Small RNA molecules play key regulatory roles in many bacterial species. However, little mechanistic data exists for the action of small regulatory RNAs in the human pathogen group A Streptococcus (GAS). Here, we analysed the relationship between a putative GAS sRNA and production of the secreted virulence factor streptokinase (SKA). SKA promotes GAS dissemination by activating conversion of host plasminogen into the fibrin-degrading protease plasmin. Homologues of the putative sRNA-encoding gene fibronectin/fibrinogen-binding/haemolytic-activity/streptokinase-regulator-X (fasX) were identified in four different pyogenic streptococcal species. However, despite 79% fasX nucleotide identity, a fasX allele from the animal pathogen Streptococcus zooepidemicus failed to complement a GAS fasX mutant. Using a series of precisely constructed fasX alleles we discovered that FasX is a bona-fide sRNA that post-transcriptionally regulates SKA production in GAS. By base-pairing to the 5′ end of ska mRNA, FasX enhances ska transcript stability, resulting in a ∼10-fold increase in SKA activity. Our data provide new insights into the mechanisms used by small regulatory RNAs to activate target mRNAs, and enhances our understanding of the regulation of a key GAS virulence factor.
Bacteria have evolved a complex array of transcriptional and post-transcriptional regulatory mechanisms that integrate internal and external signals into an optimized response. RNA-based mechanisms of regulation are a key component of a cells regulatory capacity (Gottesman, 2005; Waters and Storz, 2009). With respect to bacterial pathogens, small regulatory RNAs (sRNAs) have been described that regulate virulence factor production and fitness within the host (Toledo-Arana et al., 2007). Individual sRNAs can function through drastically different mechanisms, from sequestration of regulatory proteins (Lenz et al., 2005; Babitzke and Romeo, 2007), to base-pairing with target mRNA molecules to increase or decrease their translation and/or stability (Grieshaber et al., 2006; Morita et al., 2006; Boisset et al., 2007). For the subclass of sRNAs that activate gene expression through base-pairing (reviewed in Frohlich and Vogel, 2009), molecular mechanisms include: (i) activation of mRNA translation by induction of structural rearrangements in the 5′ untranslated region (UTR) that unmasks the ribosome binding site (Lybecker and Samuels, 2007; Soper et al., 2010) and (ii) enhancement of mRNA stability through undetermined mechanisms after binding to the mRNA 3′-UTR (Opdyke et al., 2004). Very recently, the Clostridium perfringens sRNA VR-RNA was shown to enhance target mRNA stability after binding to the 5′-UTR (Obana et al., 2010).
The human bacterial pathogen group A Streptococcus (GAS, S. pyogenes) causes a diverse array of infections ranging from self-limiting pharyngeal (strep throat) infections to severe invasive infections such as necrotizing fasciitis (the flesh-eating syndrome). Importantly, the disease potential of GAS is attributable to the co-ordinated expression of specific subsets of encoded virulence factors (Roberts and Scott, 2007; Gryllos et al., 2008; Trevino et al., 2009). A critical GAS virulence factor is the secreted protein streptokinase (SKA) (Sun et al., 2004). SKA subverts components of the host fibrinolytic system to promote GAS spread from a local fibrin-clot-encapsulated infection to a systemic infection (McArthur et al., 2008). SKA promotes bacterial spread through activation of host plasminogen into plasmin, a broad spectrum protease that dissolves the meshwork of fibrin fibres present in a blood clot (Lottenberg et al., 1994; Svensson et al., 2002).
Recently, we performed a microarray-based genome-wide search for GAS sRNAs that, together with a previous bioinformatic approach, yielded an estimate of 75 sRNAs in the genome of the serotype M1 GAS isolate MGAS2221 (Livny et al., 2006; Perez et al., 2009). To date, only three putative or proven GAS sRNAs have been investigated experimentally. The pleiotropic effect locus sRNA enhances the abundance of several virulence factor-encoding mRNAs in a strain-specific manner (Li et al., 1999; Mangold et al., 2004; Perez et al., 2009). The RofA-like protein IV regulator X sRNA enhances the abundance of mRNAs encoding the virulence factors C5a peptidase, cysteine protease and M protein (Roberts and Scott, 2007). Finally, the putative fibronectin/fibrinogen-binding/haemolytic-activity/streptokinase-regulator-X (FasX) sRNA enhances the abundance of ska mRNA, and reduces the abundance of fbp and mrp mRNAs (which encode fibronectin and fibrinogen-binding proteins respectively) (Kreikemeyer et al., 2001). Concomitant with the mRNA level differences, a wild-type GAS strain has increased SKA activity and decreased binding to fibronectin and fibrinogen compared with an isogenic fasX mutant (Kreikemeyer et al., 2001). While not studied in detail, FasX is positively regulated by the upstream three-component regulatory system FasBCA (Fig. 1A) (Kreikemeyer et al., 2001). The mechanisms by which pleiotropic effect locus, RofA-like protein IV regulator X or FasX regulate virulence factor expression are unknown.
Here, we discovered that FasX post-transcriptionally regulates SKA production through a process that requires FasX hybridization to the 5′-UTR of ska mRNA to enhance transcript stability. This is only the second description of a sRNA that stabilizes an mRNA target after binding to the 5′-UTR, and is the first for a sRNA that binds less than 30 nucleotides from the start codon, a binding location previously associated with negatively regulating sRNAs. Thus, we provide new insights into the mechanisms of sRNA-mediated positive regulation. Our data also elucidate a new layer of regulation in the expression of the critical GAS virulence factor SKA.
