli endoribonuclease involved in the maturation of 5S rRNA and the decay of many mRNAs (for a review see Cohen & McDowall 1997). The rne-1 (also called ams-1) temperature-sensitive mutation was initially described as increasing the chemical half-life of total RNA (Kuwano et al. 1977; Ono & Kuwano 1980). Another independently isolated mutation, mapped to the same gene, rne-3071, leads to the accumulation of 9S rRNA, a precursor of 5S rRNA (Ghora & Apirion 1978; Mudd et al. 1990). These two mutations were caused by amino acid substitutions in the highly homologous region between RNase E and RNase G (McDowall et al. 1993). We investigated the effect of the cafA mutation on cellular RNA metabolism and found that a precursor molecule of 16S rRNA accumulated in the cafA::cat mutant cells. We also found that an E. coli mutant, called BUMMER (Dahlberg et al. 1978), which accumulated a precursor form of 16S rRNA, named 16.3S rRNA which has 66 extra nucleotides at the 5′-end, has an 11 bp deletion in the cafA gene (Wachi et al. 1999). These results indicated that cafA gene encodes an RNase activity involved in the 5′-end maturation of 16S rRNA which is defective in the BUMMER mutant. We proposed that this activity was provided by a new enzyme named RNase G. The gene was renamed as rng. Hereafter we use ‘RNase G’ and ‘rng’ instead of ‘CafA’ and ‘cafA’, respectively.
The endoribonucleolytic activity of RNase G was confirmed by Tock et al. (2000) by using highly purified protein and synthetic oligo-ribonucleotide substrates in vitro. The action of RNase G was dependent on the nature of the 5′-end of the substrates. RNase G cleaves 5′-monophosphorylated substrates efficiently but does not cleave the 5′-hydroxylated ones at all. RNase G action was also blocked by a 5′-triphosphate group. Interestingly, the 5′-monophosphate group may only stimulate cleavage at sites which are present on the same RNA molecule. Considering these characteristics, they suggest that RNase G prefers to cut to completion 5′-monophosphorylated decay or processing intermediates rather than initiating the decay of intact 5′-triphosphorylated RNAs. In contrast, the 3′-phosphorylation status did not affect the rate of cleavage by RNase G. From the results of in vitro experiments, Tock et al. (2000) suggested that RNase G contributes in some manner to the decay of RNAs in E. coli. Our genetic evidence (Wachi et al. 1997), showing that functions of RNase E and RNase G overlap, although partially, also supports this idea. Here we present the first evidence suggesting that RNase G is involved in an in vivo degradation of mRNA.
Overproduction of AdhE protein in the rng::cat mutant
In the course of analysing E. coli rng::cat mutant strains, we found that an rng::cat mutant GM11 produced a large amount of protein with a molecular weight of about 100 kDa, when compared with the rng+ parent strain MC1061 (Fig. 1). The rng::cat strain GM11 produced about fivefold more 100 kDa protein than the wild-type strain MC1061. Most of the 100 kDa protein was recovered in the membrane fraction, although a significant amount was recovered in the soluble fraction. Overproduction of the 100 kDa protein was reversed to the wild-type level when the plasmid pMEL2, which carries only the rng gene as an intact gene (Wachi et al. 1989), was introduced into the mutant strain (Fig. 1). This indicated that overproduction of the 100 kDa protein is due to a defect in the function of the rng gene.
This phenomenon was very conspicuous when the rng::cat mutation was introduced into the MC1061 strain. In the background of the W3110 or CSH26 strains, an overproduction of the 100 kDa protein was less significant than that in the case of MC1061 (see below). Therefore, a further analysis of this phenomenon was mainly carried out using MC1061 and its rng::cat derivative GM11.
