|Home | About | Journals | Submit | Contact Us | Français|
Yersinia enterocolitica is an enteric pathogen that consists of six biotypes: 1A, 1B, 2, 3, 4, and 5. Strains of the latter five biotypes can carry a virulence plasmid, known as pYV, and several well-characterized chromosomally encoded virulence determinants. Y. enterocolitica strains of biotype 1A lack the virulence-associated markers of pYV-bearing strains and were once considered to be avirulent. There is growing epidemiological, clinical, and experimental evidence, however, to suggest that some biotype 1A strains are virulent and can cause gastrointestinal disease. To identify potential virulence genes of pathogenic strains of Y. enterocolitica biotype 1A, we used genomic subtractive hybridization to determine genetic differences between two biotype 1A strains: an environmental isolate, Y. enterocolitica IP2222, and a clinical isolate, Y. enterocolitica T83. Among the Y. enterocolitica T83-specific genes we identified were three, tcbA, tcaC, and tccC, that showed homology to the insecticidal toxin complex (TC) genes first discovered in Photorhabdus luminescens. The Y. enterocolitica T83 TC gene homologues were expressed by Y. enterocolitica T83 and were significantly more prevalent among clinical biotype 1A strains than other Yersinia isolates. Inactivation of the TC genes in Y. enterocolitica T83 resulted in mutants which were attenuated in the ability to colonize the gastrointestinal tracts of perorally infected mice. These results indicate that products of the TC gene complex contribute to the virulence of some strains of Y. enterocolitica biotype 1A, possibly by facilitating their persistence in vivo.
Three species of the genus Yersinia, Yersinia pestis, Yersinia pseudotuberculosis, and Yersinia enterocolitica, are pathogenic for humans. Y. pestis is the causative agent of bubonic and pneumonic plague and is transmitted by flea bites or respiratory aerosols. Y. pseudotuberculosis and Y. enterocolitica are intestinal pathogens that can produce symptoms such as diarrhea, fever, and abdominal pain if they are ingested in contaminated food or water (8).
Y. enterocolitica is a heterogenous species which is divided into six biotypes: 1A, 1B, and 2 through 5, on the basis of its biochemical behavior (37). Of these biotypes, only 1B and 2 to 5 ever carry the Yersinia virulence plasmid (pYV), which encodes approximately 50 proteins, including a surface adhesin, YadA, a type III secretion system, and 12 effector proteins that allow the bacteria to evade phagocytosis and killing by neutrophils and macrophages (7). Strains of biotypes 1B and 2 through 5 also carry chromosomal genes that have been implicated in virulence, including inv and ail, which mediate invasion of eukaryotic cells; myf, which encodes a fimbrial adhesin; and ystA, which encodes a heat-stable enterotoxin (Yst-a) (8). In addition to these factors, biotype 1B strains possess iron acquisition genes on a “high-pathogenicity island” and the ysa type III secretion apparatus, which also contribute to virulence (5, 18).
Y. enterocolitica strains of biotype 1A do not carry pYV and typically lack Ail, Myf, the ysa type III secretion system, and the high-pathogenicity island and seldom produce Yst-a (reviewed in reference 34). Although this suggests that Y. enterocolitica biotype 1A strains are not pathogenic, they have been isolated from patients with gastrointestinal symptoms in various countries around the world (34) and in two controlled studies were found to be significantly associated with disease (14, 25). In addition, Burnens et al. (4) reported that the duration and severity of infections with biotype 1A Y. enterocolitica are similar to those caused by pYV-bearing strains.
The epidemiological evidence that some strains of Y. enterocolitica biotype 1A are able to cause disease is supported by laboratory investigations showing that strains of this biotype can be separated into two groups: pathogenic and nonpathogenic (16, 17). Members of the pathogenic group, comprising strains isolated from humans with gastrointestinal symptoms, possess several virulence-associated properties that are absent from strains obtained from other sources. These properties include a significantly greater capacity to invade HEp-2 and Chinese hamster ovary (CHO) cells, to survive within bone marrow-derived macrophages, to egress or “escape” from HEp-2 cells and macrophages, and to persist within the gastrointestinal tracts of perorally inoculated mice for longer periods than biotype 1A strains from nonclinical sources (16, 17). The factors that allow only some biotype strains to exhibit these characteristics are not known.
