|Home | About | Journals | Submit | Contact Us | Français|
Four 3′-to-5′ exoribonucleases have been identified in Bacillus subtilis: polynucleotide phosphorylase (PNPase), RNase R, RNase PH, and YhaM. Mutant strains were constructed that were lacking PNPase and one or more of the other three ribonucleases or that had PNPase alone. Analysis of the decay of mRNA encoded by seven small, monocistronic genes showed that PNPase was the major enzyme involved in mRNA turnover. Significant levels of decay intermediates, whose 5′ ends were at the transcriptional start site and whose 3′ ends were at various positions in the coding sequence, were detected only when PNPase was absent. A detailed analysis of rpsO mRNA decay showed that decay intermediates accumulated as the result of a block to 3′-to-5′ processivity at the base of stem-loop structures. When RNase R alone was present, it was also capable of degrading mRNA, showing the involvement of this exonuclease in mRNA turnover. The degradative activity of RNase R was impaired when RNase PH or YhaM was also present. Extrapolation from the seven genes examined suggested that a large number of mRNA fragments was present in the PNPase-deficient mutant. Maintenance of the free ribosome pool in this strain would require a high level of activity on the part of the tmRNA trans translation system. A threefold increase in the level of peptide tagging was observed in the PNPase-deficient strain, and selective pressure for increased tmRNA activity was indicated by the emergence of mutant strains with elevated tmRNA transcription.
The steady-state amount of a particular mRNA in a cell is a function of its synthesis and degradation. Thus, it is understood that mRNA decay is an important factor in setting levels of gene expression. In a generally accepted model for Escherichia coli (4, 18, 21), mRNA decay initiates with endonuclease cleavage by RNase E, a 5′-end-dependent endoribonuclease (23). Such cleavage generates an upstream fragment with an unprotected 3′ end, which is rapidly degraded by the 3′-to-5′ exonucleases, RNase II or polynucleotide phosphorylase (PNPase), and a downstream fragment that begins with a monophosphate nucleoside, which is a much better substrate for subsequent binding and cleavage by RNase E than the 5′-terminal triphosphate nucleoside of the initial transcription product. Additional cycles of rapid endonucleolytic cleavage and 3′-to-5′ degradation result in the observed all-or-nothing pattern of decay, i.e., Northern blot analysis of a specific mRNA generally shows the full-length mRNA but few or no mRNA decay fragments. Rapid decay of mRNA seems to be an essential function, as either RNase II or PNPase is required for viability; inactivation of both RNase II and PNPase is lethal (11). E. coli contains six other 3′-to-5′ exoribonucleases (39). Very recently, one of these, RNase R, was reported by several groups to be involved in mRNA decay (1).
We have been studying mRNA decay in Bacillus subtilis. It was shown years ago by Duffy and colleagues that mRNA decay in B. subtilis occurs primarily phosphorolytically, rather than hydrolytically as in E. coli (14). Deutscher and Reuven, working with extracts of E. coli and B. subtilis and using a poly(A) substrate, showed that almost 90% of RNA decay in an E. coli extract is due to RNase II activity, whereas more than 98% of RNA decay in a B. subtilis extract is due to a phosphate-dependent activity (9). The B. subtilis genome contains no homologue of RNase II, and the phosphorolytic nature of RNA decay in B. subtilis is due to PNPase activity (25). Despite the apparent dominant role of PNPase in RNA decay, the B. subtilis pnpA gene, encoding PNPase (22), is not essential (33). A pnpA mutant has a number of phenotypes, including cold sensitivity, competence deficiency (22), tetracycline sensitivity, filamentous growth (33), and the more recently discovered dysregulation of trp operon expression (8). The effect of the pnpA disruption on the decay of mRNA has been studied using the plasmid-borne erythromycin resistance gene, ermC (1a, 12). ermC mRNA decay fragments that are virtually undetectable in the wild-type strain are easily observed in the pnpA mutant.
Three other B. subtilis 3′-to-5′ exoribonucleases have been identified and characterized. (i) RNase PH, the product of the rph gene, is a phosphorolytic enzyme that is not essential (7) and is involved in tRNA processing (C. Condon and D. H. Bechhofer, unpublished data), as is the E. coli RNase PH enzyme (10). (ii) RNase R, the product of the rnr gene, was purified on the basis of the remaining Mn2+-dependent exonucleolytic activity in an extract of a PNPase-deficient strain (28). The function of RNase R is not known. RNase R is able to degrade rRNA in vitro and can also degrade an artificial RNA substrate that contains a strong stem-loop structure, although this requires an extended 3′ single-stranded sequence downstream of the stem (28). Recently, RNase R of E. coli has been found to function in the quality control of rRNA (3). (iii) YhaM was identified in a strain of B. subtilis that was deficient for both PNPase and RNase R (29). YhaM requires Mn2+ or Co2+ for activity and is not active in the presence of Mg2+.