Two regions of FasX are identical across pyogenic streptococcal species
Many sRNAs function by complementary base-pairing to target mRNAs. Therefore, fasX sequence comparisons may be informative by identifying regions of conserved and/or variable nucleotides. To assess conservation of the 205 bp fasX gene in serotype M1 GAS, we sequenced fasX from 48 serotype M1 isolates that were recovered from diverse geographical locations. The fasX gene was identical in all 48 strains (Table S1). To increase the potential for sequence variation we compared the fasX genes from the 13 available GAS genome sequences, which represent 10 different serotypes. The resultant data were relatively uninformative as no more than two single nucleotide polymorphisms (SNPs) distinguished the contemporary M1 fasX allele from the fasX alleles of any other serotype (data not shown).
To further utilize available sequence information we expanded our fasX comparison with other streptococcal species. We compared the fasX allele of GAS isolate MGAS2221 with that of the Streptococcus equi isolate 4047 (Holden et al., 2009), the Streptococcus zooepidemicus isolate MGCS10565 (Beres et al., 2008), and the Streptococcus uberis isolate 0140J (Ward et al., 2009). S. equi and S. uberis are primarily opportunistic pathogens of horses and cows, respectively, although they can also cause bacteraemia and meningitis in humans. S. zooepidemicus infects a wide-range of animals that include horses, cows, pigs, sheep and dogs. The S. zooepidemicus (79% identity), S. equi (78% identity) and S. uberis (67% identity) fasX alleles all harboured multiple SNPs relative to the GAS allele (Fig. 1B). Interestingly, two regions of the predicted FasX molecule were highly conserved between the four streptococcal species (green and blue shading in Fig. 1B). The locations of these conserved regions relative to the predicted FasX secondary structure are shown in Fig. S1A and B, which were generated using the bioinformatic programs RNAalifold and RNAfold respectively (Hofacker et al., 2002; Hofacker, 2003). The paucity of variation in defined regions suggests that these nucleotides are critical to FasX function.
A S. zooepidemicus fasX allele does not complement the GAS fasX mutant strain 2221ΔfasX
To facilitate analysis of FasX-mediated regulation we created strain 2221ΔFasX, an isogenic fasX mutant of MGAS2221 (Fig. S2). Strain 2221ΔFasX was complemented by introducing plasmid pFasXC, a fasX-containing derivative of the shuttle vector pDC123 (Chaffin and Rubens, 1998). To test whether a S. zooepidemicus fasX allele could complement strain 2221ΔFasX we also introduced plasmid pZOOFasX (Fig. S3), which encodes FasX from the S. zooepidemicus strain MGCS10565. The mutation of fasX in strain 2221ΔFasX, and the restoration of FasX transcription in 2221ΔFasX containing plasmid pFasXC or pZOOFasX, was verified using Northern blot analysis (Fig. 2A).
SKA activity levels in the culture supernatants of strains MGAS2221 and 2221ΔFasX containing empty vector, and strain 2221ΔFasX containing pFasXC or pZOOFasX, were determined. As expected, plasmid pFasXC, but not empty vector, restored high-level SKA activity to strain 2221ΔFasX (Fig. 2B). The S. zooepidemicus fasX allele did not complement strain 2221ΔFasX, indicating that the two regions of fasX conserved between GAS and S. zooepidemicus are not sufficient for regulation of GAS SKA activity (Figs 1B and 2B).
Using a quantitative RT-PCR approach we identified that fasX mutation leads to a drop in ska mRNA concentration to only 10% of that observed in the parent strain (Fig. 2C). Surprisingly, the complementation plasmid pFasXC only restored ska mRNA levels to ∼40% of wild-type levels, an unexpected finding given the high level of FasX RNA transcribed from the complementation plasmid relative to the parental strain (Fig. 2A), and the restoration of 80% SKA activity in culture supernatants (Fig. 2B). To address whether overexpression of FasX detrimentally effected the level of ska mRNA we introduced plasmid pFasXC into parental strain MGAS2221. The presence of plasmid pFasXC reduced ska mRNA levels in MGAS2221 to ∼30% of that observed in MGAS2221 containing empty vector (Fig. 2C). Thus, for uncharacterized reasons overproduction of FasX RNA has an inhibitory effect on this system.
The FasX RNA is the regulatory molecule
While postulated (Kreikemeyer et al., 2001), it has yet to be experimentally confirmed that the FasX transcript, and not a protein encoded within FasX, is the regulatory molecule. To facilitate testing of whether any of the five FasX open reading frames (ORFs; Fig. S4A) were essential for regulatory activity, we performed site-directed mutagenesis to create plasmid-encoded fasX alleles mutated in one or more of the ORFs (Fig. S5). As a consequence of overlap between ORFs only three 1 bp deletion mutant alleles were required (plasmids pΔ1-3; Fig. S4A and B). The complementation plasmid pFasXC and 1 bp deletion mutant derivatives pΔ1, pΔ2 and pΔ3, were transformed into strain 2221ΔFasX and their ability to restore SKA activity was tested. Each of the four plasmids restored SKA activity to wild-type levels (Fig. S4C). Thus, the data are consistent with none of the FasX ORFs being required for SKA regulatory activity, and hence that FasX is a bona-fide sRNA.