The identification of virulence genes in Y. enterocolitica biotype 1A would contribute to our understanding of how these bacteria cause disease and provide diagnostic tools to distinguish potentially pathogenic biotype 1A strains from their less virulent counterparts. For this study, we used genomic subtractive hybridization, a technique that has been applied to many bacterial pathogens to discover novel virulence genes and targets for diagnostic purposes (38). This approach led to the identification of novel virulence-associated genes of Y. enterocolitica biotype 1A, related to the insecticidal toxin complex (TC) genes of other bacterial species.
The bacterial strains and plasmids used in this study are listed in Table Table1.1. In addition to these strains, a sample of Yersinia strains from our culture collection was screened for the presence of DNA sequences. This sample consisted of 120 Y. enterocolitica isolates (78 clinical biotype 1A, 28 nonclinical biotype 1A, and 14 non-biotype 1A strains) and 2 Yersinia aldovae, 3 Yersinia bercovieri, 4 Yersinia frederiksenii, 6 Yersinia intermedia, 4 Yersinia kristensenii, 3 Yersinia mollaretii, and 13 Y. pseudotuberculosis strains. Yersinia strains were grown in brain heart infusion (BHI; Oxoid, Hampshire, England) broth or on BHI agar at 28 to 30°C. Escherichia coli strains were cultured in Luria-Bertani broth or on Luria-Bertani agar at 37°C. When required, antibiotics were used at a final concentration of 100 μg of kanamycin, 100 μg of ampicillin, 12 μg of tetracycline, 20 μg of chloramphenicol, and 15 μg of gentamicin per ml.
DNA extractions and manipulations were performed according to established protocols (1, 29) or manufacturers' instructions. Colony blot and reverse dot blot hybridizations were performed using digoxigenin (DIG)-labeled probes at 68°C as described in the DIG Application Manual (Roche Diagnostics, Mannheim, Germany). E. coli cells were made electrocompetent by washing them with 10% (vol/vol) glycerol (29). Electrocompetent Y. enterocolitica cells were prepared as described by Conchas and Carniel (6).
Genomic subtractive hybridization was performed using the Clontech PCR-Select Bacterial Genome Subtraction kit with Y. enterocolitica T83, a clinical isolate, as the tester and Y. enterocolitica IP2222, an isolate from water, as the driver. The virulence-associated properties of these two strains have been reported previously (17). Briefly, 2-μg samples of Y. enterocolitica T83 and Y. enterocolitica IP2222 genomic DNA were each digested with RsaI, and two PCR adaptors were ligated to different aliquots of tester DNA. These preparations were denatured and hybridized at 63°C to an excess of denatured driver DNA which binds homologous sequences in the tester DNA (first hybridization). The two DNA pools were mixed together and hybridized at 63°C in the presence of more denatured driver DNA (second hybridization). Tester-specific sequences were amplified in two rounds of PCR using adaptor-specific primers. The subtracted fragments were ligated to pGEM-T Easy and electroporated into E. coli XL1-Blue.
To determine if the clones possessed Y. enterocolitica T83-specific sequences, reverse dot blot hybridization was performed. Inserts were amplified using adaptor-specific primers, applied to each of two positively charged nylon membranes (Roche Diagnostics) using a Bio-Dot Microfiltration apparatus (Bio-Rad Laboratories, Hercules, CA), and hybridized with DIG-labeled, RsaI-digested Y. enterocolitica T83 genomic DNA or DIG-labeled, RsaI-digested Y. enterocolitica IP2222 genomic DNA at high stringency. Inserts which hybridized to Y. enterocolitica T83 DNA but not to DNA from Y. enterocolitica IP2222 were deemed to possess tester-specific sequences.