Several mutants deficient in more than one of these four exoribonucleases (PNPase, RNase PH, RNase R, and YhaM) have been constructed (28, 29). In the present study, a set of double and triple RNase mutants was constructed, allowing study of the role of these exoribonucleases in the decay of endogenous mRNAs. The transcription product of the rpsO gene, which encodes ribosomal S15 protein, was chosen for detailed analysis by Northern blotting from high-resolution denaturing gels. This system allowed the mapping of specific mRNA decay products. The role of tmRNA in a strain lacking PNPase was also examined.
B. subtilis RNase mutant strains were derivatives of the parent strain BG1, which is trpC2 thr-5. The preparation and transformation of B. subtilis competent cell cultures were done as described previously (13). E. coli strain DH5α (17) was the host for plasmid constructions.
RNase mutant strains are listed in Table Table1.1. The construction of the chloramphenicol-resistant and kanamycin-resistant pnpA deletion mutants (33), the spectinomycin-resistant rnr deletion mutant (28), and the phleomycin-resistant yhaM deletion mutant (29) has been described previously. The tetracycline-resistant rnr deletion mutant was constructed by replacing an internal 303-bp BglII fragment of the previously cloned rnr gene (28) with a tetK gene contained on a BamHI fragment. Substitution of the rph coding sequence by a spectinomycin resistance protein coding sequence was accomplished as follows. A 2-kb piece of chromosomal DNA, including the rph gene and parts of the upstream gerM gene and downstream ysnA gene, was cloned into an M13mp18-replicative form. The sequence at the rph start codon was mutagenized (20) to give an NdeI restriction site. The rph coding sequence (245 codons) contains an XbaI site located at codon 212. Codons 1 to 212 of the rph coding sequence were replaced, in frame, with a spectinomycin resistance protein-coding sequence that had been amplified as an NdeI-XbaI fragment. Multiply mutant strains were constructed by transformation of RNase mutant strains with chromosomal DNA obtained from a strain with a deletion of a gene producing a different RNase.
RNA was isolated by hot phenol extraction from B. subtilis cultures grown to mid-logarithmic phase in minimal medium containing Spizizen salts with 0.5% glucose, 0.1% Casamino Acids, 0.001% yeast extract, 50 μg of tryptophan/ml, 50 μg of threonine/ml, and 1 mM MgSO4, as described previously (8). Northern blot analysis of RNA separated on 6% polyacrylamide denaturing gels or sequencing gels was done as described previously (15). The veg riboprobe was synthesized by T7 RNA polymerase (Ambion) in the presence of [α-32P]UTP, using as a template an isolated PCR fragment containing the veg coding sequence. 5′-end-labeled oligonucleotide probes were prepared using T4 polynucleotide kinase (New England Biolabs) and [γ-32P]ATP. To control for RNA loading in Northern blot analyses of rpsO decay fragments and mRNA half-life, membranes were stripped and probed for 5S rRNA, as described previously (31). The size marker for the Northern blots (see Fig. Fig.2)2) was a 50-bp ladder (Invitrogen), which was 5′ end labeled. For Northern blots shown in Fig. Fig.11 (see also Fig. Fig.55 and and7C),7C), the size marker was the TaqI-digested plasmid pSE420 (2). Size markers on sequencing gels were sequencing reactions done with single-stranded M13mp18 DNA.
Cells were grown at 37°C in 25 ml of Luria-Bertani (LB) medium containing erythromycin (1 μg/ml) in the presence or absence of isopropyl-β-d-thiogalactopyranoside (IPTG) (0.5 mM). At mid-logarithmic phase (absorbance at 600 nm, 0.5), cells were harvested by centrifugation, washed with 10 mM Tris-HCl (pH 7.8)-10 mM MgCl2-60 mM NH4Cl, and stored at −80°C. Cell pellets were resuspended in 0.1 ml of Y-PER (protein extraction reagent; Pierce) and lysed by sonication (two times for 1 min each). The protein concentration of the extract was adjusted to 3.5 mg/ml. A 25-μl sample was mixed with 12.5 μl of 3× sodium dodecyl sulfate (SDS) loading dye (150 mM Tris-HCl [pH 6.8], 300 mM dithiothreitol, 6% SDS, 0.3% bromophenol blue dye, 30% glycerol), heated at 100°C for 5 min, and applied to an SDS-15% polyacrylamide gel. After electrophoresis (30 V, 14 h), proteins were electroblotted to a polyvinylidene difluoride membrane (Millipore) and probed with Penta-His antibody (QIAGEN). Antibody was visualized by chemiluminescence of ECL Plus (Amersham Pharmacia Biotech) after binding with horseradish peroxidase-conjugated secondary antibody.
