Clostridium difficile is known to be the major cause of antibiotic-associated diarrhea and of pseudomembranous colitis. An important requirement for the diagnosis of C. difficile-associated diarrhea is toxin detection. At least two different toxins can be produced: toxin A, primarily responsible for the enterotoxic activity which results in an inflammatory process and alteration of permeability, and toxin B, which has a direct cytopathic effect by depolymerization of filamentous actin, which causes destruction of the cell cytoskeleton [1]. The overwhelming majority of toxigenic strains produce both toxins, while neither of them is produced by non-toxigenic strains. Identification of toxigenic C. difficile has traditionally been performed by determining the cytopathic effect on cell cultures. This reliable assay indirectly detects C. difficile toxin B-producing strains. Nevertheless, the cytotoxic assay is time-consuming and it can produce up to 30% false-negative results in ‘low producer’ strains [1]. Several serologic tests have been used but none of them have been successful enough to replace the cytotoxic assay. Molecular biology techniques have also been developed for this purpose, including hybridization and PCR. Although some of these techniques have focused on the toxin B gene [2], most of them have targeted toxin A sequences [3]. The aim of this study was the development of a rapid test for the detection of the toxin B gene, which is always present in toxigenic strains, based upon a nested PCR. All the results have been compared with the cytotoxicity assay, which is the present standard method.

Fifty-nine clinical isolates of C. difficile were included. C. difficile ATCC 9689 was also processed as a reference strain. C. perfringens ATCC 13124, C. ramosum clinical isolate no. 94/1391 and C. sordellii clinical isolate no. 94/4291 were processed to evaluate interspecies specificity. Twenty clinical isolates of other enteropathogenic bacteria, including Shigella dysenteriae, Salmonella enterica, Aeromonas hydrophila and Campylobacter jejuni, were also included in order to study possible cross-reactions when assaying stool specimens directly. The toxigenicity of C. difficile strains was evaluated by cell culture assay following standard procedures [4]. In cases in which direct toxin detection from stool samples was negative but a C. difficile strain was isolated, monolayers were also inoculated with supernatant filtrates of C. difficile pure broth cultures. Specificity of cytopathic effect was confirmed with a neutralizing high-titered antiserum (TechLab, VPI Research Park, Blacksburg, VA, USA).

C. difficile total DNA was extracted using a modification of previously described methods [5]. The extraction protocol was carried out in anaerobic conditions until lysis was completed. A nested PCR was designed in order to achieve optimal sensitivity when direct samples were assayed. Primers were selected from areas of the toxin B gene [6] where the mole G+C percentage was slightly increased (G+C% of C. difficile chromosome is 28%) in order to use relatively high annealing temperatures which allow specific amplification. Primers CDTB1, CDTB2 and CDTB3 (Table 1) were designed and checked for the absence of dimerization properties or internal loops by a specific software application. PCRs were carried out in 50 mL, containing 25 ng of total DNA (or 7 mL of the first amplification reaction), 2 mM MgCl2, 5 pmol of each primer (CDTB1+CDTB3 for first reaction and CDTB1+CDTB2 for the second one), 200 mM each of dCTP, dGTP, dATP and dTTP (Boehringer Mannheim, Germany) and 0.5 U of Taq DNA polymerase (Boehringer Mannheim, Germany), in Taq DNA polymerase incubation buffer. Amplifications were carried out in a thermal cycler (Perkin Elmer Cetus) for 30 cycles of 94°C/45 s, 50°C/45 s and 72°C/75 s for the first amplification reaction, and 30 cycles of 94°C/45 s, 55°C/45 s and 72°C/30 s for the second reaction. Since nested PCR assays are known to be more apt to generate false-positive results than single PCR protocols, multiple negative controls, aerosol-resistant tips and separate processing areas were used to minimize the risk of amplicon contamination. After PCRs were completed, a volume of 10 to 15 mL was subjected to electrophoresis in 1.4% agarose gels followed by ethidium bromide staining. Sensitivity, specificity and predictive values were calculated for the designed PCR assay with reference to the cytotoxicity assay in accordance with previous descriptions [4].

Table 1.  Sequences, orientations and locations of PCR primers
PrimerOrientationSequence (5′–3′)Position

The cytotoxic assay was positive with the 56 C. difficile isolates and the reference strain C. difficile ATCC 9689. One of these, isolate no. 110, produced atypical cytopathic effects which remained after the addition of the specific antiserum. Four isolates were clearly non-toxigenic.

Primers CDTB1–CDTB2 and CDTB1–CDTB3 were used in a pilot assay study for the detection of the target sequence in single amplification reactions, but the results obtained were considerably less satisfactory than those produced by the nested reaction (data not shown).

Nested PCR assay showed the expected 322-bp amplification band corresponding to the Nested reaction (primers CDTB1–CDTB2). The band of 1125 bp corresponding to the first amplification (primers CDTB1–CDTB3) was occasionally observed (Figure 1). PCR was positive for C. difficile ATCC 9689 and 54 clinical isolates of C. difficile, and negative for the remaining five isolates. Strain no. 110 (the one that produced atypical cytopathic effects) was PCR negative. All other assayed microorganisms, including other species of the genus, even C. sordellii, and the 20 assayed enteropathogens, were both PCR negative and non-cytotoxic on cell cultures. Evaluation of the PCR assay in comparison with the present standard method, the cytotoxic assay, was as follows: sensitivity 100%; specificity 100%; positive predictive value 100%; negative predictive value 100%; false-negative rate 0%; and false-positive rate 0%. Concordance between both techniques was absolute.


