The paralogous ribonucleases J1 and J2, recently identified in Bacillus subtilis, have both endoribonucleolytic and 5′-to-3′ exoribonucleolytic activities and participate in degradation and regulatory processing of mRNA. RNases J1 and J2 have partially overlapping target specificities, but only RNase J1 is essential for B. subtilis growth. Because mRNA decay is important in regulation of virulence factors of Streptococcus pyogenes (the group A streptococcus, GAS), we investigated the role of these newly described RNases in GAS. We found that conditional mutants for both RNases J1 and J2 require induction for growth, so we conclude that, unlike the case in B. subtilis, both of these RNases are essential for GAS growth, and therefore their functions are not redundant. We compared decay of representatives of the two classes of messages we had previously identified: Class I, which decay rapidly in exponential and stationary phase of growth (hasA and gyrA), and Class II, which are stable in stationary phase and exhibit a biphasic decay curve in exponential phase (sagA and sda). We report that RNases J1 and J2 affect the rate of decay of Class I messages and the length of the first phase in decay of Class II messages.
Streptococcus pyogenes, also known as the group A streptococcus or GAS, is a Gram-positive bacterium that is an exclusively human pathogen. Although GAS colonizes the skin and mucosal surfaces asymptomatically in up to 20% of the human population, it also produces multiple types of disease that vary from superficial infections of skin and throat to invasive and life-threatening infections, such as necrotizing fasciitis and streptococcal toxic shock syndrome (Tart et al., 2007; Cunningham, 2008). In addition, GAS infections can lead to complications such as rheumatic fever and rheumatic heart disease and to acute glomerulonephritis (Hahn et al., 2005; Chopra and Gulwani, 2007). The ability of GAS to produce different diseases and to infect the host asymptomatically results from its ability to regulate expression of virulence factors required for attachment to host tissues, evasion of the host immune response and spreading throughout the host (Churchward, 2007; Gryllos et al., 2007; Hondorp and McIver, 2007).
Because of its importance in disease progression, regulation of expression of virulence genes has been studied extensively and many proteins that regulate initiation of transcription have been identified (Churchward, 2007; McIver, 2009). However, in GAS, all regulatory mechanisms of transcripts are subsidiary to ‘growth phase regulation’ (McIver and Scott, 1997; Neely et al., 2003). Even if all known activators of a promoter are present and all repressors are absent, each transcript studied can only be detected at the correct phase of growth. Because the control of mRNA decay is an important mechanism of regulation of virulence proteins in bacteria, we began to investigate its role in growth phase regulation.
We found that mRNAs can be divided into two major classes in GAS depending on their stability in different growth phases (Barnett et al., 2007). Messages in Class I, which includes most of the mRNAs in GAS, are difficult to detect in stationary phase and decay rapidly in exponential phase (Barnett et al., 2007). When one of these mRNAs was expressed under a promoter active in stationary phase, it also decayed rapidly in stationary phase (see Barnett et al., 2007). Class II contains only a few messages. These are more abundant in stationary than in exponential phase because of dramatically increased stability in stationary phase. We suggested that Class II transcripts are not sensitive to the major GAS mRNA decay pathway. Furthermore, we proposed that ‘growth phase dependence’ of virulence factor gene expression, the primary regulatory mechanism that determines amounts of mRNA in GAS, might result from differential substrate specificity of the decay pathways (Barnett et al., 2007).
Decay pathways of mRNA have been studied in Escherichia coli and Bacillus subtilis, but little is known about mRNA decay in GAS. Decay of mRNA is controlled by multiple enzymes, and these enzymes are different in Gram-negative and Gram-positive bacteria. The central enzyme for mRNA decay in the Gram-negative bacterium E. coli is the endoribonuclease RNase E. This enzyme is essential for growth and, as a major component of the degradosome, is responsible for degradation of most mRNAs (Mudd et al., 1990; Babitzke and Kushner, 1991). The degradosome in E. coli is composed mainly of RNase E, a membrane-bound protein, helicase B, enolase and polynucleotide phosphorylase (PNPase) (Py et al., 1994; Vanzo et al., 1998; Khemici et al., 2008). RNase E targets single-stranded AU-rich RNA regions. In addition, RNase E has a preference for RNAs with 5′-monophosphate ends, although bacterial mRNA is synthesized with a triphosphate at its 5′-end (Mackie, 1998; Jiang and Belasco, 2004). Therefore, the first step, which is also the rate-limiting step, in mRNA decay for some messages in E. coli is the conversion of this 5′-triphosphate to a monophosphate by the pyrophosphohydrolase RppH (Celesnik et al., 2007; Deana et al., 2008). The 5′-P message that is generated is then cleaved endonucleolytically by RNase E and further digested by 3′-to-5′ exoribonucleases, including PNPase, RNase II and RNase R (Condon, 2007).
