OUP user menu

Production of actin-specific ADP-ribosyltransferase (binary toxin) by strains of Clostridium difficile

Simon Stubbs, Maja Rupnik, Maryse Gibert, Jon Brazier, Brian Duerden, Michel Popoff
DOI: http://dx.doi.org/10.1111/j.1574-6968.2000.tb09122.x 307-312 First published online: 1 May 2000


In addition to the two large clostridial cytotoxins (TcdA and TcdB) certain strains of Clostridium difficile produce an actin-specific ADP-ribosyltransferase, or binary toxin. PCR reactions were developed to detect genes encoding the enzymatic (cdtA) and binding (cdtB) components of the binary toxin and 170 representative strains were tested to assess the prevalence of the toxin. Positive PCR results (n=59) were confirmed by immunoblotting and ADP-ribosyltransferase assay. PCR ribotype and toxinotype (restriction fragment length polymorphism analysis of genes for TcdA and TcdB) correlated with possession of binary toxin genes. All strains with cdtA and cdtB belonged to toxin-variable toxinotypes and five toxin-producing groups of strains have been described according to the presence or absence of TcdA, TcdB and binary toxin. Result indicate that ca. 6.4% of toxigenic isolates of C. difficile referred to the Anaerobe Reference Unit from UK hospitals have cdtA and cdtB genes.

  • Clostridium difficile
  • Binary toxin
  • ADP-ribosyltransferase

1 Introduction

Clostridia produce various toxins that induce alterations in the actin cytoskeleton. These cytotoxins have been classified into at least three groups; two of them, the C3-like toxins and the binary toxins, are ADP-ribosyltransferases, whilst the large clostridial cytotoxins (LCT) are glucosyltransferases [14]. The primary virulence factors produced by strains of Clostridium difficile belong to the LCT group, and are known as toxin A (TcdA) and toxin B (TcdB) [2,5]. TcdA and TcdB are similar in structure and activity, and have been reported to be monoglucosyltransferases that modify the low-molecular-mass GTP-binding proteins of the Rho and Ras subfamily using UDP-glucose as a co-substrate [68].

One strain of C. difficile (CD196) has been shown to produce an actin-specific ADP-ribosyltransferase (binary toxin) which exhibits activity and structure similar to C. perfringensι-toxin [9]. The genes for the binary toxin have been identified and the enzymatic (cdtA) and binding components (cdtB) characterised [10]. It has been suggested that other strains may produce binary toxin, and that it may be an additional virulence factor [9,10].

Various typing schemes have been developed to study the epidemiology of C. difficile infection and determine the similarity of strains associated with disease. A library of distinct ‘types’ of C. difficile, based upon polymerase chain reaction (PCR) ribotyping [11,12], and a toxinotyping scheme, based upon PCR-restriction fragment length polymorphism (RFLP) analysis of the LCT genes [13], have been reported. Certain ‘types’ of C. difficile have been described that have significant changes in LCT genes whilst others lack enterotoxic activity but exhibit cytotoxicity [1215]. One such strain (1470), belonging to serogroup F (toxinotype VIII, ribotype 017), has been shown to possess a cytotoxin that is a functional hybrid between TcdB and the C. sordellii lethal toxin [6,16]. The cytotoxin from another TcdA-negative strain, CCUG 20309 (strain 8864), has been reported to be equally cytotoxic but more lethal than the TcdB from the type strain of C. difficile[17].

Typing schemes have highlighted a number of epidemiologically important factors regarding the possible clonality and spread of C. difficile and there is some evidence to suggest that they can provide a reliable indication of virulence potential [12,13]. The aims of the present study were to detect and assess the prevalence of binary toxin genes in C. difficile. The study also investigated variation in binary toxin genes and examined the relationship between the possession of binary toxin genes, changes in LCT genes and ribotype.

