INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Burkholderia pseudomallei, the causative agent of melioidosis, is a gram-negative, facultatively anaerobic, motile bacillus that is commonly found in the soil and stagnant waters in Southeast Asia and northern Australia. Infection by B. pseudomallei is often due to either direct inoculation into wounds and skin abrasions or to inhalation of contaminated material (11, 24, 30). This would explain the prevalence of the disease among rice farmers as well as helicopter pilots in the Vietnam War who developed melioidosis due to inhalation of contaminated dust (24, 47). Melioidosis may present as an acute pneumonia or an acute septicemia, which is the most severe form of the disease. The disease may also manifest as a chronic infection involving long-lasting suppurative abscesses in numerous sites in the body. Infection with B. pseudomallei may even result in a subclinical infection and remain undetected for a number of years. Both the chronic and subclinical forms generally remain undiagnosed until activated by a traumatic event or a decrease in immunocompetence (25).

Both secreted and cell-associated antigens have been identified in B. pseudomallei. Cell-associated antigens include exopolysaccharide (EPS) and lipopolysaccharide (LPS) (5, 8, 51). The EPS produced by B. pseudomallei is an unbranched polymer of repeating tetrasaccharide units with the structure -3)- 2-O-acetyl--D-Galp-(1-4)--D-Galp-(1-3)--D-Galp-(1-5)--D- KDOp-(2- (35, 40). The role of EPS in virulence is unknown, but sera from patients with melioidosis have been shown to contain antibodies against EPS (51). The LPS of B. pseudomallei has been reported to contain two types of O-polysaccharide moieties termed type I O-PS and type II O-PS (27, 41). Type II O-PS is an unbranched heteropolymer with repeating D-glucose and L-talose residues with the structure -3)--D-glucopyranose-(1-3)-6-deoxy--L-talopyranose-(1-, in which approximately 33% of the talose residues contain 2-O-methyl and 4-O-acetyl substituents, while the other L-talose residues contain only 2-O-acetyl substituents. Type II O-PS has been shown to be involved in serum resistance (17). Mutants lacking in type II O-PS were found to be sensitive to the bactericidal activities of 30% normal human serum. Type II O-PS mutants also demonstrated reduced virulence in three animal models of B. pseudomallei infection (17). Type I O-PS is an unbranched homopolymer with the structure -3)-2-O-acetyl-6-deoxy--D-manno-heptopyranose-(1-. The role for this polysaccharide in infection was previously undefined.

B. thailandensis is a nonpathogenic soil organism originally isolated in Thailand (6). Based on biochemical, immunological, and genetic data, B. pseudomallei and B. thailandensis are closely related species. However, these two organisms differ in a number of ways and have been classified into two different species (7). The rRNA sequence of B. thailandensis differs from that of B. pseudomallei by 15 nucleotides, and there are significant differences in genomic macrorestriction patterns between these organisms (10). The biochemical profiles of these two species differ in that B. thailandensis can utilize L-arabinose whereas B. pseudomallei does not (7, 62). The most distinct difference between these two species, however, is their relative virulence. The 50% lethal dose (LD₅₀) for B. pseudomallei in the Syrian hamster model of acute melioidosis is <10 organisms, whereas the LD₅₀ for B. thailandensis is approximately 10⁶ organisms (7). It has also been shown that the two species can be differentiated based on their propensity to cause disease in humans. Environmental strains isolated in Thailand that are able to assimilate L-arabinose are not associated with human infection, whereas clinical isolates are not able to utilize L-arabinose (54).

To identify the genetic determinants that confer enhanced virulence in B. pseudomallei, a method combining subtractive hybridization, insertional mutagenesis, and animal virulence studies was developed. The described method should aid in the identification of virulence factors in pathogenic bacteria and provide further insights into microbial diversity and evolution.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. The bacterial strains and plasmids used in this study are described in Table 1. B. pseudomallei, B. thailandensis, B. cepacia, and Escherichia coli were grown at 37°C on Luria-Bertani (LB) broth base (Becton Dickinson) agar plates or in LB broth. B. mallei was grown at 37°C on LB plates or in LB broth supplemented with 4% glycerol and at pH 6.8. For animal studies, B. pseudomallei and B. thailandensis cultures were grown at 37°C in TSBDC medium (6). When appropriate, antibiotics were added at the following concentrations: 50 µg of tetracycline, 100 µg of streptomycin, 100 µg of polymyxin B, 100 µg of trimethoprim, 25 µg of gentamicin, and 25 µg of kanamycin per ml for B. pseudomallei and 100 µg of ampicillin, 25 and 50 µg of kanamycin, 15 µg of tetracycline, and 1.5 mg of trimethoprim per ml for E. coli.