The ribonucleases J1 and J2, recently identified in the genomes of Gram-positive bacteria (Even et al., 2005), are functional homologues in B. subtilis to RNase E of E. coli, although these enzymes have no sequence similarity to RNase E (Even et al., 2005). RNases J1 and J2, which share 49% identical amino acid residues, have single-stranded target sites that are similar to sites recognized by RNase E in E. coli (Even et al., 2005). Very recently, a degradosome was identified in B. subtilis (Commichau et al., 2009). This complex appears to contain phosphofructokinase, enolase, PNPase, RNase J1 and the recently discovered membrane-bound endoribonuclease RNase Y. RNase Y is essential for growth and affects mRNA decay in B. subtilis, but its activities have not been well characterized yet. RNase J2 has been reported to be associated with the degradosome complex in B. subtilis through binding to RNase J1 (Commichau et al., 2009).
RNases J1 and J2 have endoribonucleolytic activity that may be independent of the phosphorylation state of the 5′-end of mRNA, so it is possible that endonucleolytic cleavage by these RNases may initiate mRNA decay (Even et al., 2005; Mathy et al., 2007; Li de la Sierra-Gallay et al., 2008). After endoribonucleolytic cleavage by RNases J1 and/or J2, the remaining mRNA fragments can be degraded by the 3′-to-5′ exoribonucleases PNPase, YhaM and RNase R, and/or the 5′-to-3′ exoribonuclease activity of RNases J1 and/or J2 (Condon, 2007; Mathy et al., 2007). Although in addition to endoribonucleolytic activity, RNases J1 and J2 are the first bacterial enzymes shown to have 5′-to-3′ exoribonuclease activity (Mathy et al., 2007), this activity is stimulated by the presence of a monophosphate at the 5′-end of an mRNA and thus is unlikely to be the initial step in mRNA decay. However, if there is a pyrophosphohydrolase in Gram-positive bacteria with a function like that of RppH of E. coli, mRNA decay may be initiated by conversion of the 5′-PPP of an mRNA to a 5′-P by this enzyme (Deikus et al., 2008). The 5′-P mRNA can then be degraded by the 5′-to-3′ exoribonuclease activity of J1 and/or J2.
Analysis of decay of many mRNAs suggested that RNases J1 and J2 have overlapping, but not identical, substrate specificities in B. subtilis (Even et al., 2005; Mader et al., 2008). The finding that RNase J1, but not RNase J2, is essential for growth of B. subtilis (Kobayashi et al., 2003) supports the idea that these enzymes have somewhat different roles in the metabolism of this organism.
In previous work in GAS, we began analysing the role of different ribonucleases by generating deletion mutants for all homologues to known 3′-to-5′ exoribonucleases (Barnett et al., 2007). We found that the exoribonucleases PNPase, YhaM and RNase R are not essential for growth in GAS as we could delete each of their genes individually. This indicates either that one of the 3′-to-5′ exoribonucleases can compensate for the absence of another or that there is an additional yet unidentified 3′-to-5′ exoribonuclease in GAS. However, we found that PNPase, and not YhaM or RNase R, is rate-limiting for decay of messages that belong to Class II (stable in stationary phase), including the messages for sagA and sda, which code for the important virulence factors streptolysin S and streptodornase (a DNase) respectively (Barnett et al., 2007). We also found that PNPase is not rate-limiting for decay of Class I (unstable) mRNAs, including the hasA message that codes for the first enzyme in the synthesis of the hyaluronic acid capsule (essential for GAS to produce significant infection), and the message for gyrA, encoding gyrase, a housekeeping enzyme (Barnett et al., 2007). These results suggest that Class I messages decay by a different pathway from Class II messages. Because the absence of PNPase only partially stabilizes Class II messages in exponential phase (T1/2 increased from 3–5 min to 20–25 min in the absence of PNPase) compared with their stationary phase stability (T1/2 > 100 min) (Barnett et al., 2007), we expect to find additional ribonuclease(s) that control growth phase-dependent mRNA stability in GAS.
In this work, we demonstrate that decay of Class I and Class II messages are affected differently by RNases J1 and J2 in exponential phase. We find that both of these RNases increase the decay rate for Class I messages (those that are degraded rapidly in exponential phase). For Class II messages, we show clearly that there is a delay phase before decay is initiated. For these messages, overproduction of both RNases J1 and J2 reduced the time before decay began with no effect on the second phase of the curve. Therefore, we conclude that these RNases are required to initiate cleavage of Class II messages, although they have no detectable effect on the rate of decay for these messages. By using conditional expression mutants for the first time in GAS, we also show that both RNases J1 and J2 are required for growth of this organism.