The deletion of single FasX nucleotides can result in the abrogation of SKA regulatory function
To conduct a systematic analysis of which FasX nucleotides were important for regulatory function we created 22 derivatives of the complementation plasmid pFasXC, each with a single nucleotide deletion. Each of the 22 pFasXC derivatives (pFXΔ plasmids) were transformed into strain 2221ΔFasX and SKA activity was tested. Significant variation was observed in the ability of the individual pFXΔ plasmids to restore SKA activity (Fig. 3A). While the majority of pFXΔ-containing 2221ΔFasX derivatives had high SKA activity levels (green bars in Fig. 3A), four derivatives had only moderate activity (orange bars in Fig. 3A), and two derivatives had low activity (red bars in Fig. 3A). To ensure that 2221ΔFasX derivatives with moderate or low SKA activity levels produced FasX we performed Northern blot analysis. No appreciable difference in FasX abundance was observed between the tested 2221ΔFasX derivatives containing pFXΔ plasmids or the complementation plasmid pFasXC (Fig. 3B). In addition, no appreciable difference in secondary structure was evident from bioinformatic predictions of the mutant FasX RNAs (data not shown). The locations of the nucleotides deleted in the 22 pFXΔ plasmids, and the effect of nucleotide deletion on SKA activity, are summarized in Fig. 3C.
Nucleotides required for regulation of SKA activity are confined to a distinct region of the FasX molecule
To further define FasX nucleotides required for regulatory activity we created four additional plasmid-encoded fasX mutant alleles. The first mutant harboured two SNPs, changing a C-repeat sequence CCCC to GGCC (plasmid pFXC45/46G; Fig. 4A). The other three mutant alleles had large deletions of 22 nt (pFXΔ72-93), 51 nt (pFXΔ155-205) and 122 nt (pFXΔ1-34,115-205) (Fig. 4A). After introduction into 2221ΔFasX, transcription of the mutant fasX genes was confirmed via Northern blot analysis (Fig. 4B), and SKA assays were performed to assess the effect of the mutations on FasX activity (Fig. 4C). Surprisingly, only plasmid pFXC45/46G failed to restore high-level SKA activity to strain 2221ΔFasX. Thus, one or both of the C nucleotides that were substituted in plasmid pFXC45/46G appear critical for FasX function. In contrast, the extensive regions of FasX deleted in plasmids pFXΔ72-93, pFXΔ155-205 and pFXΔ1-34,115-205 are dispensable (Fig. 4A).
FasX enhances ska mRNA stability
Deletion of fasX leads to a decrease in ska mRNA abundance (Fig. 2C) (Kreikemeyer et al., 2001). Thus, FasX may act at the level of transcription, increasing the number of ska transcripts produced, or at the post-transcriptional level, by increasing the stability of ska transcripts. To test the hypothesis that FasX functions by increasing the stability of ska mRNA we used a quantitative RT-PCR approach. GAS cultures were treated with rifampicin to inhibit RNA synthesis and degradation of ska (test) and proS (control) mRNA transcripts were monitored over time. The proS mRNA transcripts were degraded at similar rates in each of the three strains tested (grey lines in Fig. 5A). In contrast, ska mRNA transcripts were highly unstable in the fasX mutant strain 2221ΔFasX, but more stable in the wild-type and complemented strains (black lines in Fig. 5A). Thus, the FasX-mediated regulation of SKA activity at least partially occurs at the post-transcriptional level by enhancing the stability of ska mRNA.
We next tested the hypothesis that the inability of the fasX allele present in plasmid pFXΔ46 to restore high level SKA activity to strain 2221ΔFasX (Fig. 3A) was due to an inability to enhance ska mRNA stability. Using our quantitative RT-PCR approach we identified that while the stability of ska mRNA in strain 2221ΔFasX containing pFXΔ46 was increased relative to empty vector (Fig. 5B), the initial rate of ska mRNA degradation was similar, with 50% of the transcript being degraded after less than 1 min. In contrast, the ska transcript in strain MGAS2221 containing empty vector had a half-life of ∼7.5 min. Thus, although the mutant FasX encoded within pFΔ46 afforded some protection to ska mRNA, the level of protection was significantly lower than that afforded by wild-type FasX.
Determination of the ska transcriptional start site
To facilitate analysis of the regulatory mechanism between FasX and ska mRNA we determined the transcriptional start site of the MGAS2221 ska gene. Our data identified the GAS ska transcriptional start site as being the first of four G residues located 32 nt upstream of the start codon, and is identical to that observed in Streptococcus equisimilis (Gase et al., 1995).
In vivo confirmation of FasX hybridization to the 5′-UTR of ska mRNA
Our data are consistent with the CCCC region of FasX being important for regulatory activity. To identify potential regions of complementarity between this region of FasX and ska mRNA we performed a bioinformatic analysis using the program TargetRNA (Tjaden et al., 2006). Nine nucleotides at the extreme 5′ end of ska mRNA were identified as being perfectly complementary to the CCCC region of FasX (Fig. 6A). To facilitate testing whether this putative FasX:ska mRNA interaction occurs in vivo we created strain 2221ΔX.GCSKA, a 2221ΔFasX derivative in which the four G residues located upstream of ska were mutated to GCCG (Fig. 6B). If the residues at the 5′ end of ska mRNA base-pair with FasX, and if this interaction is required to enhance ska mRNA stability, then the low level of SKA activity produced by strain 2221ΔX.GCSKA would not be complemented by introduction of pFasXC (which contains wild-type fasX), but would be complemented by introduction of pFXC46/47G (which contains a fasX derivative in which the CCCC region has been mutated to CGGC; Fig. 6B). To test this idea we used SKA activity assays (Fig. 6C). The data are consistent with FasX interacting directly with the 5′ end of ska mRNA. That ska mRNA is stable only when FasX and the ska 5′-UTR are complementary to one another was confirmed by quantitative RT-PCR (Fig. 6D and E). Thus, FasX post-transcriptionally regulates expression of the secreted virulence factor SKA by hybridizing to the 5′-UTR of ska mRNA and increasing transcript stability.
