- Top of page
- Experimental procedures
- Supporting Information
The presence of very different sets of enzymes, and in particular the presence of RNase E and RNase J, has been used to explain significant differences in RNA metabolism between the two model organisms Escherichia coli and Bacillus subtilis. However, these studies might have somewhat polarized our view of RNA metabolism. Here, we identified a RNase J in Mycobacterium smegmatis that has both 5′-3′ exo- and endonucleolytic activity. This enzyme coexists with RNase E in this organism, a configuration that enabled us to study how these two key nucleases collaborate. We demonstrate that RNase E is responsible for the processing of the furA-katG transcript in M. smegmatis and that both RNase E and RNase J are involved in the 5′ end processing of all ribosomal RNAs. In contrast to B. subtilis, the activity of RNase J, although required in vivo for 23S rRNA maturation, is not essential in M. smegmatis. We show that the pathways for ribosomal RNA maturation in M. smegmatis are quite different from those observed in E. coli and in B. subtilis. Studying organisms containing different combinations of key ribonucleases can thus significantly broaden our view of the possible strategies that exist to direct RNA metabolism.
- Top of page
- Experimental procedures
- Supporting Information
Ribonucleases are important enzymes that intervene in the degradation, processing and quality control of all RNA species in the cell. These vital processes play an essential role in the control of gene expression and in the adaptation to environmental changes. Current knowledge suggests that the RNases E, J and Y are key players in eubacterial RNA metabolism. All known eubacteria contain at least one of these three ribonucleases and several of them have all three (Laalami and Putzer, 2011).
mRNA degradation/maturation has been extensively studied in Escherichia coli and Bacillus subtilis. Important differences in the mechanisms underlying mRNA metabolism have been demonstrated between these two model organisms and reflect the fact that they contain quite different sets of RNases. In the Gram-negative bacterium E. coli, RNase E is the key enzyme initiating mRNA decay. It generally catalyses a rate limiting endonucleolytic cleavage within an AU rich single-stranded region, causing the rapid 3′-5′ exonucleolytic degradation of the fragment upstream of the cleavage site (Carpousis et al., 2009). The strong preference of RNase E for 5′ monophosphorylated RNA substrates (Mackie, 1998) stimulates the subsequent endonucleolytic cleavage of the 3′ cleavage products. In some cases, this preference allows an alternative decay pathway in which internal cleavage by RNase E is triggered by prior conversion of the 5′ terminal triphosphate to a monophosphate by the pyrophosphohydrolase RppH (Deana et al., 2008). Inactivation of RNase E increases global mRNA half-life, and the non-catalytic C-terminal half of RNase E serves as the scaffold for the degradosome complex (Carpousis et al., 2009). RNase E also plays a major role in the maturation of ribosomal RNA (Deutscher, 2009). RNase G, a shorter paralogue of RNase E, which coexists with RNase E only in the β- and γ-subdivisions of the Proteobacteria (Condon and Putzer, 2002), plays a minor role in mRNA metabolism in E. coli (Arraiano et al., 2010). RNase E orthologues in other species are classified as RNase E/G enzymes because they usually share significant similarity with the catalytic N-terminal half of E. coli RNase E and RNase G (Condon and Putzer, 2002).
