INTRODUCTION
Clostridium perfringens is a Gram-positive, anaerobic spore-forming bacterium, and C. perfringens type A is a causative agent of food poisoning and non-foodborne human gastrointestinal diseases [Reference Rood and Cole1]. C. perfringens is commonly found in the gastrointestinal tract of mammals, as well as in soil and sewage. The diarrhoea and cramps, that comprise the typical clinical symptoms of this human gastrointestinal disease, are induced by a single 35-kDa polypeptide known as C. perfringens enterotoxin (CPE) [Reference Hauschild and Hilsheimer2–Reference Stark and Duncan4]. The knock-out mutant of cpe, i.e. the gene encoding CPE, could not induce rabbit ileal loop fluid accumulation and intestinal histopathological damage [Reference Sarker, Carman and McClane5]. Its synthesis is under the strict positive control of sporulation in the sporulation medium [Reference Sarker, Carman and McClane5–Reference Smith and McDonel7] and thus in the intestinal environment. Brett et al. reported that CPE-producing C. perfringens is responsible for 6·8% of all cases of sporadic diarrhoea [Reference Brett8]. There are multiple reports of non-foodborne human gastrointestinal diseases in elderly people [Reference Borriello9–Reference Fukao14]. It is reported that the intestinal flora of elderly people includes a higher number of C. perfringens than that of younger people [Reference Yamagishi15].
In this study, we analysed C. perfringens isolates which were derived from clusters of gastroenteritis cases in our geriatric hospital. We found that the clusters were outbreaks of enterotoxigenic C. perfringens infection. In the outbreak and sporadic isolates, cpe was plasmid borne. The experimental data showed that the same cpe-carrying plasmid had spread amongst the C. perfringens isolates.
MATERIALS AND METHODS
Clinical specimens and culture conditions
All faecal specimens from diarrhoea cases were examined routinely by Gram staining and by culture for the detection of Salmonella, Shigella, Campylobacter, Vibrio, Aeromonas and Yersinia. Faecal specimens in which round-ended Gram-positive rods containing spores were observed were analysed for CPE by using a reversed passive latex agglutination assay (RPLA) (Denka Seiken, Tokyo, Japan). The faecal specimens were mixed with 10× volume of saline and centrifuged at 3000 rpm for 20 min. Then the supernatant was serially diluted with diluent in the kit before being mixed with the latex beads. The culture method used for C. perfringens detection was as follows. The specimens were mixed with an equal volume of 99·5% alcohol and incubated for 30 min at room temperature. They were then plated on Gifu Anaerobic Medium agar plate (Nissui, Tokyo, Japan) containing 5% egg yolk and cultured at 35°C for 18–24 h in an anaerobic pouch (AnaeroPack; Mitsubishi Gas Chemical Co., Tokyo, Japan). A lecithinase-positive colony was selected from each plate and tested for haemolytic activity, motility and gas production. The resultant C. perfringens isolates were subjected to subsequent analyses.
Polymerase chain reaction (PCR)
The typing of C. perfringens isolates was performed by the established multiplex PCR [Reference Yoo16]. The cpe gene was detected by PCR with primers 3F and 4R according to the procedure of Miyamoto et al. [Reference Miyamoto, Wen and McClane17]. For distinguishing between the chromosomal and plasmid-borne cpe, cpe-IS sequences were amplified with primers cpe4F, IS1470R1.3, IS1470-likeR1.6 and IS1151R0.8 according Miyamoto et al. [Reference Miyamoto, Wen and McClane17]. A sequence in the tra locus, which corresponds to a part of the orf16 of the putative Tn916, was also amplified by primers 69219 (5′-CTTCATAGGATTGCTTCGCTC-3′) and 70672R (5′-CTTATAAATCCACATACAGACCAATACAG-3′) [Reference Miyamoto18]. The tra locus and orf16 have been renamed tcp locus and tcpF respectively [Reference Bannam19], and the new names will be used hereafter. The PCR was performed with ExTaq (Takara Bio, Otsu, Japan). The annealing temperatures were 55°C, 50°C and 45°C for the cpe, cpe-IS and tcpF amplifications, respectively.