Addition of complementary nucleotides to the 5′ end of ska mRNA enhances transcript stability in a FasX-independent manner
We hypothesized that addition of a complementary loop to the 5′ end of ska mRNA that mimics FasX binding would stabilize the transcript and uncouple ska mRNA stability from FasX-mediated regulation. In addition, we hypothesized that addition of a non-complementary sequence would result in unstable ska transcripts regardless of the FasX status of the cell. To test these hypotheses we used homologous recombination to insert complementary or non-complementary sequences at the ska transcriptional start site (Fig. 7A and B). The insertions were made in both the MGAS2221 and the 2221ΔFasX backgrounds. The stability of ska mRNA in the four strains was tested by quantitative RT-PCR following addition of rifampicin. As predicted, strains containing the complementary extension to the 5′ end of ska mRNA had highly stable ska transcripts, both in the presence (strain 2221.ska::EXY) and absence (strain 2221ΔFasX.ska::EXY) of FasX, while strains with the non-complementary extension had unstable ska transcripts in the presence (strain 2221.ska::EXN) and absence (strain 2221ΔFasX.ska::EXN) of FasX (Fig. 7C). Western blot analysis of SKA expression supported the ska transcript stability data (Fig. 7D).
FasX does not enhance ska mRNA stability by inhibiting the ribonucleases CvfA and PNPase
In Bacillus subtilis the initial, and rate-limiting, step in the mRNA decay pathway is endonucleolytic cleavage of the 5′ end of a transcript (Condon, 2003; Commichau et al., 2009). In part because of this we hypothesized that hybridization of FasX with ska mRNA blocks the access of an endoribonuclease to the ska transcript, preventing cleavage and leading to the observed enhanced transcript stability. Conserved virulence factor A (CvfA) was recently characterized in GAS as being a membrane-spanning protein with putative endoribonuclease activity (Kang et al., 2010). Deletion of the cvfA gene has a major effect on transcript levels in a growth phase and nutritional stress-dependent manner (Kang et al., 2010). To facilitate testing whether FasX enhances ska mRNA abundance by inhibiting CvfA-mediated cleavage of the transcript we created cvfA mutant derivatives of MGAS2221 and 2221ΔFasX. SKA expression and ska mRNA stability were compared between the two cvfA mutant strains and two control strains. Neither cvfA mutant strain differed from their control strain with respect to the rate of ska mRNA degradation and the level of secreted SKA protein (Fig. S6A and B). Thus, FasX does not enhance ska transcript stability by inhibiting CvfA-mediated cleavage. It is noteworthy that in contrast to ska mRNA, degradation of proS mRNA (which we monitored as a control transcript) was effected by cvfA mutation (Fig. S6A).
The exoribonuclease polynucleotide phosphorylase (PNPase; encoded by the pnpA gene) contributes to the growth phase regulation of mRNA stability in GAS (Barnett et al., 2007). Given that PNPase regulates the stability of several GAS mRNAs we tested whether FasX regulates SKA production by inhibiting PNPase activity. Similar to the cvfA data, inactivation of pnpA had no effect on SKA expression, regardless of the presence or absence of FasX (Fig. S6C and data not shown). Thus, FasX does not enhance ska transcript stability by inhibiting PNPase-mediated cleavage.
The prevalence of sRNAs in bacterial genomes indicates that they represent a fundamental mechanism of regulation (Gottesman, 2005; Waters and Storz, 2009). Not confined to a single level of regulation, sRNAs have been described in E. coli that regulate transcriptionally [e.g. the 6S RNA inhibits transcription from σ70 promoters (Wassarman, 2007)], post-transcriptionally [e.g. the sRNA RyhB represses the translation and stability of sodB mRNA (Afonyushkin et al., 2005)] and post-translationally [e.g. the sRNA CsrB sequesters the global regulatory protein CsrA (Babitzke and Romeo, 2007)]. In part due to the absence of mechanistic data regarding sRNA-mediated regulation in the human pathogen GAS, the current study examined the regulation of the key virulence factor SKA by the sRNA FasX. Research into SKA production by beta-haemolytic streptococci began more than seven and a half decades ago (Tillett and Garner, 1933), and continues today in part due to the critical role of this enzyme during infection (Svensson et al., 2002; Khil et al., 2003; Sun et al., 2004). SKA activity is negatively regulated at the transcriptional level by the CovR/S (also known as CsrR/S) two-component regulatory system (Levin and Wessels, 1998; Federle et al., 1999; Sumby et al., 2005; Gryllos et al., 2008). Here, we have shown that FasX positively regulates SKA activity post-transcriptionally by binding to the 5′-UTR of ska mRNA to increase transcript stability. The highly regulated nature of SKA expression is consistent with the level and timing of SKA expression being important during infection.
Transcription of fasX is regulated by the upstream genes fasBCA (Fig. 1A) (Kreikemeyer et al., 2001). The fasBC genes encode proteins with homology to sensor kinases, while fasA encodes a protein with homology to response regulators. Thus, fasBCA may encode a three-component system through which the predicted membrane-spanning proteins FasB and FasC recognize as as-yet-unknown signals, resulting in transduction of the signal through FasA (possibly via phosphorylation), activating fasX transcription. Other than fasBCA all being required for fasX transcription (data not shown) (Kreikemeyer et al., 2001), nothing else is known regarding the regulation of fasX transcription by FasBCA.