Construction and screening of subtractive hybridization libraries. Subtractive hybridization between B. pseudomallei and B. thailandensis was carried out using a PCR-Select bacterial genome subtraction kit (Clontech) as recommended by the manufacturer except that the hybridization temperature was increased from 63°C to 73°C due to the high G+C content in the genomes of these species. In construction of the subtractive hybridization library, B. pseudomallei genomic DNA was used as the tester and B. thailandensis genomic DNA was used as the driver. The secondary PCR products obtained were cloned into pZErO-2.1 (Invitrogen) and pPCR (Table 1) and were enriched for B. pseudomallei-specific sequences. The subtraction library was screened by sequencing of the tester-specific DNA fragments. The library containing random clones was diluted in sterile phosphate-buffered saline (PBS) to 10⁶, and 100 µl was plated on LB plates containing kanamycin (50 µg/ml) and 1 mM isopropylthio--D-galactoside (IPTG). Individual colonies were picked and grown overnight at 37°C in LB with kanamycin (50 µg/ml). Plasmid DNA was isolated using a miniprep plasmid isolation kit (Qiagen).

DNA sequencing and analysis. Automated DNA sequencing was performed by ACGT (Northbrook, Ill.) and the University of Calgary Core DNA Services. The M13 forward primer (dGTAAAACGACGGCCAGT) was used to initiate sequence reactions with the subtractive hybridization clones. DNA flanking the Tn5-OT182 insertions was sequenced using the previously described primers OT182-LT and OT182-RT (16). The DNA flanking the insertion of pSR1015 was sequenced using the pSKM11 primer (38). DNA and protein sequences were analyzed with DNASIS for IBM and with the ORF Finder program at the National Center for Biotechnology Information (NCBI). DNA sequences were analyzed for homology using the BLASTX program through GenBank at NCBI.

Cloning of a subtractive hybridization product and mobilization into wild-type B. pseudomallei. The DNA insert from pDD1015 was cloned as a KpnI-XhoI fragment into a mobilizable suicide vector, pSKM11 (Table 2). The 373-bp fragment was ligated to pSKM11 digested with the same enzymes to create pSR1015. SM10(pSR1015) was conjugated with B. pseudomallei 1026b using a previously described protocol (16).

Animal studies. The animal model of acute B. pseudomallei infection has been previously described (18). Syrian hamsters (females, 6 to 8 weeks) were injected intraperitoneally with 100 µl of one of a number of serial dilutions of logarithmic-phase cultures in sterile PBS. The five control animals were inoculated with 10¹ CFU of wild-type B. pseudomallei. The test animals (five per dilution) were inoculated with either 10¹, 10², or 10³ CFU of the mutant strain, SR1015. Blood from two of the test animals was diluted and plated on Ashdown medium with and without the addition of tetracycline (50 µg/ml) to verify the stability of pSR1015 (7). For determination of LD₅₀ for SR1015, hamsters (five per group) were inoculated with 10³, 10⁴, 10⁵, and 10⁶ CFU. After 48 h, the LD₅₀ was calculated (42).

Immunoassays. Immunogold electron microscopy was performed as previously described (17). Samples for Western blot analysis were prepared as previously described (8). Immunoassay was performed with a 1:250 dilution of the primary antibody, polyclonal rabbit antiserum raised to a B. pseudomallei O-PS-flagellin protein conjugate (5, 8). The secondary antibody used was horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin G (IgG) (Sigma). Enzyme-linked immunosorbent assays (ELISAs) for the presence of EPS were carried out as previously described, using the EPS-specific monoclonal antibody 3015 at a dilution of 1/100 (17, 51, 52). Tn5-OT182 mutants were screened by ELISA according to the same protocol with rabbit polyclonal sera specific for an O-PS-flagellin conjugate containing antibodies to type I O-PS, type II O-PS, and flagella (5).

