Development and Application of a Method for Counterselectable In-Frame Deletion in Clostridium perfringens

Abstract

Many pathogenic clostridial species produce toxins and enzymes. To facilitate genome-wide identification of virulence factors and biotechnological application of their useful products, we have developed a markerless in-frame deletion method for Clostridium perfringens which allows efficient counterselection and multiple-gene disruption. The system comprises a galKT gene disruptant and a suicide galK plasmid into which two fragments of a target gene for in-frame deletion are cloned. The system was shown to be accurate and simple by using it to disrupt the alpha-toxin gene of the organism. It was also used to construct of two different virulence-attenuated strains, ΗΝ1303 and HN1314: the former is a disruptant of the virRS operon, which regulates the expression of virulence factors, and the latter is a disruptant of the six genes encoding the α, θ, and κ toxins; a clostripain-like protease; a 190-kDa secretory protein; and a putative cell wall lytic endopeptidase. Comparison of the two disruptants in terms of growth ability and the background levels of secreted proteins showed that HN1314 is more useful than ΗΝ1303 as a host for the large-scale production of recombinant proteins.

Clostridium perfringens is a spore-forming anaerobic Gram-positive rod with a low DNA G+C content (28). It produces potent histolytic enzymes and toxins, causing various diseases ranging from mild food-borne diarrhea to fulminant histotoxic infection (3, 29, 37). Other pathogenic clostridial species also produce various toxins and enzymes (28). Some of such clostridial proteins are useful and commercially available, and others are potentially useful (15). Unfortunately, the expression of clostridial proteins in a heterologous host is often inefficient, especially when high-molecular-weight proteins are expressed, mainly because of extremely biased codon usage as to the use of A and T at silent positions (39).

To circumvent such problems, we have developed an expression system including C. perfringens using the following advantages of the organism: it grows very rapidly, its doubling time being less than 10 min under optimal conditions (19), and it is relatively aerotolerant and can be simply batch cultured (28). Furthermore, strain 13, which is used in our system, is an extremely poorly sporulating strain with mutation of the master sporulation regulatory gene and hence is suitable in terms of environmental biohazard risk (20). Our system has so far been successfully applied to NanI sialidase (37), ColH collagenase (39), and clostripain-like protease (11).

Since application of our system to the large-scale preparation of clostridial proteins would cause a biosafety concern, we have attempted to attenuate the virulence of C. perfringens by deleting multiple virulence genes to extents comparable to those in microorganisms generally regarded as safe (GRAS) (16, 22, 30). In order to efficiently delete multiple genes, we chose the GalK/galactose (Gal)-based counterselection system that was developed for Myxococcus xanthus (21, 41) and applied it to Streptococcus mutans (17). From C. perfringens strain 13, we constructed a parental strain, HN13, that has an in-frame deletion (IF) in a galKT operon. As a suicide vector, we constructed pGALK containing a drug resistance marker and the galK gene to mediate the toxicity of galactose or a galactose analog (17, 41). The rationale of the markerless in-frame deletion system is as follows. The first recombination occurs at either of the two fragments of the target gene for the in-frame deletion. These two homologously recombined integrants are positively selected by the drug resistance marker. However, when the second recombination event occurs in the integrants, the GalK-mediated toxicity allows selection of an in-frame deletion mutant and a revertant strain, which can be differentiated simply by colony PCR.

By using the HN13/pGALK system, we disrupted the virRS operon encoding the VirR/VirS two-component signal transduction system involved in the regulation of virulence genes (5, 25). In an attempt to reduce not only major virulence factors but also major secretory proteins, we disrupted the six genes encoding the α toxin (phospholipase C [PLC]), θ toxin (perfringolysin O [PFO]), κ toxin (ColA collagenase), a clostripain-like protease (CLP), CPE1281 of a 190-kDa secretory protein (10), and CPE0278 of a CwlO homologue from Bacillus subtilis, a cell wall lytic endopeptidase (43). The results presented here indicate that the six-gene mutation abolished the production of the corresponding proteins and also gave almost the same growth rate and NanI productivity as those for the wild-type strain.

MATERIALS AND METHODS

Bacteria, growth conditions, and DNA manipulation.

