Stabilization of Clostridium perfringens collagenase mRNA by VR-RNA-dependent cleavage in 5′ leader sequence

Authors


E-mail nakamura.kouji@biol.tsukuba.ac.jp; Tel. (+81) 298 53 6419; Fax (+81) 298 53 7723.

Summary

The small RNA (sRNA), VR-RNA that is directly regulated by the VirR/VirS two-component system, regulates many genes including toxin genes such as collagenase (colA) and phospholipase C (plc) in Clostridium perfringens. Although the VR-RNA 3′ region is sufficient to regulate the colA and plc genes, the molecular mechanism of toxin gene regulation by VR-RNA remains unclear. Here, we found that colA mRNA is cleaved at position −79 and −78 from the A of the first codon (ATG) in the presence of VR-RNA. The processed transcripts were stable compared with longer intact transcripts. On the other hand, colA mRNA was labile in a VR-RNA-deficient strain, and processed transcripts were undetectable. The stability and processing of colA mRNA were restored by transformation of the 3′ region of VR-RNA-expression vector. The 3′ region of VR-RNA and colA mRNA had significant complementation and interacted in vitro. These results show that VR-RNA base pairs with colA mRNA and induces cleavage in the 5′ untranslated region (UTR) of colA mRNA, which leads to the stabilization of colA mRNA and the activation of colA expression.

Introduction

Small RNAs (sRNAs) are important for the regulation of genes that should be rapidly controlled in response to environmental changes (Telodo-Arane et al., 2007). Pathogenic bacteria must adapt to changing conditions and host signals for effective infection. Several sRNAs are involved in the regulation of virulence-associated genes and are essential to the establishment of successful infection (Romby et al., 2006). In many cases, sRNAs regulate the translation and/or stability of target mRNA through forming base pairing with the target mRNA. Most of these sRNAs bind around the Shine–Dalgarno (SD) sequence in the 5′ untranslated region (UTR) of its target mRNA, which causes translational repression and frequently destabilizes the mRNA (Gottesman, 2005; Waters and Storz, 2009). Other sRNAs activate the translation and stability of their target mRNAs by base pairing far upstream of the SD sequence, which promotes ribosome entries (Fröhlich and Vogel, 2009). Thus, many sRNAs regulate the translational efficiency of target genes and affect the amount and stability of target mRNA. One established virulence-associated sRNA, Staphylococcus aureus RNAIII, which is the effecter of the agr two-component system, base pairs with at least hla, spa and rot target mRNAs (Boisset et al., 2007). It binds to the 5′ UTR of the hla mRNA where the SD sequence is sequestered by the formation of an intermolecular stem-loop structure, which prevents the inhibitory structure and results in the activation of α-haemolysin (hla product) translation (Morfeldt et al., 1995). The RNAIII base pairs with the 5′ UTRs of spa and rot, which encode protein A and the transcription factor Rot, respectively, including the SD sequence of the target mRNA, which inhibits the ribosome entry to the SD sequence and the translation of target genes (Huntzinger et al., 2005; Geisinger et al., 2006). Thus, RNAIII sRNA plays a key role in the regulation of virulence-associated genes though antisense mechanisms.

Clostridium perfringens is a Gram-positive anaerobic spore-forming bacterium that is widespread in environmental soil and sewage, as well as in animal intestines. It is also a causative agent of gas gangrene in humans, and it produces numerous extracellular enzymes and toxins (Rood and Cole, 1991; Petit et al., 1993). C. perfringens has only a few genes that encode enzymes for amino acid biosynthesis and thus needs to generate nutritional sources through host cell degradation (Shimizu et al., 2002a). Therefore, the expression of genes encoding extracellular enzymes and toxins should be tightly controlled in concert for efficient infection of host cells and proliferation. The VirR/VirS two-component system is a key global regulator of C. perfringens gene expression (Shimizu et al., 1994; Okumura et al., 2008; Ohtani et al., 2010). Phosphorylated VirR directly binds to a target promoter with two direct repeated sequences (VirR-box) and then activates the transcription of target genes including theta-toxin (pfoA) and alpha-clostripain (ccp) (Cheung and Rood, 2000; Okumura et al., 2008). The VirR/VirS system also activates the vrr gene that encodes VR-RNA, which is a 386 nt sRNA that regulates expression of the plc (α-toxin), colA (κ-toxin), cpd2 and several housekeeping genes (Shimizu et al., 2002b; Kawsar et al., 2004; Ohtani et al., 2010). Thus, the VirR/VirS system also affects the expression of many genes through regulation of the effecter molecule, VR-RNA. However, how VR-RNA regulates the target genes and which gene is the direct target of VR-RNA remains unclear.

The amount of colA mRNA was significantly reduced in a vrr-deficient strain of C. perfringens, suggesting that VR-RNA is the activator of colA (Shimizu et al., 2002b). Moreover, the 3′ region of VR-RNA (residues 291–386) is responsible for regulating both the colA and plc genes, and it plays an important role in the VR-RNA regulatory mechanism (Shimizu et al., 2002b). Previous findings suggest that the colA gene has seven 5′ ends (P1–P7), consisting of major (P1, P2, P6 and P7) and minor (P3, P4, P5) transcripts (Fig. 1A), and that P6 and P7 transcripts expression depends on the VirR/VirS system (Ba-Thein et al., 1996). Thus, VR-RNA would apparently control the transcription of one of the major promoters from which P6 and P7 transcripts are derived. However, convincing similarity to the σA consensus sequence is not found upstream of P6 and P7 sites. This suggests that the actual colA promoter has not yet been defined, and that complex post-transcriptional regulation is involved in colA expression in addition to transcriptional regulation.