Identification of RNases J1/J2 in GAS
Based on sequence similarities, the paralogous proteins RNase J1 (Bsu1455, ykqC, rnjA) and RNase J2 (Bsu1679, ymfA, rnjB) of B. subtilis belong to the β-CASP subfamily of metallo-β-lactamases (Even et al., 2005; Dominski, 2007). According to crystal structure analysis, these proteins contain three distinct domains: a β-lactamase domain that is present in all metallo-β-lactamases, a β-CASP domain that is present only in the β-CASP subfamily of β-lactamases and a C-terminal domain that is present only in RNase J orthologues (Li de la Sierra-Gallay et al., 2008). The catalytic centre of these proteins is located in the cleft between the β-lactamase and the β-CASP domains and is formed by the conserved histidines and aspartates organized into sequence motifs I–V and A–C (Li de la Sierra-Gallay et al., 2008). The locations of conserved residues of these motifs in the RNase J1/J2 sequences are shown in Fig. S1. Adjacent to the catalytic centre, the RNase J proteins have a binding pocket that can accommodate a 5′-monophosphate nucleotide whose presence might explain the preference of RNase J for a 5′-monophosphorylated RNA substrate (Mathy et al., 2007; Li de la Sierra-Gallay et al., 2008). The seven residues involved in co-ordination of a nucleotide 5′-P in this pocket are conserved in the RNase J proteins (Li de la Sierra-Gallay et al., 2008) (Fig. S1).
All sequenced GAS genomes encode two homologues of RNases J1 and J2 of B. subtilis (Even et al., 2005). In strain MGAS315 of GAS these are SpyM3_1620 and SpyM3_0657 (Beres et al., 2002). Using clustal W2 multiple sequence alignment, we found that the protein encoded by SpyM3_1620 has higher homology to RNase J1 than to RNase J2 of B. subtilis (Fig. S1; 61% identity to RNase J1 and 43% identity to RNase J2). Moreover, the protein encoded by SpyM3_1620 and RNase J1 of B. subtilis have the same key amino acids in conserved motifs I–V and A–C, and the same residues involved in the binding and co-ordination of a nucleotide 5′-P in the nucleotide 5′-P binding pocket (Fig. S1). This suggests that the protein encoded by SpyM3_1620 is RNase J1, and we will refer to this gene as rnjA.
The previous phylogenetic analysis showed that Lactobacillales, the order that includes the Streptococcaceae family, generally have one homologue that is clearly similar to RNase J1 and another that is equally similar to RNases J1 and J2 of B. subtilis (Even et al., 2005). The protein encoded by MGAS315 SpyM3_0657 has equal homology with RNases J1 and J2 of B. subtilis (39% identity to RNase J1 and 41% identity to RNase J2). However, like RNase J2 of B. subtilis, SpyM3_0657 has a polar amino acid in conserved motif C (glutamine in SpyM3_0657 and asparagine in RNase J2 of B. subtilis; Fig. S1). Therefore, we will refer to SpyM3_0657 as rnjB and its protein product as RNase J2.
The sequences of GAS RNases J1 and J2 differ from each other at several locations expected to be critical for function. In the conserved regions II and V, which are involved in formation of the catalytic centre of these enzymes (Li de la Sierra-Gallay et al., 2008), RNase J1 contains different amino acid residues from RNase J2 (boxed in Fig. S1), suggesting that these enzymes might have differences in activity. In addition, because these enzymes also differ at four out of seven amino acids involved in co-ordination of a 5′-end monophosphate of mRNA (highlighted in Fig. S1), it is possible that they might also show differences in 5′-end specificity.
RNases J1 and J2 are both essential in GAS
To determine the role of RNases J1/J2 in GAS, we tried to generate insertion–deletion mutations in both rnjA and rnjB by allelic replacement in the chromosome using homologous recombination. To replace the internal portion of rnjA or rnjB with an antibiotic resistance gene, we electroporated GAS strain MGAS315 with up to 100 µg of linear DNA containing the spectinomycin resistance gene (aad9) flanked on each side with ∼1.5 kb of sequence corresponding to the upstream and downstream regions for each gene, and selected for spectinomycin-resistant colonies as described (Barnett et al., 2007). After several attempts, we obtained only one spectinomycin-resistant colony for rnjA and two for rnjB. Although the transformants had been colony-purified, PCR analysis showed that the cells from all three colonies contained aad9 but also retained the internal region of rnjA or rnjB (data not shown). This indicates that these spectinomycin-resistant colonies were not generated by the expected double crossover but instead probably contained a mixture of wild-type and mutant cells, suggesting that neither rnjA nor rnjB could be deleted.
Therefore, we constructed conditional mutants that expressed rnjA or rnjB under the tetracycline-inducible promoter Ptet (Geissendorfer and Hillen, 1990; Wang and Kuramitsu, 2005) in the chromosome of MGAS315 (Fig. 1). Using homologous recombination, a cassette encoding an antibiotic resistance gene followed by the tetR repressor, three tet operators, and a consensus ribosome binding site (AGGAGG), was inserted directly in front of the ATG translation start codon of rnjA and rnjB to generate strain JRS7316 (PtetrnjA) and JRS7317 (PtetrnjB) (Fig. 1). These conditional mutants are designed to express rnjA or rnjB only in response to addition of an inducer, anhydrotetracycline (AHT). Both strains showed limited growth on plates in the absence of induction (Fig. 2A), suggesting that both RNases are essential in GAS.
