We analysed the region encoding VR-RNA (VirR-regulated RNA), which has been reported to be positively regulated by the two-component VirR/VirS system in Clostridium perfringens. The VR-RNA promoter identified by primer extension analysis was preceded by a probable VirR-binding site (CCAGTTNNNCAC), which resembled a repeated sequence motif present in the promoter region of the theta-toxin (pfoA) gene. A VR-RNA-null mutant, constructed by a homologous recombination, exhibited a reduced amount of transcription of the alpha- (plc) and kappa-toxin (colA) genes, which was restored by the complementation of intact VR-RNA, indicating that the VR-RNA region plays an important role in the regulation of the plc and colA genes in C. perfringens. It was found that the regulatory effect was observed even when the hyp7 gene encoded on VR-RNA was deleted or a nonsense mutation was introduced in the hyp7-coding region. We found that the small 3′-portion of VR-RNA was sufficient for the activation of toxin genes, which suggested that VR-RNA itself could act as an RNA regulatory molecule for the plc and colA genes mediating the regulatory information from the VirR/VirS system in C. perfringens.
The Gram-positive anaerobic pathogen, Clostridium perfringens, produces a number of extracellular toxins that are believed to play important roles in its pathogenicity (gas gangrene or clostridial myonecrosis) (Hatheway, 1990; Petit et al., 1999). The roles of these toxins in its pathogenicity have been well studied (Rood and Cole, 1991), and it has been shown that some of these toxins act in concert for the pathogenesis of this organism (Awad et al., 1995). Further studies of the mechanism of the genetic regulation of these toxin genes is important for understanding the virulence of C. perfringens.
A two-component regulatory system, VirR/VirS, has been reported to be involved in global regulation of the production of theta-toxin (or perfringolysin O), kappa-toxin (or collagenase), alpha-toxin (or phospholipase C), sialidase, protease, and haemagglutinin in C. perfringens (Lyristis et al., 1994; Shimizu et al., 1994). In a previous study, it was reported that the VirR/VirS system regulated the expression of the plc (alpha-toxin), pfoA (theta-toxin), and colA (kappa-toxin) genes at the transcriptional level (Ba-Thein et al., 1996). Primer extension analysis revealed the VirR/VirS-dependent and independent promoters for the pfoA and colA genes, and a single VirR/VirS-dependent promoter for plc (Ba-Thein et al., 1996). As there was no conserved DNA motif to which the phosphorylated VirR proteins could bind in their promoter regions (Ba-Thein et al., 1996), it was suggested that complex regulatory networks might be involved in toxin production in C. perfringens (Shimizu et al., 1997). Recently, three clones, that contained genes regulated by the VirR/VirS system, were identified using a differential display method; two of these were positively regulated and one was negatively regulated at the level of transcription (Banu et al., 2000). The VirR/VirS system was shown to positively regulate the ptp (protein tyrosine phosphatase), cpd (2′, 3′-cyclic nucleotide phosphodiesterase) and hyp7 (hypothetical 7 kDa protein) genes, and to negatively control the ycgJ, metB (cystathionine gamma synthase), cysK (cysteine synthase) and ygaG genes, together with other toxin genes (Banu et al., 2000; Ohtani et al., 2000). The hyp7 gene was thought to act as a secondary regulator that specifically controlled the colA and plc genes but not the pfoA gene (Banu et al. 2000). These findings indicate that the VirR/VirS system is a bifunctional regulatory system that controls many virulence and/or housekeeping genes both positively and negatively, possibly through many yet unidentified pathways in C. perfringens.
In this study, we analysed the hyp7 region in more detail and found a probable VirR-binding site in its promoter region that resembled a repeated sequence motif previously reported as a VirR-binding site in the pfoA promoter region (Cheung and Rood, 2000). Interestingly, in this study, a regulatory RNA molecule (VR-RNA), rather than the hyp7 gene product, was found to be responsible for the transcriptional regulation of the colA and plc genes in C. perfringens.