Despite its central role in mRNA stability in E. coli, RNase E is absent from many bacterial species including most Firmicutes, such as B. subtilis, and even some proteobacteria (Condon and Putzer, 2002; Laalami and Putzer, 2011). These bacteria instead contain the dual activity endo-/5′-3′ exoribonuclease RNase J (Even et al., 2005; Mathy et al., 2007) and/or the endonuclease RNase Y (Shahbabian et al., 2009). For example, in B. subtilis the essential RNase J1 (rnjA) and its paralogue RNase J2 (rnjB) have endonucleolytic cleavage specificity similar to that of RNase E (Even et al., 2005). RNase J1 and to a lesser extent RNase J2 also possess a 5′-3′ exonucleolytic activity that strongly prefers a substrate with a single phosphate at the 5′ end (Li de la Sierra-Gallay et al., 2008; Mathy et al., 2010). This property fitted well with the observation that, in B. subtilis, obstacles such as a stalled ribosome or secondary structure near the 5′ end can strongly stabilize downstream RNA for several kilobases indicative of a strong 5′-3′ directionality for mRNA degradation (Bechhofer and Dubnau, 1987; Agaisse and Lereclus, 1996; Hambraeus et al., 2002). Indeed, RNases J1 and J2, which have overlapping substrate specificities, together affect the expression levels of hundreds of genes (Mäder et al., 2008). The exonucleolytic activity of RNase J1 also participates in 16S rRNA maturation (Mathy et al., 2007). On the other hand, in a rnjA/rnjB double mutant (RNase J1 was depleted) global mRNA half-life was only slightly increased (Even et al., 2005). However, 5′ exonucleolytic degradation of a native transcript can be stimulated by a recently discovered RppH-like activity in B. subtilis (Richards et al., 2011). To what extent pyrophosphate removal from the 5′ end of original transcripts affects mRNA metabolism in B. subtilis remains to be analysed. The most significant effect on mRNA turnover in this organism can be attributed to RNase Y, a recently characterized novel endoribonuclease whose depletion increases bulk mRNA stability in B. subtilis to a similar extent as that of RNase E mutants in E. coli (Shahbabian et al., 2009). Transcriptome studies confirmed the important role of RNase Y in determining the stability of hundreds of mRNA species (Lehnik-Habrink et al., 2011) (our unpubl. results). In addition, the activity of RNase Y is significantly increased by the presence of a 5′-P group on the RNA substrate, another similarity shared with RNase E (Shahbabian et al., 2009). These observations raised the possibility that, after all, mRNA processing and degradation might be more similar between B. subtilis and E. coli than currently assumed, with an endonucleolytic cleavage being the key step in initiating mRNA decay.
In eubacteria the RNases E/G, J and Y probably represent the key enzymes initiating mRNA decay (Laalami and Putzer, 2011). In this context, mycobacteria are especially interesting to study because a genome sequence analysis shows that they contain both an RNase E/G and an RNase J but no RNase Y (Table 1). This combination has not been found in any of the organisms where RNA metabolism has been studied in some detail.
Table 1. Occurrence of ribonucleases.
| ||E. coli||B. subtilis||M. smegmatis||M. tuberculosis|
|Endoribonucleases|| || || || |
| RNase E||+||−||+||+|
| RNase G||+||−||−||−|
| RNase III||+||+||+||+|
| Mini III||−||+||−||−|
| RNase M5||−||+||−||−|
| RNase P||+||+||+||+|
| RNase Z||+||+||+||+|
| RNase I||+||−||−||−|
| RNase Y||−||+||−||−|
| RNase J||−||+||+||+|
|Exoribonucleases|| || || || |
| RNase J||−||+||+||+|
| RNase T||+||−||−||−|
| RNase R||+||+||−||−|
| RNase II||+||−||−||−|
| RNase PH||−||+||+||+|
Mycobacteria can be divided into two classes: the slow-growers that include pathogens like Mycobacterium tuberculosis and Mycobacterium leprae and the fast-growers like Mycobacterium smegmatis. It is noteworthy that the putative RNases present in M. tuberculosis and M. smegmatis are the same (Table 1). This implies that RNA metabolism is likely to be very similar and studies in M. smegmatis, which is nonpathogenic and simpler to handle than M. tuberculosis, should be pertinent for both species.
Mycobacterium smegmatis has functional orthologues for both RNase E (Zeller et al., 2007) and, as we show here, RNase J. We demonstrate that the RNase J orthologue is a dual activity 5′ exo- and endonuclease very similar to RNase J1 of B. subtilis. To analyse in more detail the relative roles of RNase J and RNase E in vivo in M. smegmatis, we created mutant strains and studied their effect on the maturation of the ribosomal RNA and processing of the furA-katG operon mRNA. Our results show that rRNA maturation in M. smegmatis is different from that in both E. coli and B. subtilis.