Because the gene downstream of rnjA is transcribed in the opposite direction from rnjA (Fig. 1), the presence of Ptet in front of rnjA is not likely to affect expression of this gene. Therefore, we conclude that RNase J1 is essential for growth of GAS.
Although the PtetrnjB strain (JRS7317) did not grow on THY plates without AHT (Fig. 2A), this might be due to a requirement for the product of a downstream gene, as these genes are transcribed in the same direction as rnjB and the insertion of Ptet in front of rnjB might affect their expression (Fig. 3A). To determine whether rnjB and the downstream gene (SpyM3_0658) are on the same transcript, we used Northern blot analysis (Fig. 3B). Using a probe for rnjB, we identified two major transcripts: one corresponds to the size expected for message made only from rnjB (∼1.7 kb), and one that was also detected with the probe for SpyM3_0658, of the size expected for a message that includes rnjB, SpyM3_0658 and possibly SpyM3_0659 (∼2.7 kb) (Fig. 3). We conclude that the insertion of Ptet in front of rnjB will affect expression of downstream genes. Therefore, we used complementation to determine which gene product is required for growth. For this experiment, we used a single copy plasmid encoding rnjB under the constitutive P23 promoter (pJRS7303) or under its own promoter (pJRS7304). Comparison of strain JRS7317 (PtetrnjB) containing the empty vector (pJRS9508) with strain JRS7317/pJRS7303 or JRS7317/pJRS7304 showed that the mutant did not grow on plates in the absence of inducer, while the complemented strains did (Fig. 2B). Therefore, we conclude that rnjB, and not a downstream gene, is essential for growth of GAS.
The inducer can be titrated to achieve maximal growth rates for the RNase J1 and J2 mutants
The RNase J2 mutant strain JRS7317 (PtetrnjB) did not grow on plates without AHT when inoculated from a plate containing 25 ng ml−1 AHT (Fig. 2A) or in broth without AHT following inoculation directly from a culture grown with 25 ng ml−1 AHT (data not shown), indicating that transfer of the inducer was minimal. However, although growth of the RNase J1 mutant strain JRS7316 (PtetrnjA) on plates was poor in the absence of inducer when the plate was inoculated from a plate containing 25 ng ml−1 AHT, some growth was visible in the heavily streaked area (Fig. 2A). The RNase J1 mutant also grew in THY broth without AHT when inoculated directly from a culture containing 25 ng ml−1 AHT (data not shown). This suggests that either RNase J1 is more stable than RNase J2 and/or that there is sufficient AHT transferred to induce production of a small amount of protein and that growth of GAS requires less RNase J1 than RNase J2.
To determine the minimal concentration of inducer required for maximal growth of the RNase J1 mutant (JRS7316) in liquid, it was necessary to minimize the amounts of AHT and RNase J1 in the cells. To do this, strain JRS7316 was grown overnight without AHT prior to induction. The cells were then collected, washed and inoculated into THY broth containing different concentrations of AHT (Fig. 4A). For the first 10 h of culture without AHT, or with the lowest concentration of AHT tested (0.01 ng ml−1), this mutant grew only to a low cell density (∼20 Klett units). AHT at 0.1 ng ml−1 induced growth to an intermediate final level, and the cells had a doubling time of 100 ± 10 min (Fig. 4A). AHT at 1 ng ml−1 and at concentrations up to 100 ng ml−1 induced growth with a doubling time of 60 ± 10 min and a final cell density of > 100 Klett units. In control experiments, AHT did not affect the growth rate of the wild-type strain MGAS315 in THY. This wild-type strain grew with a doubling time of 56 ± 2 min (data not shown), which is similar to the rate of growth of the fully induced RNase J1 mutant. Based on these results, for future experiments with the RNase J1 mutant, the ‘uninduced culture’ was grown with 0.1 ng ml−1 AHT to allow growth, and full induction was achieved by adding 25 ng ml−1 AHT.
