* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Different susceptibility of two animal species infected with isogenic
Gene desert wikipedia , lookup
Epigenetics in learning and memory wikipedia , lookup
Genome (book) wikipedia , lookup
Gene expression programming wikipedia , lookup
Gene expression profiling wikipedia , lookup
Vectors in gene therapy wikipedia , lookup
Frameshift mutation wikipedia , lookup
Genetic engineering wikipedia , lookup
Public health genomics wikipedia , lookup
Nutriepigenomics wikipedia , lookup
Point mutation wikipedia , lookup
Genomic library wikipedia , lookup
Genome evolution wikipedia , lookup
Site-specific recombinase technology wikipedia , lookup
No-SCAR (Scarless Cas9 Assisted Recombineering) Genome Editing wikipedia , lookup
Therapeutic gene modulation wikipedia , lookup
Designer baby wikipedia , lookup
History of genetic engineering wikipedia , lookup
Artificial gene synthesis wikipedia , lookup
Microevolution wikipedia , lookup
Microbiology (2003), 149, 3203–3212 DOI 10.1099/mic.0.26469-0 Different susceptibility of two animal species infected with isogenic mutants of Mycobacterium bovis identifies phoT as having roles in tuberculosis virulence and phosphate transport Desmond M. Collins, R. Pamela Kawakami, Bryce M. Buddle, Barry J. Wards and Geoffrey W. de Lisle Correspondence Desmond M. Collins AgResearch, Wallaceville Animal Research Centre, PO Box 40063, Upper Hutt, New Zealand desmond.collins@agresearch. co.nz Received 8 May 2003 Revised 15 July 2003 Accepted 28 July 2003 The Mycobacterium tuberculosis complex includes Mycobacterium bovis, which causes tuberculosis in most mammals, including humans. In previous work, it was shown that M. bovis ATCC 35721 has a mutation in its principal sigma factor gene, sigA, causing a single amino acid change affecting binding of SigA with the accessory transcription factor WhiB3. ATCC 35721 is avirulent when inoculated subcutaneously into guinea pigs but can be restored to virulence by integration of wild-type sigA to produce M. bovis WAg320. Subsequently, it was surprising to discover that WAg320 was not virulent when inoculated intratracheally into the Australian brushtail possum (Trichosurus vulpecula), a marsupial that is normally very susceptible to infection with M. bovis. In this study, an in vivo complementation approach was used with ATCC 35721 to produce M. bovis WAg322, which was virulent in possums, and to identify the virulence-restoring gene, phoT. There are two point deletions in the phoT gene of ATCC 35721 causing frameshift inactivation, one of which is also in the phoT of BCG. Knockout of phoT from ATCC 35723, a virulent strain of M. bovis, produced M. bovis WAg758, which was avirulent in both guinea pigs and possums, confirming that phoT is a virulence gene. The effect on virulence of mode of infection versus animal species susceptibility was investigated by inoculating all the above strains by aerosol into guinea pigs and mice and comparing these to the earlier results. Characterization of PhoT indicated that it plays a role in phosphate uptake at low phosphate concentrations. At least in vitro, this role requires the presence of a wild-type sigA gene and appears separate from the ability of phoT to restore virulence to ATCC 35721. This study shows the advantages of using different animal models as tools for the molecular biological investigation of tuberculosis virulence. INTRODUCTION Tuberculosis continues to be one of the most common causes of human death due to bacterial infection and is also a widespread cause of animal morbidity and mortality. In humans, the primary cause is Mycobacterium tuberculosis, although in many parts of the world a significant amount of disease is also due to infection with the very closely related organisms Mycobacterium africanum and the major animal pathogen Mycobacterium bovis. Together with a few less important species, these organisms are collectively referred to as the M. tuberculosis complex. Recent genomic studies have shown that the vast majority of genes within these species are identical, or nearly so, and that the most obvious differences between species and also between strains within species are DNA insertions and deletions encoding one or a small number of genes (Cole, 2002). While the functional Abbreviation: PPD, purified protein derivative. 0002-6469 G 2003 SGM significance of these differences as well as the many nonsynonymous nucleotide substitutions that exist between strains (Fleischmann et al., 2002) are presently unknown, some of them must account for the differences in pathogenicity observed between different species and strains. Since the identification of the first genes involved in mycobacterial virulence (Collins et al., 1995; Wilson et al., 1995), further development and improvement of molecular genetic techniques has enabled the discovery of an increasing number of such genes (de Mendonca-Lima et al., 2003; Glickman & Jacobs, 2001). While there is continuing discussion about what constitutes a true virulence gene (Barry, 2001; Wassenaar & Gaastra, 2001), it is undoubtedly helpful to know if inactivation of a gene affects the virulence of an organism in an animal model, as this potentially enables further study of the host–pathogen interactions which are affected directly or indirectly by that gene. In most cases, virulence of the parent and a mutant daughter strain Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 Printed in Great Britain 3203 D. M. Collins and others has been determined in a single model of animal virulence, usually in mice or guinea pigs. This use of a single animal model reflects practicalities in terms of cost and time of using more than one animal host under biological containment for investigating a chronic disease, as well as an underlying assumption that the features of tuberculosis in different animal hosts are sufficiently similar that most virulence mechanisms will be the same in all hosts. Nevertheless, there is long-standing evidence that different species of the M. tuberculosis complex, as well as different strains within a species, have different pathogenicities for different animal hosts. For example, strains of M. tuberculosis are not as virulent as strains of M. bovis in rabbits but both species appear similarly virulent in guinea pigs (Dannenberg & Collins, 2001); and South Indian strains of M. tuberculosis appear of similar virulence to other M. tuberculosis strains for humans but are less virulent in guinea pigs (Balasubramanian et al., 1992). Even when a single animal host is used to determine the level of virulence, the result obtained depends on the particular indices that are measured. Growth rate