Detection of binary-toxin genes (cdtA and cdtB) among Clostridium difficile strains isolated from patients with C. difficile-associated diarrhoea (CDAD) in Poland

Clostridium difficile is the aetiologic agent of antibiotic-associated diarrhoea (AAD), antibiotic-associated colitis (AAC) and pseudomembranous colitis (PMC) (Borriello, 1998; Johnson & Gerding, 1998).

Toxigenic strains can produce three toxins: TcdA, TcdB and binary toxin CDT. While TcdA and TcdB are recognized as the main virulence factors, the role of binary toxin in the pathogenesis is unclear (Rupnik et al., 2003b). TcdA and TcdB are related to each other and belong to a group of large clostridial toxins. C. difficile binary toxin CDT differs from TcdA and TcdB in structure, enzymic activity and in intracellular targets, and represents an iota-like group of clostridial binary toxins (Popoff et al., 1988; Perelle et al., 1997; Thelestam & Chaves-Olarte, 2000).

Most of the isolated C. difficile strains produce only TcdA and TcdB (A⁺B⁺CDT isolates), but the AB⁺ type is increasingly reported (Kato et al., 1998; Alfa et al., 2000; Sambol et al., 2000; Pituch et al., 2001, 2003; Kuijper et al., 2001; Johnson et al., 2001; Barbut et al., 2002; Rupnik et al., 2003a). Among AB⁺ strains, the prevalent type (toxinotype VIII, serogroup F) does not have the genes for the binary toxin, while some representatives of other AB⁺ types have both of the genes for the binary toxin (Rupnik et al., 2003b). However, the majority of strains harbouring binary-toxin genes are also A⁺B⁺ (Stubbs et al., 2000; Perelle et al., 1997; Spigaglia & Mastrantonio, 2002; Rupnik et al., 2003b). Interestingly, about 2 % of AB strains were estimated to produce binary toxin CDT (ABCDT⁺ strains) (Geric et al., 2003).

Toxins TcdA and TcdB are encoded by two genes, tcdA and tcdB, located within a 19.6 kbp pathogenicity locus (PaLoc). Based on the changes in both toxin genes, C. difficile strains can be divided into 24 groups called toxinotypes (IXXIV) (Rupnik et al., 1998, 2001, 2003a; Geric et al., 2004). Certain groups of these variant strains have been reported to possess binary toxin CDT encoded by the cdtA and cdtB genes located on the chromosome outside the PaLoc (Stubbs et al., 2000; Rupnik, 2001). Only a single isolate with tcdA and tcdB identical to the reference strain VPI 10463 and with binary-toxin genes is known (Spigaglia & Mastrantonio, 2002).

In the present study we analysed A⁺B⁺ and AB⁺ C. difficile strains isolated from Polish patients with CDAD for the presence of the cdtA and cdtB genes and examined the toxinotypes of strains with binary-toxin genes.

Bacterial isolates.
From among the C. difficile strains isolated in our diagnostic laboratory in the years 19992001 we randomly selected 58 C. difficile isolates: 41 isolates which were previously defined as A⁺B⁺ and 17 isolates which were AB⁺. Forty-six isolates were from adults hospitalized in University Hospital (Table 1). One isolate was from an adult out-patient treated at the same hospital. A further 11 isolates were from children, of whom all were hospitalized in a paediatric hospital except one out-patient. All patients, adults as well as children, suffered from AAD.

Table 1. Overview of the presence of toxin genes and toxin production in C. difficile strains isolated from patients with AAD

C. difficile reference strain VPI 10463 and a non-toxigenic strain NIH BRIGGS 8050 were used as controls in cytotoxin assay and in the PCR for the detection of non-repeating sequences of the tcdA and tcdB genes. An AB⁺ strain (GAI 95601) obtained from Dr Haru Kato (Institute of Anaerobic Bacteriology, Gifu University School of Medicine, Japan) was used as a control in the detection of repeating sequences within the tcdA gene. Strain CCUG 20309 (also 8864) was used as a control in detection of binary-toxin genes.

Culture and identification of C. difficile isolates.
Isolation of C. difficile was performed on selective Columbia agar supplemented with cycloserine-cefoxitin and amphotericin B (CCCA medium; BioMérieux) as described previously (Meisel-Mikoajczyk et al., 1999).

Toxin production.
All 58 isolates included in this study were retested for in vitro production of TcdA and TcdB. A single colony was transferred into Brain heart Infusion Broth (BHI) (Difco) and grown for 96 h. Supernatants were collected by centrifugation at 3000 g for 15 min. Toxins were determined by C. difficile toxin A test (Oxoid) and TOX A/B test (TechLab). Reactions were interpreted by visual reading: colourless results were considered negative and any yellow colour was considered positive. In vitro TcdB detection was performed with McCoy cells cultured as described previously (Meisel-Mikoajczyk et al., 1999). Tenfold serial dilutions of culture filtrate were added in duplicate to cells and incubated for 24 h. The cytopathic effect was observed with an inverted microscope. The test was considered positive if the cytopathic effect could be neutralized by the polyclonal antiserum C. difficile TOX-B Test (TechLab).