To study the growth of the RNase J2 mutant in liquid culture, the mutant strain with the empty vector (JRS7317/pJRS9508) was cultured overnight with 25 ng ml−1 AHT; cells were collected, washed and diluted into THY broth containing different concentrations of AHT. The culture doubled slowly without AHT or with 0.01 ng ml−1 AHT and stopped growing at 20 Klett units (Fig. 4B). The RNase J2 mutant with the empty vector had a doubling time of 140 ± 40 min at 0.1 ng ml−1 AHT. With 1 ng ml−1 AHT this mutant grew as well as with concentrations up to 100 ng ml−1 (doubling time 80 ± 10 min) and the culture reached > 100 Klett units. Growth of strain JRS7317 without the plasmid was the same as with the empty vector (data not shown). For future experiments, we therefore used the same concentrations of AHT as for the RNase J1 mutant: 0.1 ng ml−1 for the ‘uninduced’ culture and 25 ng ml−1 for full induction. The complemented RNase J2 mutant (JRS7317/ pJRS7304) grown without any AHT had a 70 ± 10 min doubling time, which is similar to that of the fully induced mutant (Fig. 4B). This confirms the interpretation that the growth defect of the mutant is due to the absence of RNase J2. Although the simplest interpretation is that this is a direct effect, i.e. growth rate is determined by the amount of RNase, we cannot rule out a more complex scenario in which the decrease in growth rate is an indirect effect of depletion of the RNase.
Comparison of amounts of J1 and J2 proteins
Western blot analysis with antiserum to B. subtilis RNase J1 (from C. Condon: Daou-Chabo et al., 2009) was used to measure the amounts of the J1 and J2 proteins to confirm induction (Fig. 5). The band absent in each individual uninduced J mutant served to identify the locations of RNases J1 and J2 on the gel. As expected, in the lysate of the wild-type strain, MGAS315, both proteins were present (Fig. 5, wt). The gel shows that in the conditional mutants, induction led to an increase in the amount of the respective J protein compared with the uninduced mutant.
In the induced RNase J1 strain (JRS7316), there appears to be more J1 protein than in the wild-type strain, whereas induction of the RNase J2 mutant (JRS7317/pJRS9508) resulted in an amount of J2 protein that appears to be slightly less than that in the wild-type strain (Fig. 5). In addition, the absence of RNase J2 in the uninduced conditional mutant resulted in an increase in the amount of J1 protein, suggesting that there are some regulatory feedback interactions between these proteins. The complemented J2 mutant produced substantially more RNase J2 than any other strain, so 10-fold less total protein was loaded in that lane of the gel.
Reduction of RNase J1 decreases decay of hasA and gyrA mRNA
In the wild-type strain MGAS315, two representative mRNAs from the Class I messages hasA and gyrA decayed rapidly with monophasic kinetics, and addition of AHT did not affect their decay rates (Figs 6A and 7A). To determine whether RNase J1 has a role in decay of these mRNAs, we compared their decay rates in the uninduced and induced RNase J1 mutant (strain JRS7316) in cultures grown to late exponential phase (Fig. 4A). We found that these messages decayed more rapidly when RNase J1 was induced (Figs 6B and 7B), suggesting strongly that these mRNAs are substrates for RNase J1. Furthermore, the similarity of the half-lives of these messages in the induced J1 mutant to those in the wild-type parental strain suggests that RNase J1 is normally involved in the decay of these messages. However, even in the absence of induction of RNase J1, Class I transcripts were still unstable, suggesting that other RNases can degrade these transcripts, although at a slower rate than RNase J1. Thus, RNase J1 appears to be rate-limiting for decay of these Class I transcripts.
Reduction of RNase J2 affects decay of hasA and gyrA mRNA
We determined the role of RNase J2 in mRNA decay for Class I messages using the RNase J2 conditional mutant carrying the empty vector (JRS7317/pJRS9508) so that we would be able to compare it with the complemented strain. We found that the half-lives for these messages were longer in the uninduced J2 mutant (Figs 6C and 7C) than in the wild-type (Figs 6A and 7A) or in the complemented strain (Fig. S2): 3.6 versus 1.4 min for hasA and 1.6 versus 1 min for gyrA. Thus, it appears that RNase J2, as well as RNase J1, plays a role in decay of Class I messages. Although there is more RNase J1 protein in the uninduced J2 mutant than in the wild type, decay of Class I messages is slower than in the wild type, suggesting that the roles of these enzymes in Class I message decay may differ.
In addition, although of the amount of RNase J2 in the complemented strain exceeded that in the wild type, Class I messages decayed at the same rate in both. This suggests that when the level of J2 reaches a threshold, it is no longer the limiting factor for decay of Class I mRNA. It should also be noted that the amount of protein may not reflect the amount of active RNase. For example, in E. coli, the activity of RNase E is controlled by the binding of two inhibitor proteins, RraA and RraB (Lee et al., 2003; Gao et al., 2006).