Figure 1. Amplification patterns of the assayed isolates. Lanes 1 and 10, molecular weight markers; lane 2, amplification band corresponding to C. difficile ATCC 9689; lane 3, amplification of a non-toxigenic C. difficile clinical isolate (no. 9415) DNA; lanes 4, 5 and 6, amplification of toxigenic C. difficile clinical isolates (numbers 9430, 9456 and 9490) DNAs (the first amplification product can also be seen in lane 6); lanes 7, 8 and 9, amplification of C. sordellii, C. ramosum and C. perfringens DNAs respectively.

Download figure to PowerPoint

As previously published, some low producer toxigenic strains can go unnoticed if assayed on cell cultures. Furthermore, some strains, even when having a complete genetic composition for both toxins, fail to produce toxin B, which is the one detected on the cytotoxic assay [1,7]. On the other hand, all non-toxigenic strains lack both toxin A and toxin B genes and most of the toxigenic strains have both genes. However, recently, some isolates have been encountered which have the toxin B gene but lack the toxin A gene [8], or in which the toxin A gene is incomplete [9]. All of the above reasons led us to focus on the toxin B gene in order to detect all possible genetic situations in toxigenic strains. In contrast to some other published genetic methods based on hybridizations or single amplifications, we have designed a nested PCR which has allowed us to achieve 100% sensitivity. Specificity has been optimal, since no amplifications were detected, either on other species of Clostridium or in the enteropathogens included in the study. In other published techniques, C. sordellii, whose toxin is highly homologous with the one produced by C. difficile [1], has offered false-positive results by cross-reactions. In our case a positive amplification band has never been obtained within this microorganism.

All strains included in this study were classified identically by both techniques, with the exception of strain no. 110, which was PCR negative and a toxin producer according to the cytotoxicity assay. Nevertheless, its cytopathic effect on human fibroblasts is atypical and it was not inhibited by toxin B antitoxin, so it could have been the result of a cause unrelated to C. difficile toxin B production. This finding still remains controversial; nevertheless, strain 110 was considered as non-cytotoxic and PCR negative when calculating sensitivities, predictive values, and false result rates. The 1125-kb band corresponding to the first amplification reaction was occasionally seen. Probably, the amounts of this PCR product are at the limit of detection by agarose gel electrophoresis and so it is observed occasionally in cases where the reaction yield has been optimal. Usually in these cases the amount of the main amplification product (322-bp band) is slightly reduced, probably due to inhibition of the nested amplification reaction by an excess of DNA template in it [10]. All our studied isolates of C. difficile, with the exception of C. difficile ATCC 9689, were isolated from significant clinical specimens. This may explain the high rate of toxigenicity encountered (over 93% of all C. difficile clinical isolates included).

Our approach for the detection of toxigenic strains of C. difficile is easy both to perform and to interpret and offers excellent sensitivity and specificity. Some attempts have also been made on a reduced number of fecal specimens which showed adequate yields. Further studies are now ongoing.


  1. Top of page
  2. References
  • 1
    Knoop FC, Owens M, Crocker IC. Clostridium difficile: clinical disease and diagnosis. Clin Microbiol Rev 1993; 6: 25165.
  • 2
    Gumerlock PH, Tang YJ, Weiss JB, Silva J Jr. Specific detection of toxigenic strains of Clostridium difficile in stool specimens. J Clin Microbiol 1993; 31: 50711.
  • 3
    Wren B, Clayton C, Tabaqchali S. Rapid identification of toxigenic, Clostridium difficile by polymerase chain reaction. Lancet 1990; 335: 423.
  • 4
    Murray P, Baron JOE, Pfaller MA, Tenover FC, Yolken RH. Manual of clinical microbiology, 6th edn. Washington DC : American Society of Microbiology (ASM), 1995.
  • 5
    Alonso R, Nicholson PS, Pitt TL. Rapid extraction of high purity chromosomal DNA from Serratia marcescens. Let Appl Microbiol 1993; 16: 779.
  • 6
    Barroso LA, Wang SZ, Phelps CJ, Johnson JL, Wilkins TD. Nucleotide sequence of Clostridium difficile toxin B gene. Nucleic Acids Res 1990; 18: 4004.
  • 7
    MacMillin DE, Muldrow LL, Leggette SJ, Abdulahi Y, Ekanemesang UM. Molecular screening of Clostridium difficile toxins A and B. Genetic determinants and identifications of mutant strains. FEMS Microbiol Lett 1991; 62: 7580.
  • 8
    Borriello SP, Wren BW, Hyde S, et al. Molecular, immunological and biological characterization of a Toxin A-negative, toxin B-positive strain of Clostridium difficile. Infect Immun 1991; 60: 41929.
  • 9
    Depitre C, Delmee M, Avesani V et al. Serogroup F strains of Clostridium difficile produce toxin B but not toxin A. J Med Microbiol 1993; 38: 43441.
  • 10
    InnisMA, GelfandDH, SninskyJJ, WhiteTJ, eds. PCR protocols. A guide to methods and applications. San Diego , CA : Academic Press, 1990.