Figure 1.

Northern blot analysis of plasmid-borne RNA.
A. Schematic of plasmid construction. Genetic map of the colA region is shown at top. Transcriptional start sites identified by Shimizu et al. (1994) are shown (P1–P7). Arrow indicates colA transcriptional start site identified herein. Thick lines below map indicate argR–colA intergenic region (IGR) and colA–mscL IGR cloned into pJIR418. Numbers (relative to ATG; A is +1) indicate region deleted from plasmid. Open rectangles represent multi-cloning sites (MCS) of pJIR418. Filled triangles indicate processing sites of colA mRNA.
B. Northern blot analysis with VR-RNA, colA or plasmid-specific probes. Total RNA from cultures grown at 37°C for 2 h was resolve on 6% polyacrylamide/7 M urea or 1.0% denaturing agarose. Total RNA (5 µg) was loaded to detect plasmid-coded RNA, and 2 µg for VR-RNA and chromosomal colA. Ethidium bromide-stained 5S rRNA is also shown.

Here, we found that the P6 and P7 transcripts of colA are the result of processing events from longer primary transcripts, and identified the sole colA promoter, known as a minor promoter from which the P5 transcript was derived. Processing in the colA mRNA 5′ UTR was involved in the stabilization of colA mRNA, and induced in the presence of VR-RNA. The processed transcripts were more stable than the primary message, and correlated with the total amount of colA mRNA. Moreover, VR-RNA bound to a colA mRNA 5′ UTR near the processing site in which the sequence had significant complementarity with the 3′ region of VR-RNA. These findings suggest that VR-RNA induces processing of the colA mRNA at a position corresponding to the P6 and P7 sites, and activates colA expression by stabilizing the colA message through an antisense mechanism.

Results

Identification of the colA promoter

Several transcriptional start sites of the colA gene have already been assigned by primer extension analysis (Ba-Thein et al., 1996). To verify which promoter is actively functional, the −543 to −1 (relative to ATG; A is +1) upstream region of colA was introduced into the Escherichia coli–C. perfringens shuttle plasmid pJIR418 (Sloan et al., 1992) together with 3′ UTR of colA, the +3316 to +3462 (relative to ATG; A is +1) downstream region of colA, which includes ρ-independent transcriptional terminators (Fig. 1A). We also constructed derivative plasmids with various deleted regions. The resulting plasmids pCP1, pCP6, pCP1Δ2 and pCPΔ3 were introduced into the C. perfringens wild-type strain (strain 13). The transcripts from these plasmids were smaller than the chromosomal colA mRNA since the coding regions were deleted and distinguishable from the chromosomal copy on Northern blots. Total RNAs were prepared from cells harbouring each plasmid and used for Northern blot analysis. Figure 1B shows the three species of transcripts that were detected in the RNA prepared from the cells harbouring pCP1 and pCP6. The longest major transcript was approximately 300 nt, and two minor transcripts were 280 and 240 nt. One major transcript of about 250 nt was detected in cells harbouring pCP1Δ3. The difference in the size of the major transcripts between pCP1 and pCP1Δ3 accorded with the size of region deleted from pCP1Δ3. Moreover, no transcripts were detectable in cells harbouring pCP1Δ2. These results suggest that the active promoter of colA locates between −167 and −117 (Fig. 1A).

We performed primer extension analysis to determine the 5′ ends of the colA transcript, using a plasmid sequence-specific primer that anneals to a multi-cloning site (MCS) in the plasmid sequence. The longest extension product corresponding to a 310 nt transcript was detected from cells harbouring pCP1 (Fig. 2A). The 5′ end of this transcript is A at position −141. Moreover, the 5′ end of the major transcript was also assigned to the same position in cells harbouring pCP1Δ3. The difference in size between the extension products from cells harbouring pCP1 and pCP1Δ3 corresponded to the size of the deletion in pCP1Δ3. A potential promoter sequence recognized by RNA polymerase σA[ATAAAA–(N)17–TATAAT] was observed upstream of the proposed transcriptional initiation site (Fig. 2B).

Figure 2.

Determination of 5′ end of colA mRNA.
A. 5′ ends of colA mRNA were determined using primer extension analysis. Complementary DNA was synthesized from total RNA of strain 13 harbouring pCP1 or pCP1Δ3 and 32P-labelled primer. The colA promoter sequence is shown on the right.
B. Sequence of argR–colA IGR. Transcriptional start site of colA is indicated by black arrow. Putative −35 and −10 sequence regions are boxed in dashed line. SD sequence and start codon of colA are shown in bold. Processing sites of colA mRNA are indicated by filled triangles.

