Prevalence of Enterotoxigenic Clostridium perfringens Isolates in Pittsburgh (Pennsylvania) Area Soils and Home Kitchens

Abstract

In the United States and Europe, food poisoning due to Clostridium perfringens type A is predominantly caused by C. perfringens isolates carrying a chromosomal enterotoxin gene (cpe). Neither the reservoir for these isolates nor the point in the food chain where these bacteria contaminate foods is currently understood. Therefore, the current study investigated whether type A isolates carrying a chromosomal cpe gene are present in two potential reservoirs, i.e., soil and home kitchen surfaces. No C. perfringens isolates were recovered from home kitchen surfaces, but most surveyed soil samples contained C. perfringens. The recovered soil isolates were predominantly type A, but some type C, D, and E soil isolates were also identified. All cpe-positive isolates recovered from soil were genotyped as type A, with their cpe genes on cpe plasmids rather than the chromosome. However, two cpe-positive soil isolates did not carry a classical cpe plasmid. Both of those atypical cpe-positive soil isolates were sporulation capable yet failed to produce C. perfringens enterotoxin, possibly because of differences in their upstream promoter regions. Collectively these results suggest that neither soil nor home kitchen surfaces represent major reservoirs for type A isolates with chromosomal cpe that cause food poisoning, although soil does appear to be a reservoir for cpe-positive isolates causing non-food-borne gastrointestinal diseases.

Clostridium perfringens, a gram-positive spore-forming anaerobe, is commonly present in the normal intestinal flora, feces, and soil (11). Individual isolates of this bacterium are classified into five types (A to E), based upon their production of four (alpha, beta, epsilon, and iota) typing toxins (13, 18, 19). Each C. perfringens type is associated with certain human or veterinary diseases. In addition, ∼5% of C. perfringens isolates produce another toxin, named C. perfringens enterotoxin (CPE). Although not part of the toxin typing classification scheme, CPE is biomedically important. CPE-producing type A isolates are responsible for the third most common food-borne illness in the United States and can also cause non-food-borne human illnesses such as antibiotic-associated diarrhea (12).

Recent studies (2, 3, 14, 16, 24) have shown that classical food poisoning outbreaks due to C. perfringens usually involve those type A isolates carrying their enterotoxin gene (cpe) on the chromosome. In contrast, the cpe gene is typically plasmid borne in type A isolates causing non-food-borne human illnesses (3). To date, two distinct cpe plasmid families have been identified in type A isolates (15). The first type A cpe plasmid family has an IS1151 sequence downstream of the plasmid-borne cpe gene, while the other family has an IS1470-like sequence downstream of the plasmid-borne cpe gene. Recent sequencing studies (15) showed these two type A cpe plasmid families share a conserved region corresponding to about 50% of each plasmid, while their unique variable regions carry open reading frames (ORFs) that potentially encode other toxins, bacteriocins, and metabolism-related proteins.

There has been recent progress towards understanding why the type A isolates carrying a chromosomal cpe gene are often associated with classical food poisoning outbreaks due to C. perfringens (9, 10, 20). However, the natural reservoir for type A chromosomal cpe isolates remains obscure, as does the point in the food chain where chromosomal-cpe isolates contaminate foods. A recent survey (24) of American retail foods did report that ∼1.7% of raw meat, fish, and poultry items sold in retail food stores contain type A isolates carrying a chromosomal cpe gene. Interestingly, no type A isolates carrying a plasmid cpe gene were found in any of those surveyed retail foods. Those findings suggest that meats, poultry, and seafood, which are common food vehicles for food poisoning due to C. perfringens type A in the United States and Europe, can be contaminated with type A chromosomal-cpe isolates by the time of retail purchase. However, those survey results do not preclude the possibility that foods may also become contaminated in kitchens.

Human carriers represent a potential reservoir for type A chromosomal-cpe isolates, with carriers possibly introducing these bacteria into foods during handling. Supporting this possibility, a few fecal samples from healthy people in Finland or Japan were found to contain type A chromosomal-cpe isolates (6, 22), although an ongoing American survey has not found any of these bacteria in feces from healthy people (R. J. Carman, S. Sayeed, J. Li, and B. A. McClane, unpublished observation). Collectively, these initial survey results would indicate that healthy people are not generally carriers of type A chromosomal-cpe isolates. Another possible reservoir for type A chromosomal-cpe isolates would be food animals, but this possibility has not yet been examined.

A third possibility, also not yet evaluated, is that foods might become contaminated with type A chromosomal-cpe isolates residing in environmental niches such as soil or home kitchen surfaces. C. perfringens type A isolates are commonly isolated from soil samples, but the single survey evaluating the presence of cpe-positive isolates in soils produced negative results (8). Similarly, it remains unclear whether type A chromosomal-cpe isolates are commonly present in the home kitchen environment. Since these issues are important for understanding the reservoirs and transmission of C. perfringens type A causing food poisoning and non-food-borne human gastrointestinal (GI) diseases, we surveyed soils and home kitchen surfaces in the Pittsburgh, PA, area to determine the prevalence of cpe-positive C. perfringens isolates in these two environments.