Southern hybridization. For Southern hybridization, SstI digests of genomic DNA from B. pseudomallei 1026b and SR1015, B. thailandensis E264, B. mallei NCTC 10260, B. cepacia CEP509 (genomovar I) and K56-2 (genomovar III), B. stabilis LMG14294 and LMG7000, B. vietamiensis LMG10929, and B. multivorans C5393 were transferred to GeneScreen Plus membranes (Du Pont Canada, Lachine, Quebec, Canada), and hybridization was performed at 65°C in 15 ml of 1% sodium dodecyl sulfate (SDS)-10% dextran sulfate-salmon sperm DNA (0.1 mg/ml) according to the manufacturer's recommendations. The 0.4-kb KpnI-XhoI fragment from pDD1015 was used as a probe and labeled with [³²P]dCTP using an oligonucleotide labeling kit (Pharmacia Biotech, Inc., Baie d'Urfe, Quebec, Canada).

Tn5-OT182 mutagenesis and screening for type I O-PS mutants. To screen for mutants deficient in type I O-PS, it was first necessary to create a strain that produced only type I O-PS. This was necessary as the antiserum available was polyclonal antiserum to a flagellin-O-PS conjugate that contains antibodies to flagellin, type I O-PS, and type II O-PS. Therefore, we constructed a strain that was lacking in type II O-PS and flagella, SR1001. Allelic exchange was carried out using two strains that were previously constructed in the laboratory. The donor strain, SM10pir(pPB611::Gm), has a plasmid containing a copy of the wbiE gene, involved in the synthesis of type II O-PS, which has been mutated by the insertion of a gentamicin resistance (Gm) cassette (17). The recipient strain, PB401, is a B. pseudomallei strain that has a deletion in the fliC gene. SM10pir(pPB611::Gm) was conjugated to PB401, and transconjugants were selected for by plating on LB containing gentamicin, kanamycin, and polymyxin B. Transconjugants that were Sm^r, Gm^r, and Km^s were selected (to select for loss of the vector, pKAS46), and one was designated SR1001. Transposon mutagenesis of SR1001 was performed with Tn5-OT182 according to a previously described protocol (16) except that transconjugants were selected on plates containing gentamicin and tetracycline. Transposon mutants were inoculated into 96-well plates containing 200 µl of LB with gentamicin and tetracycline and grown overnight at 37°C at 250 rpm. A negative growth control well was included for each plate. The wells of a 96-well plate were coated with 10 µl of bacteria and 90 µl of coating buffer (0.05 M carbonate buffer [pH 9.6]), and ELISA was carried out as previously described (5). The primary antibody, polyclonal rabbit antiserum to a B. pseudomallei O-PS-flagellin protein conjugate, was added at a dilution of 1:1,000. The secondary antibody, a goat anti-rabbit IgG-peroxidase conjugate (Sigma), was added at a dilution of 1:1,000. The plates were developed with an HRP color development reagent (Bio-Rad Laboratories) for 30 min. The optical density at 405 nm (OD₄₀₅) was determined using an ELISA reader. B. pseudomallei 1026b was included as a positive control, and E. coli DH5 was included as a negative control. Transposon mutants that had OD₄₀₅ readings comparable to the negative control (OD₄₀₅ = <0.100) were chosen for further analysis.

Construction of allelic exchange mutants. Allelic exchange was carried out as previously described (17). The allelic exchange vector used in these experiments was pKAS46, an allelic exchange vector based on rpsL for counterselection (49). B. pseudomallei DD503, a double mutant that contains the amrR-oprA and rpsL mutations, was the recipient strain used for all allelic exchange experiments (17, 39). All genes in these experiments were mutated by the insertion of a self-cloning trimethoprim resistance (Tp) cassette from p34EoriTp (P. J. Brett, D. DeShazer, and D. E. Woods, unpublished data). For each allelic exchange experiment, SM10pir transformed with pKAS46 containing the mutated allele was conjugated to B. pseudomallei DD503 as described above for the construction of SR1001 except that the transconjugants were plated on polymyxin B, kanamycin, and trimethoprim. The Pm^r Km^r Tp^r transconjugants were subsequently transferred to plates containing streptomycin to select for the loss of pKAS46. Mutant alleles were confirmed by self-cloning and sequencing.