The fasBCAX operon structure is conserved in GAS, S. zooepidemicus and S. equi (Beres et al., 2008; Holden et al., 2009). In contrast, S. equisimilis has one, and S. uberis has four, putative sensor kinases upstream of the fasAX genes (Steiner and Malke, 2002; Ward et al., 2009). Despite conservation of operon structure, the fasX gene of S. zooepidemicus strain MGCS10565 did not restore high-level SKA activity to GAS mutant strain 2221ΔFasX (Fig. 2B). This data identified that the two conserved regions of fasX, as identified by comparisons of fasX from four streptococcal species (Fig. 1B), were insufficient for SKA regulatory activity. Furthermore, while the S. zooepidemicus fasX allele contains a SNP within the 9 nt complementary to the ska 5′-UTR, this does not fully explain the inability of this allele to regulate SKA activity. We generated a derivative of the S. zooepidemicus fasX allele in which the SNP was substituted to that observed in GAS, and this allele still failed to restore high-level SKA activity to GAS strain 2221ΔFasX (data not shown). Thus, it appears that these 9 nt are not sufficient for regulatory activity. Interestingly, the streptokinase-encoding genes of S. zooepidemicus, S. equi, and S. uberis lack the 5′-UTR nucleotides that, in GAS, base-pair with FasX RNA. Thus, if FasX regulates ska mRNA stability in these other streptococcal pathogens then it does so through a different mechanism, or at least through different nucleotides, than in GAS.
That deletion of FasX nucleotides 42 and 46 abrogated ska regulatory function was not surprising given that they lie within the 9 nt complementary to the ska 5′-UTR (Fig. 3C). In addition, nucleotide 42 is predicted to contribute to FasX secondary structure by base-pairing with FasX nucleotide 104, providing another possible explanation of why removal of this nucleotide is detrimental to regulatory activity. Disruption of FasX secondary structure may also explain why deletion of nucleotides 61, 66 or 100 impedes regulation. However, as nucleotide 26 is located in a predicted single-stranded region of FasX then disruption of secondary structure cannot be implicated in the reduced regulatory activity of the Δ26 fasX allele (Fig. 3C). Rather than affecting secondary structure it is possible that nucleotides 61, 66 and 100 overlap with a binding site for an as-yet-unknown RNA-binding protein that is required for efficient regulation.
In addition to FasX positively regulating the abundance of ska mRNA, it also negatively regulates the abundance of the adhesin-encoding mRNAs mrp and fbp (Kreikemeyer et al., 2001). The decreased expression of adhesins, and increased expression of factors that promote bacterial spread from a site of infection, is suggestive of FasX regulating the transition of GAS from the colonization to dissemination phases of infection (Kreikemeyer et al., 2001). While the mechanisms by which FasX regulates mrp and fbp transcript levels are unknown, the library of fasX mutant alleles constructed here may facilitate their identification. There are no obvious regions of complementarity between FasX and the mrp/fbp genes. Given that a distinct conserved region of FasX appears to be required for regulation of SKA activity, it is possible that the other conserved regions of FasX mediate regulation of these other targets.
The FasX sequence complementary to the ska 5′-UTR is 5′-UCAAUCCCC-3′. The C-rich region is reminiscent of a conserved UCCC sequence motif recently identified in 11 previously uncharacterized S. aureus sRNAs (Geissmann et al., 2009). One of these sRNAs was investigated and the UCCC motif identified as directly base-pairing to target mRNAs. Similarly, the well-described virulence factor-regulating S. aureus sRNA RNAIII has UCCC sequences in three hairpin loops, and each is known to interact with target mRNA sequences (Boisset et al., 2007). It has been suggested that the UCCC motif represents a novel class of S. aureus sRNAs that target mRNAs through a common mechanism (Geissmann et al., 2009), and it is possible that the same mechanism is also present in other low GC% Gram-positive pathogens such as GAS.
RNA turnover is the primary function of the degradosome, a multi-enzyme complex that promotes RNA decay (Carpousis, 2007). While primarily studied in E. coli (Morita et al., 2004), the degradosome has also been studied in the model Gram-positive organism B. subtilis (Commichau et al., 2009). Proteins within the B. subtilis degradosome include the endoribonuclease RNase J1, the 3′ to 5′ exoribonuclease PNPase, the glycolytic enzyme enolase and the membrane-bound endoribonuclease RNase Y (Commichau et al., 2009). Initiation of mRNA decay in B. subtilis is through binding of an endoribonuclease to the 5′ end of the transcript and tracking along to a cleavage site (Condon, 2003; Commichau et al., 2009). We hypothesize that control of mRNA degradation in GAS is also 5′ end-dependent in at least some cases. Recently, CvfA was identified as the RNase Y homologue in GAS and was also shown to interact with enolase, suggesting the presence of a degradosome in this organism (Kang et al., 2010). Given this information we tested the hypothesis that CvfA was responsible for initiation of ska mRNA degradation, and that this was inhibited by FasX base-pairing to the 5′ end of ska mRNA. However, isogenic cvfA mutant strains were unchanged in their rate of ska mRNA degradation, regardless of the presence or absence of FasX (Fig. S6A and B), indicating that CvfA is not the RNase inhibited by FasX. As PNPase regulates the abundance of several GAS mRNAs we also tested whether this was the RNase inhibited by FasX (Barnett et al., 2007). Disruption of the pnpA gene also failed to enhance ska mRNA stability (Fig. S6C). Our current working hypothesis is that one or both of RNases J1 and J2, both of which are essential endoribonucleases in GAS (Bugrysheva and Scott, 2009), are responsible for initiating ska mRNA degradation via a process that is inhibited by FasX:ska mRNA hybridization.
Derivatives of MGAS2221 and 2221ΔFasX that contained a complementary 13 nucleotide extension to the 5′ end of ska mRNA had highly stable ska transcripts (Fig. 7B). The high and similar level of ska mRNA stability in strains 2221.ska::EXY and 2221ΔFasX.ska::EXY indicates that the added extension protects the mRNA in a FasX-independent manner. We believe that this is due to the extension folding back and forming a stem:loop structure at the 5′ end of the molecule as depicted in Fig. 7A. In B. subtilis the presence of a stem:loop structure at the 5′ end of an mRNA enhances its stability (Sharp and Bechhofer, 2005), which is thought to be due to blockage of endoribonuclease binding.