To identify the 100 kDa protein, the N-terminal amino acid sequencing of the protein was carried out. The sequence of 18 amino acid residues of the N-terminal portion of the 100 kDa protein was found to be AKGQSLQDPFLNALRRER, which was identical to that of the AdhE protein, the fermentative alcohol dehydrogenase, which is encoded by the adhE gene at 28 min on the E. coli genetic map (Goodlove et al. 1989), except that the N-terminal methionine residue was removed. The deduced molecular weight of the AdhE protein, 95 995 Da, was comparable with the estimated value of the overproduced protein on the SDS-polyacrylamide gel electrophoresis. It is reported that the AdhE protein forms a large complex called spirosome and that it is mainly recovered in precipitates by ultracentrifugation (Kessler et al. 1991; Imamura et al. 1992). Moreover, an adhE disruptant strain did not produce the 100 kDa protein, even when the rng::cat mutation was introduced (data not shown). Considering all this, we concluded that the 100 kDa protein was AdhE.
Effect of the rne-1 mutation on expression of the AdhE protein
As the rng gene product, RNase G, shows a great sequence homology (McDowall et al. 1993) and genetic interaction (Wachi et al. 1997) with RNase E, which is involved in mRNA degradation as well as in rRNA processing (McDowall et al. 1993), we examined the effect of a mutation of RNase E on the expression of AdhE protein. As shown in Fig. 2, an rne-1 (RNase Ets) derivative of MC1061, CH1828, did not overproduce the AdhE protein at the restrictive temperature (43 °C) as well as the permissive temperature (30 °C), while the rng::cat strain GM11 overproduced the AdhE at both temperatures (data at 43 °C not shown). These results indicate that this phenomenon is RNase G-specific.
Effects of mutations in the transcription factors involved in expression of adhE
It has been reported that the expression of adhE gene was under control of the stationary phase specific sigma factor, RpoS and the anaerobic transcription factor, Fnr (Membrillo-Hernández & Lin 1999). The consensus sequence for binding of the Fnr was found in the upstream region of the adhE (Kessler et al. 1991). It is also known that MC1061 has a wild-type allele of rpoS gene, while many other laboratory strains, including W3110, have an rpoS mutation and produce a truncated RpoS protein (Jishage & Ishihama 1997). We then examined whether the overproduction of AdhE by the rng::cat mutation depends on these transcriptional regulators, but neither the rpoS13::Tn10 mutation nor the fnr mutation demonstrated any effect on overproduction of the AdhE by the rng::cat mutation (data not shown).
Increased stability of the adhE mRNA in the rng::cat mutant
As the rng gene encodes RNase G, the effect of the rng::cat mutation on the adhE mRNA was examined by Northern hybridization. As shown in Fig. 3A, the level of the adhE mRNA in the rng::cat mutant strain GM11 was about sixfold greater than that in the rng+ strain MC1061. On the other hand, mRNA levels of the gapA gene, which is located at 39 min on the genetic map and encodes glyceraldehyde 3-phosphate dehydrogenase in the glycolysis pathway (Branlant & Branlant 1985), and the mreB gene, which is located just upstream of the rng gene on the genetic map and encodes a shape-determining protein MreB (Doi et al. 1988), were almost same in both strains (Fig. 3B,C).
To determine whether the rng::cat mutation affects transcription of the adhE gene or the stability of the adhE mRNA, a rifampicin-chase experiment was carried out. De novo synthesis of mRNAs was inhibited by the treatment of exponentially growing cells with 150 µg/mL of rifampicin, and degradation of the adhE mRNA synthesized prior to the addition of rifampicin was examined in the rng+ and rng::cat strains. As shown in Fig. 4(A,C), the adhE mRNA degraded with a half-life of 4.0 min in the rng+ cells. In contrast, it degraded with a half-life of 9.7 min in the rng::cat mutant cells. This indicates that overproduction of the AdhE protein in the rng::cat mutant strain is due to an increased stability of the adhE mRNA. Effect of the rng::cat mutation on the stability of mRNAs other than the adhE mRNA was also examined. As shown in Fig. 4(B,D), the gapA mRNA degraded with almost similar half-life times in both strains: 7.9 min in rng+ and 8.6 min in rng::cat. These results suggest that RNase G encoded by the rng gene specifically degrades the adhE mRNA, although this does not exclude the possibility that the RNase G degrades other mRNA(s) than the adhE mRNA.