The quantitation of the radioactivity of bands on Northern blots was done with a Storm 860 PhosphorImager instrument (Molecular Dynamics). rpsO mRNA half-life was determined by a linear regression analysis of the percent RNA remaining versus time. The free-energy values for predicted stem-loop structures were calculated at the Zuker RNA website (http://www.bioinfo.rpi.edu/applications/mfold/old/rna/form6.cgi), using a 37°C temperature, a 0.1 mM RNA concentration, and a 10 mM Na+ concentration.
Previous studies were done on the decay of ermC mRNA in a PNPase-deficient mutant (1a). We wished to determine the decay pattern for mRNAs encoded by endogenous genes in strains that were deficient in PNPase and other exoribonucleases. The first gene chosen for analysis was the veg gene, which encodes a small mRNA whose decay we have studied previously (34). Total RNA was isolated from a set of strains that were deficient either in PNPase alone or in combinations of three of the four B. subtilis 3′-to-5′ exoribonucleases (triple mutants) (Table (Table1).1). The RNAs were separated on a 6% polyacrylamide denaturing gel, blotted, and probed with a veg antisense riboprobe that was complementary to nucleotides (nt) 19 to 232 of the veg mRNA sequence. The results (Fig. (Fig.1)1) showed numerous decay intermediates that were detected only in strains lacking PNPase. Little accumulation of mRNA decay fragments was observed for the wild-type strain (Fig. (Fig.1,1, lane 1) or for the strain that has PNPase but lacks the other three exoribonucleases, i.e., RNase R, RNase PH, and YhaM (Fig. (Fig.1,1, lane 3). A similar pattern was found for ermC mRNA in triple mutant strains; decay fragments were detected only when PNPase was absent (data not shown). These results indicated that PNPase was the major mRNA-degrading enzyme. The patterns of decay fragments in the various pnpA strains differed somewhat, but in every case were rather complex. We sought to expand on these findings and to identify a small mRNA with a simpler pattern of decay fragments.
The decay patterns of six small, monocistronic genes, which are thought to be constitutively expressed, were probed. RNA was isolated from wild-type and pnpA strains and probed for these individual mRNAs, as well as for veg mRNA (Fig. (Fig.2).2). The probe in each case was a 5′-end-labeled oligonucleotide that was complementary to nucleotides spanning from the beginning of the coding sequence to the Shine-Dalgarno sequence. Abundant mRNA decay fragments were observed for the pnpA strain in the rpmB, rpsO, veg, and rpmE genes, and fewer were observed in the fur and rpsT genes (Fig. (Fig.2).2). The cca gene was expressed at a relatively low level (the cca lane in Fig. Fig.22 was a 5-week exposure), so the amount of decay fragments was difficult to assess. Detection of multiple fragments with 5′-proximal probes suggested that the 5′ end of these fragments was at or near the start of transcription and that the 3′ ends were at various sites within the coding sequence, as we have found previously for ermC mRNA decay fragments (1a).
The rpsO gene was chosen for further study since it was the most highly expressed (the Northern blots shown in Fig. Fig.22 were exposed for different times, with the rpsO blot having the shortest exposure), showed the clearest difference between wild-type and pnpA strains, and had distinct major bands between 100 and 200 nt. Reverse transcriptase mapping of the rpsO mRNA 5′ end was performed on RNA isolated from the wild type and pnpA strains and from one of the triple mutant strains, using as a primer the same 5′-end-labeled oligonucleotide that was used for Northern blot analysis. A single 5′ end was mapped (data not shown), suggesting that all or nearly all of the RNA fragments detected by the 5′-proximal probe began at the transcriptional start site. Thus, accurate determination of the sizes of these decay fragments would allow mapping of the 3′ ends. Separation of RNA on a high-resolution denaturing polyacrylamide gel (“sequencing gel”) followed by probing with an oligonucleotide probe is a method we have used previously to determine precisely the size of mRNA decay fragments (15). Such a Northern blot for rpsO mRNA decay fragments is shown in Fig. Fig.3A.3A. Clusters of bands at three positions were detected, migrating at about 180 nt, 133 nt, and 102 nt. The 180-nt cluster contained a set of fragments ranging from 175 to 185 nt.