DNA manipulation. Restriction enzymes and T4 DNA ligase were purchased from Life Technologies (Burlington, Ontario, Canada) and New England Biolabs (Mississauga, Ontario, Canada) and used according to the manufacturer's instructions. DNA fragments used in cloning procedures were excised from agarose gels and purified using a GeneClean II kit (Bio 101, Vista, Calif.) or Qiagen (Mississauga, Ontario, Canada) gel extraction kit. Chromosomal DNA was isolated using a previously described protocol (60). The self-cloning of B. pseudomallei flanking DNA from Tn5-OT182 mutants and from SR1015 was performed as described previously (16).

Nucleotide sequence accession number. The nucleotide sequence reported in this paper has been deposited in the GenBank database under accession number AF228583.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction and screening of the B. pseudomallei subtraction library. Subtractive hybridization was carried out between the virulent B. pseudomallei and the weakly virulent B. thailandensis in order to isolate DNA sequences encoding for virulence determinants unique to B. pseudomallei. The genomic DNA sample from B. pseudomallei containing the sequences of interest was known as the tester DNA, and genomic DNA from B. thailandensis, the reference sample, was called the driver DNA. Tester and driver DNAs were digested and subjected to two rounds of hybridization. The remaining unhybridized sequences were considered tester-specific sequences. To enrich for tester-specific sequences, excess driver DNA was added in the hybridizations. The tester-specific sequences were then amplified by PCR and cloned into pPCR or pZErO-2.1 (Table 1).

Insertional inactivation of the glycosyltransferase gene in wild-type B. pseudomallei. The 373-bp DNA insert from pDD1015 was cloned into a mobilizable suicide vector, pSKM11 (Table 1). The resulting plasmid, pSR1015, was mobilized into wild-type B. pseudomallei 1026b to create the mutant strain SR1015. Since the insert from pDD1015 was found to demonstrate homology to a glycosyltransferase from P. aeruginosa, it was postulated that it might encode a protein involved in carbohydrate synthesis.

View larger version (91K): [in this window] [in a new window]   FIG. 1.   Immunogold electron microscopy of B. pseudomallei 1026b (A) and SR1015 (B), B. thailandensis E264 (C), and B. stabilis LMG7000 (D). Bacteria were reacted with polyclonal rabbit antiserum directed against an O-PS-flagellin protein conjugate absorbed with B. thailandensis E264 to remove the antibodies directed against type II O-PS, washed, and reacted with a goat anti-rabbit IgG-gold (5 nm) conjugate. Original magnification, ×30,000.

View larger version (85K): [in this window] [in a new window]   FIG. 2.   Western blot analysis of LPS isolated from B. pseudomallei 1026b and SR1015 and B. thailandensis E264. Bacteria were reacted with proteinase K, subjected to SDS-polyacrylamide gel electrophoresis, electroblotted, and reacted with polyclonal rabbit antiserum raised to an O-PS-flagellin protein conjugate from B. pseudomallei. Lane M, prestained protein molecular weight standards (New England Biolabs); lane 1, B. pseudomallei 1026b; lane 2, B. pseudomallei SR1015; lane 3, B. thailandensis E264. The apparent molecular masses of the prestained proteins are indicated.

SR1015 is avirulent in the animal model of infection. SR1015 was tested for virulence in the Syrian hamster model of acute septicemic melioidosis. The LD₅₀ for SR1015 after 48 h was 3.5 × 10⁵ CFU, while the LD₅₀ of the parent strain, 1026b, was <10 CFU. The LD₅₀ for SR1015 was similar to that for the weakly virulent B. thailandensis (6.8 × 10⁵ CFU). This demonstrates that SR1015 is severely attenuated for virulence in this animal model of melioidosis and that type I O-PS is a major virulence determinant of B. pseudomallei.