A cosmid library of Y. enterocolitica T83 was constructed using the method described by DiLella and Woo (11). Briefly, 30- to 45-kb fragments of Sau3AI-digested Y. enterocolitica T83 genomic DNA were ligated to dephosphorylated pHC79 vector DNA and packaged into bacteriophage heads using the Packagene Lambda DNA Packaging System (Promega Corp., Madison, WI). These were subsequently used to infect E. coli LE392 cells according to the manufacturer's instructions. Approximately 800 clones were obtained and screened for the presence of subtracted DNA sequences by colony blot hybridization.
Nucleotide sequencing was performed using an ABI PRISM Big Dye Terminator v3.0 cycle sequencing kit (Applied Biosystems, Foster City, CA). Reaction products were analyzed on an Applied Biosystems ABI PRISM 377 DNA sequencer at the Australian Genome Research Facility. Vector-specific primers SP6 (5′-TATTTAGGTGACACTATAG-3′) and T7 (5′-TAATACGACTCACTATAGGG-3′) (26) were used to sequence inserts from subtracted clones. Homology searches were performed using the BLASTN, BLASTX, and BLASTP programs available at the National Center for Biotechnology Information website (www.ncbi.nlm.nih.gov). The promoter prediction program found at www.fruitfly.org/seq_tools/promoter.html was used to identify putative promoter sequences (28).
The prevalence of Y. enterocolitica T83 tcbA, tcaC, and tccC among 155 Yersinia strains was determined by colony blot hybridization. DIG was incorporated into PCR amplicons using primers 57F (5′-TAATAGTATTTCGAACGGAGAC-3′) and 57R (5′-ACGGTTAACCACACCCAGTTC-3′), primers 91′F (5′-TGCCAACAAGTCTTAATGTTCC-3′) and 91′R (5′-TGATATAGCATACCTGGTAGC-3′), and primers 8′F (5′-TCATAACTGTCACCGATCG-3′) and 8′R (5′-TACCAATTAAGCGCTGGGTC-3′) to produce probes to detect tcbA, tcaC, and tccC, respectively. Probes were hybridized at high stringency to bacteria that had been applied to Hybond N membranes (Amersham Biosciences, Buckinghamshire, England). Data were analyzed using Fisher's exact test (GraphPad Software, San Diego, CA).
RNA was isolated from Y. enterocolitica T83 grown to log phase at 30°C and 37°C using the RNeasy mini kit (QIAGEN, Valencia, CA) and the RNase-free DNase set (QIAGEN) and quantified using a spectrophotometer. Equal amounts of RNA were converted into cDNA using the Superscript First-Strand Synthesis System for RT-PCR (Invitrogen Life Technologies, Carlsbad, CA) according to the manufacturer's instructions. A 580-bp fragment of the tcbA gene was amplified using primers 57F and 57R, a 360-bp fragment of the tcaC gene was amplified using primers 91′F and 91′R, and a 560-bp fragment of the tccC gene was amplified using primers 8′F and 8′R.
The tcbA, tcaC, and tccC genes of Y. enterocolitica T83 were inactivated by partial deletion and insertion of a kanamycin resistance (Kmr) gene. Inactivation of tcbA was achieved using λ Red recombinase (9). The low-copy-number plasmid pKD46 possesses three arabinose-inducible genes, γ, β, and exo, whose products are called Gam, Bet, and Exo, respectively (9). Gam inhibits host RecBCD exonuclease V, while Bet and Exo promote recombination. We cloned a chloramphenicol resistance (Cmr) gene cassette into pKD46 so that the plasmid could be maintained in Y. enterocolitica T83, which is intrinsically resistant to ampicillin. A 1.3-kb Cmr gene cassette was removed from pBAC1 by digestion with SmaI and HincII and ligated to blunted, NcoI-digested pKD46 to produce pKD46CAT. pKD46CAT was transformed into Y. enterocolitica T83 and transformants expressing λ Red recombinase were prepared from bacteria grown in SOB (29) containing 100 mM l-arabinose and chloramphenicol.