Pulsed-field gel electrophoresis (PFGE)
Samples for PFGE were prepared as previously described [Reference Wada13]. In this analysis, 100 μm of thiourea (Sigma, MO, USA) was added to 0·5× Tris-borate-EDTA buffer in order to prevent the degradation of clostridial DNA [Reference Corkill20]. The PFGE conditions are described in the figure legends. Southern hybridization analysis was performed after PFGE or conventional gel electrophoresis. The probes were labelled by digoxigenin (Roche, Basel, Switzerland) and used for hybridization according to the manufacturer's instructions.
RESULTS
Analysis of faecal specimens and C. perfringens isolates
Our hospital has 646 beds for the primary care of elderly patients. One ward has a total of 40–43 beds: six rooms with six beds, zero or one room with four beds, and three or four rooms with one bed. Medical staff members belong to each ward in principle, but doctors have the opportunity to examine patients in other wards as consultants. Patients also have the chance of having contact with patients from other wards at the central examination unit. The average period of hospitalization is about 18 days. Usually, 0–5 isolations of C. perfringens per month are recorded. From the middle of February to the beginning of April 2006, a cluster of diarrhoea cases was observed in ward W10. During this period, 19 faecal specimens from 16 patients in this ward were examined, and 16 C. perfringens isolates were obtained from 14 patients. However, the number of C. perfringens isolations reverted to the normal level by the middle of April 2006. From the end of July to the beginning of September 2006, another cluster of diarrhoea cases was observed in ward W08. During this period, 25 faecal specimens from 18 patients in this ward were examined, and 16 C. perfringens isolates were obtained from 11 patients. Using the RPLA assay for detection of CPE, positive results were obtained in 63·2% (12/19) and 41·7% (10/24) of faecal specimens from wards W10 and W08, respectively. The positive results were obtained with 1280× or 2560× diluted supernatants of the faecal specimens. Using the PCR assay for detection of the cpe gene in the C. perfringens isolates, positive results were obtained in 71·4% (10/14) and 68·8% (11/16) of ward W10 and ward W08 isolates, respectively. These PCR results coincided with those of the Southern hybridization analysis by using a cpe-PCR product as a probe (see below). The numbers of patients whose faecal samples were positive for CPE and/or contained the cpe-positive C. perfringens isolates were 12 and nine in wards W10 and W08, respectively. No enteropathogenic bacteria other than C. perfringens were isolated in these patients. None of these patients had received antibiotic therapy before the onset of the diarrhoea. Because of the observed cpe-PCR-positive rates in our isolates, which were markedly higher than those in normal human C. perfringens isolates (see below), we consider the diarrhoea clusters to be outbreaks caused by enterotoxigenic C. perfringens.
PFGE and Southern hybridization analysis
In order to clarify the relationship between the two outbreaks, we analysed the 18 ward W10 isolates (14 isolates from the February–April outbreak plus four additional isolates during the July–September period), the 16 ward W08 isolates from the July–September outbreak, and five sporadic isolates from patients in other wards or from an outpatient. All the isolates were type A by multiplex PCR assay for typing [Reference Yoo16]. By PFGE analysis with SmaI digestion, the restriction profiles of the 28 cpe-positive isolates (14 ward W10 isolates, 11 ward W08 isolates and three sporadic isolates) could be classified into seven patterns, when minor variations were considered (Fig. 1a, lanes 1, 2, 4, 5, 7, 8, 9). All of the 11 cpe-positive ward W08 isolates exhibited identical PFGE patterns (Fig. 1a, lane 1). Of the 14 cpe-positive ward W10 isolates, nine isolates obtained in the February–April outbreak exhibited identical PFGE patterns (Fig. 1a, lane 2), whereas the four isolates obtained during the July–September period exhibited different PFGE patterns (Fig. 1a, lanes 5–7) from that of the February–April outbreak isolates (Fig. 1a, lane 2). The remaining one isolate, i.e. isolate 210, was obtained in the February–April outbreak period from ward W10, but it possessed a different PFGE pattern (Fig. 1a, lane 4) from that of the outbreak isolates from ward W10 (Fig. 1a, lane 2). The patterns of the outbreak isolates of wards W08 and W10 were very similar but distinguishable; the migration rates of the second and fourth bands slightly differed from each other (shown by white arrowheads in Fig. 1a, lane 1). The PFGE patterns of two sporadic cpe-positive isolates from other wards were different from the isolates of wards W10 and W08 (Fig. 1a, lanes 8, 9). The remaining one sporadic isolate revealed a pattern that was indistinguishable from that of the ward W10 outbreak isolate (data not shown). This isolate was from an outpatient who had been hospitalized in ward W10 (her hospitalization period overlapped with the outbreak period of ward W10).