Induction of RNase J1 affects decay of sagA and sda mRNA
To determine the role of RNase J1 in decay of Class II mRNAs, we compared the decay of two mRNAs of this Class in the RNase J1 conditional mutant (JRS7316) with and without induction, as we did for the Class I messages, and compared these with decay in the parental strain MGAS315. For sagA and sda mRNAs in the wild-type strain MGAS315, there is a delay of ≥ 10 min before decay begins (Figs 8A and 9A). In a control experiment, addition of AHT had no effect on decay of either sagA or sda message in this strain (Figs 8A and 9A). In the conditional RNase J1 mutant, the most striking effect of induction of RNase J1 on these transcripts was the change from biphasic decay kinetics (like those in the wild-type strain) in the uninduced mutant to monophasic decay in the induced mutant (Figs 8B and 9B). When RNase J1 was induced, decay of both transcripts began immediately, consistent with the finding that the amount of RNase J1 is greater in the induced RNase J1 mutant than in the wild-type strain (Fig. 5).
Because induction of RNase J1 eliminated the delay before decay begins following arrest of transcription, it also appears that this enzyme may be required for initiation of decay of Class II transcripts in GAS. However, once decay was initiated, its rate was similar to that in the uninduced mutant, suggesting that RNases other than J1 play the major role in the second phase of decay of Class II messages. A model including this is presented in the Discussion (Fig. 10).
Induction of RNase J2 also affects decay of sagA and sda mRNA
Induction of RNase J2 did not increase the rate of decay of sagA or sda mRNA compared with the rate in the parental wild-type strain (Figs 8 and 9). However, induction of RNase J2 resulted in a 10 min decrease in the time preceding initiation of decay. This suggests that like J1, RNase J2 is required for an early step in degradation of Class II mRNAs. The time preceding decay in the induced J2 mutant is similar to that in the wild type, consistent with the observation that the amount of J2 in the induced mutant is similar to that in the wild type (Fig. 5). As for RNase J1, the rate of decay, once it was initiated, was similar in the induced and uninduced J2 mutant, suggesting that RNases other than the J enzymes play the major role in the decay phase of Class II messages.
Induction of RNase J1 or J2 decreases the amount of sagA mRNA
In addition to their effect on the kinetics of decay of Class II transcripts, we observed an effect of RNases J1 and J2 on the amount of sagA mRNA: induction of the RNase reduced the amount of sagA mRNA about fivefold (Fig. S3). This is probably an indirect effect resulting from reduction by the RNase of the amount of transcript of a factor that activates the transcription of, and/or increases the stability of, the sagA message. No similar effect was seen on the level of sda mRNA or on the amounts of the Class I transcripts.
The success of a bacterial strain requires it to rapidly adapt to different growth conditions. This is particularly important for the human pathogen GAS, which can infect its host through the very different environments of respiratory mucosa or skin and which must adapt to growth in different tissues as the infection progresses. Most work on understanding the bacterial response to a changing environment has focused on studying regulation of transcription initiation. However, we learned that for GAS, mRNA decay plays an important role in determining the amount of transcript available for translation, at least for some virulence factors (Barnett et al., 2007). Rate of mRNA decay appears to be the determinant responsible for ‘growth phase regulation’ of gene expression, the predominant regulatory effect seen for all genes in GAS. We learned that, based on their stability in stationary phase, there are at least two classes of mRNAs in GAS and these appear to decay using different enzymatic pathways. For Class II mRNAs, PNPase is rate-limiting for decay, while this is not true for Class I mRNAs (Barnett et al., 2007). The work presented here was undertaken to learn about the role of the newly described RNases J1 and J2 in decay pathways for these two mRNA classes.
Class I and Class II mRNAs decay by different pathways
Class I and Class II messages were first identified in GAS because they have very different decay rates in the stationary phase of growth (Barnett et al., 2007). However, they also decay differently in exponential phase: Class II messages exhibit biphasic kinetics of decay (Figs 8 and 9). The initial absence of decay for Class II mRNAs is not due to delay of rifampicin entry into cells as the Class I messages start decaying immediately after rifampicin is added to stop transcription (Figs 6 and 7). These two differences in decay suggest that Class I and Class II mRNAs decay by different pathways, i.e. they use at least one different enzyme. We found here that both RNases J1 and J2 are involved in decay by both pathways, but their roles appear to be different: for Class I mRNAs, these RNases affect the rate of decay, and for Class II messages they affect the first phase of the biphasic curve, i.e. the delay before decay begins.
RNases J1 and J2 are not completely redundant in function
The sequences of RNases J1 and J2 are not identical, and thus, differences in function between them are expected. It has been shown that in vitro, both B. subtilis RNases J1 and J2 have endoribonucleolytic and 5′-to-3′ exoribonucleolytic activities (Even et al., 2005; Mathy et al., 2007). However, RNase J1 is essential for growth of B. subtilis, while RNase J2 is not (Kobayashi et al., 2003), indicating that their metabolic roles differ in vivo.
As expected based on sequence comparisons, we found that there are similarities and differences between the metabolic roles of RNases J1 and J2 in GAS as well as between the GAS enzymes and those of B. subtilis. We found that in GAS, a conditional mutant in either RNase J1 or in RNase J2 was unable to grow without induction. Therefore, we conclude that, unlike the situation in B. subtilis, both of these enzymes are essential in GAS. Thus, as in B. subtilis, although the roles of these RNases in GAS may overlap, they are not completely redundant.