Escherichia coli NovaBlue (Merck KGaA, Darmstadt, Germany) was used as the recipient strain for transformation and was grown in LB medium at 37°C. Clostridium acetobutylicum ATCC 824 was obtained from the Riken BioResource Center and was used as the source of the galK gene in the suicide vector. C. perfringens strain 13 (33) was used as the host strain. It was grown on 2-fold-diluted Gifu anaerobic medium (GAM/2) agar plates (Nissui Pharmaceutical, Tokyo, Japan) under anaerobic conditions using an AnaeroPack system (Mitsubishi Gas Chemical, Tokyo, Japan), unless otherwise stated. Chloramphenicol was added to the agar plates at 10 μg/ml (GAM/2-Cm10), when necessary. All recombinant strains constructed from C. perfringens strain 13 and used as IF mutants are listed in Table 1. They were grown in TY medium (3% tryptone, 2% yeast extract, 0.1% sodium thioglycolate) or the same medium supplemented with sugar additive, as follows: TY-G1 and TY-G3 with 1 and 3% glucose, respectively; TY-GA1 and TY-GA3 with 1 and 3% galactose, respectively; and TY-DGA1 with 1% 2-deoxygalactose (DGA). Cultures were performed under anaerobic conditions in tightly capped tubes at 37°C. Twofold-diluted TY (TY/2) agar plates were also used for the in-frame deletion system.

TABLE 1.

IF mutant strains derived from C. perfringens strain 13

Strain	Relevant characteristic
HN13	Strain 13 ΔgalKT
HN1301	HN13 Δplc
HN1302	HN13 ΔvirS
HN1303	HN13 ΔvirRS
HN1304	HN1302 Δplc
HN1305	HN13 Δclp
HN1306	HN13 ΔcolA
HN1307	HN13 Δpfo
HN1308	HN13 ΔCPE1281
HN1309	HN13 ΔcwlO
HN1310	HN1301 Δclp
HN1311	HN1310 ΔcolA
HN1312	HN1311 Δpfo
HN1313	HN1312 ΔCPE1281
HN1314	HN1313 ΔcwlO

DNA manipulation was performed using standard protocols (31). Transformation of C. perfringens was performed by electroporation, as described previously (12). The DNA sequences were determined with ABI Genetic Analyzers 310 and 3130, using the methods recommended by the manufacturer and double-stranded plasmid DNA as the template. All the constructs, including PCR-amplified fragments were verified by DNA sequencing.

Construction of suicide vectors for the IF system.

All the primers used in this study are listed in Table S1 in the supplemental material. A suicide vector, pGALK (Fig. 1), expressing the galactokinase (galK) and chloramphenicol resistance (catP) genes was constructed from pJIR418 (36). The 3,548-bp FspI-NaeI fragment of pJIR418 containing oriCP and rep was removed and self-ligated. Furthermore, a 1,192-bp BspHI-BstZ17I fragment containing ermBP was replaced with a 119-bp fragment of P_fdx containing the ferredoxin (fdx) gene promoter region of pFN (39). P_fdx was PCR amplified by using FN5 and FN3 as primers and pFN as the template. The resultant plasmid was designated pJIR418-P_fdx. To express the galK gene of C. acetobutylicum ATCC 824 (galK-Ca) under P_fdx control, a 1,167-bp galK-Ca-coding region was PCR amplified by using GalK-CaN and GalK-CaC as primers and was then cloned into pJIR418-P_fdx at the NdeI-BstZ17I sites. The resultant plasmid was designated pGALK.

FIG. 1.

IF system for C. perfringens. (A) Gene organization in the region containing the gal operon in C. perfringens strain 13. The dark gray- and light gray-shaded regions, which are designated N and C, respectively, are the regions used to create the GalKT-IF gene. The region denoted by Δ represents the deleted region in the galKT operon. The stem-loop structure shows a putative rho-independent transcriptional terminator. Genetic symbols are tagged with CPE numbers (33). (B) Schematic representation of the suicide vectors pGALK and pCM-GALK. Genetic symbols: oriEC, the replication origin from pUC18; catP, the chloramphenicol acetyltransferase gene from C. perfringens; galK-Ca, the galactokinase gene from C. acetobutylicum ATCC 824; P_fdx, the fdx promoter region from pFN; T_fdx, the transcriptional terminator region of the fdx gene from C. perfringens strain 13; and MCS, the multiple-cloning site. (C) Construction of the IF plasmid pGalKT-IF. (D and E) Schematic diagrams of the first (D) and second (E) recombinations with the IF system involving pGalKT-IF.