VR-RNA-dependent processing in 5′ UTR of the colA primary transcript produces the shorter transcript

Two shorter transcripts (280 and 240 nt) were detected in cells harbouring pCP1 and pCP6 (Fig. 1B). Primer extension analysis suggested that their 5′ ends are T and G at positions −117 and −116 and G and T at positions −79 and −78 respectively (Fig. 2). However, no possible promoter sequence was identified upstream of the former site, and the latter site was undetectable in cells harbouring pCP1Δ2. Therefore, these fragments might have been created by cleavage of the primary transcripts. The processed and primary types were distinguishable because the 5′ ends of the primary and processed transcripts were triphosphorylated and monophosphorylated respectively. To determine whether or not short colA mRNA was the processing product, we applied 5′ RNA ligase-mediated rapid amplification of cDNA ends (RLM-RACE). Primary transcripts were amplified by PCR when total RNA was treated with tobacco acid pyrophosphatase (TAP). In contrast, only processed transcripts were amplified from total RNA that was not treated with phosphatases, since primary transcripts with triphosphorylated 5′ ends cannot ligate to the RNA adapter and thus cannot be amplified. PCR products about 170 bp were amplified from an RNA sample that was not treated with phosphatases to detect the processed transcripts (Fig. 3, lane 3). This PCR product was not amplified in RNA digested with CIAP (Fig. 3, lanes 1 and 2). The digestion of total RNA prepared from strain 13 with CIAP/TAP resulted in amplification of PCR products about 250 bp that corresponded to the primary transcripts (Fig. 3, lane 2). In this condition, CIAP treatment was sufficient to remove 5′-monophosphate group but not 5′-triphosphate group completely. Each PCR product was cloned and sequenced (data not shown), showing that the 5′ end of the primary transcript was A at position −141 and that of the smaller processed transcript was G and T at positions −79 and −78 respectively (Fig. 2B). No PCR product with a 5′ end corresponded to the 280 nt minor transcript shown in Fig. 1B. Moreover, the primary transcripts, but not the shorter processed transcripts, were amplified in RNA samples derived from VR-RNA-deficient TS140 cells (Fig. 3, lanes 5 and 6). These results imply that the primary transcript is cleaved at −79 or −78 in wild-type cells by a VR-RNA-dependent mechanism. Thus the previously identified VirR/VirS-dependent P6 and P7 transcripts result from processing of the P5 primary transcripts (Ba-Thein et al., 1996).

Figure 3.

5′ RLM-RACE analysis of colA transcripts. Total RNAs extracted from strain 13 (lanes 1–3) or TS140 (lanes 4–6) served as substrates for RLM-RACE. PCR products were resolved on 6% polyacrylamide/7 M urea gels and stained with ethidium bromide. Total RNA treated with (+) or without (−) TAP or CIAP. PCR products corresponding to cDNAs from primary or processed colA mRNA are shown by arrows.

VR-RNA activates processing in the colA 5′ leader sequence and regulates the stability of the colA mRNA

Most sRNAs base pairing with the target mRNA regulate the stability of the mRNA (Waters and Storz, 2009). To determine whether or not VR-RNA affects the stability of target colA mRNA, we analysed the half-life of the colA mRNA. C. perfringens wild-type (strain 13/pJIR418), VR-RNA-deficient strain (TS140/pJIR418) and TS140 harbouring VR-RNA-expression vector (TS140/pVrr) were grown to the mid-exponential phase, and then incubated with rifampicin. The colA mRNA persisted for almost 4 min after adding rifampicin to strain 13 but was hardly detectable at 2 min after adding rifampicin to TS140, and this transcript stability was restored by VR-RNA expressed from the plasmid (Fig. 4A). These results suggest that VR-RNA blocks degradation and regulates the stability of the colA transcript. We examined the RNA samples described in Fig. 4A by primer extension analysis. A primer oligonucleotide was designed to anneal to the chromosomal colA gene. The longer primary transcripts were detected in all RNA samples at 0 min, including that from VR-RNA-deficient cells (Fig. 4B, lanes 1, 5 and 9). These primary transcripts immediately disappeared when transcription was blocked by rifampicin. The shorter processed transcript was detected in wild-type cells and in TS140 cells harbouring pVrr (lanes 1–4 and lanes 9–12). These processed transcripts were stable and detectable at 8 min after adding rifampicin, but were undetectable in TS140 cells (lanes 5–8). The stability of the processed transcripts reflected the total amount of colA mRNA detected by Northern blot analysis (Fig. 4A), suggesting that expression of the colA gene depended on the amount of the processed mRNA. Expression of the processed transcript was restored by transformation of the VR-RNA-expression vector, suggesting that VR-RNA accelerates colA mRNA processing and that the processed type is more stable than the primary colA mRNA.

Figure 4.

Processing within colA mRNA 5′ UTR correlates with stability.
A. Stability of colA mRNA in strain 13 and TS140. C. perfringens strain 13 or TS140 harbouring pJIR418 and TS140 harbouring pVrr were grown at 37°C for 2 h. Total RNA was isolated from the culture prior to or at indicated time after adding rifampicin (200 µg ml−1), and 5 µg of total RNA was subjected to Northern blot analysis using VR-RNA- or colA-specific probes. The 16S and 23S rRNAs on the blot detected by methylane blue staining are also indicated on bottom.
B. Processing within colA mRNA 5′ UTR depends on VR-RNA. Primer extension analysis proceeded using 32P-labelled primer and total RNA shown in (A). Promoter sequence for colA was determined by sequence reaction using same primer. Sequences of the transcriptional start and processing sites are indicated on right.