2 Materials and methods

2.1 Strains, detection of TcdA and TcdB and nucleic acid preparation

Strains (n=170) of C. difficile, belonging to 95 ribotypes [12] and 14 toxinotypes [13], have been analysed. Strains prefixed R were obtained from the Anaerobe Reference Unit (ARU) library [12]; other strains were obtained from NCTC, ATCC and CCUG. Enterotoxin and cytotoxin production were determined with the Tox A TEST immunoassay (TechLab, BioConnections, Leeds, UK) and Vero cell cytotoxicity [18]. Crude template nucleic acid was prepared using Chelex-100 (Bio-Rad, Hemel Hempstead, UK) [11].

2.2 PCR ribotyping and toxinotyping

PCR ribotyping and toxinotyping were performed according to methods described previously [11,13]. Dendrograms were produced with the cluster correlation algorithm by the unweighted pair group method using arithmetic averages (UPGMA) and GelCompar 4.0 (Applied Maths, Kortrijk, Belgium).

2.3 PCR for binary toxin genes

The primers Tim6 and Struppi6 amplify the cdd3 gene from all C. difficile strains [19] and were used as a positive PCR control. Primers designed to amplify regions of cdtA and cdtB were as follows: cdtApos 5′-TGAACCTGGAAAAGGTGATG-3′ (position, cdtA 507–526); cdtArev 5′-AGGATTATTTACTGGACCATTTG-3′ (position, cdtA 882–860); cdtBpos 5′-CTTAATGCAAGTAAATACTGAG-3′ (position, cdtB 368–389); cdtBrev 5′-AACGGATCTCTTGCTTCAGTC-3′ (position, cdtB 878–858). Template nucleic acid (5 μl) was added to a PCR mixture (total 50 μl; 50 mM KCl, 10 mM Tris–HCl pH 9.0, 0.1% Triton X-100, 1.5 mM MgCl2, 200 μM each dNTP, 0.15 μM each primer, 1 U Taq polymerase (Promega, Southampton, UK)). Reactions were subjected to 30 cycles of 94°C for 45 s, 52°C for 1 min and 72°C for 1 min 20 s. Nucleic acid preparations and PCR were done in triplicate.

2.4 Partial sequencing of binary toxin genes

PCR products, generated from 11 representative strains of selected ribotypes, were cleaned with QIAquick-spin PCR clean-up columns (Qiagen Ltd., Crawley, West Sussex, UK) and sequenced with the ABI-Prism Dye Terminator Cycle Sequencing kit (Perkin-Elmer, Warrington, UK). Sequences for short cdtA products from strains R10456 and IS51, and short cdtB products from strains R8637, IS81, R10456 and IS51 have been assigned EMBL accession numbers AJ238324, AJ238325 and AJ237817AJ237820, respectively.

2.5 SDS–PAGE, Western blotting and ADP-ribosyltransferase assay

Representative ‘type’ strains (n=13) from selected ribotypes [12] and toxinotypes [13] were cultured in brain heart infusion broth for 48 h and protein was precipitated from culture supernate with ammonium sulfate [20].

Supernatant protein (30 μg) was electrophoresed in SDS–PAGE gels (10% acrylamide) and transferred onto nitrocellulose membranes [21]. Membranes were blocked in phosphate-buffered saline containing dried milk (5%) for 1 h and incubated overnight at room temperature with rabbit immunoglobulin (1 in 5000 dilution) raised against enzymatic (Ia) and binding components (Ib) of C. perfringensι-toxin [20]. Bound antibody was detected with peroxidase-labelled protein A and the Signal Plus kit (Pierce Chemical Co., Rockford, IL, USA).