in mice is often used to assess virulence of strains of the M. tuberculosis complex but this only indicates a subset of virulence mechanisms. Commonly used strains of M. tuberculosis and M. bovis grow at similar rates in mice and by that criterion would have similar virulence, but mice infected with M. bovis die much sooner that those infected with M. tuberculosis, so by that criterion M. bovis would be more virulent (North et al., 1999). These different ways of assessing virulence have recently been emphasized by the production of a number of single-gene mutants of M. tuberculosis that have similar growth to their parents in animals but differ markedly when compared on the basis of animal survival (Kaushal et al., 2002; Steyn et al., 2002). In earlier work (Collins et al., 1995), it was shown that M. bovis ATCC 35721 carries a point mutation in the principal sigma factor gene, sigA (designated rpoV in the earlier work), that causes attenuation of its virulence in guinea pigs when it is inoculated subcutaneously (subcutaneous guinea pig model). That work used an in vivo complementation approach in which ATCC 35721 was complemented with a wild-type sigA gene (to produce M. bovis WAg320) that restored virulence in this guinea pig model. Recently, it was found that the arginine to histidine change this mutation encoded at position 515 in SigA (SigA-R515H) reduced the interaction of this sigma factor with a secondary transcription factor WhiB3 (Steyn et al., 2002). It was also shown that inactivation of whiB3 in M. tuberculosis H37Rv had no effect on its ability to grow and persist in mice infected intravenously or in a subcutaneous guinea pig model but mice infected with this mutant had significantly longer mean survival times than those infected with the parent H37Rv strain. In contrast, inactivation of whiB3 in M. bovis completely attenuated its growth in a subcutaneous guinea pig model. These results indicate a different role for some M. tuberculosis and M. bovis genes in pathogenesis generated in different animal models. 3204 In New Zealand, it has been difficult to eradicate bovine tuberculosis from cattle in some parts of the country because of continual reinfection from tuberculous populations of the Australian brushtail possum (Trichosurus vulpecula), an animal introduced into New Zealand in the nineteenth century (de Lisle et al., 2001). In the course of tuberculosis studies on possums, we discovered that both M. bovis ATCC 35721 and M. bovis WAg320, its complement with a wild-type sigA gene, are attenuated when inoculated intratracheally into these animals. This result was in stark contrast to the situation in guinea pigs, where ATCC 35721 itself is attenuated but WAg320 has virulence approaching that of a wild-type strain. In this study, we describe this discovery, its use to identify an ABC transporter gene, phoT, and the virulence properties of phoT in different animal models of tuberculosis. METHODS Bacterial strains and culture conditions. The M. bovis strains used in this work are listed in Table 1. Liquid culture of all M. bovis strains was performed in Tween-albumin broth (Kent & Kubica, 1985) or supplemented Middlebrook 7H9 (Difco) medium (Collins et al., 2002). Solid culture was performed on supplemented Middlebrook 7H11 (Difco) medium (Collins et al., 2002). Where appropriate, media also contained 50 mg hygromycin ml21 or 20 mg kanamycin ml21. Culturing of M. bovis strains in low phosphate concentrations was performed in Sauton medium (Darzins, 1958). For culture of M. bovis from guinea pigs, half the spleen and the left apical lobe of the lung was used and for culture from mice, all the spleen and all the lungs. For culturing from possums, 1 g tissue from the caudal portion of the spleen was used as well as 1 g tissue from the lung that included a lesion or, if no lung lesion was present, from the right cardiac lobe. Samples from the same animal were homogenized separately with 20 ml water for 1 min. This was filtered through sterile cheese cloth and centrifuged at 3500 g for 20 min. The pellet was resuspended in 0?5 ml water and aliquots were plated onto Middlebrook 7H11 supplemented with 0?6 ml oleic acid l21, 50 g BSA l21, 20 g glucose l21, 7?7 g NaCl l21, 0?4 % sodium pyruvate, 0?5 % lysed sheep red blood cells, 10 % bovine serum, 10 mg fungizone ml21, 200 IU polymyxin B sulphate ml21, 100 mg tarcarcillin ml21 and 10 mg trimethoprim ml21. Escherichia coli strains were grown with appropriate antibiotics in L broth at 37 uC for XL-1 Blue MR and 30 uC for x2764. Identification of the virulence-restoring gene by in vivo complementation of ATCC 35721 in possums. Genomic DNA was extracted from M. bovis WAg200, partially digested with Sau3AI and fragments of 30–50 kb were prepared using sucrose gradient centrifugation as described previously (Collins et al., 1995). The fragments were ligated to BclI-digested pUHA8, a cosmid shuttle vector containing a kanamycin resistance gene and a mycobacteriophage integration sequence, that had been constructed from pYUB178 (Pascopella et al., 1994) by positioning PacI recognition sites on either side of the BclI cloning site. The fragments were also ligated into a second vector, pUHA28, that had been constructed by ligating sigA into the KpnI site in pUHA8. Constructs were packaged into lambda heads (GigaPack II Gold, Stratagene) and transduced into E. coli x2764; kanamycin-resistant clones were pooled and cosmid DNA was prepared as described previously (Collins et al., 1995). The cosmid DNA was electroporated into ATCC 35721 (Wards & Collins, 1996) and kanamycin-resistant colonies were pooled into separate ATCC 35721(pUHA8 : : WAg200) and ATCC 35721(pUHA28 : : WAg200) libraries and inoculated into possums. Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 Microbiology 149 M. bovis phoT in phosphate transport and virulence Table 1. Bacterial strains and plasmids Strain or plasmid M. bovis ATCC 35721 ATCC 35723 WAg200 BCG WAg320 WAg322 WAg324 WAg325 WAg758 E. coli XL-1 Blue MR x2764 Plasmids pYUB178 pUHA8 pBluescript II KS pUHA9 pUHA28 pUHA30 pUHA37 pUHA38 pUHA39 pUHA43 Relevant properties Source or reference Low virulence for guinea pigs; sigA-R515H Moderate virulence for guinea pigs Wild-type New Zealand cattle isolate; high virulence for guinea pigs Pasteur strain 1173P2 ATCC 35721(pYUB178 : : sigA); virulent for guinea pigs ATCC 35721(pUHA8 : : WAg200 cosmid : : phoT) ATCC 35721(pUHA8 : : phoT) ATCC 35721(pUHA8 : : phoT) ATCC 35723 phoT : : hygr ATCC* ATCC* Collins et al. (1995) ATCC* Collins et al. (1995) This study This study This study This study Laboratory K-12 strain for in vitro cloning HB101 lysogen with l cI857 b2 redb3 S7; cosmid cloning Stratagene Jacobs et al. (1986) Mycobacterial integrating cosmid shuttle vector; Kanr pYUB178 with PacI sites either side of BclI cloning site E. coli phagemid cloning vector; Ampr pBluescript II KS with PacI–BclI–PacI cassette inserted at BamHI cloning site pUHA8 : : sigA pUHA8 containing virulence-restoring cosmid insert from WAg322 pUHA8 containing 1973 bp of phoT locus pUHA8 containing 2311 bp of phoT locus pUHA9 containing phoT–phoY2 locus from ATCC 35721 pUHA9 containing pUHA37 insert interrupted with hygr; suicide plasmid Pascopella et al. (1994) Collins et al. (1995) Stratagene Collins et al. (2002) This study This study This study This study This study This study *American Type Culture Collection. Colonies recovered from possum lesions were characterized by junction fragment analysis (Collins et al., 1995). Briefly, DNA extracted from ATCC 35721(pYUB8 : : WAg200) and ATCC 35721(pYUB28 : : WAg200) clones was characterized by restriction digestion and Southern blot hybridization using a probe of pUHA8 to reveal the size of fragments at the junction sites where the integrated vector arms join the host chromosomal DNA. The cosmid insert in a selected clone of ATCC 35721(pYUB8 : : WAg200) was recovered by digesting its chromosomal DNA with PacI, which has no sites in the chromosome of M. bovis, and cloning of the PacI fragment into pUHA9, a derivative of pBluescript II KS that contains PacI cloning sites. This fragment was partially digested with Sau3AI and fragments of 2–4 kb were prepared by sucrose gradient fractionation and ligated into the BclI cloning site of pUHA8. This mini-library was inoculated into possums, M. bovis colonies were cultured from the lesions and the inserts in two of these virulencerestoring clones were sequenced. DNA sequences were analysed using the programs of the Genetics Computer Group and compared to the GenBank (http://www.ncbi.nlm.nih.gov) and Sanger (http:// www.sanger.ac.uk) databases. et al., 1991). DNA from allelic-exchange mutants in which the phoT gene had been deleted was identified by restriction digestion and Southern blot hybridization and confirmed by PCR analysis using primers on each side of the phoT site into which the hygromycin resistance gene had been ligated. Animal models of virulence. All animal work was approved by the institution’s Animal Ethics Committee. For the subcutaneous guinea pig model, the virulence of each M. bovis strain was tested in 3–6 female Dunkin-Hartley guinea pigs, as described previously (Wilson et al., 1995). Animals were inoculated by injecting 106 c.f.u. subcutaneously into their flank and were killed at 8 weeks. The Knockout of phoT in M. bovis. Knockout of the phoT gene from ATCC 35723 was performed in a similar way to that described previously (Wards et al., 2000). Briefly, a suicide plasmid was constructed by inserting a hygromycin resistance gene into the EcoRV site of phoT (Fig. 1) in the 1973 bp insert from pUHA37. The phoT : : hygr fragment was transferred into pUHA9 to produce pUHA43, which was electroporated into ATCC 35723 using a highefficiency electroporation technique (Wards & Collins, 1996). Cells were plated onto medium containing hygromycin and resistant colonies were subcultured and their DNA was extracted (van Soolingen http://mic.sgmjournals.org Fig. 1. Annotation of the phoT locus in M. tuberculosis and M. bovis showing the identity of the fragments from pUHA37 and pUHA38 that restored virulence to M. bovis ATCC 35721 in possums. The EcoRV site denotes where the hygr gene was inserted into the fragment from pUHA37 in producing a suicide plasmid, pUHA43, for knockout of phoT. Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 3205 D. M. Collins and others assessment of virulence in early work was based on the presence of visible lesions of tuberculosis in the spleen and, in later comparative work, on the enumeration of M. bovis c.f.u. in samples from the spleen. All animals were tested for infection by determining their delayed-type hypersensitivity to tuberculin using intradermal injection of 4 units (0?1 ml) of bovine purified protein derivative (PPD; AgriQuality, New Zealand) immediately prior to inoculation with M. bovis and immediately prior to the animals being killed. For the possum model, animals were inoculated intratracheally with 105 c.f.u. of M. bovis in 200 ml. The organisms were inoculated through a 2 mm diameter vinyl cannula per os into the trachea of anaesthetized animals as described previously (Pfeffer et al., 1994) and killed after 6 weeks. The assessment of virulence in early work was based on the presence of visible lesions of tuberculosis in the lungs and, in later comparative work where groups of four animals were used, on the enumeration of M. bovis c.f.u. in samples from the lung and spleen. All possums were tested for infection before inoculation and before being killed, by using a lymphocyte stimulation assay as described previously (Skinner et al., 2002). For the aerosol guinea pig and mouse models, single cell suspensions of each M. bovis strain were prepared by sonication for 30 s and filtration through an 8 mm filter. Groups of three Dunkin-Hartley guinea pigs and four BALB/c mice were infected as described previously (Aldwell et al., 2003; McMurray et al., 1985) using an aerosol chamber which produces droplet nuclei of the size appropriate for entry into alveolar spaces. The aerosolized solution contained approximately 56106 c.f.u. ml21; based on previous empirical observations and the time of animal exposure this would have resulted in the inhalation and retention in the lungs of approximately 500 c.f.u. for guinea pigs and 30 c.f.u. for mice. Animals were killed after 8 weeks (except for guinea pigs inoculated with WAg320) and the assessment of virulence was based on the enumeration of M. bovis c.f.u. in samples from the lung and spleen. Guinea pigs were tested for infection by determining their delayedtype hypersensitivity to tuberculin using intradermal injection of 4 units of bovine PPD immediately prior to inoculation with M. bovis and immediately prior to the animals being killed. In all cases of M. bovis enumeration, statistical analyses by ANOVA were performed on log10 transformations of M. bovis c.f.u. M. bovis culture in ciprofloxacin and low-phosphate medium. For determination of the MIC of strains in ciprofloxacin, 10-fold serial dilutions of each strain were cultured in duplicate on supplemented Middlebrook 7H11 medium containing no ciprofloxacin and on the same medium containing doubling dilutions of 0?6–0?0375 mg ciprofloxacin ml21. For culture in low-phosphate medium, approximately 36106 c.f.u. of each strain in 50 ml Tweenalbumin broth were inoculated