A total of 136 fecal samples, 102 (75%) from female food handlers and 34 (25%) from male food handlers, were collected from food handlers in southern Finland during summer 2003. These persons reported no gastrointestinal symptoms at the time of sampling. Their ages ranged from 15 to 65 years (mean 30 years, median 24 years). The samples were kept at –70°C until investigated.

Each fecal sample (1 g) was divided into 2 tubes that contained freshly prepared thioglycollate (40 mL) (Oxoid, Basingstoke, UK). One tube was heated at 75°C for 20 min; the other was left unheated. After anaerobic incubation at 37°C for 20–22 h, the presence of cpe in each tube was determined by nested PCR (19). Persons with a cpe-positive fecal sample are hereafter referred to as cpe carriers.

Hydrophobic grid membrane filter-colony hybridization (HGMF-CH) (20) was used to isolate cpe+ C. perfringens from cpe carriers. In addition to probe-positive colonies, those showing no hybridization signals but having a typical color for C. perfringens colonies (usually black, occasionally gray or grayish yellow) were isolated from each sample to obtain a collection of both cpe+ and cpe– C. perfringens from a single sample and to further study the genetic relatedness of these isolates by PFGE (see below).

The isolates were subjected to PCR to determine the toxinotype (A–E) and the presence of cpe (21). C. perfringens strains NCTC 8239, ATCC 3626, CCUG 2036, CCUG 2037, and CCUG 44727 were used as positive controls.

C. perfringens isolates that possessed cpe were further studied by PCR to determine the cpe genotype of the strain. Total DNA was isolated by using Advamax beads (Edge Biosystems, Gaithersburg, MD, USA) according to the manufacturer's instructions. IS elements downstream of cpe that determine the cpe genotype (IS1151-cpe, IS1470-like-cpe, or IS1470-cpe) of each isolate were characterized by using primers and protocols described in Table 1. The location of cpe (plasmid or chromosome) was concluded according to the genotyping results (Table 1).

C. perfringens isolates were then typed by PFGE with ApaI and SmaI (New England Biolabs, Beverly, MA, USA) to study the genetic relationships between isolates (25). PFGE profiles were analyzed visually and with a computer software program (Bionumerics, version 4.5; Applied Maths, Kortrijk, Belgium). The similarities between macrorestriction patterns (MRP) were expressed by Dice coefficient correlation, and clustering by the unweighted pair-group method with arithmetic averages was used to construct a dendrogram.

To test the capability of cpe+ strains to produce CPE, 1–3 cpe+ isolates representing each MRP were sporulated (26). The final culture in modified Duncan-Strong medium (Sigma-Aldrich Chemie, Steinheim, Switzerland) was examined by phase-contrast microscopy to confirm the sporulation of the strain. The culture was then sonicated until >95% of the spores were free, as determined by phase-contrast microscopy. The culture was centrifuged at 1,500× g for 25 min at 4°C, and cytotoxicity of the supernatant was tested with a Vero cell assay according to Sandvig and Olsnes (27). The assay monitors the inhibition of protein synthesis in Vero cells after addition of toxic proteins, including CPE. The Vero cells were grown in a minimal essential medium (Gibco BRL, Paisley, UK) supplemented with 10% fetal calf serum.

The supernatant was precipitated in 80% saturated ammonium sulfate and kept at 4°C before the Vero cell assay. The strain was defined as producing CPE when the inhibition of protein synthesis of Vero cells was >20%.

The gene encoding for cpe was detected in the feces of 25 food handlers (18%). For 23 samples (92%), the heated tube showed the positive result, and for 3 of them the unheated tube also yielded a positive PCR result. For 2 samples, only the unheated tube showed a positive PCR result. No association between gender or age and carrier status of cpe was found (Table 2).

The HGMF-CH method was used to determine whether cpe+ C. perfringens was present in 23 persons; isolation was successful from 11 persons. The average number of cpe+ C. perfringens isolates carried by these persons was 2.5 × 10² CFU/g. In 10 of these persons, cpe– C. perfringens was also recovered; the average number was 1.8 × 10⁴ CFU/g. In each of these 10 case-patients, cpe– C. perfringens formed a 10- to 1,000-fold majority of the C. perfringens population. In 1 person, only cpe+ C. perfringens isolates were recovered. A total of 77 C. perfringens isolates (average of 7 isolates per person) were recovered; all possessed the gene encoding for the α toxin only, which signified that they belonged to type A (Table A1). Of these, 36 isolates were positive for cpe (Table A1).

When the relative prevalence of different cpe genotypes was determined with PCR, strains representing the IS1151-cpe type were found in 5 persons (3.7%) and strains representing the IS1470-like-cpe type were found in 4 persons (2.9%), findings that indicated that all of these persons carried C. perfringens strains with plasmidially located cpe. A strain with chromosomally located cpe, representing the IS1470-cpe type, was detected in 1 person (0.7%). Furthermore, cpe+ C. perfringens strains representing none of the aforementioned types, referred to as cpe+ strains with an unknown genetic arrangement downstream of cpe, were observed in 2 persons (1.5%) (Table A1) (Figure 1).

In PFGE analysis, all isolates were typeable with SmaI restriction enzyme, and all but 2 isolates were typeable with ApaI. The discriminatory power was equal with both enzymes used, which showed a high genetic diversity among C. perfringens. PFGE analysis showed 1–5 different MRPs in each person, with cpe+ and cpe– isolates generally being unrelated to each other. However, in 1 case, 2 isolates (CPI 57–1 and CPI 57–2) from the same person showed identical MRPs with both restriction enzymes, the former being cpe+ and the latter cpe– (Table A1; Figure 2). In 8 of 11 persons, cpe+ isolates with similar MRPs were identified, whereas 3 persons carried cpe+ isolates with 2 different MRPs. In 2 of these persons (numbers 18 and 57), both MRPs represented the same cpe genotype (IS1470-like-cpe); the third person (number 75) had MRPs with different cpe genotypes (IS1470-like-cpe and IS1151-cpe) (Table A1).

Successful sporulation was achieved with isolates representing 9 different MRPs. In 8 (89%) of these MRPs, >20% inhibition of protein synthesis was detected in Vero cells, which suggested CPE production of the strain (Table 3).