We also constructed another suicide vector for the IF system. A 113-bp transcriptional terminator region of the fdx gene (T_fdx) was PCR amplified by using FTG5 and FTG3 as primers and was then introduced at the BstZ17I-ClaI sites in pGALK. The resultant plasmid was designated pGALK-T_fdx. A 1,384-bp AvrII-NruI fragment of pGALK-T_fdx was replaced with the AvrII-BsrBI fragment in pCM3 (GenBank/EMBL/DDBJ accession no. AB562893). The resultant suicide vector was designated pCM-GALK (Fig. 1B). A 2,203-bp EcoRI-AvrII fragment containing the plc gene was cloned from pJIR418α (23) into pCM3, and the resultant plasmid was designated pPLC.

Construction of a galactose metabolism-deficient C. perfringens strain.

Using the sequence information for C. perfringens strain 13 (33) regarding galactose metabolism (Fig. 1A), two homologous regions, the upstream region of galK (N fragment, 903 bp) and the downstream region of galT (C fragment, 926 bp), were PCR amplified using the following primers: primers GalK-N5 and GalK-N3 and primers GalT-C5 and GalT-C3, respectively. Both the N and C fragments were cloned into pGALK at the BamHI-SalI sites, with pGalK-N and pGalT-C, respectively, thereby being obtained. The BglII-SalI fragment derived from pGalT-C was inserted into pGalK-N at the BglII-SalI sites. The resultant plasmid was designated pGalKT-IF and was used as the IF plasmid for disruption of the galKT operon. The IF gene (GalKT-IF) in pGalKT-IF was fused in-frame at the BglII site between Arg27 (AGA) of GalK and Ser471 (TCT) of GalT, resulting in elimination of an internal 2,615-bp region (Fig. 1C; see Table S1 in the supplemental material).

pGalKT-IF was electroporated into strain 13, and then chloramphenicol-resistant cells generated through single-crossover recombination were screened by spreading cells on GAM/2-Cm10 plates and incubation for 8 h. Theoretically, the first homologous recombination occurs at two different crossover points, the upstream and downstream homologous recombination sites, i.e., the N and C sites, respectively, as shown in Fig. 1A and C. The crossovers at the two sites, the N and C crosses (Fig. 1D), were differentiated by colony PCR, using the following four sets of primers: for the N cross, primers FW and GalK-IN and primers GalT-IC and GalT-3, and for the C cross, primers GalK-5 and GalK-IN and primers GalT-IC and RV.

Strains with the N and C crosses were individually cultured in TY medium for 4 h. The cultures were then serially diluted 10-fold with TY medium and plated onto TY/2-DGA1 plates, followed by 8 h incubation for counterselection. Theoretically, the second homologous recombination causes looping out of the plasmid-derived region and includes two types of recombination, as shown in Fig. 1D and E. One occurs at homologous regions different from those in the case of the first recombination. This generates an IF mutant. The other occurs at the same sites as those in the case of the first recombination. This generates a revertant strain. However, the revertant wild-type strain expresses its own galK gene and hence cannot grow on a DGA-containing plate. Thus, DGA-resistant and chloramphenicol-sensitive clones were selected as IF mutant HN13 after the second recombination.

Strain HN13 was identified by colony PCR using three sets of primers: primers GalK-5 and GalK-IN, GalT-IC and GalT-3, and GalK-5 and GalT-3. Colony PCR with primers FW and GalK-IN, GalT-IC and RV, GalK-CaN and GalK-CaC, and CatP-N and CatP-C (for detection of the catP gene) was also performed to confirm that HN13 was free of the plasmid. Furthermore, the IF mutations of the galKT operon were confirmed by DNA sequencing of the PCR product amplified by using GalK-5 and GalT-3 as primers and the isolated chromosomal DNA as the template. The phenotypes of HN13 obtained from the N- and C-cross strains were confirmed to be identical, as described in the Results section, and therefore, it was used as a parental strain for the IF system (Table 1).

Construction of IF mutants.

IF mutation of the genes encoding PLC (CPE0036), VirS (CPE1500), VirR (CPE1501), CLP (CPE0846), ColA (CPE0173), PFO (CPE0163), CPE1281, and CwlO (CPE0278) was performed as described above, except that 3% galactose instead of DGA was used for the counterselection. As the suicide vector, pCM-GALK was used for the construction of IF mutants other than HN1301, HN1302, and HN1303, which were constructed by using pGALK. IF genes such as PLC-IF, VirS-IF, VirRS-IF, CLP-IF, ColA-IF, PFO-IF, CPE1281-IF, and CwlO-IF were constructed by PCR amplification using the primers listed in Table S1 in the supplemental material. The N and C fragments were PCR amplified with primers N5 and N3 and primers C5 and C3, designed for each construct, respectively. The N and C fragments were ligated adjacent to each other in the suicide vector. In the case of CwlO-IF, they were fused by the overlap extension PCR method (8). The N and C crosses and the IF mutation were confirmed by colony PCR and DNA sequencing using the sets of primers listed in Table S1 in the supplemental material.