Possible involvement of the 5′-untranslated region of adhE mRNA in RNase G-dependent stability control
In order to estimate what factor(s) are involved in stability control by RNase G, an adhE-lacZ fusion gene was constructed as described in Experimental procedures. Plasmid pALF1 carrying a 1.3 kb upstream region of the adhE gene containing the promoter sequence, the 5′-untranslated region and the first 27 bp of the coding region joined with the lacZ gene in-frame was introduced into MC1061 (rng+) and GM11 (rng::cat). As shown in Fig. 5A, the AdhE-LacZ fusion protein was overproduced in the rng::cat mutant strain GM11 at almost similar levels to the intact AdhE protein encoded by the chromosomal adhE gene. β-Galactosidase activity expressed from the adhE-lacZ fusion gene in GM11 (rng::cat) was about 3.6-fold higher than that in its parental strain MC1061 (Fig. 5B). This result implies that most of the coding region of the adhE gene is not required for stability control by RNase G.
Plasmid ALF1 was also introduced into CSH26 (rng+) and its rng::cat derivative GC11. Interestingly, expression level of the AdhE-LacZ fusion is much lower (about 20%) in CSH26 even in the rng+ background, than that in MC1061. However, the ratios of increase of the β-galactosidase activity by introduction of the rng::cat mutation were almost the same, both in CSH26 and MC1061 (Fig. 5). This suggests that some unknown transacting factor(s), which probably affects the basal transcription levels of the adhE gene, may be changed between MC1061 and CSH26 (probably also W3110) and that this is indifferent to the effect of the rng::cat mutation.
RNase G is a newly identified endoribonuclease responsible for the 5′-end processing of 16S rRNA (Wachi et al. 1999; Li et al. 1999). It is encoded by the rng gene located at 71 min on the E. coli genetic map (Wachi et al. 1991) and shows a high sequence similarity with the N-terminal half of RNase E which is involved in rRNA processing as well as in mRNA stability (McDowall et al. 1993). Here we have presented evidence which suggests that RNase G is involved in the in vivo degradation of the adhE mRNA which encodes a fermentative alcohol dehydrogenase, AdhE. It was found that the rng::cat mutant strain GM11 overproduced the AdhE protein. A rifampicin-chase experiment revealed that overproduction of the AdhE was due to the increased stability of the adhE mRNA in the rng::cat mutant strain. These results indicate that, in addition to rRNA processing, RNase G is responsible, at least in part, for in vivo mRNA degradation in E. coli. This is the first demonstration of the role of RNase G in in vivo mRNA metabolism.
As far as we could test, the effect of the rng::cat mutation on mRNA stability was adhE mRNA-specific. The rng::cat mutation showed negligible effects on gapA mRNA or mreB mRNA. These results suggest that RNase G has some specific recognition sequence(s) in vivo, as was suspected from in vitro experiments (Tock et al. 2000). Analysis of the adhE-lacZ fusion gene suggests that the 5′-untranslated region of the adhE mRNA is involved in stability control by RNase G. Role of the 5′-untranslated region in mRNA stability control by RNase III is also reported in the pnp and rnc mRNAs (Portier et al. 1987; Bardwell et al. 1989). It is thought that RNase III processing triggers a decay of the transcripts downstream. Vytvytska et al. (2000) also demonstrated that the 5′-untranslated region of the ompA mRNA was involved in stability control by RNase E. Since the action of RNase G depends on the nature of 5′-end of the substrates as RNase E (Tock et al. 2000), a similar mechanism may be involved in the case of RNase G. It is very likely that the recognition sequence(s) by RNase G lies in the 5′-untranslated region of the adhE mRNA. The recognition sequence(s) and/or digestion site(s) by RNase G should be therefore determined in order to draw more specific conclusions. It would also be of interest to find RNA molecules, of which the degradation or maturation is catalysed by RNase G, other than 16S rRNA and adhE mRNA. For this purpose, a 2D-gel analysis of whole proteins and DNA array analyses are now under investigation.
