Downloaded from by guest on 13 January 2019

Transcription

1 FEMS Microbiology Letters 148 (1997) 197^202 Characterization of polymorphisms in the toxin A and B genes of Clostridium di cile Maja Rupnik a;b, Veit Braun a, Florian Soehn a, Miha Janc b, Monika Hofstetter a, Rita Laufenberg-Feldmann a, Christoph von Eichel-Streiber a; * a Verfuëgungsgebaëude fuër Forschung und Entwicklung, Institut fuër Medizinische Mikrobiologie und Hygiene, Obere Zahlbacherstr. 63, Abstract Johannes Gutenberg-Universitaët Mainz, Mainz, Germany b Department of Biology, Vecna pot 111, University of Ljubljana, Ljubljana, Slovenia Received 4 December 1996; revised 17 January 1997; accepted 21 January 1997 We have used six independent polymerase chain reactions (A1^A3 and B1^B3) for amplification of the entire sequence of the two toxin genes tcda and tcdb of several Clostridium difficile strains. With this approach we have detected (1) restriction site polymorphisms which are distributed all over the genes, and (2) deletions that could be found only in tcda. Characteristic differences between strains were mainly focused to the 5P third of tcdb (B1 fragment) and/or the 3P third of tcda (A3 fragment). The possible use of our approach for typing of C. difficile toxin genes is discussed. Keywords: Polymerase chain reaction; Clostridium di cile; Genetic polymorphism; Toxin A; Toxin B 1. Introduction Clostridium di cile, prevalently isolated in hospitalized patients, is the causative agent of antibioticassociated diarrhea (AAD) and pseudomembranous colitis (PMC) [1]. Its major virulence factors are the two high molecular weight molecules, toxin A (TcdA, 308 kda) and toxin B (TcdB, 270 kda) designated enterotoxin and cytotoxin, respectively [1,2]. They are biochemically, immunologically and structurally related to toxins produced by C. sordellii and C. novyi and all of these toxins are grouped together as large clostridial cytotoxins. For the whole group a * Corresponding author. Tel.: +49 (6131) ; fax: +49 (6131) three domain model, with the catalytic domain situated at the amino-terminal end, a central translocation and a carboxy-terminal repetitive ligand domain was established [3]. C. di cile toxins are encoded by two separate genes, tcda and tcdb. Together with three additional genes (tcdc, tcdd and tcde) they form the pathogenicity locus (PaLoc) which is found only in toxinogenic isolates ([4], Fig. 1). The enormous size of the tcda and tcdb genes and the presence of repetitive 3P domains, prone to homologous recombination, raised the question whether the toxin genes of di erent isolates might display genetical polymorphisms. Some naturally occurring C. di cile variant strains were described as TcdA-negative, TcdB-positive (8864 [5,6]; F-serotype strains [7]) due to the / 97 / $17.00 ß 1997 Published by Elsevier Science B.V. All rights reserved PII S (97)

2 198 M. Rupniket al. / FEMS Microbiology Letters 148 (1997) 197^202 lack of TcdA detection by ELISA. However, with a partial genetic analysis some regions of the tcda gene were detectable. In strain 8864a deletion of the repetitive ligand domain of tcda was discovered, leaving only the 5P end of tcda intact [5,6]. C. di cile serotype F strains carry at least parts of the tcda 5P and 3P regions [7]. The tcdb gene of strain 1470 was sequenced and variations discovered in the catalytic domain were found to give rise to a variant C. sordellii-like cytopathic e ect [8]. The e ect caused by the TcdB isolated from 8864was identically changed [9]. Here we present a polymerase chain reaction (PCR) method for rapid screening of polymorphisms in toxin genes of C. di cile isolates and show that this approach may be useful in characterization of known and for identifying novel variant strains. 