TCA precipitation, SDS-PAGE, and N-terminal amino acid sequencing.

Supernatant from an 8-h culture was obtained by centrifugation at 15,000 × g for 2 min at 4°C, followed by passage through a 0.2-μm-pore-size membrane filter and was frozen at −80°C until use. One milliliter of the culture supernatant was subjected to 20% trichloroacetic acid (TCA) precipitation. The precipitate was washed with acetone, dissolved in 20 μl of 1 M Tris-HCl, pH 8.0, plus 100 μl of SDS buffer (125 mM Tris-HCl, pH 6.8, 2.5% SDS, 5% 2-mercaptoethanol, 0.02% bromophenol blue, 20% glycerol), and then heated at 96°C for 3 min. SDS-PAGE was performed as described previously (40). Twenty-four microliters of protein sample, equivalent to 200 μl of the culture supernatant, was separated on a 14% gel and then stained with Coomassie brilliant blue R. The N-terminal amino acid sequences were determined with an automated protein sequencer (ABI 492; Perkin-Elmer, Waltham, MA), after transfer to polyvinylidene difluoride membranes as described previously (14).

Northern blotting.

Preparation of total RNA from C. perfringens cells and Northern blot analysis were carried out as described previously (6), with slight modifications. In brief, the C. perfringens strain was grown at 37°C to the mid-exponential phase of growth (optical density at 600 nm [OD₆₀₀] = 1.3), and then total RNA was prepared by the SDS/phenol method. For Northern blot analysis, 10 μg of RNA was hybridized at 55°C with a DNA probe which had been PCR amplified with primers VirR-FW and VirR-RV using C. perfringens 13 chromosome DNA as the template and labeled with digoxigenin-11-dUTP (Roche Applied Science, Mannheim, Germany). The hybridized signals were detected as Lumi-Phos 530 (Lumigen, Southfield, MI) chemiluminescence on X-ray film.

Enzyme assaying of a culture supernatant. (i) PLC.

The assaying of PLC activity was performed using p-nitrophenylphosphorylcholine (Sigma-Aldrich, St. Louis, MO), as described previously (13). PLC activity was also examined by observing the white halo formed around a colony on a GAM plate containing 5% (vol/vol) egg yolk (EY-GAM), after 1 μl of the culture was spotted and incubation under anaerobic conditions. Ten microliters of the culture supernatant was also applied to a well in an egg yolk agarose plate of EY-TBS (1% agarose, 5% egg yolk, 5 mM dithiothreitol [DTT], TBS buffer [20 mM Tris-HCl, pH 7.5, 150 mM NaCl]), followed by incubation at 37°C.

(ii) CLP.

CLP activity was assayed using benzoyl-l-arginine p-nitroanilide as described previously (11).

(iii) ColA.

Collagenase activity was measured by using 4-phenylazobenzyloxycarbonyl-l-Pro-l-Leu-Gly-l-Pro-d-Arg (Sigma-Aldrich) as a synthetic substrate, as described previously (39).

(iv) PFO.

Titration of hemolytic activity toward sheep red blood cells (SRBCs) in a culture supernatant was examined as described previously (23). The hemolytic activity of the culture supernatant was also examined on an SRBC agarose (1% agarose, 1.25% SRBC, 5 mM DTT, 5 mM CaCl₂, TBS buffer) plate, with which the hemolytic activity of PLC was also measurable. Ten microliters of the culture supernatant was applied to a well in the SRBC agarose plate, followed by incubation at 37°C for 4 h. The hemolytic activity of the colony was detected on an SRBC-TY-G1 agarose plate (1% agarose, 1.25% SRBC, TY-G1 medium), after spotting of 1 μl of the culture and anaerobic incubation at 37°C for 8 h.

(v) Caseinase.

Proteolytic activity in a culture supernatant was assayed using azocasein (Sigma-Aldrich), as described previously (39).

(vi) NanI sialidase.

Strains carrying pFFN (38) were cultured in TY-G1 medium containing 20 μg/ml of erythromycin. The sialidase activity in a culture supernatant was determined using 2′-(4-methylumbelliferyl)-α-d-N-acetylneuraminic acid, sodium salt hydrate (Sigma-Aldrich), as described previously (38).