The 3′ region of VR-RNA is required for VR-RNA-dependent processing in colA mRNA

The 3′ region of VR-RNA was responsible for colA regulation (Shimizu et al., 2002b), suggesting that colA mRNA processing requires the 3′ region of VR-RNA. We then examined the roles of the 3′ region of VR-RNA on the 5′ processing of the colA mRNA. Total RNAs were isolated from TS140 strains carrying plasmids containing VR-RNA variants in which internal parts of 3′ region were deleted (Shimizu et al. 2002b; Fig. 5A) and examined by primer extension analysis. Processing of colA mRNA occurred when VR-RNA variants had a deletion from 216 to 281 or 1 to 290 (pSB1047 or pSB1049; Fig. 5B). On the other hand, 5′ processing of colA mRNA was undetectable in cells expressing VR-RNA lacking the 3′ region (pSB1042 and pSB1043; Fig. 5B). The deletion region in pSB1043 contains a portion of transcriptional terminator structure, which might cause the low abundance of the VR-RNA expressed from pSB1043. However, the amount of the transcripts is almost same as genomic VR-RNA. Thus, undetectable processed transcripts were caused by lacking the 3′ region of VR-RNA but not low expression. The expression profiles of whole colA mRNA detected by Northern blot analysis correlated with the amount of processed colA mRNA. These data demonstrate that the 3′ region located at +292 to +316, of VR-RNA is important for processing colA mRNA, and thus retains indispensable activity for colA regulation.

Figure 5.

The 3′ region of VR-RNA is responsible for colA mRNA processing.
A. Schematic map of plasmids expressing VR-RNA with deletions. Nucleotide numbers (relative to transcriptional start site +1) indicate regions deletion from VR-RNA-coding sequence. Predicted sizes of full-length and deletion VR-RNA are indicated on right.
B. Processed colA mRNA expression is activated in cells expressing 3′ region of VR-RNA. C. perfringens strain 13 harbouring pJIR418 and TS140 harbouring pJIR418, pVrr, pSB1042, pSB1044, pSB1047 or pSB1049 were grown at 37°C for 2 h. Primer extension analysis proceeded using 32P-labelled primer and 8 µg of total cellular RNA to detect primary and processed colA mRNA (top panel and second panel from top). Total RNA (2 µg) was used for Northern blot analysis of colA and VR-RNA (third panel from top and bottom panel).

Removal of the processed region in colA 5′ leader sequence stabilizes the transcript

To examine the effect of the processing in the colA 5′ leader sequence on the mRNA stabilization, we constructed pCP1Δ5, in which transcription was initiated near position −79, and the transcript sequence was almost identical to that of the processed form (Fig. 6A). A smaller transcript about 250 nt length was expressed from pCP1Δ5 and detected by Northern blot analysis with a plasmid-specific probe (Fig. 6B, lanes 3 and 4). The transcript from pCP1Δ5 accumulated more stably in both strains 13 and TS140 compared with those in cells harbouring pCP1 (Fig. 6B). These results showed that processing within the 5′ UTR of colA mRNA is sufficient for colA mRNA stabilization.

Figure 6.

Processing stabilizes the colA mRNA 5′ UTR.
A. Schematics of pCP1 and pCP1Δ5 are as in Fig. 1A. Region +2 to +62 (relative to transcriptional start site +1) was deleted from pCP1 to generate pCP1Δ5.
B. Northern blot analysis with plasmid-specific probe. Total RNA (5 µg) from cultures grown at 37°C for 2 h was separated on 6% polyacrylamide/7 M urea or 1.0% denaturing agarose. The 5S rRNA detected by ethidium bromide staining is shown at bottom.

Effects of mutation in the colA SD sequence on the stability of the processed transcripts

The result of above figures suggested that the structure of the colA mRNA 5′ UTR related to the regulation of colA. Therefore we predicted secondary structure of the primary and processed colA mRNA 5′ UTR by mfold program (http://www.bioinfo.rpi.edu/applications/mfold/). As shown in Fig. 7, the colA SD sequence is masked by an intramolecular structure in the primary transcript (Fig. 7A). In contrast, a part of the stem-loop structure was destroyed in the processed transcript (Fig. 7B). Thus, ribosomes binding to the SD sequence and the translation might be inhibited in the primary transcripts but not in the processed transcripts in which the SD sequence is free from the stem-loop structure (Fig. 7B). It was suggested that initiating ribosomes caused the stabilization of colA mRNA in the presence of VR-RNA. Therefore, we made mutations in the SD sequence of pCP1 and pCP1Δ5 (Fig. 8A), designated as pCP1mutSD and pCP1Δ5mutSD, respectively, and detected the transcript from the plasmid (Fig. 8B). The mutation of the SD sequence decreased the amount of the processed transcripts (Fig. 8B, lanes 2 and 4), suggesting that ribosomes binding to the SD sequence stabilized the colA transcripts. It was noted that both pCP1 and pCP1Δ5 contained no start codon (Fig. 8A). Thus the stabilization of colA mRNA depends on initiating ribosomes but not the translation. We made mutations to introduce start codon in pCP1Δ5, designated pCP1Δ5plusAUG from which a 12-amino-acid peptide was translated. The amount of the processed transcripts from the pCP1Δ5plusAUG did not increase compared with that from pCP1Δ5 (Fig. 8B, lane 5). These data indicated that ribosomes binding to the SD sequence in colA mRNA stabilized the processed transcripts.

Figure 7.