- Top of page
- Experimental procedures
- Supporting Information
Mycobacterial RNA metabolism has not been well studied but these high GC Gram-positive organisms contain an RNase E/G type enzyme and a putative RNase J orthologue. In addition, a compilation of potential RNase orthologues (Table 1) suggests that mycobacteria constitute an interesting class of bacteria that combine key enzymes present in E. coli and B. subtilis. Sequence and 3D structure comparisons of the M. smegmatis protein 2685 with known RNases J of T. thermophilus and B. subtilis (Figs S1 and S2) showed good overall similarities (55% and 59%, respectively, with B. subtilis RNases J1 and J2) and the conservation of all important motifs, notably the five classical β-lactamase motifs, the three β-CASP motifs and the amino acids constituting the mononucleotide binding pocket (Li de la Sierra-Gallay et al., 2008). However, this comparison provided no clues as to the major enzymatic activity. It was important to verify that the single putative M. smegmatis RNase J orthologue (MSMEG_2685) actually is an RNase and whether it has endo- and/or exoribonucleolytic activities. In fact, in B. subtilis both RNase J1 and RNase J2 show a similar endoribonucleolytic activity (Even et al., 2005), but RNase J2 contrary to RNase J1 has a very weak 5′-3′ exoribonucleolytic activity (Mathy et al., 2010). In contrast, archeal RNase J orthologues appear to have 5′ exonuclease but no endonuclease activity (Clouet-d'Orval et al., 2010; Hasenohrl et al., 2011). In vitro assays with the purified mycobacterial protein showed that it actually has both activities, 5′-3′ exo- and endonucleolytic. The exonucleolytic activity is comparable with that of B. subtilis J1 when assayed on the same 5′ monophosphorylated B. subtilis thrS leader substrate (Li de la Sierra-Gallay et al., 2008). The Asp85Lys and His86Ala mutant protein has almost completely lost its 5′ exonuclease activity, confirming that the activity attributed to M. smegmatis RNase J is genuine and not due to contamination during protein purification.
Using the B. subtilis thrS leader region in its 5′-triphosphorylated form we showed that M. smegmatis RNase J also has endonucleolytic activity. The major cleavage site was found at position +245 immediately upstream of the terminator structure, which is also the principal site of cleavage by B. subtilis RNase J1/J2, as described previously (Even et al., 2005). Interestingly, significant cleavage by the M. smegmatis as well as the B. subtilis enzyme was also observed within the first four to five 5′ proximal nucleotides that are part of a region previously shown to be single-stranded (Luo et al., 1998). RNase J1 has been shown to cleave single-stranded sequences non-specifically under high enzyme concentrations (Daou-Chabo and Condon, 2009). However, under the conditions used here RNase J cleaved the thrS leader with a clear bias towards two distinct sites, i.e. upstream of the leader terminator and at sites near the 5′ end. Insertion of three to ten A residues after position +2 of the thrS RNA also did not create new cleavage sites. On the contrary, even in these modified thrS leader substrates, with extended 5′ single-stranded regions, cleavage by RNase J in vitro occurred predominantly within the first four 5′ proximal nucleotides (Fig. S3).
Interestingly, the Asp85Lys and His86Ala mutant protein that is almost devoid of 5′ exonucleolytic activity retained considerable endonucleolytic activity. This was evident not only for the cleavage of the 5′ proximal residues but also for the cleavage upstream of the terminator structure (data not shown). Because the double mutation concerns two residues that co-ordinate to one of the catalytic Zn+2 ions, it is unlikely that the mutant protein binds both Zn+2 ions. This configuration is reminiscent of that encountered in RNase J2. This enzyme has three active site residues that are not conserved with respect to the paralogous RNase J1, all of which are involved in the co-ordination of the equivalent Zn+2 ion. Similar to the mutated mycobacterial protein, B. subtilis RNase J2 has endo- but almost no exonucleolytic activity.
The capacity of RNase J to endonucleolytically remove the 5′ terminal triphosphorylated nucleotide of a native transcript could play an important role in initiating mRNA decay by creating an entry site for the 5′ exonucleolytic activity of RNase J. In the most straightforward model, RNase J would simply switch from endo- to exonucleolytic mode without releasing the RNA (Li de la Sierra-Gallay et al., 2008). Even though pyrophosphate removal by an RppH-like activity has also been observed in B. subtilis (Richards et al., 2011), RNase J cleavage close to the 5′ end might well constitute an alternative pathway to initiate 5′ exonucleolytic decay of native transcripts.
The mode of action by which RNase J releases the first nucleotide(s) must be endonucleolytic, because co-ordination of the γ-phosphate of the terminal nucleotide in the phosphate binding pocket would place the scissile bond out of phase (Li de la Sierra-Gallay et al., 2008). The recent discovery that the RNA entry channel of RNase J actually extends past the active site (Dorléans et al., 2011; Newman et al., 2011) created a rationale for how the catalytic cleft could directly accommodate an RNA in endonucleolytic mode (Fig. S5). Direct access to the site of cleavage is thus likely to be the principal mode of action to cleave an RNA far from the 5′ end.