RESULTS

Construction of an IF system for C. perfringens.

Genome analysis of C. perfringens strain 13 (Fig. 1A) showed that the operon (mglBAC) for the Gal ABC transporter system is present upstream of the galM-galKT cluster. Therefore, Gal uptake seems to be mediated by the Gal ABC transporter system. Furthermore, C. perfringens strain 13 possesses all genes corresponding to the enzymes involved in the Leloir pathway for Gal utilization (9): GalM (Gal mutarotase), GalK (galactokinase), GalT (Gal-1-phosphate uridylyltransferase), GalE (UDP-Gal-4-epimerase), and GalU (UTP-glucose-1-phosphate uridylyltransferase). The ability of strain 13 to use Gal as a sole carbon source was examined by growing it in TY medium with and without Gal. Strain 13 could grow in TY-GA1 as well as in TY-G1. Moreover, strain 13 was vulnerable to the toxicity of DGA, a nonmetabolizable Gal analog, which exhibits toxicity upon being taken up and phosphorylated. Thus, the development of a GalK/galactose-based IF system for C. perfringens requires the preparation of a parental mutant strain (Fig. 1) in which the galKT operon is disrupted. It also requires the construction of a suicide vector that confers drug resistance on transformants and that simultaneously allows constitutive expression of GalK after being integrated into the chromosome. All data with respect to the growth of strain 13 and galKT-disruptant strain HN13 in various media are shown in Fig. S1 in the supplemental material.

We constructed a suicide vector, pGALK (Fig. 1B), for the IF system which carries galK-Ca fused to the fdx promoter (39). galK-Ca was chosen on the basis of the following assumption: the galK genes of C. perfringens and C. acetobutylicum are phylogenetically close enough to be well expressed and functional but different enough in identity (61.2%) at the DNA level to prevent homologous recombination. The galK-Ca gene in pGALK was suggested to be functional by the finding that galKT-disruptant strain HN13 (hereinafter named strain HN13) was obtained efficiently from the wild-type strain using pGalKT-IF and DGA.

To confirm the phenotypic change in Gal utilization by strain HN13, we examined its growth in TY-GA1. It grew well in TY-G1 and similarly to the wild-type strain, but not in TY-GA1, unlike the wild-type strain. Furthermore, it was insensitive to the DGA toxicity, indicating that strain HN13 can be used as a parental strain for the IF system.

plc disruptant of strain HN13.

PLC, one of the major virulence factors of C. perfringens, was disrupted using the IF system to examine its accuracy and efficiency. The transformation efficiency of strain HN13 was approximately 10⁷ CFU per μg of pJIR418 DNA, with this being comparable to that of strain 13. Approximately 5,000 chloramphenicol-resistant transformants were obtained per μg of pPLC-IF (see Table S1 in the supplemental material) upon the first recombination. Of these transformants, 32 clones were randomly chosen and replica plated onto GAM/2-Cm10 and TY/2-GA3 plates (Fig. 2A). Twenty-nine clones grew well on the GAM/2-Cm10 plates but not on the TY/2-GA3 plates. Sixteen of these 29 clones were then subjected to colony PCR to confirm the crossover point, with the clones with the N and C crosses being named strains N and C, respectively. Eight and seven clones were confirmed to be strains N and C, respectively, whereas the rest remained unidentified. Strains N and C were individually grown in TY medium for 4 h and then plated onto TY/2-G3 and TY/2-GA3 plates, respectively. As shown in Fig. 2B, Gal-resistant colonies were formed on TY/2-GA3 plates at a level of approximately 0.1% of the total number of CFU obtained on TY/2-G3 plates. The streaking of a colony of strain N or C on a TY/2-GA3 plate also allowed the isolation of a single Gal-resistant colony (Fig. 2A). Sixteen Gal-resistant clones derived from each of strains N and C were replica plated onto TY/2-GA3 and GAM/2-Cm10 plates. In both cases, 15 clones were Gal resistant and chloramphenicol sensitive, so they should be either an IF mutant or a revertant strain (strain R). Colony PCR revealed that seven and five clones were the IF mutants derived from strains N and C, respectively, and that the remaining clones were strain R. The two IF clones derived from strains N and C were confirmed to be the plc IF mutant by DNA sequencing of the PCR products and were used as HN1301. All data with respect to analysis of PCR products are shown in Fig. S2 in the supplemental material.

FIG. 2.