MATERIALS AND METHODS

Collection and processing of kitchen samples.

The kitchen survey was conducted in two phases, each using slightly different sampling techniques. In the first phase, a total of 300 samples were collected from the kitchens of 30 homes in the Pittsburgh region during the period from August to October 2006. These kitchen samples were collected with dry swabs (Sarstedt Co.) from floors or the surfaces of kitchen counters, tables, sinks, stoves, refrigerators, dishwashers, and microwaves. The samples were returned quickly to the laboratory, where a 5-ml aliquot of differential reinforced clostridial broth medium (DRCM; Merck Co.) was added to each tube containing a kitchen swab. After gentle mixing, a 1-ml aliquot of each DRCM suspension (containing a kitchen sample) was added to two tubes, each containing 9 ml of sterile DRCM. To identify, and encourage the germination of, C. perfringens spores present in the kitchen samples, one tube of each paired set was heat shocked at 75°C for 20 min before incubation at 43°C for 16 h (no anaerobic incubator was needed since DRCM is sufficiently reducing for C. perfringens growth in the presence of air). The other tube was directly incubated at 43°C for 16 h to enrich for C. perfringens vegetative cells or self-germinating spores that had been present in the sample. A color change in the medium from pink to black was indicative of the possible presence of C. perfringens, as confirmed by inoculating DRCM with known C. perfringens laboratory isolates.

In phase 2 of the kitchen survey, a total of 120 samples were collected from the kitchens of 12 homes in the Pittsburgh area during August and September 2007. These kitchen samples were collected by moist swabbing of an 100-cm² area on refrigerator handles, food preparation surfaces, cutting boards, sinks, sink faucets, floors, oven handles, microwave oven handles, the inside surfaces of microwave ovens, and the kitchen eating table. Sterile swabs were moistened with DRCM prior to sample collection. After use on a kitchen surface, the swab was placed into a 5-ml tube of DRCM and returned immediately to the laboratory. The tube contents were then divided in half, with one portion directly cultured at 43°C and the other heat shocked at 75°C (to promote spore germination) for 20 min prior to incubation at 43°C. A color change in the medium from pink to black was indicative of the possible presence of C. perfringens, as confirmed by inoculating DRCM with known C. perfringens laboratory isolates.

Testing the ability of our kitchen survey method to detect the presence of C. perfringens on impervious surfaces.

To confirm that our survey methods can detect C. perfringens when present on surfaces, two approaches were used. First, surveys were conducted of various locations in our laboratory using the same methods described above for collecting kitchen samples. The second approach involved intentionally adding C. perfringens to a nonporous (glass) surface. For this experiment, both a vegetative culture and a sporulating culture of C. perfringens strain NCTC8239 were prepared as described previously (10). Those cultures were then serially diluted, and 10 μl of each culture dilution was spread on the glass surface, which was left overnight on the laboratory bench. The next morning, the surfaces were swabbed with either a dry or moist swab, and samples were processed as described for the kitchen survey.

Collection of soil samples.

Samples were collected on 32 days, extending from November 2005 to October 2006, with two or three samplings per month. Sampled sites were various locations in the Pittsburgh region, including agronomic, home garden, residential, wild/natural, riparian, and industrial soils. A total of 502 soil samples were collected throughout the study. Each sample contained >1 g of soil, which was collected in a 15-ml sterile plastic tube.

Preparation of soil samples.

A 10-ml aliquot of sterile DRCM was added to 1 g of each soil sample in a 15-ml sterile tube. After gentle mixing, 1-ml aliquots of a DRCM suspension containing a soil sample were added to two tubes, each containing 9 ml of sterile DRCM. To identify, and encourage the germination of, C. perfringens spores present in the soil samples, one tube of each paired set was heat shocked at 75°C for 20 min before incubation at 43°C for 16 h. The other tube was directly incubated at 43°C for 16 h to enrich for C. perfringens vegetative cells or self-germinating spores that had been present in the sample.

MPN estimation of C. perfringens levels per gram of soil.

As we had used previously to evaluate C. perfringens levels in foods (24), a three-tube (five-dilution) most-probable-number (MPN) method was employed to estimate C. perfringens levels in soil samples. Briefly, a soil suspension (prepared as described above) was diluted by 10-fold increments (from 10⁻¹ to 10⁻⁵) in DRCM. After incubation for 16 h at 37°C, cultures testing presumptively positive for C. perfringens produced a black precipitate. Statistical analyses were performed as previously described (24).

Multiplex PCR assay to determine the toxin genotype of C. perfringens soil isolates.