into 5 ml Sauton medium containing serial dilutions of phosphate. The final concentration of phosphate for each culture was calculated taking into account the phosphate added in the 50 ml inoculum. RESULTS Virulence in possums of M. bovis ATCC 35721 complemented with wild-type sigA (WAg320) M. bovis ATCC 35721 and its complement with wildtype sigA (WAg320) were assessed for their ability to cause disease in comparison to a wild-type M. bovis strain (WAg200) in an intratracheal possum model of virulence. The results are given together with the genotypes of the strains in Table 2, in comparison to the results previously obtained with a subcutaneous guinea pig model (Collins et al., 1995). Using an assessment of virulence based on the ability of a strain to produce visible tuberculous lesions, the complementation of ATCC 35721 with wildtype sigA restored a moderate level of virulence for guinea pigs but had no ability to restore virulence for possums. Subsequently, this result was confirmed by repeating the experiment and enumerating the c.f.u. in the lungs and spleen of possums and the spleens of guinea pigs (Table 3). Identification of a wild-type gene complementing ATCC 35721 to virulence for possums In order to identify a gene or genes that might restore virulence to ATCC 35721 for possums, two integrated cosmid libraries of recombinant M. bovis ATCC 35721 were prepared: ATCC 35721(pUHA8 : : WAg200) and ATCC 35721(pUHA28 : : WAg200). The only difference between the two libraries was the presence of sigA in pUHA28. Each library was inoculated into the trachea of two possums. In each case, one of the two possums developed Table 2. Virulence and sigA genotype of M. bovis strains in guinea pigs inoculated subcutaneously and in possums inoculated intratracheally Strain Genotype sigA-R515HD WAg200 ATCC 35721 WAg320 WAg322 WAg324+325 ATCC 35723 WAg758 + + + + Virulence* sigA Guinea pig spleen lesions Possum lung lesions + +++ 2 ++ ++ ++ ++ + +++ 2 2 +++ +++ +++ + + + + *Virulence: 2, no lesions; +, 1–5 lesions; ++, 6–50 lesions; +++, >50 lesions or (in possums) consolidated lung lobe. DsigA-R515H: mutant allele of sigA encoding histidine instead of arginine at position 515 in SigA. 3206 Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 Microbiology 149 M. bovis phoT in phosphate transport and virulence Table 3. Virulence of M. bovis strains in various animal models The results are means±SE; the number of animals per group (n) is shown for each model. Values in each column with the same superscripts (a,b,c) are not significantly different (P<0?05). Strain Genotype* sigA phoT Virulence Guinea pig (log10 c.f.u. per organ) Subcutaneous (n=3–6) Spleen ATCC 35721 R515H 2 WAg320 R515H, + 2 WAg324 R515H 2, + ATCC 35723 + + WAg758 + 2 2?4b±0?1 4?9a±0?4 4?7a±0?4 5?2a±0?4 2?3b±0?5 Possum, intratracheal (log10 c.f.u. g”1) (n=4) Mouse, aerosol (log10 c.f.u. per organ) (n=4) Aerosol (n=3) Lung Spleen 6?1b±0?4 3?7b±0?5 7?9a±0?1 5?5a±0?4 3?8c±0?7 3?0b±0?5 5?2bc±0?5 4?0b±0?3 2?2d±0?4 1?0cD±0?0 Lung Spleen Lung Spleen 1?8b±0?3 2?0b±0?1 5?4a±0?8 6?6a±0?2 3?0b±0?4 1?8b±0?1 1?7bD±0?0 4?7a±1?0 4?9a±0?3 1?7bD±0?0 5?9b±0?1 6?6a±0?1 5?5b±0?2 3?6c±0?3 3?1c±0?2 5?1a±0?1 5?0a±0?2 4?7a±0?1 2?2b±0?3 2?0b±0?2 *R515H, mutant allele of sigA encoding histidine instead of arginine at position 515 in SigA; +, wild-type allele; 2, inactive phoT. DThis is the maximum log10 c.f.u. possible as no organisms were isolated for these animals. a small lung lesion. Subsequent analysis of the recombinants from these lesions showed that their cosmid inserts largely overlapped. Reinoculation of one of the ATCC 35721(pUHA8 : : WAg200) recombinants (WAg322) into a possum caused multiple lung lesions and WAg322 was also moderately virulent in guinea pigs (Table 2). The cosmid-sized insert in WAg322 was recovered from genomic DNA of this strain as an E. coli plasmid, pUHA30. The insert was subcloned, using partial Sau3AI digestion, back into the same mycobacterial integrating shuttle vector, pUHA8, and this mini-library was electroporated into ATCC 35721 to produce ATCC 35721 (pUHA8 : : pUHA30). Inoculation of this library into a single possum produced numerous lesions from which individual colonies were isolated. From 70 isolates analysed by restriction digestion and Southern blot hybridization, the two most common fragment patterns were found in 32 and 10 isolates respectively. Strains with both these predominant patterns were isolated from lesions in both the lung and spleen. Representative cultures (WAg324 and WAg325) with each of these patterns were combined together and inoculated into two possums. This mixed inoculum produced extensive lesions in both animals, and subsequent culture followed by DNA extraction, restriction digestion and Southern blot hybridization showed that both WAg324 and WAg325 were present in the lesions (Table 2). Subsequently, this result with WAg324 was confirmed in four possums by enumerating the c.f.u. in the lungs and spleen (Table 3). The subcloned fragments of pUHA30 in WAg324 and WAg325 were recovered by PacI digestion of genomic DNA, cloned into the E. coli sequencing vector pUHA9 to produce pUHA37 and pUHA38 respectively, and sequenced. The sequences were 1973 bp and 2311 bp in length and overlapped by 1973 bp (Fig. 1). Comparison of this 1973 bp sequence to that of http://mic.sgmjournals.org M. tuberculosis strains H37Rv and CDC1551 and M. bovis strain AF2122/97 revealed that WAg200 had an identical sequence to that of M. bovis AF2122/97 and that this sequence differed from that of the two M. tuberculosis strains by two nucleotides. Annotation of the 1973 bp region into ORFs was identical in the genome sequences of both M. tuberculosis H37Rv and CDC1551 and is shown in Fig. 1. The sequence contains only two complete ORFs, for genes designated phoT and phoY2 in H37Rv. The two nucleotide differences between M. bovis and M. tuberculosis were in the coding region of phoT (Fig. 2); one change was neutral and the other caused a change from phenylalanine to leucine at position 35 on PhoT. The equivalent region to this 1973 bp sequence was cloned from M. bovis ATCC 35721 to produce pUHA39. Sequencing of the insert in this plasmid revealed no mutations in phoY2 but two well-separated point deletions in the phoT ORF as shown in the alignment of phoT ORFs in Fig. 2. Subsequent sequencing of part of phoT in M. bovis BCG revealed that it too has one of the same frameshift mutations (Fig. 2). PhoT is an ABC transporter with a high similarity to proteins of the phosphate transport system, usually designated PstB. There are close homologues of M. bovis PhoT in other slow-growing mycobacteria (Mycobacterium leprae, 91 % identical; Mycobacterium