Gal counterselection with the IF system. (A) Selection by streaking. Strain HN13 with pPLC-IF integrated was streaked on both GAM/2-Cm10 and TY/2-GA3 plates. (B) Selection by spreading. The strain was also cultured in TY medium for 4 h, followed by 10⁴-fold dilution with the medium. One hundred microliters of the diluted culture was spread onto both TY/2-G3 and TY/2-GA3 plates. The plates were incubated anaerobically for 12 h at 37°C.

The PLC production of HN1301 was examined by SDS-PAGE. No band corresponding to PLC was detectable for the culture supernatant of HN1301, unlike for the culture supernatants of strains 13, HN13, and R (Fig. 3A). On the other hand, there was no significant difference between these four strains in the levels of other major secretory proteins, such as CLP (heavy and light chains [11]), ColA, PFO, and 190-kDa hypothetical protein CPE1281 (10), which was identified by N-terminal amino acid sequencing. This indicates that HN1301 is disrupted properly and also suggests that the IF system does not affect the productivity of secretory proteins except for a disrupted one. Complete loss of the PLC production of HN1301 was also shown by the egg yolk test involving a culture and a culture supernatant (Fig. 3B and C) and also by assaying of PLC activity (Table 2). HN1301 transformed with pPLC carrying the plc gene produced and secreted a large amount of PLC (Fig. 3D), indicating that HN1301 retains the ability to produce and secrete secretory proteins. Thus, HN1301 was proved phenotypically to be the plc IF mutant.

FIG. 3.

Expression of PLC in C. perfringens strains 13, HN13, and HN1301. (A) SDS-PAGE analysis of culture supernatants of C. perfringens strains 13, HN13, and HN1301 and the revertant strain (R). The OD₆₀₀s of the cultures are shown underneath the gel. Major secreted proteins are indicated by arrowheads on the left; CPE1281 (187.0 kDa), ColA (121.4 kDa), PFO (52.7 kDa), PLC (42.6 kDa), and the heavy chain (HC) and two light chains (LC and LC′) of CLP (CLP-HC, 32.8 kDa; CLP-LC′, 15.4 kDa; CLP-LC, 14.9 kDa [11]). (B) PLC activity exhibited by colonies on EY-GAM plates. (C) PLC activity in the culture supernatants on EY-TBS agarose plates. The samples applied to the wells corresponded to those applied to the lanes in panel A. (D) Complementation test for HN1301. Strain HN1301 transformed with pPLC or pCM3 was grown in TY-G1-Cm10 for 8 h, and then the culture supernatants were examined for PLC activity and subjected to SDS-PAGE analysis. All assays were carried out as described under Materials and Methods.

TABLE 2.

PLC, CLP, ColA, PFO, and caseinase activities in culture supernatants of the IF mutants

Strain	Activitya
	PLC	CLP	ColA	PFO	Caseinase
13	421 ± 21	64.5 ± 0.31	1719 ± 6.5	923.9 ± 4.7	7.08 ± 0.15
HN13	439 ± 12	64.4 ± 0.31	1722 ± 4.6	930.9 ± 6.1	7.11 ± 0.08
HN1301	—	ND	ND	ND	ND
HN1302	39 ± 1	—	37 ± 0.0	<0.9	<0.09
HN1303	29 ± 4	—	24 ± 5.1	—	<0.09
HN1305	ND	—	ND	ND	<0.09
HN1306	ND	ND	—	ND	ND
HN1307	ND	ND	ND	—	ND
HN13014	—	—	—	—	<0.09

^aThe activities of PFO and the other enzymes are expressed in U and mU/ml of culture supernatant, respectively. Values are expressed as means ± SDs for triplicate determinations. — and ND, undetected and not determined, respectively.

virR/virS disruptant of strain HN13.

Production of virulence factors such as PLC, CLP, ColA, and PFO by C. perfringens is regulated at the transcriptional level by the VirR/VirS two-component signal transduction system (34). Furthermore, it also regulates the expression of many virulence-related housekeeping genes through the function of VirR-regulated RNA (24, 35). Thus, the disruption of the virRS operon with the IF system can be expected to greatly attenuate the virulence.

By means of the IF system, we constructed virS IF mutant HN1302 and virRS IF mutant HN1303 (Table 1; see Table S1 in the supplemental material). Met11 (ATG) and His399 (CAT) were used as the IF fusion sites in the case of HN1302. Met1 (ATG) of VirR and His399 (CAT) of VirS were chosen as the IF fusion sites in the case of HN1303. As expected, Northern blot analysis showed that strains HN13 and HN1302 expressed a single 2.1-kb virRS transcript (2) and a 0.9-kb virR transcript, respectively, whereas no hybridized band was detected for HN1303 with the virR-specific probe (Fig. 4A). The levels of virRS transcripts in strain HN13 were almost the same as those in strain 13 (data not shown). However, the level of virR transcripts in HN1302 seemed higher than that of the virRS transcripts in strain HN13, probably due to the higher level of stability of the shorter transcript.