2. Materials and methods 2.1. Clostridium di cile strains The 17 strains from a pool of 200 isolates were selected from a previous study due to their di erences in Southern hybridisation [10]. Strain 8864was obtained from the Swedish type culture collection (strain No ). Strain 1470 and strain 2022 (standard strains of the serogroups F and G, respectively) were isolated in the laboratory of M. Delmeèe (Brussels, Belgium). Strain 2022 was added to this study because of its toxinogenic nature and its lack of in vivo pathogenicity [11]. The two nontoxinogenic strains (C1075 and CD39) served as negative controls in all experiments throughout. Standard cytotoxicity testing of toxigenic culture supernatants was performed with CHO cells as described earlier [8] PCR ampli cations PCRs were performed in a total volume of 50 Wl using 300 ng of chromosomal DNA, 200 WM of each dntp, 15 pmol of each of the paired primers and 1U Taq-polymerase (Fermentas, St. Leon-Rot, Germany). The bu er contained 10 mm Tris-HCl, 50 mm KCl, 0.08% Nonidet P40, 3 mm MgCl 2, and 0.1 mg/ml BSA. For ampli cation of A3 fragments TMA (tetramethylammonium chloride, Sigma) was added at a nal concentration of M. The reaction mixtures were overlaid with 100 Wl of mineral oil. Ampli cation was done for 30 cycles (35 for the A3 fragment) using the OmniGene Temperature Cycler from Hybaid (MWG; Ebersberg, Germany). Two-step PCR programs were as follows: initial denaturation for 93³C/3 min; annealing and extension for 8 min either at 56³C (fragments A1 and A2) or at 47³C (fragments A3, B1, B2, B3), followed by a 4 s denaturation at 93³C. After the nal cycle the probes were incubated at 56³ or 47³C for additional 10 min to end the ampli cation process. Primer sequences (MWG, Ebersberg, Germany, or Pharmacia, Freiburg, Germany) and PCR fragments are summarized in Table 1. Ampli ed fragments were visualized on 1% agarose to check for size di erences. Additionally, the fragments were subjected to restriction enzyme digestions (see Table 1) to identify restriction fragment length polymorphism (RFLP). In all experiments C. di cile VPI10463 served as the reference strain. 3. Results 3.1. Ampli cation of fragments covering the tcda and tcdb genes The toxin genes tcda (8130 bp) and tcdb (7098 bp) were divided into three parts according to our three domain model [3]. The 5P fragments A1 and B1 encode for the catalytic domain of the respective toxins, A2 and B2 their translocation and A3 and B3 the repetitive ligand domains (Fig. 1). A set of 12 primers corresponding to the domain borders were chosen and are listed in Table 1. Speci c products were obtained in all toxinogenic strains, but none were present in nontoxinogenic isolates. Ampli cation with the two-step program led to the best results. A further modi cation of this protocol was only needed to obtain the A3 product. Here the addition of TMA to the reaction mixture and an increased number of cycles were necessary Ampli ed fragment length polymorphisms Fig. 2A summarizes PCR ampli cations of the ve representatitve strains 10463, 2022, 100, 1470, and

3 M. Rupnik et al. / FEMSMicrobiology Letters 148 (1997) 197^ Fig. 1. A: Partial map of the pathogenicity locus of C. di cile: tcda^e the ve open reading frames (ORF; open arrows) of the pathogenicity locus are indicated; the other eight ORFs have been presented elsewhere [4]. Hatched areas represent the repetitive domains of tcda and tcdb, respectively. Restriction sites depicted are: HindIII, H; HincII, Hc; PstI, P. B: Relative position of the PCR products. Six fragments A1^A3 and B1^B3 corresponding to the putative catalytic (A/1B1), the translocation (A/B2) and the ligand (A/B3) domains [3] of the two toxins are shown; primers (sequences in Table 1) for their ampli cation are designated and indicated by black arrow heads In the case of the tcdb gene the product lengths within each group (B1^B3) were identical