For ADP-ribosyltransferase assay, the supernatant protein (10 μg) was incubated at 37°C for 1 h in 50 μl of a solution of 50 mM triethanolamine–HCl, pH 7.5, 5 mM MgCl2, 10mM dithiothreitol, 10 mM thymidine containing brain extract as a source of actin (10 μg) and [32P]NAD (106 cpm per reaction). Protein was precipitated with 20 μl of a solution of bovine serum albumin (1 mg ml−1) and SDS (10% w/v) and 0.5 ml of trichloroacetic acid (10% w/v); reactions were incubated on ice for 1 h. The precipitate was immobilised on GFC filters (Millipore, Watford, UK), washed twice in 10 ml of trichloroacetic acid (10% w/v), dried and counted for radioactivity.

3 Results

3.1 Detection of cdtA and cdtB by PCR

PCR reactions with nucleic acid from CD196 (ribotype 027, toxinotype III; Table 1) resulted in products (622, 375 and 510 bp) with sequences corresponding to the cdd3, cdtA and cdtB genes (Fig. 1). PCR results for a further 169 strains belonging to 95 ribotypes and 14 toxinotypes are summarised in Table 1. PCR indicated the presence of cdtA and cdtB in 59 strains belonging to 16 ribotypes and nine toxinotypes.

View this table:
Table 1

PCR ribotype, toxinotype and results of PCR detection of cdtA and cdtB for strains of C. difficile

PCR ribotypeaNumber of strains testedToxinotypeaPCR detection (cdtA/cdtB)PCR ribotypeaNumber of strains testedToxinotypeaPCR detection (cdtA/cdtB)
  • aSee [1517] for complete PCR ribotype and toxinotype data.

  • bNT: non-toxigenic strains (tcdA- and tcdB-negative).

  • cNew toxinotype (to be described elsewhere).

Figure 1

Individual PCR products for (a) cdtA and (b) cdtB generated with representative strains: CD196, R6786, R8637, IS51, CCUG 20309, R5989, IS93 and IS58 (lanes 1–8, respectively). Lanes L refer to 100-bp ladder (300–700 bp).

3.2 Correlation between ribotype, tcdA/tcdB toxinotype and binary toxin PCR

A correlation between binary toxin PCR results, toxinotype and ribotype was observed (Table 1). All strains within a ribotype (and toxinotype) consistently produced results that indicated the presence or absence of cdtA and cdtB. Only strains belonging to variant toxinotypes [13] that have significant changes in LCT genes when compared to strain VPI 10463 (toxinotypes III, IV, V, VI, VII, IX, X, XI and XIV) possessed binary toxin genes. The tcdA-negative, tcdB-positive strains (including strain 1470) belonging to variant toxinotype VIII did not possess binary toxin genes. All strains in toxinotypes 0, I, II, XII and XIII that have only minor changes in tcdA and tcdB when compared to VPI 10463 do not possess cdtA and cdtB. Comparative analysis of ribotype profiles from certain binary toxin PCR-positive strains highlighted a similarity in profiles (Fig. 2).

Figure 2

UPGMA dendrogram depicting similarities in PCR ribotype patterns for strains of C. difficile possessing cdtA and cdtB.

Of the 2246 strains constituting the Anaerobe Reference Unit C. difficile collection [12], 1902 strains produce TcdA and TcdB, and 123 strains (5.5% of total and 6.4% of toxigenic strains) have been assigned to the 16 binary toxin-positive ribotypes. Strains of these ribotypes have been isolated from patients in 18 of the 42 hospitals that have referred isolates to the Anaerobe Reference Unit. Strains have also been isolated from patients in the USA and Europe, and from veterinary and environmental sources.