FIG. 4.

Northern blot analysis of virR mRNA and expression of PLC in the virRS IF mutants. (A) Northern blot analysis of the virR transcript. Ten micrograms of total RNA was subjected to Northern blot analysis using a digoxigenin-labeled virR-specific 520-bp DNA probe, as described under Materials and Methods. The calculated sizes of mRNA and the RNA size markers are given on the right and left, respectively (b, bases). (B) SDS-PAGE analysis of culture supernatants of strains HN13, HN1301, HN1302, HN1303, and HN1304. (C) PLC activity in the culture supernatants on EY-TBS agarose plates. The samples applied to the wells corresponded to those applied to the lanes in panel B.

SDS-PAGE analysis of culture supernatants (Fig. 4B) showed that the levels of PLC, CLP, ColA, and PFO produced by HN1302 and HN1303 were far lower than those produced by strain HN13. The enzyme assay (Table 2) showed that no CLP or PFO activity and only marginal ColA and caseinase activities were detectable in the culture supernatant of HN1303 and that the inhibitory effect on toxin production was more prominent in HN1303 than in HN1302. A likely explanation is that HN1302 expresses more virR than strain HN13, so the virS-defective phenotype could be partially complemented by the plasmid-encoded virR (2). In contrast to these proteins, CPE1281 and several minor unidentified proteins seemed to be produced at increased levels in culture supernatants of HN1302 and HN1303. Large amounts of low-molecular-weight substances, which correspond to those detectable in growth medium (data not shown), were detectable in the supernatants of the two disruptants. This seems to be due to loss of CLP protease regulated by the VirR/VirS system. The PLC productivity of the two disruptants was completely abolished when the plc gene was disrupted (Fig. 4B and C). The growth of the mutants in TY-G1 was slightly impaired, as shown by comparison of the OD at 600 nm between the cultures of the mutants and the wild-type strain (Fig. 4A).

Sextuple disruptant of strain HN13.

Another approach for constructing GRAS strains from C. perfringens is to attenuate the virulence and potential virulence determinants by deleting each relevant gene. By means of the IF system, we disrupted the genes encoding PLC, CLP, ColA, and PFO, which are regulated by the VirR/VirS system. Besides these virulence factors, two major secretory proteins, CPE1281 and a putative cell wall lytic endopeptidase, CwlO, were also deleted to reduce host extracellular proteins for purification of recombinant secretory proteins. Our initial attempt to construct CLP-IF in pGALK involving E. coli was unsuccessful. However, pCM-GALK (Fig. 2B), which lacks a possible intrinsic strong promoter near the multiple-cloning site, allowed cloning of C. perfringens DNA with potential toxicity. pCM-GALK was used to generate the following mutants (Table 1): single-IF mutants HN1305, HN1306, HN1307 and HN1308; a double-IF one, HN1310; a triple-IF one, HN1311; a quadruple-IF one, HN1312; and a quintuple-IF one, HN1313.

SDS-PAGE analysis of the culture supernatants of these IF mutants (Fig. 5) showed that the expression of the relevant genes was successfully abolished. In the culture supernatants of clp IF mutant HN1305 and its derivative mutants, low-molecular-weight substances were observed, as was the case for the virRS IF mutant, HN1303. Furthermore, we found that three proteins that migrated as 45-, 30-, and 15-kDa bands on SDS-PAGE had increased levels or were newly detected in the culture supernatants of HN1305 and its derivatives. The N-terminal amino acid sequences of the 45-, 30-, and 15-kDa proteins were determined to be TPLTDDQK, TPLTDDQK, and GGDVNSG, respectively. Interestingly, these three proteins are derived from the same gene product, CPE0278, a 432-amino-acid hypothetical protein, which exhibits homology to B. subtilis CwlO, a cell wall lytic endopeptidase belonging to the NLPC/P60 family (43). Thus, we constructed cwlO IF mutant HN1309 and sextuple-IF mutant HN1314, of which the phenotypic changes were confirmed by SDS-PAGE analysis (Fig. 5). Both strains exhibited normal growth, like the other mutant strains (Fig. 5).

FIG. 5.