A Y. enterocolitica tcbA::Kmr construct, ST7, was made as follows. A 2.3-kb region of tcbA was amplified from Y. enterocolitica T83 genomic DNA using primers TcbAF (5′-TTCTACAGTAATGACTCTATGC-3′) and TcbAR (5′-TCGAACTGGGTGTGGTTAAC-3′) and ligated to pGEM-T Easy. A 525-bp Eco47III fragment was removed from tcbA and replaced with a Kmr gene cassette from pUC4KIXX. The 3.2-kb tcbA::Kmr construct was excised from this plasmid by digesting it with NotI. Approximately 1.2 μg of linear DNA was electroporated into Y. enterocolitica T83 expressing λ Red recombinase. A Kmr transformant, ST7, was obtained and subcultured at 37°C, on medium lacking chloramphenicol to eliminate the temperature-sensitive plasmid pKD46CAT. The partial deletion of native tcbA and subsequent integration of the Kmr gene cassette were confirmed by PCR.
A tcaC mutant of Y. enterocolitica T83, ST5, was constructed by amplifying a 1.5-kb segment of tcaC using primers 91′F and TcaCR1 (5′-ACCAACTGCAACTGAGCGAC-3′) and ligating this fragment to pGEM-T Easy. A 478-bp fragment of tcaC was removed following digestion with EcoRV and replaced with a Kmr gene cassette on a 1.4-kb SmaI fragment from pUC4-KIXX. The tcaC::Kmr construct was excised from the resultant plasmid using NotI and ligated to Klenow-treated, BamHI-digested pRK404. This plasmid was transformed into E. coli SM10 (λpir) and transferred to Y. enterocolitica T83 by conjugation. A double-crossover mutant, ST5, was obtained by mating the transconjugants with SM10 (λpir) carrying the gentamicin-resistant (Gmr) plasmid pPH1JI, which is incompatible with pRK404, and selecting on Yersinia selective (CIN) agar (Oxoid) containing kanamycin and gentamicin. Mutant ST5 was Kmr, Gmr, and tetracycline sensitive (Tcs) and was shown by PCR to have lost a 480-bp segment of tcaC and to have gained the Kmr gene cassette.
A tccC mutant of Y. enterocolitica T83, ST6, was obtaining by using primers TccCFm (5′-ACTAAGCATTCGCACACTGG-3′) and Tc3′ (5′-ATGTCGTCCAACGCGAAGC-3′) to amplify a 2.6-kb product containing part of the tccC gene. This product was ligated to pGEM-T Easy, after which a 1.6-kb BglII/BclI fragment was removed and replaced with a Kmr gene cassette. The tccC::Kmr construct was transferred to pRK404 and used to inactivate tccC by homologous recombination as described above. The resultant mutant, ST6, was confirmed to possess a deletion or a Kmr gene insertion.
Quantitative assays of bacterial invasion of CHO cells were performed as described previously (17). Briefly, semiconfluent CHO cell monolayers were prepared in 24-well tissue culture trays (Nunc, Roskilde, Denmark). CHO cells were cultured in αMEM (Trace Biosciences, Melbourne, Victoria, Australia) containing 2 mM glutamine, 20 mM HEPES, and 10% (vol/vol) heat-inactivated (56°C for 30 min) fetal calf serum (JRH Biosciences, Lenexa, KS). Approximately 2 × 107 CFU of bacteria grown overnight without shaking were added to each well, and the trays were centrifuged at 860 × g for 8 min. Following incubation for 3 h at 37°C in 5% CO2, nonadherent bacteria were removed and the cells were washed three times with phosphate-buffered saline (PBS). Cells were incubated for a further 90 min in tissue culture medium containing 100 μg/ml gentamicin to kill extracellular bacteria. This medium was subsequently removed, and the cells were washed with PBS as before. The CHO cells were then lysed with 200 μl of 0.1% (wt/vol) digitonin (Sigma-Aldrich Corp., St Louis, MO) for 5 min, followed by the addition of 800 μl of BHI broth and vigorous pipetting to disrupt the CHO cells. Bacteria released from the CHO cells were then enumerated on agar plates.
Six-week-old female BALB/c mice were inoculated by gavage with 100 μl of a 10% (wt/vol) solution of sodium bicarbonate, followed by 6 × 108 CFU of bacteria suspended in 200 μl of PBS. Two days after inoculation, the mice were killed by CO2 inhalation and the ileum, cecum, and colon were removed aseptically. The samples were weighed and diluted 1 in 10 (wt/vol) in PBS. Each sample was homogenized using a Polytron homogenizer (Kinematica, Lucerne, Switzerland), and the homogenates were spread on duplicate CIN agar plates to determine the number of viable bacteria.