avium, 89 % identical) as well as a pstB gene in M. tuberculosis that is 49 % identical to M. bovis PhoT. Recent sequencing of the genome of M. bovis AF2122/97 (Garnier et al., 2003) has revealed that, compared to M. tuberculosis pstB, M. bovis pstB has a frameshift mutation, the functional significance of which has yet to be elucidated. Because of the possible functional relatedness of phoT and pstB, the pstB genes in the M. bovis strains used for this study were also sequenced. ATCC 35721, ATCC 35723 and BCG Pasteur were all Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 3207 D. M. Collins and others (a) 1 2 3 4 5 6 phoT knockout Wild-type phoT (b) 1 2 3 4 5 6 phoT knockout Wild-type phoT Fig. 2. Alignment of phoT from: M. tb, M. tuberculosis H37Rv and CDC1551; M. bov, M. bovis AF2122/97 and WAg200; 35721, M. bovis ATCC 35721; BCG, M. bovis BCG Pasteur; * denotes identical nucleotide to M. tb; % denotes point deletion. found to have the same frameshift mutation in pstB as that present in AF2122/97. The presence of the pstB and phoT frameshift mutations in BCG Pasteur has now been confirmed by comparison to recently available sequence from the genome project for BCG Pasteur (http://www.sanger. ac.uk/Projects/M_bovis/). Knockout of phoT in M. bovis In order to determine the role of phoT in a virulent strain of M. bovis with a wild-type sigA genotype, a suicide plasmid approach was used to inactivate the phoT gene in M. bovis ATCC 35723 by allelic exchange. A Southern blot hybridization representing two successful knockout recombinants, a single homologous recombinant and three 3208 Fig. 3. (a) Southern blot hybridization after BamHI digestion and probing with phoT of DNA from six hygromycin-resistant recombinants: lanes 1–3, illegitimate recombinants; lane 4, single homologous recombinant; lanes 5 and 6, allelicexchange phoT knockout recombinants. The expected size of wild-type and knockout fragments is indicated by arrows. (b) Agarose gel of PCR products from phoT knockouts of M. bovis in comparison to reference products: lane 1, molecular size markers; lane 2, M. bovis ATCC 35723; lane 3, M. bovis phoT knockout (WAg758); lane 4, second M. bovis phoT knockout; lane 5, pUHA43 plasmid used for knockout of phoT; lane 6, pUHA37 plasmid with wild-type phoT. illegitimate recombinants is shown in Fig. 3(a). An agarose gel of PCR products from these knockouts in comparison to reference products is shown in Fig. 3(b). Compared to its moderately virulent parent, M. bovis ATCC 35723, the phoT knockout strain (WAg758) was found to have reduced virulence when assessed in the subcutaneous guinea pig model on the basis of spleen lesions (Table 2) and confirmed by enumeration of c.f.u. in the spleen (Table 3). One possible explanation for some of the differences in Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 Microbiology 149 M. bovis phoT in phosphate transport and virulence virulence observed between the strains used in this study was their mode of administration to animals: into the lungs in the intratracheal possum model and subcutaneously and eventually into the draining lymph circulation in the subcutaneous guinea pig model. In order to further investigate the importance of these modes of administration on the assessment of virulence, a range of strains was administered to guinea pigs and mice by aerosol into the lungs and the results compared to those obtained when the same strains were administered subcutaneously into guinea pigs and intratracheally into possums. Virulence was assessed by the mean numbers of c.f.u. in the spleen and lungs of groups of infected animals. In the case of subcutaneously infected guinea pigs, none of the groups had lesions in their lungs and lung tissues were not submitted for culture of M. bovis. The results are shown for all four animal models in Table 3. sigA, phoT (WAg324) was unable to restore growth in lowphosphate medium. At the lowest phosphate concentration, strains with wild-type sigA and knockout of phoT (WAg320 and WAg758) did not grow, in contrast to ATCC 35723 with wild types of both sigA and phoT, which did grow. This indicates that in wild-type strains of M. bovis, phoT plays a role in acquiring phosphate when it is at very low concentrations. The MIC of ciprofloxacin for various M. bovis strains is also given in Table 4. There was no difference between any of the strains except for WAg758, which had half the MIC of the other strains. This indicates that phoT may play a role in phosphate transport in ATCC 35723, as resistance to ciprofloxacin has been shown to be associated with phosphate transport in Mycobacterium smegmatis (Bhatt et al., 2000). DISCUSSION WAg320, which contains sigA, was significantly more virulent than its parent, ATCC 35721, in both the subcutaneous and aerosol guinea pig models and in the lungs of mice but both strains were highly attenuated in possums. WAg320 was particularly virulent in the aerosol guinea pig model and the animals were sufficiently clinically affected after 29 days to be killed on ethical criteria. In contrast, WAg324, which contains phoT, was significantly more virulent than its parent in possums and in the subcutaneous guinea pig model but was significantly less virulent than its parent in the aerosol guinea pig model and of similar virulence to its parent in the mouse model. The phoT knockout WAg758 was significantly less virulent than its parent ATCC 35723 in all the animal models except mice, although even in mice the mean c.f.u. was 0?5 log lower. In this study, we observed that the wild-type principal sigma factor gene, sigA, which was able to restore virulence to M. bovis ATCC 35721 in a subcutaneous guinea pig model, was unable to do so for an intratracheal possum model. In order to discover a gene or genes that might be important for the virulence of M. bovis in a natural wildlife host and to identify additional mutations in ATCC 35721, we exploited this observation by employing an in vivo complementation approach. A genomic cosmid library of a virulent M. bovis strain was used to identify a virulencerestoring cosmid and then a mini-library of this virulencerestoring cosmid was used to identify a 1973 bp DNA sequence that restored virulence. Analysis of this wild-type M. bovis sequence and the homologous sequence in ATCC 35721 showed that a gene in ATCC 35721, phoT, had two frameshift mutations. Since these frameshift mutations are in the core conserved domain of this ABC transporter (Higgins, 2001), it can be concluded that the gene in ATCC 35721 is inactivated. In vitro growth of M. bovis strains in ciprofloxacin and low-phosphate medium The growth of M. bovis strains in liquid medium containing