One loopful of a positive (i.e., black) DRCM culture from a soil sample was streaked onto a tryptose sulfite cycloserine (TSC; Shahidi-Ferguson perfringens agar base [Difco] plus 0.04% d-cycloserine [Sigma]) selective agar plate. After overnight incubation in an anaerobic jar (BBL) at 37°C, three isolated colonies from each TSC plate were streaked onto individual brain heart infusion (Difco)-agar plates. From the resultant colonies, template DNA was prepared for the multiplex PCR toxin genotyping assay, as described previously (24). The multiplex PCR assay used in this study is capable of detecting the presence of genes encoding alpha toxin (plc or cpa), beta toxin (cpb), beta2 toxin (cpb2), epsilon toxin (etx), CPE (cpe), and the iota toxin activity component (iap). For this multiplex PCR, the sequences of the six primer pairs and the amplification conditions have been described previously (5).

When isolates were identified by the multiplex PCR as belonging to type C, D, or E, their genotypes were confirmed using a second PCR amplifying different sequences present, as appropriate, in genes encoding beta toxin, epsilon toxin, or the Ia component of iota toxin. Primers for those confirmatory PCRs are as follows: for beta toxin, Beta-F (TAAATGATATAGGTAAAACTACTACTA) and Beta-R (CTAAATAGCTGTTACTTTGTGAGTAAG); for epsilon toxin, Epsilon-F (TTTTAACTTGGGTTTTGTCG) and Epsilon-R (GGTTCTCCATCTAATTCAAC); and for iota toxin, Iota-F (ACATCCTCTGTAAGTAATCG) and Iota-R (GGTTCTCCATCTAATTCAAC). The template DNA was obtained from C. perfringens colony lysates, as described previously (24). Each PCR mixture contained 2 μl of template DNA, 17 μl of Taq complete 1.1× Master Mix (Gene Choice, Frederick, MD), and 1 μl of each primer pair (1 μM final concentration). The reaction mixtures, with a total volume of 20 μl, were placed in a thermal cycler (Techne) and subjected to the following amplification conditions: 1 cycle at 95°C for 2 min; 35 cycles at 95°C for 30 s, 54°C for 1 min, and 72°C for 1 min; and a single extension of 72°C for 10 min. PCR products were then electrophoresed on a 1% agarose gel, which was stained with ethidium bromide.

Multiplex PCR to determine the cpe genotypes of cpe-positive type A soil isolates.

Soil isolates identified as cpe-positive C. perfringens type A isolates by the multiplex PCR toxin genotyping assay were then tested by a multiplex PCR cpe-specific genotyping assay to establish whether they carry a chromosomal or a plasmid-borne cpe locus. The multiplex PCR assay used for this purpose has been described previously (16) and is capable of distinguishing between isolates carrying a chromosomal cpe locus, a plasmid cpe locus with downstream IS1151 sequences, and a plasmid cpe locus with downstream IS1470-like sequences. The primers and amplification conditions used for this assay were described previously (16).

PFGE Southern blotting analysis of cpe plasmid type A soil isolates.

Pulsed-field gel electrophoresis (PFGE) Southern blotting was carried out as described by Fisher et al. (4). Digoxigenin-labeled cpe probes were prepared as described previously (2). CSPD {disodium 3-(4-methoxyspiro[1,2-dioxetane-3,2′{5′-chloro}tricyclo{3.3.1.1^3/7}decan]-4-yl) phenyl phosphate} substrate (Roche Applied Science) was used for detection of hybridized cpe probes, as directed by the manufacturer.

Western immunoblot analysis of CPE expression by cpe-positive type A soil isolates.

A 0.2-ml aliquot of a fluid thioglycolate culture of selected cpe-positive type A soil isolates was inoculated into 10 ml of Duncan-Strong (DS) medium. After incubation at 37°C for 8 h, those DS cultures were examined with a phase-contrast microscope to ensure the presence of spores. At that point, DS cultures containing spores were sonicated until >95% of vegetative cells had lysed. The sonicated samples were then centrifuged to remove debris and unlysed cells. Supernatants from the DS culture lysate were analyzed for the presence of CPE using a previously described CPE Western immunoblot procedure (7).

PCR analyses to evaluate the presence of the pCPF5603 and pCPF4969 conserved and variable regions in cpe-positive type A soil isolates.

For the short-range overlap PCRs, template DNA was obtained from C. perfringens colony lysates, as described previously (24). Each PCR mixture contained 2 μl of template DNA, 17 μl of Taq complete 1.1× Master Mix (Gene Choice, Frederick, MD), and 1 μl of each primer pair (1 μM final concentration). Primers used to evaluate whether cpe plasmids of soil isolates carry the variable or conserved regions of cpe plasmid pCPF5603 or pCPF4969 from type A isolates F5603 and F4969, respectively, have been previously described (15). The reaction mixtures, with a total volume of 20 μl, were placed in a thermal cycler (Techne) and subjected to the following amplification conditions: 1 cycle at 95°C for 2 min; 35 cycles at 95°C for 30 s, 58°C for 1 min, and 72°C for 100 s; and a single extension of 72°C for 10 min. PCR products were then electrophoresed on a 1% agarose gel, which was stained with ethidium bromide.