Statistical analysis was performed using InStat version 3.05 (GraphPad Software Inc., San Diego, CA). A two-tailed P value of < 0.05 was taken to indicate statistical significance.
The Y. enterocolitica T83 TC gene sequences were submitted to GenBank and assigned accession number AY647257.
Subtractive hybridization of Y. enterocolitica strain T83 (the tester strain) with strain IP2222 (the driver strain) yielded a pool of PCR products which were cloned into the vector pGEM-T Easy and electroporated into E. coli XL1-Blue to generate a subtracted DNA library. One hundred seventy-three clones obtained in this way were tested for the presence of Y. enterocolitica T83-specific inserts by reverse dot blot hybridization. Fifty-three (31%) of these hybridized with DNA from Y. enterocolitica T83, but not IP2222, and were deemed to possess Y. enterocolitica T83-specific sequences.
Tester-specific inserts were one-pass sequenced from pGEM-T Easy using vector-specific primers SP6 and T7. The sequences obtained ranged from 359 bp to approximately 1.7 kb, with an average size of 680 bp. Three sequences in the subtracted library were present twice, and one was present three times. Therefore, only 48 unique sequences were identified, of which 23 showed homology to genes encoding known proteins, 16 resembled genes for hypothetical proteins, and 9 had no significant matches in the databases.
Four distinct Y. enterocolitica T83-specific sequences showed homology to several insecticidal TC genes which encode high-molecular-weight insecticidal toxins (3) that were first identified in the bacterium Photorhabdus luminescens. Sequence SH57 (579 bp) showed 36% (68/185) amino acid identity to toxin A from P. luminescens, sequence SH8′ (563 bp) showed 66% (124/186) amino acid identity to a putative insecticidal toxin from Y. pestis CO92, sequence SH62′ (376 bp) showed 61% (59/96) amino acid identity to a putative toxin subunit from Y. pestis KIM, and sequence SH91′ (359 bp) showed 59% (71/119) amino acid identity to an insecticidal TC from Y. pestis CO92. To characterize the genetic regions containing these fragments, a cosmid library of Y. enterocolitica T83 was screened for the presence of these sequences by colony blot hybridization. One cosmid clone, 1B12, was identified which hybridized with all four sequences. A 20-kb region encompassing the four fragments was sequenced in each direction. Analysis of this sequence revealed three large open reading frames (ORFs) that were surrounded by smaller ORFs (Fig. (Fig.1).1). BLASTP homology searches showed that the three large ORFs were homologous to insecticidal TC proteins and were named tcbA, tcaC, and tccC (Table (Table22).
P. luminescens strains W14 and TT01 possess multiple copies of TC genes scattered over four TC loci, tca, tcb, tcc, and tcd (3, 13). However, these genes can be divided into three gene families: (i) tcaAB- or tcb/tcdA-like genes, (ii) tcaC- or tcdB-like genes, and (iii) tccC-like genes (36). Y. enterocolitica T83 possessed only one copy of the TC gene complex, as determined by Southern hybridization (data not shown). In this respect, Y. enterocolitica T83 is more similar to insect-associated Y. pestis and to the insect pathogens Serratia entomophila (which causes amber disease in the New Zealand grass grub, Costelytra zealandica) and Xenorhabdus nematophila, a close relative of P. luminescens (23, 24, 27) (Fig. (Fig.11).
The prevalence of tcbA, tcaC, and tccC from Y. enterocolitica T83 among 155 Yersinia strains was determined by colony blot hybridization. The results showed that the TC genes of Y. enterocolitica T83 were significantly more prevalent among 78 clinical biotype 1A strains than among 77 other Yersinia strains, including 28 nonclinical biotype 1A strains (Table (Table3).3). None of the TC genes was in Y. enterocolitica IP2222 or the well-studied pYV-bearing Y. enterocolitica strains W22703 (biotype 2, serotype O:9) and 8081 (biotype 1B, serotype O:8). Furthermore, BLAST analysis of the Y. enterocolitica 8081 genome sequence (available at http://www.sanger.ac.uk/Projects/Y_enterocolitica/) revealed that this strain does not possess any TC gene homologues.
