To analyse the transcription pattern of the five tcdA-E genes of the pathogenicity locus (PaLoc) of Clostridium difficile a protocol was established to purify RNA from strain VPI10463. Transcription analysis of the five tcdA-E genes showed that they were all transcribed. In the early exponential phase, a high level of tcdC and low levels of tcdA,B,D,E transcripts were detectable; this was inverted in the stationary phase, suggesting that TcdC might have a negative influence on transcription of the other genes. Three transcription initiation sites, one for tcdA and two for tcdB were determined by primer extension analysis. Readthrough transcripts from outside the locus were not obtainable, so that parts of the transcription of tcdD, tcdB, tcdA and tcdC must occur by monocistronic transcription. Within the locus all possible intergenic readthrough transcripts were detectable except that between tcdC and tcdA, a stretch of DNA interrupted by a functional transcription terminator. Thus we found mono-and polycistronic transcription of tcdA and tcdB to occur which should lead to production of a surplus of tcdA over tcdB transcripts. This would explain the surplus of TcdA over TcdB expression observed in vitro. Due to its basic nature and similarity to BcnA of Clostridium perjringens and to Orf-22 of Clostridium botulinum, TcdD is most probably a regulatory protein with DNA-binding properties. On the basis of the presented study we discuss a model for the growth-phase-related, coordinate regulation of toxin expression wherein tcdC has a negative and tcdD a positive regulatory function on transcription of the tcdD,B,E and tcdA genes.

The pathogenicity locus (PaLoc) of Clostridium difficile isolate 8864 was investigated to locate genetic rearrangements that would explain the exceptional pathogenicity of this particular isolate. Two major changes were defined: an insertion of 1.1 kb between the two genes tcdA and tcdE, coding for the enterotoxin and an accessory protein of unknown function, respectively, and a deletion of 5.9 kb encompassing the 3' ends of tcdA and tcdC. Transcription of the tcdA-E genes is severely affected by both rearrangements, explaining the demonstrated complete lack of TcdA polypeptide. We present a model of coordinate, growth-related transcription of the tcdA-E genes that confirms our previous findings in strain 10463. Recombinant TcdA-8864 had UDP-glucose-glucosyltransferase activity, proving that the N-terminal 698 amino acids of the polypeptide represent the catalytic domain. However, this truncated TcdA molecule lacks a ligand and translocation domain. To assess the catalytic domain of TcdB-8864, the sequence of the 5' end of its gene was determined. TcdB-8864 shows high homology to TcdB-1470 but lower homology to TcdB-10463 within this domain. This fits well with the altered glucosylation specificity of TcdB-8864 (Rac1, Rap2 and Ra1). Having defined the variations of transcription, expression and enzymatic activity of toxins A and B, implications for the pathogenic potential of strain 8864 are discussed.