Detailed analysis of the hyp7 region
In a previous study, the VirR/VirS system of C. perfringens was found to regulate the transcription of several genes other than toxin genes (Banu et al., 2000). One of the genes, the hyp7 gene on clone pSB10291, was shown to be positively regulated by the VirR/VirS system and seemed to be transcribed on a 0.4 kb RNA (designated VR-RNA, VirR/VirS-regulated RNA) (Fig. 1A). The hyp7 gene was also transcribed on a VirR/VirS-regulated 1.2 kb mRNA that was thought to be a read-through transcript from the same promoter of VR-RNA, encoding both the hyp7 and hyp23 genes (Fig. 1A) (Banu et al., 2000). The hyp7 gene seemed to encode a peptide of 72 aa, starting from a rare start codon TTG, and a putative ribosome-binding site was found in the 5′-region of the hyp7 coding region (Fig. 1B). There was another gene, mutT, which seemed to be transcribed divergently from VR-RNA (Fig 1A and B). To identify the transcription start sites of mutT, VR-RNA, and hyp23, a series of primer extension experiments was carried out (Fig. 1C). The transcription start site of VR-RNA (P2) was located at G in position 755 (Fig. 1B). The extended product of VR-RNA was abundant in wild-type strain 13 but not detected in the virR/virS mutant strain TS133 (Fig. 1C), confirming that the transcription from the P2 start site was positively regulated by the VirR/VirS system. On the other hand, we detected start sites for the mutT and hyp23 genes (P1 and P3 respectively; Fig. 1B), although the amount of the extended products was the same in strains 13 and TS133 (Fig. 1C). From these observations, it was concluded that only the transcription from P2 of VR-RNA was positively regulated by the VirR/VirS system. The corresponding consensus promoter sequence (–35 and/or –10) was deduced from each of the mRNA start sites (Fig. 1B), and we noticed that the sequence located just upstream of the –35 region of the P2 VR-RNA promoter was quite similar to the upstream repetitive sequence of the promoter of pfoA (Ba-Thein et al., 1996). A sequence motif of CCAGTTNNNCAC identified as the VirR-binding site of the pfoA gene was also found at almost the same position in the VR-RNA promoter region as in the pfoA promoter region, though the motif being not repeated in the former (Fig. 1D). These data suggested that this DNA sequence motif found in the VR-RNA promoter might be a consensus VirR-binding site that was recently reported in the pfoA promoter region (Cheung and Rood, 2000).
Effect of the VR-RNA on toxin expression
In a previous study, it was shown that the VirR/VirS-regulated hyp7 gene is involved in the transcriptional activation of the colA and plc genes but not of the pfoA gene in C. perfringens (Banu et al., 2000). To investigate the regulatory function of the hyp7 gene on VR-RNA more precisely, we attempted to make an isogenic mutant of the VR-RNA region using a double cross-over homologous recombination in the wild-type strain 13. The region corresponding to VR-RNA was deleted from pSB10291 (position 800–1152 in Fig. 1B), and the erythromycin-resistant gene (ermBP), amplified from pJIR418 by polymerase chain reaction (PCR), was replaced with the VR-RNA region. The resulting suicide plasmid (constructed on pUC19) was electroporated into the wild-type C. perfringens strain 13. The transformants were selected on erythromycin–blood agar plates and several colonies that showed weak alpha-toxin-induced haemolysis were selected. Southern analysis using the VR-RNA probe re-vealed that the haemolysis-weak mutant (named TS140) did not contain the VR-RNA region in its chromosome (Fig. 2A), and the size of the HindIII fragment was longer in TS140 as a result of the replacement of the VR-RNA region (0.5 kb) with the ermBP gene (1.1 kb) (Fig. 2A). These findings confirmed that the VR-RNA region of the mutant strain TS140 was deleted from its chromosome by a double cross-over event. Northern analysis of the mutant strain showed that the transcrip-tion of the colA and plc genes was greatly reduced in TS140/pJIR418 compared with that in the wild-type strain 13/pJIR418 (Fig. 2B, lanes 1 and 3). The level of expression of the colA and plc genes in TS140/pJIR418 was almost the same as in the virR mutant strain TS133/pJIR418 (Fig. 2B, lanes 2 and 3). Transformation of the intact VR-RNA region into TS140 (TS140/pSB1031) restored the expression of the colA and plc genes (Fig. 2B, lane 4). However, the expression of the pfoA gene was not greatly affected in TS140/pJIR418 and TS140/pSB1031, whereas the pfoA mRNA was not seen in TS133/pJIR418 (Fig. 2B, lane 2), indicating the expression of pfoA was dependent on the VirR/VirS system but not on VR-RNA. These data clearly indicated that TS140 had a deletion of the VR-RNA region in its chromosome, which led to the reduced expression of the colA and plc genes.