Homologues of RNase G or the N-terminal endoribonucleolytic domain of RNase E are found in most of the Gram-negative and some of the Gram-positive bacteria that have been completely sequenced (Fig. 6). Most of the Gram-negative bacteria have both RNase E-type and RNase G-type enzymes, but Rickettsia prowazekii has only one homologue that is more closely related to RNase E, although it does not have a long C-terminal domain. The homologues of the Gram-positive bacteria seem to belong to G-type enzymes, judging from sequence homology and their sizes, although the Mycobacterium tuberculosis enzyme has an extra N-terminal domain. Since RNase G can make cleavages that resemble those of RNase E (Tock et al. 2000), a careful assignment of the contribution of these two enzymes in RNA decay or processing is required in organisms that possess both enzymes. It is also interesting to investigate the role of RNase G or E in organisms that only have one homologue.
Bacterial strains and media
The bacterial strains used are listed in Table 1. Cells were grown at 30 °C in L broth containing 1% Bactopeptone, 0.5% yeast extract, 0.5% NaCl and 0.1% glucose (pH 7.0). Appropriate antibiotics were added for culturing the cells carrying plasmids.
Table 1. Bacterial strains
National Institute of Genetics, the Bacterial Stock Center, Mishima 411-8540, Japan.
Cells suspended in sodium phosphate buffer (50 mm, pH 7.0) were disrupted by sonication. After removing the unbroken cells, cell lysates were fractionated by ultracentrifugation (100 000 g, 30 min, 4 °C). The supernatant was collected as soluble fractions. The precipitates were suspended in the same buffer and collected as membrane fractions. Proteins were separated by SDS–polyacrylamide gel electrophoresis and gels were stained with Coomassie Brilliant Blue. AdhE bands were quantified by measuring band intensity using NIH-Image software after serial dilutions.
Isolation of total cellular RNA and analysis of mRNA
Total cellular RNA was isolated and analysed as previously described (Wachi et al. 1999). RNAs were separated by formaldehyde–agarose gel electrophoresis (6.5% formaldehyde, 1 × MOPS buffer [20 mm 3-(N-morpholono)-propane sulphonic acid pH 7.0, 5 mm sodium acetate, 1 mm EDTA], 1% agarose). RNAs were transferred on to positively charged nylon membrane (Hybond-N+ Amersham, Buckinghamshire, UK) by capillary method. For the hybridization probes, a 2.1 kb HindIII fragment containing a part of the adhE gene and a 0.6 kb PstI fragment containing a part of the mreB gene were used. For the gapA gene, a DNA fragment was amplified by polymerase chain reaction (PCR) using the set of primers, 5′-GACTATCAAAGTAGGTATCAACGGT-3′ and 5′-GATGTG AGCGATCAGGTCCAGAACT-3′. Northern hybridization was carried out using a Gene Images kit (Amersham). RNAs were quantified by measuring band intensity from the NIH-Image software after serial dilutions.
For the rifampicin-chase experiment, exponentially growing cells were treated with 150 µg/mL rifampicin to inhibit the de novo synthesis of RNA, and after the indicated times, total RNA was extracted and analysed as described above.
Construction of the adhE-lacZ fusion gene
A 1.3 kb upstream region of the adhE gene containing the promoter sequence, the 5′-untranslated region and the first 27 bp of the coding region was amplified by PCR using the set of primers, 5′-GATGTGGCGAAGTTAACATGATGG-3′ and 5′-GCTCTACGAGTGCGTTAACTTACGCGAC-3′, in which the HpaI sites (underlined) were introduced for cloning. Amplified DNA was digested with HpaI and joined in-frame with the lacZ gene at the 9th codon of adhE and the 6th codon of lacZ on a mini-F plasmid carrying the ampicillin-resistance gene. The resultant plasmid, named pALF1, was introduced into the rng::cat mutant strains and their parent strains. Expression of the AdhE-LacZ fusion was examined by SDS–polyacrylamide gel electrophoresis or by measuring β-galactosidase activity. β-Galactosidase activity was determined by the method of Miller (1972).
The authors thank Dr Hiroji Aiba for a critical reading of the manuscript. We also thank Kyowa Hakko Kogyo Co. Ltd for N-terminal amino acid sequencing of the AdhE protein. This work was partly supported by a Grant-in-Aid for Scientific Research (12660069 to M.W.) and a Grant-in-Aid for Scientific Research of Priority Areas (11227203 to K.N.) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.