3.3 Sequence characterisation of short region of binary toxin genes

Similarity values for short regions of cdtA and cdtB and predicted amino acid sequences are shown in Table 2. Sequences of cdtA (327 bp) for 11 strains clustered into three groups containing strains with identical DNA sequence: group 1a, CD196, CCUG 20309, R8637, IS81 and IS93; group 2a, IS51, IS58, R6786 and R7605; group 3a, R10456 and R5989. Translated sequence resulted in two groups, one containing CD196, CCUG 20309, R8637, IS81, IS93, IS58, R6786, IS51, R7605 and another group of R5989 and R10456. The short amino acid sequences had 84–87% similarity with analogous regions of Ia of C. perfringens and the C. spiroforme toxin (Sa). DNA and translated protein sequences for cdtB (451 bp) clustered the 11 strains into five groups (Fig. 3) with identical sequence: group 1b, CD196 and CCUG 20309; group 2b, IS51, IS58, R6786 and R7605; group 3b R10456 and R5989; group 4b, R8637; group 5b, IS81 and IS93. The short amino acid sequences had 64–66% similarity with analogous regions of Ib of C. perfringens and the C. spiroforme toxin (Sb).

View this table:
Table 2

Comparison of partial nucleotide and protein sequences of cdtA and cdtB from 11 strains of C. difficile

StrainSimilarity (%)
CD196 and CCUG20309R8637IS81 and IS93R10456 and R5989IS51, IS58, R6786 and R7605
CD196 and CCUG20309100.0 (100)a100.0 (100)a97.9 (97.3)a98.8 (100)a
R863799.8 (99.3)b100.0 (100)a97.9 (97.3)a99.8 (100)a
IS81 and IS9399.6 (99.3)b99.3 (98.6)b97.9 (97.3)a99.8 (100)a
R10456 and R598998.7 (98.6)b98.4 (98.0)b98.7 (99.3)b97.9 (97.3)a
IS51, IS58, R6786 and R760598.4 (96.6)b98.2 (96.0)b98.4 (97.4)b98.4 (98.0)b
  • aSimilarity values (%) for a 327-bp (109-aa) region of the cdtA gene (amino acid similarity values in parentheses) from 11 strains (within five groupings) of C. difficile.

  • bSimilarity values (%) for a 451-bp (149-aa) region of the cdtB gene.

Figure 3

UPGMA dendrogram depicting similarities in predicted protein sequences for a short region (150 aa) of cdtB from strains of C. difficile.

3.4 Western blotting, immunodetection and ADP-ribosyltransferase assay

Expression of the binary toxin genes for 13 representative strains was assessed immunologically and activity measured by ADP-ribosyltransferase assay (Table 3). Seven PCR-positive strains reacted with anti-Ia and anti-Ib. Strains IS58, R6786, IS51 and R7605 (belonging to sequence groups 2a/2b and ribotypes 033, 045, 066 and 078, respectively) were positive by PCR but failed to react with anti-Ia, anti-Ib or both. These strains grouped together when analysing cdtA and cdtB sequences and ribotype patterns (Figs. 2 and 3).

View this table:
Table 3

PCR ribotype, production of large clostridial cytotoxins TcdA and TcdB, toxinotype, and binary toxin detection by PCR, immunoblotting and ADP-ribosyltransferase assay for 13 strains of C. difficile

PCR ribotypeaType strainaLCT analysisBinary toxin analysis
ToxinotypebEnterotoxin TcdACytotoxin TcdBPCRWestern blot analysisADP-ribosyltransferase activity (U)
cdtAcdtBCDTa (using anti-Ia)CDTb (using anti-Ib)
036CCUG 20309X+++++7588c
  • aPCR ribotype and strain data are detailed in [12].

  • bToxinotyping of toxin genes tcdA and tcdB is detailed in [13].

  • cSignificant ADP-ribosyltransferase activity compared to control.

  • dStrain does not produce detectable enterotoxin or cytotoxin but contains part of the tcdA gene.

All PCR-positive strains, with the exception of IS51, produced significant actin-specific ADP-ribosylating activity. The two strains that were negative in PCR reactions failed to react with anti-Ia and anti-Ib and did not produce ADP-ribosylating activity. Strain IS58 does not produce active TcdA or TcdB, but toxinotyping results indicate that a short region of the tcdA gene is present. IS58 did not react with anti-Ib. However, this strain gave cdtA/cdtB PCR products, reacted with anti-Ia and had ADP-ribosyltransferase activity indicating the presence of binary toxin.