Derivatives of MGAS2221 and 2221ΔFasX that contained a non-complementary 13 nucleotide extension to the 5′ end of ska mRNA had unstable ska transcripts (Fig. 7B). The inability of FasX to enhance ska transcript stability in strain 2221.ska::EXN indicates that nucleotides at the extreme 5′ end of ska mRNA, rather than downstream nucleotides, must be base-paired to protect against RNase-mediated degradation. This is supported from work in B. subtilis, which identified that stem:loop structures could only enhance mRNA stability when they were located less than five nucleotides from the 5′ end of the transcript (Sharp and Bechhofer, 2005). In strain 2221.ska::EXN the FasX:ska base-pairing would start 14 nucleotides from the 5′ end of the transcript.
We propose that FasX is the second of what may be a large class of sRNAs that enhance the stability of their mRNA targets by formation of base-paired structures at, or very close to, the mRNA 5′ end. Interestingly, while FasX and the C. perfringens sRNA VR-RNA both enhance target mRNA stability after binding to the 5′-UTR their mechanisms differ. FasX binds to the 5′ end of ska mRNA to create secondary structure predicted to inhibit endoribonuclease-mediated degradation. In contrast, base-pairing between VR-RNA and its mRNA target leads to a cleavage event downstream of the base-pairing site, resulting in a processed mRNA that lacks 62 nucleotides at the 5′ end relative to the full-length transcript (Obana et al., 2010). Unlike the full-length transcript, which has a predicted 33 nucleotide single-stranded region at the 5′ end, the 5′ end of the processed transcript is present within a protective stem:loop structure. A potentially important distinction between FasX and VR-RNA is that FasX must remain bound to its mRNA target to enhance stability while VR-RNA does not. The reversible nature of the FasX:ska mRNA interaction implies that the positive regulation afforded by FasX could be removed by decreasing FasX transcription, or increasing FasX turnover.
FasX base-pairs with ska mRNA nucleotides −32 to −24 (relative to the first nucleotide of the start codon). That FasX binding to this region increases SKA expression was surprising given the current convention that sRNA hybridization to an mRNA between nucleotides −35 to +15 leads to inhibition of translation (Storz et al., 2004; Morita et al., 2006; Marzi et al., 2007; Sharma et al., 2007; Bouvier et al., 2008; Frohlich and Vogel, 2009). While we have not ruled out that FasX also alters translation of ska mRNA, we do not believe that this is the case as the drop in ska mRNA abundance following fasX mutation (10% of wild-type; Fig. 2C) is similar to the drop in SKA activity (10% of wild-type; Fig. 6C). Perhaps the FasX:ska mRNA interaction is sufficiently stabile to interfere with RNase cleavage but not ribosome binding. Thus, re-evaluation of our knowledge with respect to the consequences of sRNA binding in close proximity to an mRNA ribosome binding site may be warranted. In summation, our data provide new insights into the mechanisms used by sRNAs to positively regulate mRNA targets, as well as uncovering a new layer of regulation of a key GAS virulence factor.
Bacterial strains and culture conditions
MGAS2221 is representative of the highly virulent M1T1 GAS clone responsible for significant morbidity and mortality since the mid-1980s in the USA, Canada and Western Europe (Sumby et al., 2005). GAS were grown in Todd-Hewitt broth with 0.2% yeast extract (THY broth) at 37°C (5% CO2). Chloramphenicol (4 µg ml−1) and/or spectinomycin (150 µg ml−1) were added when required.
Construction of isogenic fasX mutant strain 2221ΔfasX
An isogenic fasX mutant of parental strain MGAS2221 was created by replacement of the fasX gene with a non-polar spectinomycin resistance cassette via a previously described PCR overlap extension method. PCR primers used in the construction of mutant strains are listed in Table S2. Confirmation of isogenic mutant strain construction was gained via PCR, sequencing and Southern blot analyses (Fig. S2). The Southern blot was generated using EcoRV-digested genomic DNA. The blot was probed with a labelled PCR product generated using primers FasXC and FasXD (Table S2).
Complementation of isogenic mutant strain 2221ΔfasX
To complement isogenic mutant strain 2221ΔFasX we introduced the wild-type fasX allele into the E. coli– GAS shuttle vector pDC123 (Chaffin and Rubens, 1998; Sumby et al., 2006). Forward and reverse primers used to amplify fasX contained BglII and NsiI restriction enzyme sites respectively (Table S2). PCR products digested with BglII NsiI were ligated into similarly digested pDC123. All inserts were sequenced to ensure no spurious mutations had arisen during PCR and cloning.
Cloning of fasX from S. zooepidemicus
The fasX gene from S. zooepidemicus strain MGCS10565 was amplified using primers FasXZOOF/R (Table S2), and cloned into pDC123 to create pZOOFasX. The plasmid insert was sequenced to ensure no spurious mutations had arisen during PCR and cloning.