Secondary structure of primary or processed colA mRNA 5′ UTR.
A. Predicted structure of colA mRNA from transcriptional start site to AUG start codon corresponding to primary colA mRNA 5′ UTR. Secondary structure was predicted using mfold web server. Predicted ΔG values (kcal mol−1) are indicated on top. SD sequence and start codon are shown.
B. Predicted secondary structure of processed colA mRNA 5′ UTR containing −79 to AUG start codon. Details are as described in (A).

Figure 8.

Initiating ribosomes are required to stabilization of processed transcripts.
A. Sequences around SD sequence of chromosomal colA mRNA, pCP1 and pCP1 derivatives. The mutated bases are indicated in bold forms and underlined.
B. Northern blot analysis with plasmid-specific probe. Total RNA (5 µg) from cultures grown at 37°C for 2 h was separated on 6% polyacrylamide/7 M urea and used for detection of plasmid-borne transcripts. The 5S rRNA detected by ethidium bromide staining is also shown.

VR-RNA-dependent processing activates ColA protein expression

Processing in the colA mRNA 5′ UTR is suggested to activate ribosomes binding to the SD sequence and the translation. To further analyse the function of VR-RNA on post-transcriptional colA regulation, the colA promoter, colA 5′ UTR and part of the colA ORF containing collagenase signal sequence (region −543 to +135 relative to ATG; A is +1) were fused in-frame to the GST gene (Fig. 9A, pCPE33). Because C. perfringens collagenase encoded by the colA gene is cleaved to produce mature extracellular protein, we performed Western blot analysis to detect GST fusion proteins with whole cell proteins and culture supernatants. A 30 kDa precursor protein and a 25 kDa mature protein were detected among intracellular proteins, whereas only the latter was detected among extracellular proteins (Fig. 9B, lanes 1 and 5). The expression of both the intracellular and extracellular ColA–GST fusion proteins was obviously decreased in TS140 compared with strain 13 (Fig. 9B, lanes 2 and 6). Furthermore, when the region from −142 to −80 was deleted from the colA 5′ UTR of pCPE33 same as pCP1Δ5 (Fig. 9A, pCPE30), from which the processed mRNA was transcribed, the amounts of expressed GST proteins considerably increased in strain 13 and TS140 (Fig. 9B, lanes 3, 4, 7 and 8). We also constructed a colAgst transcriptional fusion in which the colA promoter was fused to Bacillus subtilis spoVG (−47 to +45; +1 is A of the ATG start codon) and the GST coding region. Expression of the GST fusion protein in this construct was almost identical in both strain 13 and TS140 (Fig. 9C). These results indicated that VR-RNA regulated colA expression post-transcriptionally and that processing in the 5′ leader sequence of colA mRNA activated translation within the mRNA.

Figure 9.

VR-RNA affects expression of ColA–GST fusion protein.
A. Schematics show pCPE33, pCPE30 and pCPE36 vectors. The colA promoter, 5′ UTR and coding region (−543 to +135; relative to ATG; A is +1) were fused to gst in pCPE33. Arrow indicates the colA transcriptional start site. Number indicates deleted region in pCPE30. The colA promoter region (−408 to +1; relative to transcriptional start site +1) and Bacillus subtilis spoVG sequence (−47 to +45; +1 is first A of ATG start codon) were fused to gst in pCPE36.
B. Western and Northern blot analysis of strain 13 and TS140 carrying pCPE33 or pCPE30. Each lane was loaded with 0.02 A280 units of protein or 0.5 µg of total RNA from cells at mid-exponential growth phase. GST fusion protein was probed with anti-GST antibody (top panel). An arrow head and asterisks indicate premature and mature ColA–GST fusion proteins respectively. The gst mRNA was detected by probing with specific probes (bottom panel).
C. Western and Northern blot analysis of strain 13 and TS140 carrying pCPE36. Western blot shown here was exposed for longer than that shown in Fig. 7B to detect GST fusion protein from pCPE36. (Results of Western and Northern blot analysis are representative of at least three experimental trials).

Base pairing of VR-RNA and 5′ UTR of colA mRNA

Here, we showed that colA mRNA processing and stabilization requires the 3′ portion of VR-RNA (Fig. 5). A comparison of the colA mRNA 5′ UTR and the 3′ region of VR-RNA (residues 291–386) identified a complementary sequence (Fig. 10A), suggesting that VR-RNA regulates the expression of colA through interaction with the 5′ UTR of colA mRNA. We examined this notion using gel mobility shift assays. Radiolabelled colA mRNA (1 nM) was incubated with various amounts of unlabelled VR-RNA probe at 37°C for 30 min. The full-length VR-RNA probe bound to the colA mRNA probe (Fig. 10B, lanes 1–4). Moreover, a VR-RNA variant position from 281 to 374 interacted with colA mRNA (Fig. 10B, lanes 11–13), but not VR-RNA 1–96 and VR-RNA 103–201 (Fig. 10B, lanes 5–7 and lanes 8–10). Thus, the 3′ portion of VR-RNA is important for the regulation of colA expression and it is required to form a moderately stable complex with the 5′ leader sequence of colA mRNA.

Figure 10.