Sequencing of the cpe gene and PCR amplification of the cpe promoter region in selected type A soil isolates.

Template DNA was obtained from C. perfringens colony lysates, as described previously (24). The primers cpe-F (5′-ATGCTTAGTAACAATTTAAATCC-3′) and cpe-R (5′-ATTTTTGAAATAATATTGAATAAGG-3′) were designed to amplify the cpe gene, based upon a previous study (15). Each PCR mixture contained 5 μl of template DNA, 43 μl of Taq complete 1.1× Master Mix (Gene Choice, Frederick, MD), and 2 μl of each primer pair (1 μM final concentration). The reaction mixtures, with a total volume of 50 μl, were placed in a thermal cycler (Techne) and subjected to the following amplification conditions: 1 cycle at 95°C for 2 min; 35 cycles at 95°C for 30 s, 56°C for 40 s, and 72°C for 1 min; and a single extension of 72°C for 10 min. PCR products were then electrophoresed on a 1% agarose gel, which was stained with ethidium bromide. PCR products were then excised from the gel, ligated into a pCR2.1-TOPO vector (Invitrogen), and sequenced by the Genomics and Proteomics Core Laboratory of the University of Pittsburgh School of Medicine.

To amplify the cpe promoter region, primers pro-F (5′-GCTTAACTATTCTTCTTGATAGTTATCT-3′) and pro-R (5′-GCATTTTCGAACACCATTGGATTT-3′) were used. PCR conditions were as described above for amplifying the cpe gene.

RESULTS

Prevalence of Clostridium perfringens in home kitchens.

Although a number of different kitchen surfaces (see Materials and Methods) were surveyed in many different Pittsburgh area homes, no C. perfringens spores or vegetative cells were recovered using the methods described here. About two-thirds of those surveyed kitchens were located in homes with children, while about one-third were located in homes with pets (cats or dogs).

To ensure that our survey methods could identify C. perfringens in kitchens, if it was present, two confirmatory experiments were performed. First, we surveyed our laboratory bench surfaces for the presence of C. perfringens. Without the heat shocking of samples, this survey grew C. perfringens (presumptively vegetative cells since heat shocking promotes spore germination) in 3 of 20 samples, while C. perfringens spores surviving heat shocking were detected in 5 of 20 samples. All of the C. perfringens isolates recovered in our laboratory survey were cpe-positive type A isolates, with some carrying a chromosomal cpe gene but others a plasmid cpe gene.

Second, we intentionally contaminated a glass plate with dilutions of C. perfringens vegetative or sporulating cultures. After an overnight incubation, C. perfringens presumptively vegetative cells were detected, using either dry or moist swabs, for the sample containing 10⁷ cells/ml. In contrast, dry swabs were positive for C. perfringens in samples containing 10³ spores/ml while moist swabs were even more sensitive, detecting C. perfringens in samples containing 10² spores/ml. Since 10-μl aliquots of that dilution had been plated onto the glass surface, our moist-swab survey method was essentially capable of recovering and detecting the single spore deposited on the plate for the sample containing 10² spores/ml.

Initial estimation of Clostridium perfringens levels in soil.

A three-tube MPN method using DRCM was employed to estimate the levels of C. perfringens present in soil from various locations in the Pittsburgh area. Those analyses suggested that, in our soil samples, the number of C. perfringens presumptively vegetative cells (i.e., cells growing without heat shocking) per gram ranged from 0 to 9,300 and the number of C. perfringens spores per gram (i.e., cells growing after heat shocking) ranged from 0 to 4,300.

Multiplex PCR toxin genotyping of DRCM-positive soil samples.

Each DRCM-positive soil sample was plated onto TSC agar, and three randomly chosen colonies were assessed by a multiplex PCR toxin genotyping assay that classifies C. perfringens isolates into types A to E. This analysis confirmed that ∼80% of DRCM-positive TSC colonies are C. perfringens, i.e., these isolates carry the plc gene that is diagnostic for C. perfringens (Fig. 1A). This determination implies that the actual numbers of C. perfringens presumptively vegetative cells present in our soil sample ranged from 0 to ∼7,500, while the actual numbers of C. perfringens spores present in those samples ranged from 0 to ∼3,500.

FIG. 1.