We hypothesized that the accumulation of mRNA decay fragments in the pnpA strain was due to the inability of exoribonucleases other than PNPase to degrade through RNA secondary or tertiary structure. Thus, the 3′ ends of the observed decay fragments should coincide with the base of such structures. The sequence of the first 200 nt of the rpsO transcription unit was analyzed by the mfold RNA folding program of Zuker (38). Two structures with significant stem base-pairing were predicted to form (Fig. (Fig.3B),3B), and the 3′ base of these structures correlated well with the observed 180-nt and 133-nt decay products. The presence of the 102-nt decay product correlates with a pseudoknot structure that is predicted for the rpsO 5′-proximal regulatory region (32; also see Discussion).
Larger rpsO mRNA decay intermediates between 250 and 350 nt in size were also visible in the rpsO panel of Fig. Fig.2.2. Two of these were mapped by Northern blot analysis (Fig. (Fig.3C,3C, lane 1), using sequencing gels run out much further than the one shown in Fig. Fig.3A.3A. Strikingly, the sizes of these intermediates (approximately 320 and 350 nt) corresponded well with the downstream base of two other stem-loop structures that were predicted by mfold to occur in rpsO mRNA (Fig. (Fig.3D).3D). One other stem-loop structure is predicted for rpsO (Fig. (Fig.3D)3D) and indeed, a decay intermediate of 260 nt was mapped by Northern blot analysis (data not shown).
To assess the effect of the lack of PNPase on rpsO mRNA half-life, Northern blot analysis of RNA isolated at times after rifampin addition, to inhibit new transcription, was performed (Fig. (Fig.4).4). Surprisingly, the half-life of full-length rpsO mRNA was similar in the wild-type and pnpA strains: 5.6 min and 6.7 min, respectively (averages of two experiments). (The half-lives of the decay intermediates could not be determined, since we assume that their concentrations are simultaneously increasing by the decay of full-length RNA and decreasing by degradation.) This result indicated that the initiation of decay was not significantly affected by the loss of PNPase; rather, the degradation of products generated subsequent to attack on the full-length mRNA was slowed.
To study the function in mRNA decay of ribonucleases other than PNPase, rpsO decay fragments were probed in strains that were lacking PNPase and one of the other three known exoribonucleases: RNase R, RNase PH, or YhaM. RNA was separated on a denaturing polyacrylamide gel of intermediate size (i.e., about half the size of a sequencing gel). Experiments were done in duplicate, and the Northern blots were subsequently stripped and probed for 5S rRNA to allow for quantitation. The strains in which PNPase and either RNase PH or YhaM were missing (Fig. (Fig.5A,5A, lanes 4 and 5) showed slight differences (within twofold) (Fig. (Fig.5C)5C) in the quantities of decay fragments from the strain missing PNPase only. The strain that was missing PNPase and RNase R, however, showed a significant increase in the amount of 180- and 133-nt fragments, as well as a surprisingly large increase (20-fold) in the amount of 102-nt fragment. These results showed that RNase R had a role in degrading mRNA fragments when PNPase was not present.
To explore this finding further, rpsO mRNA decay fragments were probed in strains lacking three ribonucleases, i.e., PNPase and two of the other three. As can be seen from the Northern blot shown in Fig. Fig.5B5B and the quantitative data (Fig. (Fig.5D),5D), when RNase R alone was present (Fig. (Fig.5B,5B, lane 4) there was a lower level of decay fragment accumulation than in the singly mutant pnpA strain (Fig. (Fig.5B,5B, lane 2). However, when RNase PH alone was present (Fig. (Fig.5B,5B, lane 5), there was a dramatic increase in the amount of 102-nt fragment and a significant increase in the amounts of 180- and 133-nt fragments. The increase in the amount of decay fragments was not as great when YhaM alone was present (lane 6), although there was still a more than twofold increase in the amount of the 180-nt fragment and a large increase in the amount of 102-nt fragment.
It was possible to construct a quadruple mutant strain that was missing all four of the identified 3′-to-5′ exoribonucleases. This strain grew much more slowly than any of the other mutants (Table (Table1).1). The quadruple mutant RNA was probed for rpsO mRNA (Fig. (Fig.6,6, lane 5), and the same three groups of decay fragments were observed, suggesting that a similar mechanism of decay was occurring in this strain. This was likely due to yet another unknown 3′-to-5′ exoribonuclease. However, there was a pronounced shift in the migration of the 180- and 102-nt decay fragments, which would correspond to a block to exonucleolytic decay further downstream than in the triple mutants (compare lanes 4 and 5 in Fig. Fig.6).6). A similar shift was observed for the 320-nt set of bands in Fig. Fig.3C3C (compare lanes 1 and 2).