Deletion and functional analysis of VR-RNA
To confirm whether the hyp7 gene product was involved in the regulation of toxin genes, we carried out a series of deletion experiments against the VR-RNA region on plasmids. Various portions corresponding to the internal parts of VR-RNA were deleted, and the resulting plas-mids were introduced into strain TS140 to observe their complementation effects (Fig. 3). Strikingly, pSB1040, pSB1046, and pSB1047, whose VR-RNAs had a deletion of positions 806–951, 969–998, and 969–1036, respectively, could activate the transcription of the colA and plc genes in the same manner as the intact VR-RNA (Fig. 3, lanes 1, 7 and 8). Furthermore, pSB1048 and pSB1049, which had large deletions in positions 755–1036 and 755–1045, respectively, still had the ability to activate both toxin genes in TS140 (Fig. 3, lanes 9 and 10). These data indicated clearly that the active region of VR-RNA lay within a small 3′-portion of VR-RNA (position 1045 to putative terminator), and that the positive regulatory effect was not derived from a hypothetical 7 kDa product of the hyp7 gene.
To verify the conclusion that the VR-RNA-encoded polypeptide (Hyp7) is not involved in the toxin regulation, a nonsense mutation was created in the coding region of the hyp7 gene (position 828, A to T, creating a TAA stop codon in place of the TTA codon encoding leucine) by a site-directed mutagenesis (Fig. 4 and Fig. 1B). The resulting VR-RNA on pSB1050 was introduced into strain TS140, and the mutated VR-RNA was successfully ex-pressed in TS140/pSB1050 (Fig. 4, lane 3). Strikingly, the mutant VR-RNA could still activate the transcription of the colA and plc genes in just the same manner as the intact VR-RNA on pSB1031 (Fig. 4, lanes 2 and 3). Taken together, these facts obviously indicated that VR-RNA itself could act as a regulatory RNA molecule for the colA and plc genes, and that the active portion is located in a small 3′-portion of VR-RNA.
Effect of VR-RNA on other VirR/VirS-regulated genes
As the VirR/VirS system regulates several genes other than the toxin genes in C. perfringens (Banu et al., 2000), we examined the regulatory effect of VR-RNA on the ptp, cpd and ycgJ-metB-cysK-ygaG genes. The transcription of the ptp and cpd genes decreased in the VR-RNA mutant TS140/pJIR418, and strain TS140/pSB1031, having the intact VR-RNA, restored the expression of ptp and cpd to the level of the wild-type strain (Fig. 5). In addition, the expression of the ycgJ-metB-cysK-ygaG operon apparently increased in TS140/pIJR418, whereas strain TS140/pSB1031 showed reduced level of ycgJ-metB-cysK-ygaG transcription compared with the VR-RNA mutant strain (Fig. 5), indicating that the operon is negatively regulated by VR-RNA. These data clearly indicated that VR-RNA also regulates the expression of ptp, cpd and ycgJ-metB-cysK-ygaG genes just in the same manner as the VirR/VirS system does (Banu et al., 2000).