in all strains studied. In most cases the sizes of A1, A2, and A3 fragments were also identical to those of the VPI10463 control reactions. Exceptions were mainly found within the A3 fragments. Five F-serotype strains, checked in this study, carried a 1.7 kb deletion leaving a shortened 1.3 kb repetitive ligand domain (Fig. 2A). Identical deletions were found in two strains from serogroup X (23016 and 12934; data Table 1 Primers used for ampli cation of tcd domains PCR product Primer Sequence Fragment length (kb) not shown). In strain 8864 an extended deletion in a similar region was found. This strain totally lacks the A2and A3 fragments and ampli cation of the A1 product was obtained only when the primer A2NK was used, located closer to the 5P end of tcda gene (Figs. 1 and 2A) Restiction fragment length polymorphisms Minor di erences in the nucleotide sequences give Ampli cation temp. (³C) Restriction enzymes a B1 B1C 5P-AGAAAATTTTATGAGTTTAGTTAATAGAAA-3P HincII, HindIII B2N 5P-CAGATAATGTAGGAAGTAAGTCTATAG-3P B2B2C 5P-ATAGACTTACTTCCTACATTATCTGAA-3P HindIII B3N 5P-CATCTGTATAAATATTTGGTGAAATTAC-3P B3 B3C 5P-AATTTCACCAAATATTTATACAGATG-3P HindIII, MunI B4N 5P-ATTTAACATATTTTTATCTATTCA-3P A1 A1C 5P-GGAGGTTTTTATGTCTTTAATATCTAAAGA-3P HindIII, PstI A2N 5P-CCCTCTGTTATTGTAGGTAGTACATTTA-3P A2NK 5P-TATTTCTTTTGTAATATGTTCGCT-3P A2A2C 5P-TAAATGTACTACCTACAATAACAGAGGG-3P (MnlI, XbaI) A3N 5P-CTTGTATATAAATCAGGTGCTATCAATA-3P A3 A3C 5P-TATTGATAGCACCTGATTTATATACAAG-3P 35x, TMA (SpeI), EcoRI A4N 5P-TTATCAAACATATATTTTAGCCATATATC-3P a The indicated enzymes were typically used to characterize the assigned toxin fragments; parentheses mark enzymes that did not give rise to RFLPs.

4 200 M. Rupnik et al. / FEMS Microbiology Letters 148 (1997) 197^202 Fig. 2. A: Generation of the six domain PCRs. The six PCR products A1^A3 and B1^B3 were generated and separated on agarose gels. Lanes: lambda, lambda DNA digested with Eco1301 (sizes fromtop to bottom: ; 7.743; 6.223; 4.254; 3.472; 2.690; 1.882; 1.489; 925; 421 bp); lanes 1^5: the ve representative C. di cile strains: 1, VPI10463; 2, 2022; 3, 100; 4, 1470; 5, Note that fragments of the individual domains were of similar size except for 8864 (A1 shortened, no A2 and A3 fragments) and 1470 (A3 fragment shortened). B: Restriction digest with HincII and PstI. A1 and B1 fragments were the same as those from panel A. Strains 1470 and 8864 show RFLPs. rise to RFLPs. Therefore we subjected the A1^A3 and the B1^B3 products to digestions with various restriction enzymes. RFLPs obtained with representatitve enzymes are shown in Figs. 2B and 3. The particular restriction patterns of the B1 and A1 fragments were typical for all F serogroup strains and were also observed in some of the serotype X isolates (23016 and 12934). However, in a second set of three other serotype X strains (like strain 100 in Fig. 2B) a VPI10463 type of RFLP was found. Additionally, these two types of X-strains belong to different hybridisation groups. HincII and HindIII digestions of the B1 fragments seem to be indicative for a VPI10463 or an 8864/1470 `toxinotype' of the corresponding tcdb genes. Similarly, digestions of the appropriate A1 fragments with PstI or the A3 fragments with EcoRI and SpeI di ered typically between the strains (Fig. 3). Restriction enzymes used to characterize the six fragments A1^A3 and B1^B3 are listed in Table Discussion The major obstacle to identify polymorphisms of the tcda and tcdb genes is their large size of about bp. PCR ampli cations described so far covered only parts of the toxin genes and are not suitable for identifying gene polymorphisms [12^16]. Variant strains lacking TcdA production were initially identi ed by biochemical means [7,9]. At that time a partial genetical analysis was done, which indicated the existence of deletions without delineating them[5^7].