The seven genes whose mRNA decay patterns are shown in Fig. Fig.22 constituted about 0.07% of the B. subtilis genome. We estimated that approximately 150 to 200 different RNA species were detected by probes specific for these seven genes. It would be expected that similar patterns would be observed for any gene that is transcribed under the growth conditions used; many of the genes are in operons, which would likely result in even more complex patterns. Thus, it is likely that many thousands of such decay intermediates are present in the PNPase-deficient strain at any time. As these fragments begin at the 5′ end, they would represent broken mRNAs that are templates for initiation of translation but do not contain a stop codon for the release of the ribosome. We hypothesized that the pool of functional ribosomes would be severely depleted by the quantity of broken mRNAs in the pnpA strain unless there was increased activity of the tmRNA system, which unloads ribosomes from broken mRNA fragments by providing a template for trans translation (19, 27, 36). trans translation of tmRNA (encoded by the ssrA gene) results in addition of a 15-amino-acid peptide tag, which includes an Ala-Ala sequence at the carboxy terminus that is a signal for proteolysis (35).
Wild-type and pnpA mutant strains were constructed that contained the ssrA gene under control of the pspac promoter (26). The pnpA pspac-ssrA strain grew very poorly in the absence of IPTG, so much so that it was difficult to obtain a reliable doubling time when grown in liquid media. To estimate differences in growth rates, strains were grown overnight in LB liquid medium in the presence of IPTG, and the cells were washed, diluted, and plated on LB solid media with or without IPTG present. Colony sizes were measured at various times after plating. The results shown in Fig. Fig.7A7A are after 24 h of growth at 37°C. For the pnpA+ strain, the colonies were slightly larger in the presence of IPTG, likely reflecting an improved fitness when the tmRNA system was active. For the pnpA strain, we observed large (4-mm) and small (1.1-mm) colonies on the plates containing IPTG. In the absence of IPTG (i.e., no ssrA expression), two colony types were also observed, but both were very small, and only a slight difference in colony size was found (Fig. (Fig.7A).7A). The same growth differences were observed in a pnpA strain that contained the protease-resistant Asp-Asp sequence at the end of the ssrA-encoded peptide tag, rather than the wild-type Ala-Ala sequence (data not shown). This latter result indicated that the proteolysis of truncated proteins via tmRNA peptide tagging was not as important for cell growth as the tmRNA-mediated release of ribosomes itself, as has been found for other organisms (36).
The large and small colonies that arose when the pnpA strain was grown in the presence of IPTG were of interest. Examination of the frozen stock used to grow up the strain for plating revealed that cells of the large-colony type were present at a frequency of 10−3. Large- and small-colony types were propagated as separate strains, and the pspac-ssrA locus was amplified and sequenced. A point mutation in the lac operator sequence was found in the large-colony strain, changing the seventh position of the operator from an AT to a GC base pair (Fig. (Fig.7B).7B). The expression of ssrA RNA in the large- and small-colony strains was analyzed by Northern blot analysis. RNA isolation was from strains that were grown in the absence of IPTG overnight, diluted to a ratio of 1:20, and then grown with or without IPTG until mid-logarithmic phase. The results in Fig. Fig.7C7C show that, relative to that in the wild-type strain, expression of ssrA RNA is increased slightly in the small-colony strain but is increased significantly in the large-colony strain in the absence of IPTG (4.7-fold relative to that in the uninduced wild type) and even more so in the presence of IPTG (7.3-fold relative to that in the induced wild type). We suppose that at some point in the growth of the pnpA strain transformed with the pspac-ssrA construct, the absence of ssrA expression put a stress on cells such that a mutation that allowed improved ssrA transcription was selected.
To observe peptide tagging directly, additional wild-type and pnpA strains were constructed with a version of the ssrA peptide tag coding sequence in which six of the codons were replaced with histidine codons (Fig. (Fig.7D).7D). The final two codons of the peptide tag were either the wild-type Ala-Ala (AA) sequence or the proteolysis-resistant Asp-Asp (DD) sequence (16). Strains were grown in the presence or absence of IPTG to induce ssrA expression, and tagged peptides were detected by Western blot analysis using an anti-His antibody (Fig. (Fig.7E).7E). As expected, no tagged peptide was observed in strains containing the ssrA RNA with the AA terminal sequence. In the wild-type strain containing the ssrA RNA with the DD end, tagged proteins were observed. In the strain with a deletion of pnpA, the level of tagged proteins was severalfold higher than in the wild-type strain. Quantitative Western blot analysis gave an estimate of an approximately threefold-higher level of tagged protein in the strain with a deletion of pnpA (data not shown). In the experiment shown in Fig. Fig.7E,7E, no tagged peptide was detected in the pnpA strain with the pspac-ssrA construct if IPTG was not added (i.e., ssrA transcription was not induced). In some experiments (not shown), tagged peptide could be detected even in the absence of IPTG, and this was likely due to the sporadic occurrence of lac operator mutants in the pnpA strain.