In this study, we analysed the VR-RNA region that was identified as one of the VirR/VirS-regulated clones in C. perfringens (Banu et al., 2000). The VirR/VirS-regulated promoter of VR-RNA (P2) was identified, and the se-quence motif (CCAGTTNNNCAC) found in the upstream region of the VR-RNA promoter was quite similar to that in the upstream region of the pfoA gene, which was shown to be a direct binding site of VirR protein (Cheung and Rood, 2000). The presence of a VirR-binding site in both VR-RNA and pfoA may facilitate the study of the genetic regulation in C. perfringens, i.e. by screening of the VirR-binding site on the C. perfringens whole genome sequencing in progress in our laboratory, to identify other VirR/VirS-regulated genes in silico.
The most striking finding in this study is that VR-RNA itself, rather than the hyp7 product, was found to be responsible for the transcriptional regulation of the colA, plc, ptp, cpd and ycgJ-metB-cysK-ygaG genes. We analysed the secondary structure of the 386 nt VR-RNA from the transcription start site to the poly T tract of the putative rho-independent terminator using a computer program (Fig. 6). The predicted structure of whole VR-RNA had a tight and compact secondary structure overall, and its 5′- and 3′-ends were predicted to pair and to have a strong axis of symmetry (Fig. 6). However, from the deletion analysis of VR-RNA, the active portion was located in a small 3′-region (Fig. 6), suggesting that not all but only the 3′-region of VR-RNA is sufficient to act as a positive regulator for alpha- and kappatoxin genes. The clone that had the 5′-region but not the 3′-region of VR-RNA (pSB1044) could not activate the toxin genes (Fig. 3), indicating that a sequence of this region, rather than the secondary structure of VR-RNA, seems to be responsible for its regulatory function. The significance of the 3′-region sequence in the regulatory function of VR-RNA should be elucidated by further experiments.
Much like VR-RNA, the regulatory RNAs for virulence factors have been also reported in Staphylococcus aureus and Streptococcus pyogenes, both of which are Gram-positive pathogens (Novick et al., 1993; Kreikemeyer et al., 2001). A regulatory RNA molecule, RNAIII of S. aureus, has been reported to be involved in the regulation of the production of many toxins and surface proteins. RNAIII is a 0.5 kb RNA molecule en-coded by the agr locus in S. aureus. In the agr locus, there is another transcriptional unit named RNAII that transcribes agrB, agrD, agrC and agrA, in that order. The products of agrA and agrC highly resemble a response regulator and sensor histidine kinase of a bacterial two-component system respectively. The transcription of RNAIII has been shown to be positively regulated by the AgrA/AgrC system and the direct effector of the agr system has been shown to be RNAIII itself, rather than the delta-haemolysin encoded on RNAIII. The predicted structure of RNAIII has a strong axis of symmetry and its 5′- and 3′-ends are predicted to pair (Novick et al., 1993). Recently, the fasX RNA in S. pyogenes was reported to regulate fibronectin/fibrinogen-binding protein and other secreted virulence factors in a growth-phase-dependent manner (Kreikemeyer et al., 2001). From the point of view of genetic evolution, it is very interesting to investigate how Gram-positive pathogens that produce numerous extracellular toxins have come to control their virulence through these unique regulatory RNA molecules.
In C. perfringens, it is now clear that the two-component VirR/VirS system involves a regulatory cas-cade. The VirR/VirS system positively regulates the transcription of VR-RNA, possibly by direct binding of the VirR protein to the consensus VirR-binding site. Then VR-RNA positively regulates the transcription of the colA, plc, ptp and cpd genes, and negatively controls the ycgJ-metB-cysK-ygaG genes, mediating the regulatory signal from the VirR/VirS two-component system (Fig. 7). Compared with colA, the transcription of plc is only partially regulated by VR-RNA, which may be as a result of the fact that the plc gene has single constitutive promoter whose expression is enhanced by the VirR/VirS system (Ba-Thein et al., 1996). The transcription of the pfoA gene would be regulated directly by the VirR protein via a direct binding of the VirR protein to its promoter, independently of the VR-RNA cascade (Fig. 7). Thus, C. perfringens has a regulatory cascade to control its virulence and housekeeping genes, and it would be interesting to investigate other regulatory networks controlling the pathogenicity of this organism.