RT-PCR was used to determine if tcbA, tcaC, and tccC were expressed by Y. enterocolitica T83. The results of this analysis showed that the TC genes are expressed at both 30°C and 37°C (Fig. (Fig.22).
To determine if the Y. enterocolitica T83 TC genes contribute to virulence, three different allelic exchange mutants, ST7 (tcbA::Kmr), ST5 (tcaC::Kmr), and ST6 (tccC::Kmr), were constructed and tested for loss of virulence. Firstly, we examined the ability of the mutants to grow in BHI broth at 30°C and in αMEM supplemented with 10% heat-inactivated fetal calf serum and 20 mM HEPES at 37°C in 5% CO2. Samples taken at 4-h intervals over a 28-h period showed that the TC mutants, ST7, ST5, and ST6, exhibited the same growth kinetics as the wild-type strain under both sets of conditions (data not shown).
We have previously shown that biotype 1A strains of clinical origin are able to invade CHO cells in higher numbers than strains obtained from the environment (16, 17). We tested the ability of the tcbA mutant, ST7, to invade CHO cells using a quantitative gentamicin protection assay. The invasive capacity of ST7 (1.83% ± 0.98%) was not significantly different from that of the wild-type strain (1.98% ± 0.05%; P = 0.80; Student's t test, two tailed). In addition, the ability of mutant ST7 to escape from HEp-2 cells was not different from that of the parent strain (data not shown).
We previously reported that Y. enterocolitica biotype 1A strains of clinical origin colonize the gastrointestinal tracts of mice for significantly longer periods than strains isolated from the environment (17). For this part of the study, we administered approximately 6 × 108 CFU of Y. enterocolitica T83 and the isogenic tcbA, tcaC, and tccC mutants to 10 6-week-old BALB/c mice. Two days later, the mice were killed and the numbers of bacteria that had colonized the ileum, cecum, and colon were determined. This time point was chosen because previous investigations had shown that Y. enterocolitica biotype 1A strains of clinical origin are at their most abundant in these tissues 2 days after infection (17). All three Y. enterocolitica TC mutants generally showed a decreased ability to colonize the intestinal tracts of mice compared to wild-type strain Y. enterocolitica T83 (Fig. (Fig.33).
Although Y. enterocolitica biotype 1A strains were once regarded as nonpathogenic, we reported previously that they can be divided into two groups: one comprising strains of clinical origin that possess virulence-associated characteristics and another made up of strains obtained from nonclinical sources that lack virulence-associated characteristics (17). In this study, we used subtractive hybridization to detect DNA sequences in a clinical biotype 1A strain, Y. enterocolitica T83, which were absent from a biotype 1A strain isolated from water. Among the genes identified in this way were tcbA, tcaC, and tccC, which were expressed by Y. enterocolitica T83, and were significantly more prevalent among clinically sourced biotype 1A strains, suggesting that these genes may contribute to the virulence of some Y. enterocolitica strains of biotype 1A.
In P. luminescens, the TC genes encode high-molecular-weight toxins capable of killing insects (3). P. luminescens colonizes the folds between the extracellular matrix and basal membrane of the midgut epithelium of insects (31). During occupation of this specific niche, the bacteria express TC A (Tca), which induces rounding up of the cells of the midgut epithelium, causing them to detach from the basal membrane and accumulate in the gut lumen (2, 31). The exact role of the different TC proteins in pathogenesis is uncertain. Previous studies have indicated that members of the tcaAB- or tcb/tcdA-like gene family encode the active toxins, whereas the tcaC- and tccC-like gene families encode proteins involved in toxin activation. There is also evidence to suggest that when multiple tcaAB- or tcb/tcdA-like genes are present in one bacterium, the toxins they encode act on different targets. For example, when XptA1, XptB1, and XptC1 of X. nematophila are combined, they act on larvae of the cabbage white butterfly species Pieris brassicae and Pieris rapae, whereas the combination of XptA2, XptB1, and XptC1 is toxic for the tobacco budworm Heliothis virescens (30). These findings support the hypothesis that TcaC- and TccC-like proteins are involved in the delivery or activation of TcaAB- or Tcb/Tcd-like toxins with different host range specificities.