Potential colA mRNA 5′ UTR and VR-RNA base pairing and interaction in vitro.
A. Potential base pairing between colA mRNA and VR-RNA 3′ region. Numbers represent region of interaction (relative to ATG; A is +1). Filled triangles indicate processing sites on colA mRNA.
B. Interaction between colA mRNA 5′ UTR and VR-RNA 3′ region in vitro. Gel mobility shift assay with 1 nM of radio-labelled colA mRNA 5′ UTR and various amounts of VR-RNA (5, 10, 50 nM) or mutant VR-RNA (5, 10, 100 nM). Thick lines indicate free probe and colA mRNA 5′ UTR–VR-RNA complex.

Discussion

Our results showed that mRNA encoded by the collagenase gene (colA) is a direct target of the VR-RNA in C. perfringens and that the abundance of colA mRNA and protein is modulated by cleavage within the 5′ leader sequence of colA mRNA. They also showed that this cleavage requires the small regulatory RNA, VR-RNA, and regulates the stability of the colA transcript (Fig. 4B). We compared the sequence of the 3′ region (+291 to +386) of VR-RNA, which is required for VR-RNA function, with that of the 5′ leader of colA mRNA, and found significant complementarity between them (Fig. 10). Processing sites were located proximal to the ends of this complementary sequence (Fig. 10A). Therefore, the ribonuclease that functions in the processing of colA mRNA seemed to recognize the edges of single/double-stranded RNA.

Although several 5′ ends of the colA mRNA (P1–P7) had been identified previously (Ba-Thein et al., 1996), we indicated the only colA promoter which is corresponding to P5. We performed three independent primer extension analyses of colA using primers, CP1 (complementary to MCS in pCP1; Fig. 2), CP2 (complementary to +6 to +35 relative to the colA start codon; Fig. 4B) and CP3 (complementary to +1 to +28 relative to the colA start codon; data not shown). While some extension products upstream P5 were observed in the analyses using primers CP2 or CP3 (Fig. 4B and data not shown), only P5 and P6/7 5′ ends were detected in both analyses. Moreover, no extension products upstream P5 were detected from pCP1 using the primer CP1 (Fig. 2). Thus, the previous mappings of promoters P1 to P4 upstream P5 are thought to be non-specific extension products generated by mis-priming.

The processing within the colA 5′ leader sequence could cause a conformational change of the RNA secondary structure (Fig. 7). In B. subtilis, the mRNAs which have a 5′-proximal secondary structure beginning less than 6 nt from the 5′ end, followed by an SD sequence close to the secondary structure, are thought to be very stable (Bechhofer, 2009). The structure of processed colA mRNA applies to this example, but in the primary transcript, the stem-loop structure was too far from the 5′ end (33 nt). As shown in Fig. 4, processed colA mRNA is apparently more stable than the primary transcripts. In addition, the predicted secondary structure of the 5′ leader of primary transcript showed that an SD sequence is located in an intermolecular structure (Fig. 7A), which might inhibit translation. Processing within the 5′ leader sequence might lead to destruction of part of the stem-loop structure which would allow the ribosome to access to the SD sequence. In fact, the −140 to −80 region of the colA leader sequence was deleted from colAgst translational fusion constructs, and protein expression was obviously increased in VR-RNA-deficient cells (Fig. 10). These data indicated that translational efficiency is upregulated in the processed form of colA mRNA. Moreover, mutation of the SD sequence in pCP1 vector destabilized the processed transcript (Fig. 8) but addition of AUG start codon did not increase the stability of transcripts. This finding is consistent with the previous observation that ribosome binding to the SD sequence helps to protect mRNA from ribonuclease attack (Bechhofer, 2009; Dreyfus, 2009). The VR-RNA-dependent processing event could promote such protection by changing the accessibility of the ribosome to SD sequence, leading to the upregulation of colA in C. perfringens.

Frequently, sRNAs repress the target gene expression by base pairing with the SD sequence, which blocks ribosomes from binding to the target mRNAs (SgrS, OxyS, MicC, ArcZ and CyaR). On the other hand, another group of sRNAs that includes RyhB, DsrA, RprA, Qrr, RNAIII and GlmZ (Lease and Belfort, 2000; Majdalani et al., 2002; Hammer and Bassler, 2007; Prévost et al., 2007; Urban and Vogel, 2008) activates target genes through interaction with the 5′ leader sequence of mRNA as anti-antisense RNAs. For example, rpoS mRNA forms a highly stable stem-loop structure in the 5′ leader sequence and this inhibits ribosome entry to the SD sequence. Cleavage of this structure by RNase III promotes the rapid degradation of rpoS mRNA (Resch et al., 2008). The annealing of expressed DsrA to the 5′ leader sequence of rpoS in an anti-antisense manner induces a conformational change in the rpoS mRNA and ribosome entry, and the stability of the mRNA increases (Lease and Belfort, 2000). Alternatively, the DsrA–rpoS mRNA duplex is also cleaved by RNase III, which apparently prevents DsrA recycling but does not affect the stability of rpoS mRNA (Resch et al., 2008). Thus, continued DsrA–rpoS mRNA interaction is essential for rpoS activation. On the other hand, in VR-RNA–colA regulation, VR-RNA-dependent processing within colA mRNA significantly stabilizes the mRNA, and translational activity is activated in subsequently generated and processed colA mRNA. These results suggest that processing within colA mRNA contributes to make the translational activation becoming irreversible. Moreover, VR-RNA that binds to the 5′ leader sequence can be recycled after processing. If the role of processing is also to allow VR RNA recycling, then VR-RNA and colA mRNA would not be simultaneously cleaved. However, although whether or not VR-RNA is coincidentally cleaved remains unclear, full length and short length of VR-RNA were detected in the Northern blot analysis (Fig. 4A). Besides, since only the base pairing of VR-RNA with colA mRNA might cause the conformational change of colA 5′ UTR, in which the RBS is accessible, it remains to be determined that translational activation and mRNA stabilization of colA mRNA require only base pairing of VR-RNA or subsequent processing event. To further understand this regulatory mechanism and to address above issues, the factors involved in colA mRNA processing should be identified.