In what concerns cleavage very close to the 5′ end we would like to suggest an alternative model in which RNase J would function as a ‘sliding endonuclease’. In this mechanism, the native triphosphorylated RNA enters the RNA entry channel of the dimer and is threaded towards the active site in the same way as 5′ monophosphorylated RNA. Because the 5′ PPP moiety cannot interact productively with the phosphate binding pocket, we propose that the RNA can slide past the catalytic centre and be cleaved endonucleolytically at the downstream nucleotides. This sliding mode would be expected to be very sensitive to secondary structure and probably not go beyond a few nucleotides. Indeed, the cleavable phosphodiester fits into the monophosphate binding pocket much in the same way as the 5′ terminal monophosphate (Dorléans et al., 2011). It is therefore likely that once the 5′ PPP group has slided past the active site one the following phosphodiester groups approaching the catalytic centre should readily snap into the phosphate binding pocket. This would explain why the observed endonucleolytic cleavage is essentially observed for the 5′ proximal nucleotides.
RNase E and RNase J1 are essential for viability in E. coli and B. subtilis respectively. These ribonucleases are clearly very important for RNA metabolism but the precise reason for their essentiality has not been established, neither for RNase E nor for RNase J1. In M. smegmatis RNase E is essential while RNase J is not, implying that the 5′-3′ exonucleolytic activity of RNase J is dispensable. A similar configuration exists in the α-proteobacterium Sinorhizobium where the RNase J orthologue is dispensable. However, it is not known whether RNase J in this organism has 5′ exo-, endo- or both nucleolytic activities (Madhugiri and Evguenieva-Hackenberg, 2009).
Processing of mRNA precursors has also been observed in Mycobacteria but available data are extremely scarce. In fact, the sole known case is the furA-katG operon encoding a Fur-like transcriptional repressor (furA) and a catalase peroxidase (katG) involved in the activation of isoniazide, an antitubercular drug (Milano et al., 2001; Sala et al., 2003). The major katG transcript in M. smegmatis is produced by processing of the native bicistronic furA-katG mRNA close to the end of the furA ORF (Sala et al., 2008). Here we showed that the ribonuclease responsable for this cleavage is RNase E. Processing of the bicistronic furA-katG mRNA is also observed in M. tuberculosis and, similarly, prevented by a depletion of M. tuberculosis RNase E (data not shown). Processing occurs 17 nucleotides upstream of the furA stop codon 3′ to two A residues in a single-stranded region between two potential secondary structures. This is similar to the sequence and structural context identified for ectopic cleavage sites for E. coli RNase E (Mackie, 1992; Baker and Mackie, 2003). Maturation of the native transcript thus generates a relatively stable katG mRNA and an unstable furA transcript. Because FurA negatively autoregulates its own transcription from pfurA, this allows the cell to produce a sufficient amount of KatG peroxidase and at the same time keep FurA repressor biosynthesis low.
Processing of ribosomal RNAs
In Mycobacteria, rRNA processing has not been studied in much detail (Ji et al., 1994a,b) and the nucleases involved were unknown. The M. tuberculosis single rRNA operon (rrnA) and the two M. smegmatis operons (rrnA and rrnB) contain the genes for the 16S, 23S and 5S rRNA, co-transcribed as a single precursor molecule (Ji et al., 1994a; Gonzalez-y-Merchand et al., 1996). Co- or post-transcriptional folding of the precursor rRNAs is predicted to involve extensive base pairing between the leader and spacer-1 region (16S), spacer-1 and spacer-2 (23S and 5S) as suggested by the secondary structures predicted by the M-fold algorithm (Fig. 6). The M. smegmatis rrnA and rrnB operons have identical sequences downstream of position −165 with respect to the 16S rRNA structural gene. Thus the folded precursor structures that are the substrates for a series of maturation reactions to produce the mature rRNAs should be the same for both operons. The large processing stalks are thus a common feature shared by most if not all eubacteria. They are generally cleaved by the double-stranded specific RNase III or a related nuclease that physically separate the individual rRNA precursors. We have not formally shown that RNase III is responsible for the initial cleavage of these processing stalks in M. smegmatis. However, we have identified 5′ ends in the different rRNA precursors that when the stalks are folded, are in proximity on the opposite strand and near a bulge. Such a configuration is very characteristic of an RNase III cleavage site. Thus, initial separation of the 16S, 23S and 5S rRNA most likely occurs through two RNase III cleavages (Fig. 6).