Cloning and sequencing of the genetic loci required for type I O-PS production and export. Two methods were used to clone the genes involved in the production and export of type I O-PS. The DNA flanking the insertion of pSR1015 was cloned from SR1015 and sequenced. We also used transposon mutagenesis to clone the genes involved in production of the polysaccharide; this was done to obtain any unlinked genes that may be involved in polysaccharide production. Approximately 1,300 transposon mutants were screened for loss of type I O-PS by ELISA. Six mutants were identified, and the DNA flanking the transposon insertion was cloned and sequenced. The Tn5-OT182 mutants SLR5, SLR8, SLR13, SLR18, and SLR19 mapped to the same region of the chromosome (Fig. 3). Sequence analysis of the cloned fragments revealed the presence of 20 potential open reading frames involved in the synthesis and export of type I O-PS (Fig. 3). The open reading frames that predicted proteins involved in polysaccharide biosynthesis were found to demonstrate homology to proteins involved in the synthesis of a polysaccharide structure composed primarily of mannose (Table 3). The other reading frames in the locus predicted proteins involved in the transport of capsular polysaccharides in a variety of bacteria, particularly those that produce group 2 and group 3 capsular polysaccharides (Table 3 and reference 59). The genes responsible for the production of type I O-PS was found to be similar to other loci encoding for capsular polysaccharides in that they are divergently transcribed (Fig. 3 and reference 44). The gene cluster involved in the production of this polysaccharide is also similar to group 3 capsule gene clusters in that there are no genes encoding KpsF and KpsU, which are present in group 2 capsule gene clusters (59). However, the organization of the B. pseudomallei type I O-PS gene cluster differs in that it does not contain two export regions flanking a single biosynthetic region as seen in other group 3 capsule polysaccharide clusters (12). The biosynthetic genes identified thus far are not organized into one continuous transcriptional unit; instead, wcbB, manC, and wcbP are separated from the rest of the biosynthetic genes. Another interesting feature is that kpsC is usually found next to kpsS in other group 2 and 3 clusters, unlike the case for wcbA and wcbO in B. pseudomallei (Fig. 3 and reference 59). The promoter sequences of the transcriptional regions of the type I O-PS cluster have yet to be identified. The overall G+C content of this region is about 58%, lower than the G+C content of the rest of the chromosome (68%). The low G+C content in these clusters suggests that polysaccharide genes have a common origin and may have been transferred horizontally between species (21).

View larger version (7K): [in this window] [in a new window]   FIG. 3.   Organization of the chromosomal region containing the genes responsible for the synthesis and export of type I O-PS in B. pseudomallei. The upper part shows the locations of the genes. The direction of transcription is represented by arrows, and gene names are indicated. The locations of Tn5-OT182 insertions are represented by triangles. Mutants constructed by allelic exchange are shown. The straight line indicates insertion of the Tp cassette into the gene of interest. The horizontal line below the genetic map represents B. pseudomallei chromosomal DNA; the locations of relevant restriction endonuclease recognition sites (Bg, BglII; E, EcoRI; H, HindIII; K, KpnI; B, BamHI) are shown.

The type I O-PS is also present in B. mallei and the B. cepacia complex but not in B. thailandensis. Southern blot analysis using a probe containing the A-T-rich glycosyltransferase fragment from pDD1015 confirmed that the fragment was present in B. pseudomallei 1026b and SR1015 but not in B. thailandensis (Fig. 4). B. mallei and the B. cepacia complex were also tested for the presence of this fragment. It was found that the probe hybridized to an SstI fragment in B. mallei and the B. stabilis (formerly genomovar IV) strains LMG7000 and LMG14294. The B. cepacia complex has recently been divided into two genomovars (B. cepacia genomovar I and genomovar III) and three species (B. multivorans, B. vietnamiensis, and B. stabilis) (55, 56). None of the strains tested from B. cepacia genomovars I and III, B. multivorans, or B. vietnamiensis were found to contain this DNA fragment. Southern blot analysis was carried out on five other B. stabilis strains: CEP0717, CEP0467, J687, CEP0726, and LMG14291. All of the B. stabilis strains tested hybridized to the probe from pDD1015 (data not shown). The presence of type I O-PS was confirmed in the B. stabilis strain LMG7000 by immunoelectron microscopy. As seen in Fig. 1D, B. stabilis LMG7000 showed reactivity to the type I O-PS antibodies, but lacked a uniform distribution of the polysaccharide on the cell surface, and therefore appears to produce less of this polysaccharide than B. pseudomallei 1026b.