Most base-pairing sRNAs require the RNA chaperone Hfq to form stable sRNA–target mRNA hybrids or to maintain their own stability (Waters and Storz, 2009). Northern blot analysis showed that the C. perfringens hfq gene is expressed during the exponential phase when high levels of VR-RNA and colA are also expressed (data not shown). However, the expression of VR-RNA and colA did not differ in an hfq deletion mutant (data not shown), suggesting that Hfq protein is not involved in either colA regulation by VR-RNA or the stability of VR-RNA. In S. aureus, RNAIII regulates the expression of several genes, but Hfq protein is not required for the regulation (Boisset et al., 2007). Hfq protein is not generally thought to be important in sRNA regulation in Gram-positive bacteria.

The VirR/VirS two-component system in C. perfringens is a key factor in global gene regulation that directly activates the VR-RNA, ccp, pfoA, virT and virU genes (Okumura et al., 2008). A VirR/VirS–VR-RNA regulatory cascade appears to regulate over 100 genes containing those for several toxins (colA, plc, pfoA, nanI, nanJ and nagL; Ohtani et al., 2010). However, target genes that are regulated directly by VR-RNA have not been identified. The present findings indicate that VR-RNA directly regulates the colA gene by base paring with its 5′ leader sequence and subsequently processing colA mRNA. Other direct target genes might be significantly complementary to VR-RNA. The phospholipase C-coding gene, plc, is an essential toxin gene required for virulence that is considered to be positively regulated by VR-RNA (Shimizu et al., 2002b). While the stability of plc mRNA is regulated by VR-RNA (data not shown) and the 3′ region of VR-RNA is important for plc regulation (Shimizu et al., 2002b), no similar or complementary sequences between the 5′ leader of plc and the 3′ region of VR-RNA have been identified, suggesting that plc expression is regulated through different mechanisms. Although the study described herein is not intended to address details of the mechanism underlying the control of gene expression by VR-RNA, the data favour the notion that such control is exerted post-transcriptionally at the level of protein abundance. We believe that our findings provide new insights into the molecular mechanisms through which sRNAs activate their targets in pathogenic bacteria.

Experimental procedures

Bacterial strains

Clostridium perfringens strain 13 (Mahony and Moore, 1976) and its derivative strain TS140 (Shimizu et al., 2002b) harbouring plasmids were cultured under anaerobic conditions using an Anaeropack (Mitsubishi Gas Chemical, Tokyo, Japan) at 37°C in Gifu anaerobic medium (GAM) broth or BHI-sheep blood agar plates (Difco and Nippon Biotest Laboratories) containing 25 µg ml−1 chloramphenicol. E. coli JM109 was grown at 37°C in LB medium supplemented with either 50 µg ml−1 ampicillin or 25 µg ml−1 chloramphenicol.

Plasmids and oligonucleotides

Plasmids used in this study were constructed as described in Supporting information. Oligonucleotide primers used in this study were listed in Table S1.

RNA extraction

Total RNAs of C. perfringens were extracted using phenol and glass beads. Cell pellets were suspended in 1 ml of LETS buffer [100 mM LiCl, 10 mM EDTA, 10 mM Tris-HCl pH 7.5 and 1% (w/v) sodium dodecyl sulphate] and mixed with 1 ml of phenol-chloroform-isoamyl alcohol and 500 µl of glass beads (SIGMA). The cell suspensions were vortexed for 4 min. After repeated extraction of the aqueous phase with phenol-chloroform-isoamyl alcohol, the nucleic acids were recovered by precipitation with 2.5 volumes of ethanol and a 1/10 volume of 1 M LiCl. The pellets were resuspended in 300 µl of water, and then mixed with 900 µl of 4 M NaOAc. The samples were incubated at −30°C for at least 2 h, and then the nucleic acids were collected by centrifugation. The resulting total RNA precipitates were treated with RNase-free DNase I (Promega) according to the manufacturer's instructions. The RNA was quantified at A260 and the quality was checked by agarose gel electrophoresis.

Northern blot analysis

Total RNA mixed with two volumes of loading buffer (50% formamide, 6% formaldehyde, 20 mM 3-morpholinopropanesulphonic acid, 5 mM NaOAC, 1 mM EDTA, 0.05% bromophenol blue and 10% glycerol) was incubated at 65°C for 10 min, and then resolved on 1% agarose gels containing 2% formaldehyde. Short RNA species were analysed by separating total RNA on 6% polyacrylamide/7 M urea/TBE-gels and then transferred onto Gene Screen Plus nylon membranes (PerkinElmer). The membranes were baked at 80°C for 2 h, and pre-hybridized at 42°C in hybridization buffer [7% SDS, 50% deionized formamide, 50 mM sodium phosphate (pH 7.0), 2% blocking solution, 0.1% N-lauroylsarcosine, 0.75 M NaCl and 75 mM sodium citrate]. Digoxigenin-labelled DNA probes were hybridized and detected according to the supplier's instructions (DIG Application Manual, Roche).