RNase E was found to be involved in 16S rRNA maturation. Its partial depletion caused a twofold increase in the putative RNase III processing product (position +148) but also a significant accumulation of the +36 nt precursor. This is consistent with RNase E being able to create the mature 5′ end of 16S rRNA. The putative cleavage site is composed of a U rich single-stranded region (Fig. 6A) compatible with the known requirements for RNase E cleavage in E. coli (Ehretsmann et al., 1992; McDowall et al., 1994) and M. tuberculosis (Zeller et al., 2007). In contrast, the complete inactivation of RNase J had no effect on 16S rRNA except when RNase E was depleted at the same time. The stabilization of two precursors with the 5′ end close to the mature 5′ end, suggested that the RNase J collaborates with RNase E to complete the maturation of the 16S rRNA.
We propose the following model for 16S rRNA maturation (Fig. 7A): (i) ribonucleolytic cleavages at +148 and +1977/79, likely performed by RNase III, release the 16S rRNA precursor from the 30S original transcript; (ii) an unidentified nuclease cleaves at position +265 leaving 36 nt upstream of the mature 16S rRNA 5′ end; and (iii) this shortened precursor is then processed by RNase E and RNase J to create the mature 16S rRNA 5′ end.
When we compare M. smegmatis 16S rRNA maturation with that of E. coli and B. subtilis (Fig. 7A) we note that in all cases, maturation appears to occur in a two-step mechanism: the initial RNase III precursors are processed by RNase E (E. coli, +66 nt) or unknown endoribonucleases (B. subtilis, +38 nt and M. smegmatis, +36 nt) apparently to create a new 5′ end free from the processing stalk that could increase the efficiency of the final maturation steps at both the 5′ and 3′ extremities.
RNase J plays a major role in 23S rRNA maturation: in the Δrnj mutant strain no mature 23S rRNA was present in the cell. Because the strain is viable, the absence of a fully mature 23S rRNA does not compromise ribosome activity. This indicates that the precursor molecule, although 14 nucleotides longer at its 5′ end than the mature 23S rRNA, is able to efficiently assemble into a functional ribosome. It is likely that the +14 nt precursor is then trimmed by the RNase J 5′ exonuclease activity, but we cannot rule out an endonucleolytic cleavage by RNase J to create the mature 23S rRNA in a single step.
Ribonuclease E is also involved in 23S rRNA maturation, although its role is less relevant: notably the RNase III generated precursor 23S rRNA (from cleavages at positions +148 and +1977/79) is increased when RNase E is depleted and the precursor pattern of the rnj/rne double mutant shows an additional band corresponding to +26 nt precursor (Fig. 5C, lane 4).
Maturation of the 23S rRNA in M. smegmatis is thus quite different when compared with both E. coli and B. subtilis (Fig. 7B). In fact, RNase J, the nuclease involved in the final steps of 23S rRNA maturation in M. smegmatis, does not participate in the processing of 23S rRNA in a B. subtilis wild-type strain (Redko et al., 2008).
In the absence of RNase J, maturation of 5S rRNA is normal, whereas in the RNase E mutant several precursors (91 nt and 37 nt from the mature 5′ end) accumulate. However, the fact that in the rnj/rne double mutant the amount of mature 5S rRNA is clearly reduced suggests that both ribonucleases are involved in its maturation. The data are compatible with a direct maturation of the 5S rRNA 5′ end by RNase E using the +37 nt precursor and possibly also longer precursors. However, it is likely that another nuclease besides RNase E is involved in the final processing steps. This unknown nuclease might also account for the +37 nt precursor. Again, the maturation pathway in M. smegmatis differs from 5S rRNA processing in E. coli and B. subtilis (Fig. 7C).
More data are needed to fully understand rRNA processing in Mycobacteria particularly at the 3′ end. Nevertheless, our experiments clearly show that the mechanisms and enzymes involved in the 5′ end processing of all three rRNAs are significantly different compared with what we know from the two model organisms E. coli and B. subtilis. From an evolutionary and mechanistic point of view it will be interesting to better understand how combinations of RNases present in a given organism influence and direct RNA metabolism.