SDS-PAGE analysis of culture supernatants of the IF mutants. (A) Culture supernatants of the single-IF mutants, HN1301, HN1305, HN1306, HN1307, HN1308, and HN1309, were subjected to SDS-PAGE analysis. (B) The culture supernatants of strains HN13, HN1301, HN1303, HN1305, HN1310, HN1311, HN1312, HN1313, and HN1314 were also subjected to SDS-PAGE analysis. Secreted proteins derived from CwlO are indicated by arrowheads on the right.

Enzyme assaying of the culture supernatants showed that each single-IF mutant produced no relevant protein and also that HN1314 produced no PLC, CLP, ColA, or PFO activity (Table 2). Furthermore, we examined the hemolytic activities of the culture supernatants using SRBC agarose plates (Fig. 6A). virS IF mutant HN1302, virRS IF mutant HN1303, and pfo IF mutant HN1307 exhibited the hemolytic activity of PLC, and plc IF mutant HN1301 exhibited the hemolytic activity of PFO, while HN1314 did not show the hemolytic activity of either one. This was also the case for hemolysis of colonies formed on SRBC-TY-G1 agarose plates (Fig. 6B).

FIG. 6.

Hemolytic activity of the IF mutants on SRBC agarose plates. (A) Hemolytic activity in the culture supernatants of strains 13, HN13, HN1302, HN1303, HN1307, HN1314, and HN1301 and the revertant strain (R). (B) Hemolytic activity of strains HN13 and HN1314. These assays were performed as described under Materials and Methods. No clear halo was detected when TY-G1 medium was used.

In order to assess the recombinant protein productivity of strain HN1314, 74-kDa NanI sialidase was expressed in strains 13, HN13, and HN1314 transformed with pFFN (38). Their growth was almost the same, and the NanI productivity differed only slightly, i.e., only to such an extent that the NanI activities of strains 13, HN13, and HN1314 grown for 8 h were 120, 125, and 141 units/ml of culture supernatant, respectively, and those of strains grown for 12 h were 124, 128, and 142 units/ml of culture supernatant, respectively. SDS-PAGE analysis showed the presence of a large amount of NanI in the culture supernatant of HN1314 but the absence of other major secretory proteins (data not shown). Thus, sextuple-IF mutant HN1314 can be regarded as being more suitable than virRS IF mutant HN1303 in terms of the residual levels of toxins and having an inhibitory effect on cellular growth.

DISCUSSION

In this study we developed a precise in-frame allelic exchange system for C. perfringens involving galK as a counterselectable marker. Genetic markers employed for IF systems so far have been levansucrase (42), phosphoribosyltransferases (18), and galactokinase (21, 41), which mediate the toxicity of their substrates or nonmetabolizable analogs in merodiploids that have not undergone the second recombination. Among them, we employed the GalK system because C. perfringens possesses a chromosomally encoded galKT operon, utilizing galactose as a sole carbon source, and also because the C. acetobutylicum galK homologue can be maintained stably on a suicide vector without recombination between the two homologues. We confirmed the excellent efficiency and reliability of our IF system, on the basis of the results of deletion of the plc genes, and applied it to the disruption of the virS and virRS genes and also to the disruption of six genes encoding extracellularly secreted proteins.

In the case of the virRS IF mutant, the production of major toxins decreased markedly, and thus it can be used as a virulence-attenuated host, depending on the purpose. However, the toxin productivity of the mutant was not completely abolished. Furthermore, it cannot be ruled out that the attenuated virulence may be restored under certain circumstances where cross talk with other signaling pathways evokes the expression of the virulence factors. In contrast, the sextuple-IF mutant was completely deficient in the toxins, it thus being more suitable for the large-scale preparation of recombinant proteins, especially when the products are to be used for medical purposes.

In the postgenomic era, comparative and functional genomics have been progressing (26). Aided by high-throughput methods for proteomic (1, 34) and transcriptomic (32, 34, 44) analyses of C. perfringens, a genome-wide search for factors involved in the networks of such biological processes as genetic exchange, development, virulence, and quorum sensing will reveal an increasing number of potentially key molecules in each network.

Our method has several advantages over other methods available for the mutagenesis of C. perfringens, including intron-mediated mutagenesis (4, 7): a selection marker can be reused for disruption of multiple genes, it does not involve a polar effect, and it allows allelic exchange of chromosomal regions with fragments of interest (27). The galK-based IF system will facilitate the construction of C. perfringens strains that are more useful for the biomedical and bioindustrial sciences.