Multiplex PCR genotyping results for C. perfringens soil isolates. (A) Toxin typing analysis of selected C. perfringens soil isolates, identifying cpe-negative type A isolates (S102-2, SS95-13); cpe-positive, cpb2-negative type A isolates (S292-3, S350-1, SS151-1); cpb2-positive, cpe-negative type A isolates (S37-1); type C isolates (SS32-1); type D isolates (S153-2, SS211-1); and type E isolates (S155-1 and SS372-2). (B) cpe genotyping results for cpe-positive type A isolates identifying type A soil isolates carrying a plasmid cpe locus with downstream IS1470-like sequences (S194-1, SS151-1), type A soil isolates carrying a plasmid cpe locus with downstream IS1151 sequences (S37-1, SS44-1), and non-genotypeable cpe-positive type A soil isolates S292-3 and S350-1. Isolate SS167-1 could not be genotyped using crude culture lysates, as shown, but (using purified DNA) was identified as carrying a cpe plasmid with downstream IS1151 sequences (not shown). For comparison, a type A food poisoning-causing derivative with a chromosomal cpe gene (SM101), a type A non-food-borne GI disease-causing isolate carrying a plasmid cpe locus with downstream IS1151 sequences (F5603), and a type A non-food-borne GI disease isolate carrying a plasmid cpe locus with downstream IS1470-like sequences (F4969) are also shown. M, marker lane.

This multiplex PCR assay further showed that, regardless of origin, ∼70 to 75% of the surveyed soil samples contained C. perfringens (Table 1). The vast majority (Fig. 1A and Table 2) of the C. perfringens isolates recovered from soil were classified as type A, i.e., they carried the plc gene but not the etx, cpb, or iap gene. However, type C, D, and E isolates were found in a few soil samples. Those non-type A soil isolates were confirmed as type C, D, or E by a second independent PCR for the cpb, etx, or iap gene (not shown). Some of those non-type A soil isolates were present as spores, others as vegetative cells. No type B soil isolates were identified during this survey.

TABLE 1.

Presence of C. perfringens in soil samples in the Pittsburgh area

C. perfringens form	No. (%) of samples positive for C. perfringens froma:
	City (136)	Suburbs (309)	Farm (57)	Total (502)
Presumptively vegetative cells	100 (73.5)	204 (66.0)	39 (68.4)	343 (68.3)
Spores	101 (74.3)	231 (74.8)	45 (78.9)	377 (75.1)
cpe-positive presumptively vegetative cells	5 (3.7)	4 (1.3)	2 (3.5)	11 (2.2)
cpe-positive spores	7 (5.1)	17 (5.5)	1 (1.8)	25 (5.0)

^aNumbers of parentheses after sites are numbers of samples obtained.

TABLE 2.

Analysis of C. perfringens toxin types in Pittsburgh soils

Type	No. (%) of samples positive for indicated form of C. perfringens/total no. of samples
	Presumptively vegetative cells	Spores
A	339/343 (98.8)	374/377 (99.2)
B	0/343 (0)	0/377 (0)
C	0/343 (0)	1/377 (0.3)
D	8/343 (2.3)	1/377 (0.3)
E	4/343 (1.2)	3/377 (0.8)

Presence of cpe-positive C. perfringens isolates in the soil.

The multiplex PCR toxin genotyping assay also demonstrated that ∼7% of the surveyed soil samples contained C. perfringens isolates carrying the enterotoxin (cpe) gene. Those cpe-positive soil isolates were all classified as type A, with some being present in soil as presumptively vegetative cells but others as spores.

A multiplex cpe-specific genotyping PCR analysis was then applied to further genotype the type A cpe-positive isolates recovered from soil. That assay detected no cpe-positive soil isolates (either presumptively vegetative cells or spores) carrying a chromosomal cpe gene. Most cpe-positive soil isolates recovered as presumptively vegetative cells from soil were found to carry a cpe plasmid belonging to either of the two previously identified cpe plasmid families. Interestingly, two cpe-positive type A isolates that had been present as presumptively vegetative cells in soil did not amplify any of the cpe-specific genotyping multiplex PCR products that should indicate carriage of a chromosomal or a classical plasmid-borne cpe gene (Table 3 and Fig. 1B).

TABLE 3.

Genotyping results for cpe-positive C. perfringens type A isolates in Pittsburgh soil samples

Genetic element	No. (%) of samples positive for indicated form of C. perfringens/total no. of samples
	Presumptively vegetative cells	Spores
Chromosomal cpe	0/12 (0)	0/25 (0)
Plasmid cpe	12/12 (100)	25/25 (100)
IS1151	6/12 (50.0)	8/25 (32.0)
IS1470-like	4/12 (33.3)	17/25 (68.0)
Unknown	2/12 (16.7)	0/25 (0)

Similar multiplex PCR cpe genotyping results were obtained for the cpe-positive isolates that had clearly been present in soils as heat-resistant spores (Table 3 and Fig. 1B). That is, none of the cpe-positive spores recovered from soil carried a chromosomal cpe gene, but all cpe-positive spores present in our soil samples carried a cpe plasmid belonging to one of the two previously identified cpe plasmid families. For unknown reasons, one cpe-positive type A spore recovered from soil initially failed to give a multiplex PCR cpe-specific genotyping reaction using crude lysates but, upon retesting using purified DNA as the template, was identified as a plasmid cpe isolate with IS1151 sequences downstream of its cpe gene.