Fig. 1. (a) Comparison of SmaI-digested chromosomal fragments of the representative C. perfringens isolates. The names of isolates are as follows: lane 1, 101; lane 2, 201; lane 3, 202; lane 4, 210; lane 5, 215; lane 6, 216; lane 7, 218; lane 8, 302; lane 9, 304; lane 10, 305. Isolate 101 is from ward W08; isolates 201, 202, 210, 215, 216 and 218 are from ward W10; and isolates 302, 304 and 305 are from other wards. All isolates except isolates 202 and 305 were cpe positive by PCR assay. Two fragments of isolate 101 whose migration rates differ from those of the corresponding fragments of isolate 201 are shown by white arrowheads. ‘M’ indicates a size maker of lambda concatemers. The sizes of the marker are shown on the left side of the image. The strength of the electric field was 6 V/cm. The switching interval of the electric field was ramped from 20 s to 25 s for 18 h. (b) Comparison of NotI-digested DNA fragments of C. perfringens isolates. The isolate in each lane is the same as panel (a). The strength of the electric field was the same as panel (a). The switching interval of the electric field was ramped from 0·5 s to 5 s for 20 h. (c) Southern hybridization analysis of the NotI-digested DNA fragments shown in panel (b). The cpe gene of isolate 101 was amplified with primers 3F and 4R, and the resulting PCR product was used as a probe.
Subsequent Southern hybridization analysis using a cpe-PCR product from isolate 101 as a probe showed that none of these SmaI fragments reacted with the probe. Instead, the origins of the cpe-positive isolates yielded positive signals (data not shown). When NotI was used for digesting the C. perfringens DNAs, 75-kb fragments were obtained in all but one cpe-positive isolate (Fig. 1b). The cpe probe in turn reacted with these 75-kb fragments (Fig. 1c). The one exceptional isolate, i.e. isolate 218, had a slightly larger NotI fragment (Fig. 1b, lane 7), which was also positive for the cpe hybridization signal (Fig. 1c, lane 7). The NotI fragment of isolate 202 was slightly smaller than 75 kb (Fig. 1b, lane 3), but it was negative for the cpe-hybridization signal (Fig. 1c, lane 3). Sporadic isolate 305, whose SmaI–PFGE pattern (Fig. 1a, lane 10) was indistinguishable from that of the outbreak isolates from ward W08 (Fig. 1a, lane 1), did not have a large NotI fragment (Fig. 1b, lane 10) and was also negative for the cpe-hybridization signal (Fig. 1c, lane 10). Not by SmaI digestion, but by NotI digestion, the DNA molecules on which the cpe gene was located were linearized and came to migrate into the gel. This observation denoted that the cpe gene is not located on the chromosome but on the 75-kb plasmid in our isolates.
PCR analysis of IS sequence
The cpe gene on the chromosome is typically associated with IS1470, whereas this gene on plasmids has been reported as located in two forms: one associated with IS1470-like and the other with IS1151 [Reference Cornillot21, Reference Miyamoto22]. Miyamoto et al. have developed a PCR test for the identification of the IS1470-associated chromosomal cpe, the IS1151-associated episomal cpe and the IS1470-like-associated episomal cpe [Reference Miyamoto, Wen and McClane17]. In this PCR test, the cpe gene was shown to be associated with IS1151 in all our outbreak and sporadic isolates (data not shown). According to the published nucleotide sequence of a large plasmid harbouring cpe-IS1151, this plasmid has a single NotI site [Reference Miyamoto18], which is comparable with our PFGE analysis result.
Comparison of the 75-kb plasmids
To clarify whether or not the 75-kb plasmids in the various C. perfringens isolates were identical, we performed conventional gel electrophoresis with PvuII digestion of the genomic DNAs, and subsequent Southern hybridization analysis by using the 75-kb NotI fragment of isolate 101 as a probe. The probe hybridized to the same four bands of the cpe-positive isolates (Fig. 2). The PvuII restriction profile of isolate 218, which had a slightly larger plasmid than the others (Figs 1b, c, lane 7), was indistinguishable from that of the 75-kb plasmid (Fig. 2, lane 7). However, isolate 202 (Fig. 2, lane 3), which has a cpe-negative plasmid (see above), showed a different hybridization pattern from the others.