We compared the sequence of the active 3′-region of VR-RNA with those of the promoter regions of VR-RNA-regulated genes but could not find any similar sequences, which implies that VR-RNA does not regulate the expression of genes by annealing to the complementary DNA or RNA of their promoter region. Instead, it is highly possible that unknown proteinaceous factors are involved in the regulation through formation of a ribonucleoprotein complex, just as proposed in RNAIII (Novick et al., 1993) and that the 3′-region of VR-RNA is important for an interaction with the putative proteins. We examined whether the hyp23 gene encodes a VR-RNA-binding protein, as hyp23 is partly transcribed in the 1.2 kb mRNA, together with VR-RNA (Fig. 1), and was thought as a candidate for a possible VR-RNA-binding protein. A hyp23 mutation in strain 13 was introduced by a single cross-over homologous recombination, but no influence of the hyp23 mutation on the expression of colA and plc was observed, although slight reduction of the pfoA mRNA was seen in the mutant (data not shown). These facts indicate that the hyp23 gene is not involved in the VR-RNA-mediated regu-lation, leaving a possibility that it has other regulatory function on pfoA. Further extensive analysis will be needed to elucidate the precise regulatory mechanism of VR-RNA on the virulence in C. perfringens, and an extensive search for the putative VR-RNA binding protein is currently undertaken in our laboratory.
General recombinant DNA techniques were performed as described by Sambrook and colleagues (Sambrook et al., 1989) unless otherwise noted. C. perfringens strains were transformed by an electroporation-mediated transformation as described previously (Shimizu et al., 1994). Site-directed mutagenesis was performed using a Takara LA PCR in vitro mutagenesis kit (Takara Shuzo) according to the manufacturer’s instructions. Deletions on pSB1031 (Banu et al., 2000) were made as follows. The appropriate primer set was designed to inversely amplify the pSB1031, including vector pJIR418, and the inverse PCR was performed. The resulting PCR product was treated with a BKL kit (Takara Shuzo) to fill up and phosphorylate both termini, and then self-ligated to give a circular plasmid that lacked the region between the primers. The deletion end-points were confirmed by nucleotide sequencing with reverse or universal primers using a Big-Dye terminator reaction kit and ABI 310 se-quencer (Applied Biosystems).
Primer extension experiments were performed basically as described previously (Ba-Thein et al., 1996) using synthetic oligonucleotide primers, VR-RNA-1 (5′-TCTCCCCTTTTTAAGTTTTTCTT-3′), mutT-1 (5′-AATTTTTTCTTTTCCCTAAACTCTC-3′) and hyp23–1 (5′-CAAATAAGCGCCTTACAAA-3′), which were complementary to the VR-RNA, the mutT gene (Banu et al., 2000), and the hyp23 gene (Banu et al. 2000) respectively. Of each primer, 10–50 pmoles was end-labelled with 50 μCi of [γ-32P]-ATP (>7000 Ci mmol–1, Amersham Pharmacia Biotech) and 10 units of T4 polynucleotide kinase. The primer was annealed to 50 μg of total RNA, and the annealed primer was extended by AMV reverse transcriptase (Bethesda Research Laboratories) with 0.25 mM dNTPs at 42°C. The extension products were separated by electrophoresis on a sequencing gel, together with a nucleotide sequence reaction that had been carried out with the appropriate plasmid template using the same primer.
This work was supported by a Grant-in Aid for Scientific Research on Priority Areas (C) ‘Genomic Biology’ from the Ministry of Education, Science, Sports and Culture of Japan. K.O. is a Research Fellow of the Japan Society for the Promotion of Science.