Differences between RNases J1 and J2 in degree of induction required for growth
Although RNases J1 and J2 were expressed using identical Ptet promoters and with identical ribosome binding sites, the conditional mutant for RNase J1, but not that for J2, grew without induction when inoculated directly from a culture grown with inducer. Analysis of the amount of each protein by reaction with B. subtilis anti-J1 antiserum indicates that, if the antiserum recognizes both proteins equally, there is more J2 in the wild-type strain than J1. This suggests that less J1 than J2 is needed for growth. However, because both enzymes are essential for growth, their metabolic roles are not identical.
A model for decay of Class I and Class II messages
We found that decay of Class I messages began immediately at the time transcription was stopped, and, by 10 min, the amount of transcript was reduced at least 100-fold for the Class I messages we studied. However, Class II messages remained stable for ≥ 10 min before they began to decay. We also found that RNases J1 and J2 are involved in decay of both message classes. The simplest explanation of this involvement is that both classes of messages are substrates for both enzymes, although this has not been shown biochemically. To explain the different effects on Class I and Class II messages, we propose that Class I messages are preferred substrates for RNases J1 and J2 and therefore their decay is immediately apparent. Because most GAS messages are members of Class I, we propose that degradation of these transcripts engages most of the RNases J1 and J2 molecules in the cell until some of the preferred substrates are completely degraded (Fig. 10). Within this time, degradation of Class II messages is not detected. Only after degradation of the messages in Class I are the RNases J1 and J2 released to act on Class II messages, for which they have a lower affinity. In our experiments, this occurred 10 min after transcription arrest for the sda transcript and 20 min for the sagA transcript (Figs 8 and 9). The time of onset of decay of a Class II mRNA after transcription arrest might be determined by the affinity of the J1 and J2 RNases for the specific transcript.
For Class II messages, because induction of RNases J1 and J2 decreases the delay in the onset of decay after transcription arrest, we suggest that these RNases are required to initiate decay. It seems likely that this would involve the endonucleolytic activity of RNases J1 and J2, and that this cleavage generates the substrate for further decay of the Class II messages. Accumulation of decay intermediates for sagA mRNA in the induced RNase J1 and J2 mutants (Figs 8B and C) and for sda in the induced RNase J1 mutant (Fig. 9B) is consistent with the idea of endonucleolytic cleavage. Because induction of J1 and J2 did not change the rate with which Class II messages decay once it has been initiated, we suggest that the degradation of these messages is mediated by exoribonucleolytic activity of other RNases (Fig. 10). In agreement with this idea, we demonstrated previously that PNPase is rate-limiting for decay of Class II messages (Barnett et al., 2007).
Our studies measured mRNA decay in GAS after transcription was stopped, while in growing cells transcripts are continuously synthesized. Our model suggests that in exponentially growing cells, RNases J1 and J2 would be sequestered on Class I messages whose supply is constantly replenished by active transcription. This sequestration would therefore be expected to reduce the availability of RNases J1 and J2 to efficiently degrade Class II messages for which they have a lower affinity. In stationary phase, the amount of Class I transcripts is drastically reduced. This might result either from efficient degradation or from a lower rate of synthesis. The latter is likely to be correct based on an experiment in which the stable sagA transcript was synthesized at lower levels from Phas than from Psag in stationary phase (Barnett et al., 2007). If transcription activity is low in stationary phase, more RNases J1 and J2 should be available to degrade Class II messages than in exponential phase. However, Class II messages are remarkably stable at stationary phase in GAS, suggesting the possibility that RNases J1 and J2 are not active at stationary phase, as we also suggested for PNPase (Barnett et al., 2007). If this is correct, the inactivation of these RNases will be one of the critical factors underlying ‘growth phase regulation’. Further experimental work will be needed to test the suggestions proposed above and to provide a greater understanding of the mechanisms affecting activity of the RNases important for mRNA decay.
In this work, we have highlighted major differences in the decay pathways of the two message classes of GAS and identified roles for RNases J1 and J2 in both decay pathways. We have also proposed a model for growth phase regulation of transcripts in GAS. In addition, using conditional mutants for the first time in GAS, we have learned that RNases J1 and J2 have different metabolic roles in GAS, and that both are essential for growth of this organism.
Bacterial growth conditions
Group A streptococcus strain MGAS315 (serotype M3: Beres et al., 2002) and its derivatives were grown in Todd-Hewitt medium with 0.2% yeast extract (THY) at 37°C. E. coli Top10 cells (Invitrogen), used for cloning, were grown in LB medium. For E. coli cultures, concentrations of antibiotics were: spectinomycin, 50 µg ml−1; kanamycin, 50 µg ml−1; chloramphenicol, 20 µg ml−1. For GAS cultures, antibiotics were used at the following concentrations: spectinomycin, 75 µg ml−1; kanamycin, 200 µg ml−1; chloramphenicol, 5 µg ml−1. AHT was used at 25 ng ml−1 unless noted otherwise.