various low concentrations of phosphate is given in Table 4. Growth was clearly associated with the presence of a wildtype sigA gene. In the absence of the wild-type version of Work to this point had established that phoT was necessary to restore virulence to ATCC 35721 in the intratracheal possum model, that it also restored virulence in Table 4. In vitro growth of M. bovis strains in low-phosphate medium and in ciprofloxacin Strain Genotype* sigA ATCC 35721 WAg320 WAg324 ATCC 35723 WAg758 R515H R515H, + R515H + + GrowthD at phosphate concentration (mM) of: Ciprofloxacin MIC (mg ml”1) phoT 4?0 0?62 0?29 0?14 2 2 2, + + 2 + + NG NG NG + + NG NG NG NG NG + + + + + + NG + 0?30 0?30 0?30 0?30 0?15 *R515H, mutant allele of sigA encoding histidine instead of arginine at position 515 in SigA; +, wild-type allele; 2, inactive phoT. DNG, No growth; +, growth. http://mic.sgmjournals.org Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 3209 D. M. Collins and others the subcutaneous guinea pig model and that ATCC 35721 has mutations in both its sigA and phoT genes. Clearly, ATCC 35721 is substantially different from a wild-type M. bovis strain and might even have further mutations perhaps of a compensatory nature, so it was desirable to confirm the virulence properties of phoT in a virulent M. bovis strain. In order to do this, the gene was inactivated by allelic exchange from the moderately virulent M. bovis strain, ATCC 35723, to produce WAg758. WAg758 was significantly less virulent than its parent in both the subcutaneous guinea pig and intratracheal possum models, establishing that phoT activity is necessary to maintain the virulence of ATCC 35723 in these models. The ability of phoT to restore virulence to ATCC 35721 for both the subcutaneous guinea pig and intratracheal possum models contrasts with the ability of sigA to restore virulence to ATCC 35721 only for the subcutaneous guinea pig model. This raised the question of whether the difference between the models with respect to sigA was due to differences in the host response to infection of guinea pigs and possums or was related to the mode of administration. It was possible that sigA might be sufficient for restoring virulence to ATCC 35721 when the strain is inoculated subcutaneously but not when animals are infected via the lungs. To investigate this possibility and also the effect of the presence or absence of phoT, the same group of five strains was inoculated into two additional animal models: aerosol models of the mouse and guinea pig. An important result from this work was the finding that WAg320 (ATCC 35721 complemented with sigA) was significantly more virulent than ATCC 35721 in both the subcutaneous and aerosol guinea pig models as well as in the mouse model. The original finding that WAg320 is virulent in guinea pigs (Collins et al., 1995) therefore reflects a general host susceptibility of guinea pigs to this strain that is not dependent on route of administration. The corollary of this is that because WAg320 is not virulent when administered into the lungs of possums, there is a clear difference in susceptibility to this strain between possums and both guinea pigs and mice. At the molecular level, it has now been shown that the mutation in SigA-R515H in ATCC 35721 affects its binding to an accessory transcription factor, WhiB3, which is thought to regulate genes that influence the immune response of the host (Steyn et al., 2002). If this is the case, then regulation of these genes by WhiB3 does not appear to be important for influencing the immune response of possums. Inoculation of ATCC 35723 and its phoT knockout strain (WAg758) in four different models gave consistent results. In three of the models, WAg758 was significantly less virulent than its parent while in the mouse model the trend was in the same direction but the difference was not significant with the small number of animals used. Clearly, phoT is important for the virulence of M. bovis in a range of animals. ATCC 35721 recombinants complemented with phoT were also found to be virulent when inoculated 3210 in the subcutaneous guinea pig model irrespective of the presence of the wild-type allele of the principal sigma factor gene, sigA. In our earlier work in which sigA was identified in a similar in vivo complementation approach to that used here for phoT, we analysed the predominant virulencerestoring clone from the library which represented 80 % of the recombinant M. bovis isolates recovered from guinea pigs (Collins et al., 1995). In light of the present finding that phoT also restores virulence to ATCC 35721 in the subcutaneous guinea pig model, we would now conclude that some of the remaining 20 % of virulence-restoring recombinants from the first study probably contained cosmids incorporating the phoT locus. The most likely explanation for their unequal representation in the earlier study is the under-representation of cosmids containing phoT in the library used in that study. Another possible explanation is that phoT restores ATCC 35721 (WAg324) to a slightly lower degree of virulence than does sigA (WAg320) and is outcompeted by WAg320 when both are inoculated together. However, far greater numbers of animals per group would be needed than those used throughout this study in order to determine if there is a significant degree of difference in virulence between these two strains. Sequencing of a PCR product made from genomic DNA of M. bovis BCG showed that one of the frameshift mutations in the phoT of ATCC 35721 was also present in BCG. It is now well established (Wards et al., 2000; Lewis et al., 2003) that the absence of the RD1 region from BCG contributes greatly to its attenuation and that much of this attenuation can be accounted for by the loss of the esat-6 gene (Wards et al., 2000), one of the nine genes deleted from BCG in the RD1 region. The identification of a frameshift mutation causing inactivation of the phoT gene, which has virulence properties and is very distant from RD1 on the chromosome, indicates that phoT may also be contributing to the attenuation of BCG. Comparative DNA analysis of M. bovis PhoT shows that it is essentially identical to M. tuberculosis PhoT and encodes an ATP-binding cassette (ABC) protein with a probable role in phosphate transport (Cole et al., 1998). Bacteria generally have an active phosphate transport system that includes two membrane-spanning proteins, a substratebinding protein and an ABC protein designated PstB which hydrolyses ATP and provides the energy required for transport (van Veen, 1997; Linton & Higgins, 1998). Normally, these proteins are in a single operon or at the same locus on the genome. The M. tuberculosis