4 Discussion

It has been generally accepted that C. difficile produces two major toxins (TcdA and TcdB) that belong to the LCT family. The production of another toxin, an actin-specific ADP-ribosyltransferase (binary toxin), by a single strain of C. difficile (CD196) was first reported in 1988 [9]. In the present study 170 representative strains, which have been characterised previously by ribotyping [11,12] and PCR-RFLP analysis of tcdA and tcdB (toxinotyping) [13], were analysed and 59 strains yielded PCR products with primers for binary toxin components. A correlation was observed between ribotype, toxinotype and possession of cdtA and cdtB, and this relationship permitted the reliable prediction of which strains possess binary toxin.

Expression of CDTa and CDTb by strains representing 13 ribotypes was assessed with antibodies raised to the ι-toxin of C. perfringens (anti-Ia and anti-Ib), and also by ADP-ribosyltransferase assay. The culture supernate from most PCR-positive strains reacted with anti-Ia and anti-Ib, but four strains (IS58, R6786, IS51 and R7605) exhibited ambiguous results. Analysis of short DNA and predicted amino acid sequences of binary toxin components revealed that these four strains group peripheral to CD196 and other binary toxin-producing strains (Fig. 3, Table 2). These differences in sequence could explain the lack of reaction with anti-Ia or anti-Ib [9] and actin-specific ADP-ribosyltransferase assays support this theory since only one PCR-positive strain (IS51) failed to yield activity.

Binary toxin genes were observed only in strains that possess some part of the pathogenicity locus containing the genes for TcdA and TcdB (Tables 1 and 2). The binary toxin was not detected in C. difficile VPI 10463 [9] or in the large proportion of strains with similar LCT toxinotypes [13]. Interestingly, binary toxin genes were only detected in strains that have significant changes in the genes for TcdA and TcdB when compared to VPI 10463 – the so called variant toxinotypes [13]. Toxinotype VIII is the only group of strains with significantly changed LCT genes that does not possess the genes for binary toxin. However, strains of toxinotype VIII are TcdA-negative and have been shown to possess a cytotoxin that is a functional hybrid between TcdB and the C. sordellii lethal toxin [6,16].

The results of the present study show that C. difficile can be divided into at least five different toxin-producing groups: (1) LCT producers, (2) LCT and binary toxin producers, (3) TcdB-only producers, (4) TcdB and binary toxin producers and (5) binary toxin-only producers. These groupings add further support to the theory that C. difficile is divided into stable subpopulations that have evolved from common ancestors [13].

The role of binary toxins in the pathogenesis of intestinal infections is unclear. Strains of C. perfringens that produce binary toxins also produce additional toxins [4]. However, C. spiroforme would seem to produce only binary toxin and is involved in gastrointestinal disease in animals and humans, whilst the C. botulinum C2 toxin has been found to induce haemorrhagic enteritis in animals [3]. In C. difficile infections, the binary toxin may act in synergy with LCT to depolymerise the actin cytoskeleton by a complementary mechanism. C. difficile CD196 was investigated because of the severity of clinical symptoms [9] and it is conceivable that the production of this additional toxin may exacerbate symptoms.

Results indicate that 6.4% of toxigenic isolates referred to the Anaerobe Reference Unit from UK hospitals contain cdtA and cdtB. Strains of the predominant ribotypes acquired nosocomially in the UK [12] do not possess binary toxin, indicating that it does not play a major role in the aetiology of antibiotic-associated diarrhoea. However, binary toxin-producing strains have been isolated from patients in 18 of 42 UK hospitals that refer strains for epidemiological analysis. Binary toxin-positive isolates have also been obtained from patients in the USA and throughout Europe indicating that they are widespread.


The authors would like to thank Tamara Majstorovic (University of Ljubljana) for contributing certain DNA samples and PCR results.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
View Abstract