Growth conditions for J1 and J2 mutants and J2 complemented strains
Strains JRS7316 (RNase J1 mutant), JRS7317 (RNase J2 mutant) and the complemented derivatives of strain JRS7317 (generated as described in Methods of Supporting information) were grown overnight with AHT. However, to reduce the amount of AHT in the uninduced JRS7316 culture, this strain was grown overnight without AHT prior to induction. For induction of expression of RNase J1/J2, cells from the overnight culture were collected by centrifugation, washed twice with THY without AHT, resuspended 1:20 into fresh THY without AHT and divided into equal parts for each mutant. AHT was added to each part to a final concentration of 0.1 ng ml−1 or 25 ng ml−1 for mRNA decay experiments or as indicated in the Results section for growth curve studies. Cells were grown in stoppered side-arm flasks, and cell growth was monitored using a Klett-Summerson Photoelectric Colorimeter (Bel-Art products) with the red filter. For analysis of growth on plates, cells from a single colony grown on a THY plate with AHT were streaked first on a THY plate without AHT and then on a THY plate with AHT. Cells were grown overnight at 37°C and photographed.
Rifampicin (1 mg ml−1) was added to late exponential phase cultures (see Fig. 4) to stop transcription. Cells were collected after rifampicin was added at time intervals indicated in the Results section, and total RNA was isolated through a CsCl2 gradient (McIver and Scott, 1997). DNA was removed by treating RNA samples with TURBO DNA-free kit (Ambion) as recommended by the manufacturer. The concentration and the quality of RNA was determined by optical density at range 220–350 nm, and RNA integrity was confirmed by the presence of distinct bands for 16S and 23S rRNA after resolving RNA samples in an agarose gel.
For Northern blot analysis for sagA, RNA was resolved in 5% polyacrylamide-8M urea gels and electroblotted to Zeta-Probe GT Genomic Tested Blotting Membrane (Bio-Rad) in 0.5× TBE. For gyrA, hasA and sda, rnjB and SpyM3_0658, RNA was resolved in 1% agarose–formaldehyde gels and transferred to Zeta-Probe membrane by capillary transfer in 10× SSC. Membranes were hybridized to an antisense 32P-UTP-labelled RNA probe generated by in vitro transcription (Supporting information and Barnett et al., 2007). The membranes were exposed to a phosphoimager cassette, and the intensity of the bands specific for each gene was quantified using the ImageQuant programme (Molecular Dynamics).
Preparation of cell lysates
Cell lysates were prepared by a modification of the method described previously (Zahner and Scott, 2008). Cells were grown to late exponential phase, collected by centrifugation, washed twice with saline and resuspended in lysine buffer A (50 mM ammonium acetate, 10 mM CaCl2, 1 mM dithiothreitol; pH 6.2) containing 200 U ml−1 recombinant phage lysin from bacteriophage B30 of Streptococcus agalactiae (Pritchard et al., 2004). Cells were digested by incubation at 30°C for 30 min and disrupted by sonication. SDS-PAGE sample buffer was added to the protein lysates, samples were boiled for 10 min, centrifuged (1 min at 16 000 g), and the protein concentration was determined in supernatants using the Pierce 660 nm Protein Assay with Ionic Detergent Compatibility Reagent (ThermoScientific).
For Western blot analysis, cell lysates were separated on a NuPAGE 7% Tris-Acetate Gel (Invitrogen) and electrotransferred to a PVDF membrane (PerkinElmer). The membrane was blocked with 5% skim milk powder in TBS containing 0.05% Tween-20 for 1 h at room temperature and incubated with rabbit polyclonal anti-B. subtilis RNase J1 antibodies (from C. Condon) overnight at 4°C. Following washes with Tween-TBS, the membrane was incubated with goat anti-rabbit IgG conjugated with horseradish peroxidase (ThermoScientific) for 1 h at room temperature, washed and incubated with SuperSignal West Pico Chemiluminescent Substrate (ThermoScientific) for 5 min at room temperature. To visualize the protein bands, the membrane was exposed to CL-XPosure Film (ThermoScientific).
We thank Wolfgang Hillen for advice about the use of the tet promoter constructs, which originated in his laboratory and to Howard Kuramitsu for providing the pTet plasmid. We are especially grateful to Ciaran Condon for the generous gift of the anti-RNase J1 antiserum. We thank Joel Belasco for his suggestions, and Gordon Churchward, Charles Moran and Bernard Weiss for critically reviewing this manuscript. This work was supported in part by grant RO1-20723 from the National Institutes of Health to JRS, and JVB was partially supported by Fellowship 0725554B from the American Heart Association.