complex is unusual in having three putative pst operons that between them contain three phosphate-binding proteins and four membrane-spanning proteins (Braibant et al., 2000) and it is thought that these multiple copies might enable subtle biochemical adaptations of these organisms to different phosphate-limiting conditions during the infectious cycle (Sarin et al., 2001). pstB is the only gene encoding an ABC protein in these operons, but phoT, which is 130 kb from pstB on the chromosome, is also an ABC protein gene and Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 Microbiology 149 M. bovis phoT in phosphate transport and virulence PhoT has higher homology to PstB in some other organisms (http://genolist.pasteur.fr/TubercuList) than does PstB of M. tuberculosis. In particular, PhoT has very high homology to what appear to be PstB-encoding genes surrounded by other phosphate transport genes in the as yet unannotated, unfinished genome sequences of both M. smegmatis and M. avium (http://www.ncbi.nlm.nih.gov/sutils/genom_table. cgi?). Comparison of the two M. smegmatis sequences in TIGR_1772 and AAC15686 to each other and to the PstB proteins in other organisms indicates that they are almost certainly from the same gene but that the earlier sequence (AAC15686) is translated from a gene sequence that has a small number of sequencing errors (comparisons not shown). To investigate whether PhoT has a role in phosphate transport, the five strains of this study were cultured in decreasing concentrations of phosphate. Results of this work showed that growth at lowered phosphate concentrations required the presence of the wild-type principal sigma factor SigA but that, even with this factor present, growth did not occur at the lowest phosphate concentration tested without the presence of an active PhoT. Recently, it was shown that disruption of the pst operon in M. smegmatis reduced its phosphate uptake ability and doubled its sensitivity to the fluoroquinolone ciprofloxacin (Bhatt et al., 2000). This group also showed that resistance to ciprofloxacin involved an efflux system (Banerjee et al., 2000). Testing of the sensitivity of the strains used in this study to ciprofloxacin similarly showed that inactivation of phoT in ATCC 35723 doubled its sensitivity as well as reducing its ability to scavenge phosphate. Bhatt et al. (2000) concluded that there is strong evidence that pstB plays a role in phosphate importing as well as having a second role in exporting. In this study, PhoT also appears to have two different functions: a phosphate transporting activity at low phosphate levels requiring SigA and a virulence role in the presence of either SigA or, for some animal models, also SigA-R515H. While the ciprofloxacin results and analogy to the M. smegmatis system suggest that this virulence role may involve the export of one or more substances, this remains to be determined. At this stage, these conclusions on the role of PhoT in M. bovis must be accepted with caution as the determination of growth in low phosphate concentrations measures an in vitro phenotype and virulence is an in vivo phenotype. In better-studied pathogenic bacteria, it has been shown that uptake of phosphate has its own regulatory system and that the relationship of this regulation with virulence is only partly understood (Rao et al., 2003; von Kruger et al., 1999). An additional complicating factor in interpreting these results in the broader context of other species of the M. tuberculosis complex is the very recent finding that pstB in M. bovis, including the strains used in this study, has a frameshift mutation that would be expected to affect the expression of a functional PstB. Since PstB and PhoT may have functionally overlapping roles and pstB is not frameshifted in M. tuberculosis, it is possible that the effects on either virulence or phosphate scavenging that occur when phoT is inactivated in M. bovis may not occur in M. tuberculosis. http://mic.sgmjournals.org Virulence of pathogenic organisms is a complex phenotype that depends on the function of many genes in the pathogen as well as on a multitude of responses by the infected host. In the case of tuberculosis, the importance of the disease has encouraged the development and use of many different animal models. No single model exactly replicates the situation in humans or in important economic species such as cattle, but different models have advantages for different studies. Mouse models have been especially useful in immunological studies of tuberculosis but gave less clear-cut results for the strains used in this study than did the other models. In particular, ATCC 35721 grew to higher numbers in mice than did ATCC 35723, even though this latter strain is much more virulent that ATCC 35721 in guinea pigs and possums. These results emphasize the caution that should be exercised in using a single animal model to study gene effects on a phenotype as complex as virulence even when isogenic strains are being used. In this study, the susceptibility of different animals to different strains of M. bovis enabled the discovery of the virulence role of phoT, the likely separation of this role from its phosphate-scavenging ability, and the particular importance of this gene for the virulence of M. bovis in a natural wildlife host. These discoveries illustrate the advantages of using different animal models as tools for the molecular biological investigation of tuberculosis virulence. ACKNOWLEDGEMENTS This study was supported by a grant from the New Zealand Foundation of Research Science and Technology. REFERENCES Aldwell, F. E., Tucker, I. G., de Lisle, G. W. & Buddle, B. M. (2003). Oral delivery of Mycobacterium bovis BCG in a lipid formulation induces resistance to pulmonary tuberculosis in mice. Infect Immun 71, 101–108. Balasubramanian, V., Wiegeshaus, E. H. & Smith, D. W. (1992). Growth characteristics of recent sputum isolates of Mycobacterium tuberculosis in guinea pigs infected by the respiratory route. Infect Immun 60, 4762–4767. Banerjee, S. K., Bhatt, K., Misra, P. & Chakraborti, P. K. (2000). Involvement of a natural transport system in the process of effluxmediated drug resistance in Mycobacterium smegmatis. Mol Gen Genet 262, 949–956. Barry, C. E. (2001). Interpreting cell wall ‘virulence factors’ of Mycobacterium tuberculosis. Trends Microbiol 9, 237–241. Bhatt, K., Banerjee, S. K. & Chakraborti, P. K. (2000). Evidence that phosphate specific transporter is amplified in a fluoroquinolone resistant Mycobacterium smegmatis. Eur J Biochem 267, 4028–4032. Braibant, M., Gilot, P. & Content, J. (2000). The ATP binding cassette (ABC) transport systems of Mycobacterium tuberculosis. FEMS Microbiol Rev 24, 449–467. Cole, S. T. (2002). Comparative and functional genomics of the Mycobacterium tuberculosis complex. Microbiology 148, 2919–2928. Cole, S. T., Brosch, R., Parkhill, J. & 38 other authors (1998). Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 393, 537–544. Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 3211 D. M. Collins and others Collins, D. M., Kawakami, R. P., de Lisle, G. W., Pascopella, L., Bloom, B. R. & Jacobs, W. R. (1995). Mutation of the principal protein-deficient guinea pigs against respiratory challenge with virulent Mycobacterium tuberculosis. Infect Immun 50, 555–559. sigma factor causes loss of virulence in a strain of the Mycobacterium tuberculosis complex. Proc Natl Acad Sci U S A 92, 8036–8040. North, R. J., Ryan, L., LaCource, R., Mogues, T. & Goodrich, M. E. (1999). Growth rate of mycobacteria in mice as an unreliable Collins, D. M., Wilson, T., Campbell, S., Buddle, B. M., Wards, B. J., Hotter, G. & de Lisle, G. W. (2002). Production of avirulent mutants of Mycobacterium bovis with vaccine properties by the use of illegitimate recombination and screening of stationary phase cultures. Microbiology 148, 3019–3027. indication of mycobacterial virulence. Infect Immun 67, 5483–5485. Pascopella, L., Collins, F. M., Martin, J. M., Lee, M. H., Hatfull, G. F., Stover, C. K., Bloom, B. R. & Jacobs, W. R., Jr (1994). Use of in vivo complementation in Mycobacterium tuberculosis to identify a genomic fragment associated with virulence. Infect Immun 62, 1313–1319. Dannenberg, A. M., Jr & Collins, F. M. (2001). Progressive pulmonary Pfeffer, A., Buddle, B. M. & Aldwell, F. E. (1994). Tuberculosis in the tuberculosis is not due to increasing numbers of viable bacilli in rabbits, mice and guinea pigs, but is due to a continuous host response to mycobacterial products. Tuberculosis 81, 229–242. brushtail possum (Trichosurus vulpecula) after intratracheal inoculation with a low dose of Mycobacterium bovis. J Comp Pathol 111, 353–363. Darzins, E. (1958). The Bacteriology of Tuberculosis. Minneapolis, Rao, P. S. S., Lim, T. M. & Leung, K. Y. (2003). Functional genomics MN: University of Minnesota Press. de Lisle, G. W., Mackintosh, C. G. & Bengis, R. G. (2001). approach to the identification of virulence genes involved in Edwardsiella tarda pathogenesis. Infect Immun 71, 1343–1351. Mycobacterium bovis in free-living and captive wildlife, including farmed deer. Rev Sci Tech 20, 86–111. Sarin, J., Aggarwal, S., Chaba, R., Varshney, G. C. & Chakraborti, P. K. (2001). B-subunit of phosphate-specific transporter from de Mendonca-Lima, L., Bordat, Y., Pivert, E., Recchi, C., Neyrolles, O., Maitournam, A., Gicquel, B. & Reyrat, J. M. (2003). Mycobacterium tuberculosis is a thermostable ATPase. J Biol Chem 276, 44590–44597. The allele encoding the mycobacterial Erp protein affects lung disease in mice. Cell Microbiol 5, 65–73. Skinner, M. A., Keen, D. L., Parlane, N. A., Yates, G. F. & Buddle, B. M. (2002). Increased protection against bovine tuberculosis in the Fleischmann, R. D., Alland, D., Eisen, J. A. & 23 other authors (2002). Whole-genome comparison of Mycobacterium tuberculosis brushtail possum (Trichosurus vulpecula) when BCG is administered with killed Mycobacterium vaccae. Tuberculosis 82, 15–22. clinical and laboratory strains. J Bacteriol 184, 5479–5490. Steyn, A. J. C., Collins, D. M., Hondalus, M. K., Jacobs, W. R., Kawakami, R. P. & Bloom, B. R. (2002). Mycobacterium tuberculosis Garnier, T., Eiglmeier, K., Camus, J-C. & 19 other authors (2003). The complete genome sequence of Mycobacterium bovis. Proc Natl Acad Sci U S A 100, 7877–7882. WhiB3 interacts with RpoV to affect host survival but is dispensable for in vivo growth. Proc Natl Acad Sci U S A 99, 3147–3152. Glickman, M. S. & Jacobs, Jr, W. R. (2001). Microbial pathogenesis van Soolingen, D., Hermans, P. W. M., de Haas, P. E. W., Soll, D. R. & van Embden, J. D. A. (1991). Occurrence and stability of insertion of Mycobacterium tuberculosis: dawn of a discipline. Cell 104, 477–485. Higgins, C. F. (2001). ABC transporters: physiology, structure and mechanism – an overview. Res Microbiol 152, 205–210. Jacobs, W. R., Barrett, J. F., Clark-Curtiss, J. E. & Curtiss, R., III (1986). In vivo repackaging of recombinant cosmid molecules for analyses of Salmonella typhimurium, Streptococcus mutans, and mycobacterial genomic libraries. Infect Immun 52, 101–109. Kaushal, D., Schroede, B. G., Tyagi, S. & 8 other authors (2002). Reduced immunopathology and mortality despite tissue persistence in a Mycobacterium tuberculosis mutant lacking alternative sigma factor, SigH. Proc Natl Acad Sci U S A 99, 8330–8335. Kent, P. T. & Kubica, G. P. (1985). Public Health Mycobacteriology, a Guide for the Level III Laboratory. Atlanta, GA: US Department of Health and Human Service, Centres for Disease Control. Lewis, K. N., Liao, R., Guinn, K. M., Hickey, M. J., Smith, S., Behr, M. A. & Sherman, D. R. (2003). Deletion of RD1 from Mycobacterium tuberculosis mimics bacille Calmette-Guerin attenuation. J Infect Dis 187, 117–123. sequences in Mycobacterium tuberculosis complex strains: evaluation of an insertion-sequence dependent DNA polymorphism as a tool in the epidemiology of tuberculosis. J Clin Microbiol 29, 2578–2586. van Veen, H. W. (1997). Phosphate transport in prokaryotes: molecules, mediators and mechanisms. Antonie van Leeuwenhoek 72, 299–315. von Kruger, W. M., Humphreys, S. & Ketley, J. M. (1999). A role for the PhoBR regulatory system homologue in the Vibrio cholerae phosphate-limitation response and intestinal colonization. Microbiology 145, 2463–2475. Wards, B. J. & Collins, D. M. (1996). Electroporation at elevated temperatures substantially improves transformation efficiency of slow-growing mycobacteria. FEMS Microbiol Lett 145, 101–105. Wards, B. J., de Lisle, G. W. & Collins, D. M. (2000). An esat6 knockout mutant of Mycobacterium bovis produced by homologous recombination will contribute to the development of a live tuberculosis vaccine. Tubercle Lung Dis 80, 185–189. Wassenaar, T. M. & Gaastra, W. (2001). Bacterial virulence: can we Linton, K. J. & Higgins, C. F. (1998). The Escherichia coli ATP- draw the line? FEMS Microbiol Lett 201, 1–7. binding cassette (ABC) proteins. Mol Microbiol 28, 5–13. Wilson, T. M., de Lisle, G. W. & Collins, D. M. (1995). Effect of inhA McMurray, D. N., Carlomagno, M. A., Mintzer, C. L. & Tetzlaff, C. L. (1985). Mycobacterium bovis BCG vaccine fails to protect and katG on isoniazid resistance and virulence of Mycobacterium bovis. Mol Microbiol 15, 1009–1015. 3212 Downloaded from www.microbiologyresearch.org by IP: 88.99.165.207 On: Sat, 13 May 2017 04:15:59 Microbiology 149