Total RNA isolation
For the described Northern blot and quantitative RT-PCR analyses GAS strains were grown to mid-exponential phase (OD600 = 0.5) and aliquots added to two volumes of RNAprotect (Qiagen). After incubating at room temperature for 5 min samples were centrifuged for 10 min at 4°C and 5000 g. Cell pellets were quick frozen in liquid nitrogen and stored at −80°C until ready for use. To isolate total RNA frozen GAS cell pellets were resuspended in 100 µl TE buffer and transferred to 2 ml tubes containing fine glass shards (lysing matrix B tubes, MP Biomedicals). Tubes were placed into a glass bead beater (FastPrep machine, THERMO 101) and processed for 15 s at speed 4. Tubes were centrifuged for 5 s at 14 000 g to reduce foaming and an additional processing in the FastPrep machine performed following addition of 650 µl of buffer RLT (Qiagen). Samples were centrifuged for 30 s at 14 000 g to collect contents and 600 µl transferred to a 1.5 ml tube containing 900 µl 100% ethanol. RNA samples were subsequently bound to, washed on, and eluted from, RNeasy columns (Qiagen) as per the manufacturers' miRNeasy protocol. Contaminating genomic DNA was removed from eluted RNA samples via four 30 min incubations at 37°C with 2 µl TURBO DNase-free (Applied Biosystems), with DNA removal being verified by PCR.
Northern blot analysis
Total RNA was isolated as described and 6 µg from each GAS strain was loaded onto a 5% TBE-Urea gel before separating by electrophoresis. RNA was transferred to nylon membrane via electroblotting, UV cross-linked, and probed overnight with an in vitro transcribed probe complementary to fasX or to the 5S RNA (loading control). In vitro transcribed probes were generated using the Strip-EZ T7 kit (Applied Biosystems). DNA templates for in vitro transcription reactions were generated by PCR, with one primer containing the T7 promoter sequence (Table S2). RNA probes were labelled with biotin before hybridization (Brightstar psoralen-biotin labelling kit, Applied Biosystems). Following washes Northern blots were developed (Brightstar biodetect kit, Applied Biosystems) and exposed to film.
Indirect SKA activity assay
GAS strains were grown to the mid-exponential phase of growth (OD = 0.5), pelleted by centrifugation, and the supernatants filter-sterilized using 0.22 µm syringe filters. Filtered supernatants were stored at −20°C until ready for use. SKA activity within supernatants was measured indirectly via the SKA-induced cleavage of human plasminogen into plasmin, cleavage of the chromogenic substrate S-2251 by plasmin, and measuring the light emitted using a spectrophotometer set at 405 nm. Specifically, 70 µl of filtered GAS culture supernatants was added to wells of a 96-well microtiter plate and S-2251 (Diapharma Group) added to 400 µM. Subsequently, 2 µg of purified human plasminogen (Molecular Innovations) was added to each well and the plate placed in a temperature-controlled (37°C) spectrophotometer. The absorbance at 405 nm was measured every 2 min for 3 h. Control reactions were utilized that lacked plasminogen (negative control for SKA activity) or contained known amounts of purified streptokinase from GCS (Sigma-Aldrich; positive control for SKA activity).
Construction of plasmid-encoded mutant fasX alleles
Construction of plasmid-encoded 1 bp deletion mutants of fasX was through use of an overlap extension PCR method as outlined in Fig. S3. Briefly, using the complementation plasmid pFasXC, we used two vector embedded primers (DC123ECORV and DC123BGLII; Table S2) and two complementary mutagenesis primers (e.g. FDEL1F and FDEL1R; Table S2) to amplify the fasX gene from plasmid pFasXC while at the same time deleting 1 bp as dictated by the sequence of the mutagenesis primers. The resultant PCR product was digested with BglII NsiI and ligated into plasmid pDC123 that had been cut with the same enzymes. Ligations were transformed into E. coli (plated onto LB agar containing 20 µg ml−1 chloramphenicol) and minipreps performed on transformants. After PCR and sequencing to confirm the correct plasmid sequences they were transformed into strain 2221ΔFasX. Plasmid inserts were resequenced once transformed into GAS to ensure no spurious mutations had arisen.
Mutant fasX alleles containing deletions larger than 1 bp were constructed in a similar manner to the 1 bp deletion mutant plasmids. However, for pFXΔ155-205 an unrelated terminator sequence was added to the 3′ end of fasX to replace the terminator deleted in this mutant. This was required as in the absence of a terminator the mutant fasX allele failed to complement strain 2221ΔFasX, presumably due to transcript instability. For plasmid pFXΔ1-34,115-205, because the promoter and terminator region of the fasX allele were deleted we added a terminator to the 3′ end and placed the Pspac promoter at the 5′ end. The Pspac promoter is a low-level constitutive promoter that has been previously described (Biswas et al., 2008). Primers containing the terminator/Pspac promoter are listed in Table S2.
Western blot analysis
GAS strains were grown to mid-exponential phase (OD600 of 0.5) in THY broth at which point 10 ml was centrifuged to pellet the bacteria. The supernatant was filtered through 0.22 µm syringe filters and the proteins precipitated via addition of 3.5 volumes of ice-cold ethanol. After 4 h at −20°C the samples were centrifuged and the protein pellets resuspended in 500 µl of SDS-PAGE loading buffer. Western blots were made using standard protocols and probed with rabbit polyclonal antibodies raised against SKA and Spd3. Goat anti-rabbit horseradish peroxidase-conjugated secondary antibodies were used to detect primary antibody binding and to generate signals.
Quantitative RT-PCR analysis of ska mRNA and FasX RNA concentration
cDNA was synthesized from total GAS RNA using the reverese transcriptase Superscript III (Invitrogen) as per the manufactures' instructions. TaqMan quantitative RT-PCR was performed using an ABI 7500 Fast System (Applied Biosystems). Gene transcript levels present in different strains were compared with that of parental strain MGAS2221 using the ΔΔCT method as described (Shelburne et al., 2008). TaqMan primers and probes for the genes of interest, and the internal control gene proS, are listed in Table S2.