5 M. Rupnik et al. / FEMS Microbiology Letters 148 (1997) 197^ Fig. 3. Comparison of toxin genes of representative C. di cile isolates. The ve ORFs tcda^e of the C. di cile VPI10463 pathogenicity locus (PaLoc) are indicated, with the two toxin genes tcda and tcdb and the three accessory genes tcdc^e. The A1^A3 and B1^B3 fragments of strains 2022, 1470 and 8864 are aligned to the VPI10463 scheme; for these strains the PaLoc is reduced to only those fragments that were investigated in the present study; deletions are indicated by a v; the restriction sites indicated (EcoRI, E; HindIII, H; HincII, Hc; PstI, P; SpeI, S) were used during RFLP analysis of the di erent fragments (see text). The PCR approach described here allows for the rst time to check the complete toxin sequences in order to characterize polymorphisms by genetical means. By applying this approach deletions and typical restriction site polymorphisms in variant toxin genes were de ned more precisely. According to the results presented, strain 1470 has lost 1.7 kb of the A3 domain. The A3 fragment of strain 8864 was undetectable. In fact all but a shortend A1 fragment are missing from tcda Both deletions in tcda of 1470 and 8864 cover the immunodominant repetitive domains [2] and could easily serve as an explanation for the lack of detection of the TcdA molecules in ELISA tests [5^7]. Analysing the 2499 nt of the 3P end of the tcda gene of strain VPI10463 we had de ned highly repetitive nucleotide sequences (SRONs) of 20^41 nt, which were repeated 7^30 times [17]. At that time we already supposed that the tcda repetitive sequence should be a region subject to variation through homologous recombination [17]. As anticipated the major deletions in both strains 1470 and 8864 described here occur in the repetitive A3 fragment of the tcda gene. The RFLPs that we observed were distributed all over the two toxins. Dealing with genes of such a large size this was not surprising. Two variant restriction patterns, however, missing HincII sites in the 5P third of tcdb and an additional PstI site found in the rst 3P third of their tcda, seem to have some predictive character. This type of pattern is characteristic for the TcdA-negative strains 1470 and 8864, whose cytotoxins exert a cytopathic e ect di ering from that induced by the standard TcdB of VPI10463 [3,9]. Thus it seems possible to use our approach for typing C. di cile toxin genes. Conclusions drawn from comparison of the genotypes with the in vivo e ects of strains 8864 versus 1470, and VPI10463 versus 2022 seem to be contradictory. The apathogenic phenotype of strain 1470 [7] could be explained by the loss of most of its A3 ligand domain described above, since this would hinder TcdA-1470 to bind to the target cells [3]. However, strain 8864 kills hamsters under the typical

6 202 M. Rupnik et al. / FEMS Microbiology Letters 148(1997) 197^202 signs of PMC, despite carrying even a larger deletion in tcda gene which includes the complete A3 domain (Fig. 3). Similarly, of the two serotype G strains 2022 and VPI10463, only the latter displays pathogenicity [11]. These strains seem to carry identical A1^A3 and B1^B3 domains. The major di erence between the two pathogenic strains 8864 and versus the apathogenic isolates 1470 and 2022 could reside in their level of toxin production and VPI10463 are high level toxin producers. From these data it might be concluded that di erences in regulatory elements are responsible for the di erences in the pathogenic potential of the isolates. However, since 8864 is highly pathogenic, but lacks a functional TcdA protein required for hamster pathogenicity [18], there must be other yet unknown di erences residing in TcdB Acknowledgments Parts of the data presented here will be contained in the doctoral thesis of M.R., V.B., F.S. and M.H. The work was supported by grants from the DFG (Ei206/3-2) and Naturwissenschaftlich-Medizinische Zentrum