Analysis of the decay pattern of specific mRNAs in a variety of RNase mutant backgrounds has allowed some conclusions regarding the role of the four known B. subtilis 3′-to-5′ exoribonucleases in mRNA decay. In the analysis of veg mRNA and rpsO mRNA, the only RNA detected at significant levels in the wild type and in the triple mutant that contained only PNPase was the full-length mRNA (Fig. (Fig.1,1, lanes 1 and 3, and Fig. Fig.5B,5B, lanes 1 and 3). Furthermore, analysis of the decay pattern of small, monocistronic mRNAs in wild-type and PNPase-deficient strains (Fig. (Fig.2)2) showed that the absence of PNPase alone resulted in an accumulation of mRNA decay fragments that was nearly undetectable in the presence of PNPase. This suggested that PNPase was the major activity required for the complete turnover of mRNA. Yet, despite the dominance of PNPase in degrading mRNA, the half-life of full-length rpsO mRNA in the wild-type strain was similar to that in the pnpA mutant strain (Fig. (Fig.4).4). This striking result is important in considering the mechanism of mRNA decay initiation in B. subtilis. Studies on the effect of 5′-proximal elements on mRNA half-life (6) point to the 5′ end as the major determinant in mRNA half-life. This suggests the existence of an as yet unidentified RNase E-like activity in B. subtilis which would initiate decay by endonucleolytic cleavage, as reviewed in the introduction. Our results here further support this hypothesis. If 3′-to-5′ exoribonucleolytic activity from the native 3′ end is a major element in the initiation of decay, we would expect that knocking out PNPase, the dominant 3′-to-5′ exonuclease, would result in an increased concentration of full-length mRNA at steady state, as well as an increased mRNA half-life. The fact that this was not observed suggests that decay initiation by endonuclease cleavage was still occurring normally in the pnpA strain, but that the resultant decay intermediates were not being degraded efficiently. (As we have not measured rpsO mRNA half-life in the full set of RNase mutants, there is still the formal possibility that the initiation of decay does occur from the 3′ end, but by a 3′-to-5′ exoribonuclease other than PNPase. However, we consider this possibility unlikely, since it would be difficult to explain how other exoribonucleases that are blocked by upstream secondary structure would be able to degrade past the strong terminator structure.)
Another result that bears on the mechanism of decay initiation is the similarity between the steady-state decay pattern detected by a 215-nt riboprobe complementary to most of the veg coding sequence (Fig. (Fig.1)1) and that detected by a 21-nt DNA oligomer complementary to positions 4 to 24 of the veg transcription unit (Fig. (Fig.2).2). This similarity demonstrated that only 5′-proximal fragments were observed for veg mRNA. The smallest decay fragments detected by all seven gene probes were between 100 and 150 nt long. Not shown in Fig. Fig.22 is the bottom of the gel, in which a 50-nt marker band was present. No fragments were detected between 40 and 100 nt. This result is compatible with the initiation of decay by an endonucleolytic cleavage distal (more than 100 nt) to the 5′ end. It is also possible that the initial cleavage occurs closer to the 5′ end but that RNA structures that block 3′-to-5′ processivity, and give rise to decay intermediates (see below), are not present near the 5′ end.
We chose the rpsO mRNA for further study based on the absence of detectable decay fragments in the wild type, the intensity of the 180-nt band, and the relatively uncomplicated decay pattern (Fig. (Fig.2).2). (Although the decay of E. coli rpsO mRNA has been studied thoroughly by Régnier’s group [reference 24 and references therein], this does not provide a model for B. subtilis, since there is no conservation at the nucleic acid sequence level between the E. coli and B. subtilis rpsO genes.) Two of the three prominent decay intermediates with sizes of between 100 and 200 nt were mapped to the base of predicted local secondary structures (Fig. 3A and B). There was a qualitative correlation between the amount of accumulated fragments (as measured by intensities of the bands) and the predicted strength of the secondary structure, with the 180-nt set mapping at the base of a structure with a ΔG° of −10.6 kcal/mol and the 133-nt set mapping at the base of a structure with a ΔG° of −3.9 kcal/mol. The 5′ end of E. coli rpsO mRNA has been shown to be a site of translational regulation based on an alternative stem-loop or pseudoknot structure (30). A similar form of regulation involving pseudoknot formation has been predicted for the B. subtilis rpsO mRNA (32). The pseudoknot structure for B. subtilis rpsO mRNA is predicted to end at nucleotide 97, which fits well with the observed 102-nt set of decay fragments.