View larger version (118K): [in this window] [in a new window]   FIG. 4.   Southern hybridization analysis of genomic DNA from Burkholderia spp. digested with SstI. A 0.4-kb KpnI-XhoI fragment from pDD1015 was used as a probe. Lane 1, B. mallei NCTC 10260; lane 2, B. pseudomallei 1026b; lane 3, B. pseudomallei SR1015; lane 4, B. thailandensis E264; lane 5, B. vietnamiensis LMG10929; lane 6, B. cepacia CEP509 (genomovar I); lane 7, B. cepacia K56-2 (genomovar III); lane 8, B. stabilis LMG14294; lane 9, B. stabilis LMG7000; lane 10, B. multivorans C5393.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Although melioidosis is less common outside of Southeast Asia and northern Australia, it may be underdiagnosed in other regions, and it poses a concern due to increased travel and military involvement in regions where the disease is endemic (14). Recently, our attention has been focused on the identification of genetic determinants that contribute to the pathogenesis of B. pseudomallei infections. To obtain virulence determinants unique to B. pseudomallei, we used subtractive hybridization between this organism and a related nonpathogenic organism, B. thailandensis.

Analysis of the subtractive hybridization library revealed that B. pseudomallei contains a number of DNA sequences that are not found in B. thailandensis (Table 2). One of the subtraction clones, pDD1015, demonstrated weak homology to a glycosyltransferase, WbpX, from P. aeruginosa (45). The insert from pDD1015 was cloned into a mobilizable suicide vector, pSKM11, for insertional inactivation of the glycosyltransferase gene in wild-type B. pseudomallei. The resulting strain, SR1015, was markedly less virulent than the parent strain in an animal model. This demonstrated that B. pseudomallei contains DNA sequences encoding for virulence determinants that are not found in B. thailandensis and that the glycosyltransferase gene may encode an important virulence determinant in B. pseudomallei. Using antibodies to type I O-PS, we determined that SR1015 harbored a mutation in a glycosyltransferase gene involved in the production of type I O-PS.

Sequence analysis of the DNA flanking the glycosyltransferase gene revealed the presence of at least 20 open reading frames involved in the synthesis and export of type I O-PS (Fig. 3; Table 3). The genes identified encode for proteins that are similar to proteins involved in the biosynthesis and export of capsular polysaccharides, particularly those involved in the production of group 3 capsular polysaccharides. Group 3 capsules include the E. coli K10 capsule and may also include the H. influenzae group b capsule and the capsule produced by N. meningitidis serogroup B (59). Group 3 capsules are always coexpressed with O serogroups, are not thermoregulated, are transported by an ABC-2 exporter system, and do not contain the kpsU and kpsF genes, and usually the gene clusters map near the serA locus (59). Thus far, no serA locus that is associated with the type I O-PS cluster has been identified, but this polysaccharide is coexpressed with O antigen and lacks the kpsU and kpsF genes, and genes encoding for a putative ABC-2 transporter have been identified. The genes involved in the production of group 3 capsules are organized into regions and are divergently transcribed. Regions 1 and 3 are generally conserved and contain genes involved in export of the polysaccharide. These regions flank region 2, which contains the biosynthetic genes and is not conserved between serotypes (44). The genetic organization of the type I O-PS is also similar to that of other capsule gene clusters in that the genes are organized into more than one transcriptional unit and appear to be divergently transcribed. However, the organization of the B. pseudomallei type I O-PS cluster differs in that the biosynthetic genes identified thus far are not organized into one biosynthetic region. yafJ and yggB are likely not involved in the production of type I O-PS since they have a high G+C content (62 to 65%), they are present in B. thailandensis, and a mutation in yafJ (SR203::Tp) did not reduce virulence in hamsters (data not shown). We are also currently constructing a mutant in the polyketide synthase gene that lies downstream of the wcbP gene in order to define this end of the type I O-PS cluster.

The polysaccharide with the structure -3)-2-O-acetyl-6-deoxy--D-manno-heptopyranose-(1- was originally isolated and characterized as an O-PS component of LPS in B. pseudomallei and was designated type I O-PS (41). However, our results suggest that this polysaccharide is a capsule rather than an O-PS moiety. The genes involved in the production of this capsule demonstrate strong homology to the genes involved in the production of capsular polysaccharides in many organisms, including N. meningitidis, H. influenzae, and E. coli. In addition, the export genes associated with this cluster are not associated with the previously characterized O-PS gene cluster (17). Western blot analysis of proteinase K cell extracts (Fig. 2) and silver staining (data not shown) have shown that this polysaccharide has a high molecular mass (200 kDa) and lacks the banding pattern seen with O-PS moieties. Studies by our laboratory have indicated that mutants in the production of the core oligosaccharide of the LPS are still capable of producing this polysaccharide (9). Based on the above criteria and the genetic similarity to group 3 capsules, we propose that this polysaccharide is a group 3 capsule.