Quantitative RT-PCR analysis of ska mRNA stability
GAS strains were grown to mid-exponential phase (OD600 of 0.5) in THY broth before the addition of rifampicin to 1 mg ml−1 to inhibit RNA synthesis. After addition of rifampicin 3 ml samples were recovered after 0, 1, 2, 4, 5 and 20 min. Recovered samples were added to two volumes of RNA protect, incubated at room temperature for 5 min, pelleted by centrifugation and quick frozen in liquid nitrogen. RNA was isolated from each sample, converted into cDNA, and used in Taqman analysis. TaqMan quantitative RT-PCR was performed using an ABI 7500 Fast System (Applied Biosystems). TaqMan primers and probes for ska, and for the control gene proS, are listed in Table S2. Data are presented as percent proS (control gene; blue lines) or ska (red lines) transcript levels relative to the amount of these transcripts at time point zero. Experiment was performed in triplicate with mean values shown.
ska transcriptional start site determination
An overview of the protocol used is shown in Fig. S7. Approximately 3 µg of total RNA isolated from mid-exponential phase GAS (OD600 0.5) was used in a cDNA synthesis reaction with primer SKATMR (Table S2). RNA was removed via RNase digestion and the cDNA purified using the MinElute PCR purification kit (Qiagen). Purified cDNA was 5′ phosphorylated using T4 polynucleotide kinase (New England Biolabs), and circularized using T4 RNA ligase 1 (New England Biolabs). Circularized cDNA was purified using a second MinElute column and used in a PCR reaction with primers GSP1 and SKATMF (Table S2). A control reaction using uncircularized cDNA was also performed. PCRs were separated by electrophoresis and the band present in the test sample but absent from the control sample was extracted, TA cloned, transformed into E. coli, and sequenced from a dozen transformants. While this protocol does not distinguish primary transcript 5′ ends from internal 5′ processing sites the data were the same as that observed in S.equisimilis, which was determined using a discriminatory method (Gase et al., 1995).
Construction of double mutant strain 2221ΔX.GCSKA
Overlap extension PCR using primer pairs SKAMUT1/2 and SKAMUT3/4 (Table S2) was used to produce an ∼2 kb product spanning the ska promoter region in which the two targeted nucleotides had been substituted. The PCR product was digested with BamHI and cloned into the BamHI site of the suicide vector pBBL740 (Zhu et al., 2009), and introduced into strain 2221ΔFasX. To select for loss of the integrated plasmid, and hence potential replacement of the chromosomally derived GGGG sequence with the plasmid-derived GCCG sequence, chloramphenicol-resistant transformants were grown for five passages in THY broth without antibiotics. Four of the five passages were for 4 h each in fresh THY broth, while one passage was grown overnight in fresh THY broth. After passaging, cultures were serially diluted and plated onto blood agar plates. Individual colonies were patched onto THY agar plates with and without chloramphenicol. To test whether chloramphenicol-sensitive colonies contained a wild-type or mutant 5′ska region we performed PCR and sequencing.
Construction of MGAS2221 and 2221ΔFasX derivatives containing 5′ extensions to the ska mRNA transcript
MGAS2221 and 2221ΔFasX derivatives containing two different 13 bp extensions inserted at the ska transcriptional start site, and therefore incorporated at the 5′ end of ska mRNA, were constructed. Overlap extension PCR was used to create 2 kb regions spanning the ska transcriptional start site, with the required 13 bp sequences inserted via sequences constructed into the PCR primers (Table S2). These PCR products were cloned into the suicide vector pBBL740 to create plasmids pSKA::EXY (the transcribed 13 bp insert is complementary to the 5′ end of wild-type ska mRNA) and pSKA::EXN (the transcribed 13 bp insert is not complementary to the 5′ end of wild-type ska mRNA). Insertion of the 13 bp sequences into the GAS genome was performed by allelic exchange as described for creation of strain 2221ΔX.GCSKA. Strains were verified by PCR and sequencing.
Construction of cvfA mutant and non-mutant derivatives of MGAS2221 and 2221ΔfasX
To create cvfA mutants of strains MGAS2221 and 2221ΔFasX we disrupted the cvfA gene by insertional inactivation. A central region of the cvfA gene was amplified by PCR, cloned into the suicide vector pBBL740 (creating plasmid pCVFAKO) and transformed into the two GAS strains. Disruption of the cvfA gene in chloramphenicol-resistant transformants was confirmed by PCR and sequencing. To ensure that mutant strain phenotypes were due to cvfA disruption and not due to a non-specific effect of pBBL740 integration into the chromosome we created a second pBBL740-based plasmid (pCVFAOK), this time containing the entire 3′ end of the cvfA gene. The design of pCVFAOK was such that integration of this construct into the chromosome did not disrupt the cvfA gene. Note that these plasmids were based upon those made in the published cvfA study (Kang et al., 2010).
Construction of pnpA mutant derivatives of MGAS2221 and 2221ΔfasX
To create pnpA mutants of strains MGAS2221 and 2221ΔFasX we disrupted the pnpA gene by insertional inactivation. A central region of the pnpA gene was amplified by PCR, cloned into the suicide vector pBBL740 (creating plasmid pPNPAKO) and transformed into the two GAS strains. Disruption of the pnpA gene in chloramphenicol-resistant transformants was confirmed by PCR and sequencing.
This research was funded in part by Grant 0865139F from the American Heart Association – South Central Affiliate (to P. Sumby), and by Grants R21AI078159 and R01AI087747 from the National Institute of Allergy and Infectious Diseases (to P. Sumby). We thank B. Lei for providing plasmid pBBL740. We also thank J.M. Musser, K.J. Pflughoeft and K. Stockbauer for critical reading of the manuscript.