Thus, blocks to rpsO mRNA decay in the 3′-to-5′ direction by ribonucleases other than PNPase correspond to the downstream sides of predicted secondary structures. The simplest model for rpsO mRNA decay is that endonuclease cleavages at sites in the coding sequence are followed by 3′-to-5′ exonucleolytic decay by PNPase. The indicated secondary structures are not an obstacle to PNPase, so that decay intermediates do not accumulate in the wild type or in the triple mutant strain containing PNPase. In strains lacking PNPase, however, decay intermediates whose 3′ ends map to the downstream sides of secondary structures do accumulate. The difference between PNPase and the other 3′-to-5′ exonucleases may be due to the superior processivity of PNPase or an interaction between PNPase and a poly(A) polymerase activity that would enhance the degradation of stem-loop structures by cycles of polyadenylation and 3′-to-5′ degradation (5). Further study, using 3′-proximal probes, will be required to determine how the extreme 3′ end of rpsO mRNA is degraded, since the transcription terminator structure (ΔG° = −14.6 kcal/mol) is more stable than even the predicted structure that gives rise to the 180-nt decay intermediate.
Results with double and triple RNase mutants were revealing in terms of the secondary roles of exoribonucleases other than PNPase. The data in Fig. Fig.5A,5A, lanes 2, 4, and 5 (quantitation in Fig. Fig.5C),5C), showed that there was little difference in the amounts of the three decay intermediates in strains that were missing YhaM or RNase PH in addition to PNPase. On the other hand, the mutant that was missing PNPase and RNase R (Fig. (Fig.5A,5A, lane 3) showed a substantial increase in the amount of the 102-nt fragment, as well as significant increases in the levels of 180-nt and 133-nt fragments. These results suggested that, in the absence of PNPase, RNase R was capable of degrading past secondary structure. The absence of RNase R in the pnpA background left little exoribonuclease processivity to degrade through the predicted structures. The reason for the disproportionately large increase in the 102-nt fragment set is not clear, although it is possible that the absence of RNase R could affect regulation of rpsO expression, which is a function of the 5′-proximal pseudoknot formation. Since it has been shown that RNase R is required by E. coli for the quality control of rRNA (3), we speculate that expression of rpsO (encoding a ribosomal protein) is down-regulated in a strain that has an imbalance in fully processed rRNA content, and thus formation of the inhibitory pseudoknot is enhanced, resulting in a greater block to RNase processivity.
The ability of RNase R to degrade past secondary structure was demonstrated in a positive way using the triple mutant strains. In the strain containing only RNase R, few decay intermediates were observed (Fig. (Fig.5B,5B, lane 4). Thus, RNase R is also capable of degrading mRNA in vivo.
Surprisingly, the amount of decay fragment that accumulated when only RNase R was present (Fig. (Fig.5B,5B, lane 4) was significantly less than that which accumulated when RNase R was present with RNase PH (Fig. (Fig.5A,5A, lane 4) or with YhaM (Fig. (Fig.5A,5A, lane 5). One might expect that the presence of an additional exoribonuclease would correlate with less, not more, decay intermediates. We hypothesize that the ability of RNase R to degrade past secondary structure may be compromised when RNase PH or YhaM is present. The less-processive RNase PH and YhaM might hydrolyze nucleotides at the downstream side of a stem structure, leaving few single-stranded nucleotides at which RNase R can bind. We showed previously in vitro that RNase R could degrade past a strong stem structure that was followed by a 40-nt single-stranded tail, but was unable to do so when the tail consisted of only 12 nt (28). Interestingly, the growth rate for the triple mutant strain containing only RNase R was higher than those for all other strains containing more than one RNase mutation and was similar to the growth rate for the single pnpA mutant (Table (Table1).1). This result may also be a reflection of the superior ability of RNase R to degrade mRNA when other exoribonucleases are not present.
The involvement of YhaM in mRNA decay was indicated by the decreased accumulation of decay products for the strain containing YhaM alone, relative to that for the strain containing RNase PH alone (compare Fig. Fig.5B,5B, lanes 5 and 6, and Fig. Fig.6,6, lanes 2 and 4). The suggestion that any of the known exoribonucleases can participate in mRNA decay was evident as well in the result obtained from the quadruple mutant, which was lacking all four of the 3′-to-5′ exoribonucleases (Fig. (Fig.6,6, lane 5). In this case, decay intermediates accumulated to levels similar to that for the triple mutant strain containing RNase PH only (Fig. (Fig.6,6, lane 2), but the intermediates were larger than those of the triple mutant. Whatever activity is responsible for 3′-to-5′ degradation in the quadruple mutant apparently cannot approach stem structures as closely as the other ribonucleases can. The conspicuously low growth rate of the quadruple mutant (Table (Table1)1) also indicates that the presence of any one of the four exoribonucleases is sufficient to support a growth rate that is closer to that of the wild type than to that of the quadruple mutant.