Mainz. M.R. was supported by Grant C /II of the Slovenian Ministry of Science and Technology. We are indebted to M. Weidmann for critically reading the manuscript and for his constructive comments. We thank P. Leukel for perfect technical assistance. References [1] Knoop, F.C., Owens, M. and Crocker, I.C. (1993) Clostridium di cile: clinical disease and diagnosis. Clin. Microbiol. Rev. 6, 251^265. [2] Von Eichel-Streiber, C. (1993) Molecular biology of the Clostridium di cile toxins. In: Genetics and Molecular Biology of Anaerobic Bacteria (Sebald M., Ed.), Springer, Berlin. [3] Von Eichel-Streiber, C., Boquet, P., Sauerborn, M. and Thelestam, M. (1996) Large clostridial cytotoxins: a family of glycosyltransferases modifying small GTP-binding proteins. TIM 4, 375^382. [4] Braun, V., Hundsberger, T., Leukel, P., Sauerborn, M. and Von Eichel-Streiber, C. (1996) De nition of the single integration site of the pathogenicity locus in Clostridium di cile. Gene 181, 29^38. [5] Borriello, S.P., Wren, B.W., Hyde, S., Seddon, S.V., Sibbons, P., Krishna, M.M., Tabaqchali, S., Manek, S., Price, A.B. (1992) Molecular, immunological, and biological characterization of a Toxin A-negative, Toxin B-positive strain of Clostridium di cile, Infect. Immunol. 60, 4192^4199. [6] Lyerly, D.M., Barroso, L.A., Wilkins, T.D., Depitre, C. and Corthier, G. (1992) Characterization of a Toxin A-negative, Toxin B-positive strain of Clostridium di cile. Infect. Immun. 60, 4633^4639. [7] Depitre, C., Delmee, M., Avesani, V., L'Haridon, R., Roels, A., Popo, M. and Corthier, G. (1993) Serogroup F strains of Clostridium di cile produce toxin B but not toxin A. J. Med. Microbiol. 38, 434^441. [8] Von Eichel-Streiber, C., Meyer zu Heringdorf, D., Habermann, E. and Sartingen, S. (1995) Closing in on the toxic domain through analysis of a variant Clostridium di cile cytotoxin B. Mol. Microbiol. 17, 315^321. [9] Torres, J.F. (1991) Puri cation and characterisation of toxin B from a strain of Clostridium di cile that does not produce toxin A. J. Med. Microbiol. 35, 40^44. [10] Von Eichel-Streiber, C. and Sauerborn, M. (1995) Molecular variations of toxins A and B: implications for the diagnosis of Clostridium di cile-induced disease. In: Medical and Dental Aspects of Anaerobes (Duerden, B.I., Wade, W.G., Brazier, J.S., Eley, A., Wren, B. and Hudson, M.J., eds.), Science Reviews, Northwood. [11] Delmeèe, M. and Avesani, V. (1990) Virulence of ten serogroups of Clostridium di cile in hamsters. J. Med. Microbiol. 33, 85^90. [12] Wren, B.W., Clayton, C.L. and Tabaqchali, S. (1990) Nucleotide sequence of Clostridium di cile toxin A gene fragment and detection of toxigenic strains by polymerase chain reaction. FEMS Microbiol. Lett. 70, 1^6. [13] Fluit, A.C., Wolfhagen, M.J., Verdonk, G.P., Jansze, M., Torensma, R. and Verhoef, J. (1991) Nontoxigenic strains of Clostridium di cile lack the genes for both toxin A and toxin B. J. Clin. Microbiol. 29, 2666^2667. [14] Kato, N., Ou, C.-Y., Kato, H., Bartley, S.L., Brown, V.K., Dowell, Jr. V.R. and Ueno, K. (1991) Identi cation of toxigenic Clostridium di cile by the polymerase chain reaction. J. Clin. Microbiol. 29, 33^37. [15] Boondeekhun, H.S., Gurtler, V., Odd, M.L., Wilson, V.A. and Mayall, B.C. (1993) Detection of Clostridium di cile enterotoxin gene in clinical specimens by polymerase chain reaction. J. Med. Microbiol. 38, 384^387. [16] Gumerlock, P.H., Tang, Y.J., Weis, J.B., and Silva, Jr. J. (1993) Speci c detection of toxinogenic strains of in stool specimens. J. Clin. Microbiol. 31, 507^511. [17] Von Eichel-Streiber, C. and Sauerborn, M. (1990) Clostridium di cile toxin A carries a C-terminal repetitive structure homologous to the carbohydrate binding region of streptococcal glycosyltransferases. Gene 96, 107^113. [18] Lyerly, D.M., Saum, K.E., MacDonald, D.K., Wilkins, T.D. (1985) E ects of Clostridium di cile toxins given intragastrically to animals. Infect. Immun. 47, 349^352.