Detection of toxin genes tcdA and tcdB by PCR.
Crude template DNA was prepared using Genomic DNA PREP-PLUS (A & A Biotechnology) according to the manufacturer's instructions (Meisel-Mikoojczyk et al., 1999). For detection of the non-repeating regions of the tcdA and tcdB genes we used the primer pairs YT28/YT29 and YT17/YT18, respectively, described by Gumerlock et al. (1993) and Kuhl et al. (1993). Changes in the repeating regions of the tcdA genes were detected with the NK9/NKV011 primer pair (Kato et al., 1998).

Detection of binary-toxin genes.
Primers described by Stubbs et al. (2000) were used for amplification of binary-toxin genes cdtA and cdtB. Template nucleic acid (2 µl) was added to 22.5 µl Supermix (Gibco BRL) and 1 µl of each primer solution.

Toxinotyping.
For toxinotyping, fragments B1, covering the 5'-end of tcdB gene, and A3, covering the 3'-end of tcdA, were amplified as described previously (Rupnik et al., 1997) and subsequently cut with restriction enzymes AccI and HincII (for B1) or EcoRI (for A3). According to the combination of obtained restriction patterns in B1 and A3 PCR fragments the toxinotype was determined (Rupnik et al., 1998).

We investigated C. difficile isolates randomly selected from the large collection of University Hospital Warsaw to detect the presence of binary-toxin genes in A⁺B⁺ and AB⁺ strains. Retesting of toxin production confirmed that all 58 studied isolates were cytotoxic for McCoy cells and positive in the toxin A/B test. Seventeen isolates previously classified as AB⁺ were, as expected, negative for the toxin A test. Additionally, in all isolates non-repetitive regions of both toxin genes, tcdA and tcdB, could be amplified (Table 1).

AB⁺ strains are a well-characterized subpopulation of C. difficile. The majority belong to a single group characterized by a 1.8 kbp deletion in the repetitive region of tcdA (Rupnik et al., 1998; Kato et al., 1998; Sambol et al., 2000) and a nonsense mutation at amino acid position 46, which is responsible for the A phenotype (von Eichel-Streiber et al., 1999). Within this prevalent group of AB⁺ strains, also called toxinotype VIII, two serogroups (F and X), two PCR-ribotypes (017 and 047) and several REA types could be distinguished (Johnson et al., 2003). Additionally, several other types of AB⁺ strains were described but are each represented only by a single isolate: toxinotype X (strain 8864; Johnson et al., 2003; Rupnik et al., 1998), toxinotypes XVI and XVII (strains SUC 36 and J9965, respectively; Rupnik et al., 2003a), toxinotype V-like (Geric et al., 2003) and an unknown toxinotype (strain R7771; Johnson et al., 2003). The groups of AB⁺ strains differ from each other in the presence or absence of the binary-toxin genes and in changes within the PaLoc region. Though the differences are distributed along both toxin genes, the simplest way to differentiate between the most prevalent toxinotype VIII and other rare types of AB⁺ strains is amplification of repetitive regions of gene tcdA.

To investigate the possible heterogeneity of AB⁺ isolates included in this study, repetitive regions of tcdA were amplified with primers described by Kato et al. (1998). But as shown in Table 1, all AB⁺ isolates had an identical deletion resulting in a short 700 bp fragment of repetitive regions, identical to the toxinotype VIII (F-like) control strain GAI 95601. Therefore, all AB⁺ isolates were assigned to toxinotype VIII.

The presence of binary-toxin genes was described as a good marker to detect the C. difficile strains with variant PaLoc (toxinotypes) (Rupnik, 2001). Some strains seem to have genetic information for only one of the two components required for fully active binary toxin CDT (cdtA and cdtB) (Perelle et al., 1997). Therefore, we amplified both genes in all 58 investigated isolates. None of the AB⁺ isolates had binary-toxin genes, which is in agreement with their identification as toxinotype VIII (Rupnik, 2001; Stubbs et al., 2000). Five A⁺B⁺ isolates (8.6 %) harboured binary-toxin genes, a prevalence similar to that reported by Goncalves et al. (2004). Binary-toxin-positive strains were isolated from adult patients hospitalized at different times (between 1999 and 2001) and in different units, and were considered epidemiologically unrelated. Surprisingly, all were typed according to the changes in PaLoc as toxinotype IV. Toxinotype IV could be further differentiated into four PCR-ribotypes (Rupnik et al., 2001). When checked with the PCR-ribotyping method described by Stubbs et al. (1999) all five binary-toxin-positive strains were of the same ribotype (data not shown).

This is the first report on the occurrence of the binary-toxin genes among C. difficile isolates from patients with CDAD in Poland. The majority of studies on the prevalence of A⁺B⁺CDT⁺ strains have investigated isolates collected from different hospitals (Spigaglia & Mastrantonio, 2002; Rupnik et al., 2003a; Goncalves et al., 2004), and the only report on strains from a single hospital isolated over a 5 year period showed marked heterogeneity of such strains distributed into six toxinotypes (Geric et al., 2004). Strains included in the present study were not subsequent isolates and the random selection from the collection could have introduced the bias. However, the results do suggest that toxinotype IV might dominate amongst the population of C. difficile binary-toxin-positive isolates in our institution.