DIG-labelled DNA Probes used for Northern blot analysis were generated by DIG-High Prime according to the supplier's instructions (DIG Application Manual, Roche). Template DNA for generation of colA, VR-RNA and transcripts from pCP1 probes were amplified by PCR using primers C17/C18, V101/V102 and C3/C4 from the strain 13 genomic DNA.

Primer extension

The colA transcriptional start sites on pCP1 and pCP1Δ3 were determined using the CP1 primer (Table S1) and the 5′ ends of colA mRNA transcribed from the genomic DNA were determined using the CP2 primer (Table S1). Primer (10 pmol) was end-labelled with 1 µl of T4 polynucleotide kinase (PNK), 2 µl of 10× PNK buffer and 5 µl of [γ-32P]-ATP in 20 µl of reaction buffer at 37°C for 30 min, and then T4 PNK was inactivated by heating at 95°C for 2 min. The reaction mixture (10 µl) containing 8 µg of total RNA, 1 nmol of deoxynucleoside triphosphate (dNTPs) and 0.5 pmol of labelled primer was incubated at 95°C for 1 min and 65°C for 5 min and then cooled on ice. The extension reaction was incubated at 42°C for 1 h with 20 units of RNase inhibitor (Takara Bio) and 200 units of PrimeScript reverse transcriptase (Takara Bio). Complementary DNA products precipitated with ethanol were separated on 6% polyacrylamide/7 M urea/TBE-gels. Sequence ladders were generated using the fmol DNA sequencing system (Promega) according to the supplier's instructions.

5′ RLM-RACE

The phosphorylation state of 5′ ends of colA mRNA was determined as previously described (Bensing et al., 1996; Bralley and Jones, 2004). Total RNA (10 µg) was incubated with or without 30 units of CIAP at 37°C for 1 h, extracted with phenol/chloroform/isoamyl alcohol and then precipitated with ethanol. Thereafter, CIAP-treated RNA (3 µg) was incubated with or without 20 units of TAP in 10 µl of reaction buffer at 37°C for 1 h, and then 2 µl of CIAP/TAP-treated RNA was ligated to the 5′ RACE adaptor with 5 units of T4 RNA ligase (Takara Bio) at 37°C for 1 h. The ligated RNA was used as a template for reverse transcription using PrimeScript reverse transcriptase (Takara Bio).

We amplified colA cDNA by nested PCR using the 5′ outer primer and the 3′ outer colA primer, and the 5′ inner primer and the 3′ inner colA primer for the first and second rounds of PCR, respectively, under the following cycling conditions: 2 min at 98°C and then 30 cycles of 10 s at 98°C, 5 s at 55°C and 30 s at 72°C. The amplified DNA products were analysed by 6% polyacrylamide gel electrophoresis and cloned into E. coli using the pUC18 vector.

Western blot analysis

Whole proteins were extracted using glass beads. Cell pellets suspended in LETS buffer were mixed with an equal volume of glass beads and then vortexed for 4 min. After centrifugation, supernatants were collected. Extracellular proteins in culture supernatants were precipitated with 10% (w/v) trichloroacetic acid, washed with cold acetone and resuspended in LETS buffer. The equivalent volume of 0.02 A280 units of protein samples were separated by SDS-PAGE, and then electroblotted onto polyvinylidene difluoride membranes. The membranes were blocked with 2.5% skim milk in Tris-buffered saline containing 0.2% Tween 80, and then probed with anti-GST (Wako) diluted 1:2000. Horseradish-peroxidase-conjugated anti-mouse secondary antibodies (GE Healthcare) were used at a dilution of 1:50 000, and then bound antibodies were detected using Immunostar LD (Wako).

Synthesis and labelling of RNA in vitro

Probe RNAs were transcribed using SP6 RNA polymerase (Takara Bio) or T7 RNA polymerase (EPICENTRE Biotechnologies). EcoRI-digested pNOE10, pNOE12, pNOE13 and pNOE11 were used as templates for the transcription of VR-RNA variants in vitro, and pNOE20 digested with HindIII served as the template for the colA 5′ UTR. The samples were resolved by 6% urea-denaturing PAGE and extracted, and then purified RNAs were dissolved in sterile double-deionized water.

Gel mobility shift assays

Interaction between VR-RNA and colA mRNA was analysed as described (Møller et al., 2002). We radiolabelled colA mRNA synthesized in vitro with [γ-32P]-ATP using the KinaseMax kit (Invitrogen) according to the supplier's instructions. Unlabelled VR-RNA (0.05, 0.1, 0.5 or 1 pmol) and 5′ end-labelled colA mRNA (10 fmol) were mixed in 10 µl of reaction mixture containing 20 mM HEPES pH 7.9, 100 mM KCl, 1 mM MgCl2, 1 mM dithiothreitol and 1 µg of tRNA and incubated at 37°C for 30 min. After adding 5 µl of loading dye containing 50% glycerol, 0.1% bromophenol blue and 0.1% xylene cyanol, the reaction mixtures were resolved on 4% non-denaturing acrylamide gels in 1× TBE buffer.

Acknowledgements

We are especially grateful to T. Shimizu for gifts of C. perfringens strains and plasmids. We also thank to N. Foster for critical reading of the manuscript.

Ancillary