Fig. 2. Southern hybridization analysis of PvuII-digested genomic DNAs of C. perfringens isolates. The DNAs were separated in 0·7% agarose gel by conventional gel electrophoresis and hybridized with the digoxigenin-labelled 75-kb NotI fragment of isolate 101. The isolate in each lane is the same as that of Figure 1a. The sizes of the marker are shown on the left side of the hybridization panel.
The tcpF gene PCR
We used two primers, namely, 69219 and 70672R, for assessing the presence of the tcpF gene in our isolates. This gene has been shown to be essential for the conjugative transfer of pCW3 [Reference Bannam19]. All of the cpe-positive isolates were also tcpF positive, whereas isolates 202 and 305, which were negative for cpe, were also negative for tcpF (Fig. 3).
Fig. 3. Amplification of tcpF-specific DNA fragments. The isolate in each lane is the same as that of Figure 1a. The size of the PCR product is shown on the left side of the image.
DISCUSSION
CPE-producing C. perfringens has been identified as a major cause of not only food poisoning but also non-foodborne gastrointestinal diseases including sporadic diarrhoea [Reference Brett8, Reference Mpamugo, Donovan and Brett23], antibiotic-associated diarrhoea [Reference Borriello9, Reference Borriello10] and diarrhoea in geriatric wards [Reference Borriello10–Reference Fukao14]. In 1993, our geriatric hospital experienced a large non-foodborne outbreak of gastroenteritis affecting 38 patients [Reference Wada13]. Since then, a limited number of C. perfringens isolation has been recorded. In 2006, increased numbers of diarrhoea cases were observed in two wards at different periods, and cpe-positive C. perfringens isolates were obtained from these cases. Although only one colony was examined for each culture plate, the positive rates of the cpe gene in these isolates were 71·4% and 68·8% in ward W10 and ward W08 isolates, respectively. These rates seem considerably higher than the reported positive rate of the cpe gene (about 5%) in human C. perfringens isolates [Reference Kokai-Kun24, Reference Daube25]. In addition, no enteropathogenic bacteria other than cpe-positive C. perfringens were isolated from patients in these wards. These observations lead us to suspect that the diarrhoea clusters should correspond to outbreaks that were caused by enterotoxigenic C. perfringens.
The chromosomal SmaI PFGE patterns of the outbreak isolates were mostly identical but differed in regard to four bands (Fig. 1a, see above). In addition, their small-plasmid profiles also differed (data not shown). The common feature is the 75-kb cpe-positive plasmid. These observations suggested that these outbreak isolates had been derived from the same strain, but had acquired different small plasmids in the hospital.
A plasmid carrying the IS1470-like-associated cpe has been shown to possess the tcp locus [Reference Miyamoto18], and it can actually transfer by conjugation [Reference Brynestad26]. Another plasmid possessing the IS1151-associated cpe shares very conserved tcp sequences with the plasmid possessing IS1470-like-associated cpe [Reference Miyamoto18] and with the conjugative plasmid pCW3 [Reference Bannam19], suggesting that the plasmid harbouring the IS1151-associated cpe is also conjugative [Reference Miyamoto18]. The 75-kb plasmids in our isolates carry the IS1151-associated cpe gene and the tcp locus, suggesting that this plasmid could have horizontally spread via conjugation to various C. perfringens isolates to cause nosocomial and sporadic infection. However, the site and time of the transfer are unclear. Comparative studies between the 1993 outbreak isolates and the 2006 outbreak/sporadic isolates in our hospital are now underway.
The cpe gene on the chromosome is reported to be associated with food poisoning, whereas this gene on plasmids is associated with non-foodborne gastrointestinal diseases [Reference Cornillot21, Reference Collie and McClane27]. Accordingly, our C. perfringens isolates, containing the cpe gene on the plasmids, caused nosocomial outbreaks of non-foodborne diseases in our geriatric hospital. However, C. perfringens isolates having the cpe gene on the plasmid have previously caused foodborne outbreaks in Japan [Reference Tanaka28, Reference Tanaka29] and in Europe [Reference Lahti30]. In order to clarify the pathogenicity of C. perfringens isolates carrying the cpe gene on plasmids, extended epidemiological and molecular studies are required.
ACKNOWLEDGEMENTS
This work was supported by a grant (H18-Shinkou-11 from the Ministry of Health, Labour and Welfare, Japan) to A. Wada. We sincerely thank Satoko Matsunaga for her excellent technical assistance.
DECLARATION OF INTEREST
None.