Analysis of atypical C. perfringens cpe-positive type A soil isolates.

We investigated the two atypical type A cpe-positive soil isolates that failed to amplify any product in the multiplex PCR cpe genotyping assay. PFGE Southern blot analysis (4, 15) was first performed to determine whether the cpe gene is present on the chromosome or plasmids in those atypical cpe-positive type A soil isolates. This approach clearly showed that both of the atypical cpe-positive type A soil isolates carry a plasmid-borne cpe gene. However, the cpe plasmid in these two atypical isolates was smaller (∼60 kb) than the ≥70-kb classical cpe plasmids of type A isolates F4969 and F5603 (15). Similar PFGE Southern blotting (Fig. 2) of selected typical cpe-positive type A soil isolates confirmed the multiplex PCR cpe-specific genotyping results (Fig. 1B). Consistent with Fig. 1B results, PFGE Southern blots (Fig. 2) showed that SS151-1 and S194-1 both carry a cpe plasmid that comigrates with the ∼70-kb pCPF4969 (15). Similarly, consistent with Fig. 1B multiplex PCR cpe-specific genotyping results, type A soil isolates S37-1 and SS44-1 were found (Fig. 2) to carry a cpe plasmid that comigrates with the ∼73-kb pCPF5603. However, another soil isolate (SS167-1) whose PCR genotype was similar to that of F5603 has a much smaller (∼48-kb) cpe plasmid than the ∼73-kb pCPF5603.

FIG. 2.

PFGE Southern blot analysis of cpe gene location in selected cpe-positive C. perfringens type A soil isolates. The blot was hybridized with a cpe probe, and size markers are shown to the left of the blot. Also included for comparison are cpe-negative isolate ATCC 3624 (to demonstrate probe hybridization specificity) and type A non-food-borne GI disease-causing isolates F4969 and F5603, which have been shown by sequencing to carry cpe plasmids of ∼70 kb and ∼73 kb, respectively.

Since both classical cpe plasmid families share an ∼35-kb conserved region (15), an overlapping PCR was performed to investigate whether the cpe plasmid in the two atypical soil isolates or SS167-1 (the soil isolate with a very small cpe plasmid) carries this conserved region (see Fig. S1 in the supplemental data). This assay showed that isolates SS167-1 and S292-1 both carry substantial portions of the pCPF5603/pCPF4969 conserved region but that only a limited portion of this region is present in isolate S350-1. Since the two isolates carrying atypical cpe plasmids were PCR positive for IS1151 (see Table S1 in the supplemental data), a second overlapping PCR assay that is specific (see Fig. S2 in the supplemental data) for the variable region of the pCPF5603 cpe plasmid family carrying IS1151 sequences (15) was run. For the atypical cpe-positive soil isolates S292-3 and S350-1, this variable-region assay amplified only limited products (see Fig. S2 in the supplemental data). A similar overlapping PCR assay that is specific (see Fig. S3 in the supplemental data) for the variable regions of pCPF4969 cpe plasmids carrying IS1470-like sequences (15) failed to amplify most of this pattern for the two atypical isolates, except for amplification of the products from reactions 9 to 19 (from a region containing a putative spa bacteriocin locus) from isolate S292-3. This assay also amplified substantial portions of the expected pattern for SS167-1, but not the expected products from reactions 19 to 27.

PCRs were performed for individual ORFs present in the conserved or variable regions of pCPF5603/pCPF4969 to determine whether those ORFs are present in the atypical cpe-positive soil isolates but arranged differently from the previously characterized cpe plasmids (15). However, these individual PCRs were only intermittently positive with the two atypical soil isolates (Table 4). Finally, an attempt was made to link IS1151 to the cpe gene in the two atypical cpe-positive soil isolates using the standard cpe primer of the multiplex cpe genotyping PCR assay and a second IS1151 primer, but this was unsuccessful, even using purified DNA.

TABLE 4.

Analysis of unknown cpe-positive type A isolates from Pittsburgh soils

Isolatea	Resultb for:							IS1151-cpe linkage	CPE production
	Overlap PCR				Individual PCR		cpe promoter region PCR
	F5603 conserved region	tcp genes	F5603 variable region	F4969 variable region	Conserved ORFs	Variable ORFs
S292-3	±	+	±	±	Like pCPF5603	−	−	−	−
S350-1	±	−	−	−	Like pCPF5603	−	−	−	−

^aFor both isolates, the location of cpe is a 60-kb plasmid.