Two other insect pathogens that have TC gene homologues capable of killing insects are S. entomophila and X. nematophila (23, 24). The insect-associated bacterium Y. pestis has also been shown to possess TC gene homologues which are hypothesized to contribute to survival of the bacterium in its arthropod vector (10, 27). However, the tcaB gene of Y. pestis CO92 possesses a frameshift mutation and the tcaC gene has an internal deletion (27). These attenuations are thought to be advantageous to Y. pestis, as they may allow the bacteria to survive in the flea midgut without killing the insect. The tcaC gene, however, is only believed to possess an internal deletion based on comparisons to TC genes from other bacteria and could conceivably encode an active protein. Furthermore, although the tcaB gene of Y. pestis CO92 possesses a frameshift mutation, the corresponding tcbA homologues in Y. pestis strains KIM and 91001 do not (10, 33). The role of these genes in the flea is therefore uncertain.
Interestingly, some bacteria with no known association with insects also possess TC gene homologues, including the enteric pathogens Y. enterocolitica biotype 1A and Y. pseudotuberculosis, the plant pathogen Pseudomonas syringae, the cellulolytic bacterium Fibrobacter succinogenes, and the dental pathogen Treponema denticola (15, 20). It is possible that the TC genes in these bacteria do not encode insecticidal toxins but serve a different purpose.
Compared to Y. enterocolitica T83, isogenic tcbA, tcaC, and tccC deletion mutants exhibited a decreased ability to colonize the ilea, ceca, and colons of mice perorally inoculated with these bacteria 2 days earlier. This was not due to differences in growth kinetics or invasive ability (in the case of the tcbA mutant, ST7). Attempts to complement the Y. enterocolitica T83 TC mutants with wild-type copies of the TC genes in high-copy-number vectors were unsuccessful, possibly due to toxic effects on E. coli intermediates. Difficulties in cloning tcaAB- or tcb/tcdA-like genes in E. coli have been reported by other workers (23, 35). Although we were able to clone the three TC genes together on cosmid 1B12, this clone was unstable and could not be used to complement the TC mutants. Despite our inability to complement the TC mutants, it is apparent that the TC genes of Y. enterocolitica T83 play a role in the persistence of the bacterium in the gastrointestinal tracts of mice, given that each of the tcbA, tcaC, and tccC insertion-deletion mutants displayed a similar phenotype. The exact roles of the TC proteins in vivo, however, are unknown. We have shown that the tcbA gene is not involved in invasion of cells in vitro. Furthermore, Y. enterocolitica T83 and other TC-positive biotype 1A strains do not exhibit any cytotoxic effects on mammalian cells in vitro (reference 16 and unpublished data). Possibly, the TC proteins contribute to the enterotoxic activity of some Y. enterocolitica biotype 1A strains we have reported previously (17).
In conclusion, we have used subtractive hybridization to identify homologues of insecticidal TC genes in Y. enterocolitica biotype 1A. These genes were more prevalent among clinical biotype 1A yersiniae than other Yersinia strains but were present in only 27% of clinical biotype 1A strains, indicating that they are not essential for virulence and that they are unsuitable for use in diagnostic applications to identify virulent strains of Y. enterocolitica biotype 1A. We also showed that inactivation of the TC genes of Y. enterocolitica biotype 1A resulted in mutants which were attenuated in the ability to colonize the gastrointestinal tracts of perorally inoculated mice. Further investigations of these proteins should provide novel insights into the pathogenic mechanisms of some strains of Y. enterocolitica biotype 1A.
We are grateful to S. Fenwick, A. J. Pittard, and G. Wauters for the gift of bacteria and plasmids that were used in this study. We thank Danijela Krmek for assistance with animal experiments.
This work was supported by a grant from the Australian National Health and Medical Research Council.
Editor: J. B. Bliska