A considerable amount of mRNA decay fragments accumulated in the pnpA strain. In the case of rpsO, we found that 81% of the total RNA detected was shorter than full length (average of four experiments). While the results shown in Fig. Fig.22 suggest that this fraction might be lower for other genes, the burden of broken mRNAs is clearly substantial in the pnpA strain, and even greater in strains with additional RNase mutations. As expected, the results in Fig. Fig.77 show that the tmRNA system operates at a higher level in the pnpA strain than in the wild type. Somewhat surprisingly, though, we detected only a threefold increase in the level of tagged peptide in the pnpA strain relative to that in the wild type. We suggest that, due to the efficiency of transcription and of mRNA decay, mRNA fragments are relatively rare in a wild-type strain and that the tmRNA system is designed merely to avoid the slight depletion of free ribosomes that would result from the translation of such fragments. The level of accumulated mRNA decay fragments in a PNPase mutant, however, is so high that it would overwhelm the tmRNA system, and the level of peptide tagging would not reflect the level of mRNA fragment accumulation. Other components of the tmRNA system besides ssrA RNA, such as SmpB protein or the requirement for charging by alanyl-tRNA synthetase, may be limiting. It would also be of interest to measure the level of peptide tagging in the large-colony-type pnpA pspac-ssrA strain, which had almost fivefold more ssrA expression in the presence of IPTG than did the small-colony-type strain (Fig. (Fig.7C7C).
Although Muto et al. found previously that ssrA expression is induced under various stress conditions—up to 10-fold during heat shock and 4- to 6-fold in the presence of ethanol or cadmium chloride (26)—we could not detect a high level of induction in the pnpA strain. By the use of Northern blot analysis of three independent RNA isolations, the levels of ssrA RNA in wild-type and pnpA mutant strains were compared (data not shown). In each case, there was a slight increase in the amount of ssrA RNA (1.2- to 1.4-fold). Somewhat higher levels of ssrA RNA were found in the multiply mutant strains (1.5- to 1.8-fold increases). The weak responses we observed may be similar to the approximately 1.5- to 2-fold inductions found in the presence of elevated sodium or sucrose (26). Thus, it appears that there is no robust mechanism to induce ssrA expression in response to an accumulation of mRNA decay fragments.
It has been shown for E. coli that the tmRNA system facilitates the degradation of truncated crp mRNA, presumably by releasing protective ribosomes at the 3′ end of the mRNA fragments (37). In preliminary experiments, comparing pnpA, ssrA, and pnpA ssrA mutant strains, we have found an increased accumulation of ermC mRNA decay fragments in the pnpA ssrA double mutant, relative to that in either single mutant. However, we have not observed significant differences in mRNA decay half-lives between these strains (D. H. Bechhofer, unpublished data). Based on the detailed information about rpsO mRNA obtained in the present study, it may be informative to study rpsO mRNA decay in strains that contain both RNase and ssrA mutations.
More work needs to be done to understand the individual functions of exoribonucleases that have similar activities. The apparent redundancy of exoribonuclease activities is even greater in E. coli, where there are eight exoribonucleases (39), than it is in B. subtilis. On the other hand, Mycoplasma appears to have only one 3′-to-5′ exoribonuclease, RNase R (39). Our finding that RNase R can participate in mRNA turnover suggests that this enzyme is likely to do the same in organisms that do not have PNPase. Although the primary function of RNase R in B. subtilis might be in the quality control of rRNA, it might also serve as the chief backup mRNA decay enzyme and perhaps play an important role during growth in phosphate-limiting conditions that might reduce the activity of the phosphorolytic PNPase. Further in vitro work with these enzymes will be required to understand how they differ in their abilities to degrade past RNA secondary structure. In addition to requiring the exoribonucleases noted in the present work, the degradation of mRNA in B. subtilis requires the participation of a 5′-end-dependent endoribonuclease (6) and likely that of a poly(A) polymerase and a helicase. Identification of the genes encoding such activities will be required to gain a better understanding of the B. subtilis mRNA decay pathway.
This work was supported by Public Health Service grant GM-48804 from the National Institutes of Health to D.H.B. and grants-in-aid from the Ministry of Education, Science, Sports, and Culture of Japan (no. 14035201) and the Japan Society for the Promotion of Science (no. 14380322) to A.M.