^b+, positive for the entire region or sequence; −, negative for the entire region or sequence; ±, positive for part of the region or sequence.

CPE expression by cpe-positive type A soil isolates.

CPE Western blotting was performed to assess CPE production by 12 selected soil isolates carrying classical cpe plasmids, isolate SS167-1 with the small pCPF5603-like cpe plasmid, and the two soil isolates with atypical cpe plasmids. This analysis demonstrated the presence of CPE in sporulating cultures of all 12 soil isolates with typical cpe plasmids and SS167-1 (Fig. 3 and data not shown). However, neither of the two atypical cpe-positive soil isolates gave positive Western blot results for CPE expression. The failure of isolates S292-1 and S350-1 to produce CPE was not due to an inability to sporulate (CPE expression is strongly sporulation associated) since both those isolates sporulated well in DS medium.

FIG. 3.

Western blot analysis of CPE production by cpe-positive C. perfringens type A soil isolates. Isolates were grown for 8 h in DS sporulation medium until phase-refractile spores were present in all cultures, as confirmed by phase-contrast microscopy. Culture lysates were then subjected to Western blot analysis using CPE antibodies. Shown are results for seven selected cpe-positive type A soil isolates (S37-1, SS44-1, S194-1, SS151-1, S292-3, S350-1, and SS167-1). Also shown for comparison are CPE Western blot results for sporulating culture lysates of known CPE-positive isolate F5603 and cpe-negative isolate ATCC 3624.

To assess whether their failure to express CPE is due to the presence of a premature stop codon, the cpe ORFs of atypical isolates S292-1 and S350-1 were sequenced. However, no differences were found relative to the functional cpe ORF of F4969 or F5603. A PCR was then performed using primers hybridizing with sequences in the upstream promoter region of the functional cpe gene of NCTC8239. This reaction generated the expected products from F4969 or F5603 but no products from S292-3 or S350-1, suggesting that differences exist between the cpe promoter regions of these two isolates versus those of typical type A isolates expressing CPE.

DISCUSSION

This survey did not detect C. perfringens in the home kitchen environment, suggesting that foods do not commonly become contaminated with isolates carrying chromosomal cpe that cause food poisoning from home kitchen surfaces. An experiment where we intentionally contaminated a nonporous surface offers one possible explanation for our failure to detect C. perfringens on kitchen surfaces, i.e., C. perfringens vegetative cells do not persist well in this environment, possibly due to their anaerobic nature. Follow-up studies should survey the presence of C. perfringens food poisoning-causing isolates in home kitchens in other geographic locations and in commercial kitchens, given the common incidence of large food poisoning outbreaks due to C. perfringens type A in institutional settings. However, it is notable that a survey of a hospital kitchen in Brazil also failed to identify C. perfringens (17). The previous detection of cpe-positive C. perfringens on surfaces of hospital rooms (17) could reflect high-level, continual contamination with these bacteria, as also occurs in our laboratory. Consistent with that possibility, cpe-positive isolates were more commonly recovered from hospital rooms with patients suffering from CPE-associated non-food-borne disease.

Given previous reports (11, 21, 23) that C. perfringens is commonly found in soil, it was also conceivable that soils could be a major reservoir for C. perfringens type A causing food poisoning, with foods becoming contaminated at processing plants via direct contact with soil or soil-containing dust. However, to our knowledge there had been only a single survey using the reliable molecular epidemiologic techniques that can definitively identify cpe-positive isolates, and that survey had failed to identify any cpe-positive soil isolates (8).

The current study demonstrates that C. perfringens spores, as well as presumptively vegetative cells, are present in soils, a subject not addressed in that other recent survey. While C. perfringens has been thought to predominantly exist in soils as vegetative cells (21), we found nearly as many C. perfringens spores as presumptively vegetative cells in our soil samples. This finding could indicate that vegetative cells in soils reflect fairly recent contamination of soil and that those bacteria will eventually sporulate due to poor growth conditions. In some soil samples, this C. perfringens contamination could result from the common usage of animal manures as fertilizers. Alternatively, it is possible that C. perfringens can grow in soil. Given the little that is known about the growth or persistence of C. perfringens vegetative cells in soil, this issue clearly requires further study.

Our PCR-based results also somewhat contradict a common belief (21), based upon results from classical typing neutralization approaches, that only type A isolates of C. perfringens are present in soil. Although multiplex PCR toxin typing results from our study confirmed that type A isolates are the predominant isolates in our soil samples, type C, D, and E soil isolates were also identified. Interestingly, those non-type A isolates had been present in soil samples obtained mainly from residential and home garden soils in urban/suburban areas rather than soils from woodlands, stream banks, or farms (although it should be noted that we surveyed agronomic soils on farms rather than barnyard areas containing high levels of fresh manure; the presence of cpe-positive C. perfringens in human and animal feces is the subject of a separate ongoing study in our laboratory). The presence of non-type A isolates in soils from urban and suburban areas might conceivably reflect fecal contamination from nonfood animals such as pets or rodents. Additional surveys should explore this hypothesis.