Capsule production has been correlated with virulence in many bacteria, particularly those causing serious invasive infections of humans (4). Our studies have demonstrated that this capsule is critical for the virulence of B. pseudomallei. However, its specific role in infection has yet to be elucidated. A number of functions have been suggested for polysaccharide capsules: prevention of desiccation for transmission and survival, adherence for colonization, resistance to complement-mediated phagocytosis and complement-mediated killing, and resistance to specific host immunity due to a poor antibody response to the capsule (44). Preliminary studies have shown that type I O-PS is not involved in serum resistance. SR1015 was tested for resistance to killing by 30% normal human serum and was found to be resistant to killing (data not shown). Studies to define the role of the capsule in infection are under way.

Genomic DNAs from B. mallei NCTC 10260 and seven strains of B. stabilis were shown to hybridize to the glycosyltransferase probe from pDD1015. Immunoelectron microscopic analysis demonstrated that B. stabilis LMG7000 contained this capsule (Fig. 1). Interestingly, B. stabilis LMG7000 was noted to produce less of this polysaccharide than B. pseudomallei 1026b and lacked a uniform distribution of the polysaccharide on the cell surface. The importance of the capsule in infection by B. stabilis has yet to be elucidated. The results of our study demonstrating the presence of this capsule in B. stabilis corresponds with its recent classification as a novel species (56). This capsule may be an additional tool to aid in the identification of B. stabilis strains.

Virulence genes of a number of pathogenic bacteria are located on pathogenicity islands (PAIs), regions on the bacterial chromosome that are present in the genome of pathogenic strains but rarely present in those of nonpathogenic strains. The PAIs may range in size from about 30 kb to 200 kb and often differ in G+C content from the remaining bacterial genome; the PAIs are often associated with the carriage of many virulence genes. These genetic units are often flanked by direct repeats and may be associated with tRNA genes or insertion sequence (IS) elements at their boundaries. They may also be associated with the presence of mobility genes, such as IS elements, integrases, transposases, and origins of plasmid replication. These DNA regions are considered to be unstable in that they may be subject to deletion with high frequency or undergo duplications and amplifications (23). A number of PAIs have been described for both gram-positive and gram-negative bacteria, and the application of subtraction hybridization has been used to successfully identify such genetic elements (23, 32). The subtractive hybridization that was carried out between B. pseudomallei and B. thailandensis led to the identification of a number of sequences that were found to be A-T rich compared to the rest of the B. pseudomallei chromosome. This, combined with the fact that insertional mutagenesis of the glycosyltransferase gene identified by this method resulted in an avirulent strain, suggests that we may have identified DNA sequences from a putative PAI and that the capsular polysaccharide gene cluster may be located on this island. It is possible that B. pseudomallei, B. mallei, and B. stabilis acquired DNA encoding for capsule as well as other potential, yet unidentified virulence factors by horizontal transfer recently in evolution. B. pseudomallei is known to contain IS elements that are present in B. cepacia but not in B. thailandensis (31). However, IS elements have not yet been identified in association with the capsule gene cluster. Further studies are under way to determine whether a PAI exists in these organisms and whether the capsule gene cluster is located on such a genetic element.

The identification of bacterial virulence genes has traditionally relied on empirical predictions of putative virulence determinants and inactivation of the genes encoding for these putative virulence determinants by any number of methods, followed by comparisons of virulence between mutant and wild-type infection models (19). Tools such as in vivo expression technology and differential fluorescence technology have been developed to facilitate the identification of expressed sequences under a given set of circumstances within a test host; however, these approaches do not necessarily lead to the identification of virulence determinants (53). The method for identification of virulence genes described herein should be applicable to a broad range of pathogenic bacteria. The combination of PCR-based subtractive hybridization, insertional mutagenesis, and an animal infection model provides for the efficient detection of virulence genes. While we have applied the method to the pathogen B. pseudomallei in our current studies, it could be applied to any species and for which only a few prerequisites are in place. These prerequisites include related virulent and avirulent strains, suitable suicide vectors for insertional inactivation, and an infection model for differentiation of virulent and avirulent strains. The described method should lead to the identification of relevant virulence determinants for a number of bacterial species and further the understanding of molecular pathogenesis.