Contrary to the recent PCR-based survey identifying no cpe-positive C. perfringens isolates in soils collected from throughout the United States (8), our study recovered several cpe-positive isolates from Pittsburgh area soil. Those cpe-positive soil isolates were all classified as type A but carried a plasmid-borne cpe gene rather than the chromosomal cpe gene most commonly associated with food poisoning. These results suggest that soil is not a major reservoir for C. perfringens isolates that cause food poisoning; however, additional surveys should further test this possibility. However, the presence of type A isolates carrying a plasmid cpe gene in soils does identify soil as a potential reservoir for CPE-associated non-food-borne GI diseases.

PFGE Southern blot analysis revealed that cpe-positive soil isolates often carry cpe plasmids similar in size to pCPF5603 or pCPF4969. However, one cpe-positive soil isolate, SS167-1, carries a relatively small cpe plasmid of ∼50 kb that is similar in size to the cpe plasmids in two unusual non-food-borne GI disease-causing isolates (15). The smaller cpe plasmids in all three of these isolates resemble pCPF5603, but with different deletions. These findings show that, even within the pCPF5603 cpe plasmid family, genetic variations can develop due to processes such as deletions.

Prior to this study, the multiplex PCR cpe genotyping assay used crude culture lysates to successfully cpe genotype every one of nearly 100 cpe-positive type A isolates in our laboratory collection (25). However, this assay did not work using crude lysates of three cpe-positive type A soil isolates, although one of those isolates (SS167-1) was later cpe genotyped using purified DNA. Because of the uniqueness of those two atypical cpe-positive type A isolates, PFGE analysis was performed on them to determine if they carry a plasmid-borne or chromosomal cpe gene. Both atypical cpe-positive isolates were shown to carry their cpe genes on plasmids of ∼60 kb. Overlapping PCR findings indicate that atypical cpe-positive type A soil isolate S292-3 has a cpe plasmid carrying most of the pCPF5603/pCPF4969 conserved region but that it generally lacks the variable region of either of those two cpe plasmids. The exception is the presence of a putative bacteriocin ORF cluster on the S292-2 cpe plasmid that is also present in pCPF4969. If produced, that bacteriocin might be helpful for the survival of S292-2 in soil. The other atypical type A soil isolate is clearly unique, carrying neither pCPF5603 nor pCPF4969 conserved- or variable-region ORFs. This unusual plasmid may have arisen from insertion of a genetic element carrying cpe sequences onto a plasmid backbone different from pCPF5603 or pCPF4969.

Another recent study recovered, from feces of healthy people, two cpe-positive type A isolates that did not amplify products using the multiplex PCR cpe genotyping assay (6). It would be interesting to determine whether those cpe-positive normal fecal isolates are related to the atypical cpe-positive soil isolates recovered in the current study. However, it is noteworthy that the atypical cpe-positive soil isolates recovered in our study failed to express CPE, despite being able to sporulate. This contrasts with the putative CPE-producing ability of at least one of the atypical normal fecal isolates identified in the previous study.

The two atypical cpe-positive soil isolates recovered in this study represent, to our knowledge, the first identification of sporulation-capable type A isolates with silent cpe sequences. However, silent cpe sequences are commonly found in type E isolates (1), possibly due to insertion of an iota toxin-encoding DNA element near the cpe promoter. Given PCR results indicating an unusual cpe promoter region, a DNA element may also have inserted in the cpe promoter regions of the two newly identified atypical type A soil isolates, rendering their cpe genes silent.

Finally, despite the occasional failures now documented in two recent studies, the current multiplex PCR cpe genotyping assay has proven to be a powerful molecular epidemiologic tool. Our results suggest that, when crude lysates of a cpe-positive isolate test negative with this assay, the isolate should be retested using purified DNA. However, even by testing with purified DNA, cpe isolates will occasionally not be genotyped with the current assay. This may not represent a significant problem, since three of the four atypical soil isolates identified to date clearly do not produce CPE, at least in vitro. If this pattern holds in future studies, atypical cpe-positive isolates may have minimal epidemiologic importance and the inability to genotype them may not represent a major concern. However, if a reasonable percentage of cpe-positive atypical isolates can produce CPE, then the cpe plasmids in those isolates will require further study so the multiplex PCR cpe genotyping can be revised to allow for their detection. Regardless, the discovery of even rare type A isolates with apparently silent cpe genes demonstrates that different conclusions can be obtained about cpe-positive type A isolates using genotypic versus phenotypic characterization methods, and this should always be considered during epidemiologic investigations or diagnostic studies.