A major and critical virulence determinant of many Gram-negative bacterial pathogens is the Type III Secretion Systems (T3SS). T3SS3 in Burkholderia pseudomallei is critical for bacterial virulence in mammalian infection models but its regulation is unknown. B. pseudomallei is the causative agent of melioidosis, a potentially fatal disease endemic in Southeast Asia and northern Australia. While screening for bacterial transposon mutants with a defective T3SS function, we discovered a TetR family regulator (bspR) responsible for the control of T3SS3 gene expression. The bspR mutant exhibited significant virulence attenuation in mice. BspR acts through BprP, a novel transmembrane regulator located adjacent to the currently delineated T3SS3 region. BprP in turn regulates the expression of structural and secretion components of T3SS3 and the AraC family regulator bsaN. BsaN and BicA likely form a complex to regulate the expression of T3SS3 effectors and other regulators which in turn affect the expression of Type VI Secretion Systems (T6SS). The complete delineation of the bspR initiated T3SS regulatory cascade not only contributes to the understanding of B. pseudomallei pathogenesis but also provides an important example of how bacterial pathogens could co-opt and integrate various regulatory motifs to form a new regulatory network adapted for its own purposes.
Bacterial pathogens sense environmental cues to regulate the expression of their virulence genes. Among the many virulence mechanisms present, the Type III Secretion System (T3SS) has been documented to play a critical role in the virulence of many Gram-negative pathogens (Hueck, 1998). T3SS cluster of genes encode a needle-like apparatus that could deliver bacterial effectors directly into host cell cytosol. Pathogens exert tight control over the expression and secretion of their T3SSs via many layers of hierarchical and cross-regulation (Francis et al., 2002). Only upon encounter with host cells and under appropriate conditions would these systems be triggered. Current knowledge on regulation of T3SS in various pathogens reveals largely different mechanisms of regulation, reflecting the unique environmental niche of each pathogen (Francis et al., 2002; Ellermeier and Slauch, 2007).
Melioidosis, a serious disease endemic in Southeast Asia and northern Australia is caused by the Gram-negative bacillus Burkholderia pseudomallei (Currie et al., 2000; Leelarasamee, 2000). Since its recognition as a potential bioweapon, interest in B. pseudomallei has intensified and reports of melioidosis continue to expand beyond traditional areas of endemicity due to heightened awareness and better diagnostics. Melioidosis can present with varied clinical manifestations ranging from asymptomatic seroconversion to disseminated septicaemic disease involving multiple organs, and septic shock (Chaowagul et al., 1989). The overall mortality for melioidosis remains unacceptably high at about 50% in north-east Thailand and 20% in northern Australia (Cheng et al., 2003; White, 2003). Current understanding of B. pseudomallei pathogenesis is very limited. T3SS3, one of the three T3SS B. pseudomallei has, resembles the SPI-1 of Salmonella and mxi-spa of Shigella (Stevens et al., 2002). T3SS3 has been shown to be important for intracellular replication, endosomal escape (Stevens et al., 2002), the induction of caspase-1-dependent cell death (Sun et al., 2005), virulence in the mouse (Stevens et al., 2004) and hamster model of infection (Warawa and Woods, 2005). Although the T3SS gene clusters in B. pseudomallei have been identified, nothing is known about their regulation.
We believe that understanding how T3SS is regulated would reveal key aspects of B. pseudomallei virulence and pathogenesis. As T3SS3 is absolutely required for the induction of caspase-1-dependent oncotic death in macrophages by B. pseudomallei (Sun et al., 2005), macrophage-killing ability is a good surrogate marker for T3SS3 function. By screening for mutants with decreased cytotoxicity from our library of Himar1 transposon mutants, we identified a TetR family regulator controlling T3SS3 and T6SS5 expression through a regulatory cascade involving intermediate regulators located in or within the vicinity of the T3SS3 locus.
Identification of a TetR family regulator mutant defective in T3SS function and virulence
We screened a large pool of Himar1 transposon mutants of B. pseudomallei for their ability to cause macrophage cytotoxicity. The identified 17B9 mutant showed no defect in growth rate and motility but a 50% attenuated cytotoxicity to murine macrophage RAW264.7 cells (Fig. 1A). The Himar1 insertion site in this mutant corresponds to the 52nd codon of the BPSL1105 (Gene ID: 3093813) gene on chromosome 1. 17B9 also showed attenuated cytotoxicity to human THP-1 cells and mouse peritoneal macrophages compared with wild-type bacteria (Fig. 1A). Complementation of BPSL1105 into 17B9 significantly increased the cytotoxicity to RAW264.7 cells (Fig. 1B). Since a good correlation exists between cytotoxicity and T3SS function, the attenuated cytotoxicity of 17B9 could be due to defective T3SS function. To verify T3SS function, the expression of one of the known T3SS3 effector BopE in the 17B9 mutant was examined by Western blot. BopE was not produced in 17B9 but its production was restored in the BPSL1105-complemented mutant (Fig. 1C). BPSL1105 encodes a putative protein containing an N-terminal DNA-binding domain of TetR regulator family. Hereafter, we refer to this gene as bspR (bsa-specific regulator). We generated two new bspR mutants in two clinical strains of B. pseudomallei, KHW and K96243. Both bspR mutants showed significant decrease in macrophage cytotoxicity as well as BopE production compared with the respective wild-type strain (data not shown). Inactivation of bspR (ΔbspR::zeo) in strain KHW also greatly attenuated its virulence to BALB/c mice (Fig. 1D). All mice infected with a 100 cfu dose of the wild-type bacteria died within 15 days. The mean survival time for wild-type bacteria-infected mice was 11 days. However, the mean survival time of the bspR mutant-infected mice was extended to 30 days, indicating that bspR is important for virulence during in vivo infection.
bspR regulates T3SS3 but not T3SS1
bspR may be affecting T3SS3 function by regulating the expression of T3SS3 genes. We compared the transcriptome of the 17B9 mutant with the wild-type strain KHW grown in RPMI medium using microarray based on the fully annotated genome of strain K96243 (Holden et al., 2004). This microarray covers 100% of the 5728 protein-coding genes in the K96243 genome (details in Experimental procedures). The expression of 38 genes located adjacent and within the T3SS3 locus showed greater than 50% decrease in the mutant compared with that of the wild type (Fig. 2A). In addition, four T6SS5 genes (bimE, virG, virA and tssM) showed greater than 50% decrease in 17B9 mutant in comparison to that of wild-type bacteria (Table S1). To verify the microarray data, we compared the expression of at least one gene from each of the putative operons in T3SS3 between the bspR mutant (ΔbspR::zeo) and wild-type bacteria by real-time PCR (Fig. 2B). Consistent with microarray data, the expression of representative T3SS3 (bprC to bsaM) genes in the ΔbspR::zeo deletion mutant was significantly downregulated (Fig. 2B). In contrast, the expression of genes in T3SS1 (BPSS1397) was unchanged (Fig. 2B) and expression of T3SS2 was undetectable under our growth conditions (data not shown). Microarray experiments were also performed on bacteria cultured in LB medium, and on strain K96243 and its deletion mutant K9bspR with similar results (Table S1), ruling out the effect of culture media and strain specificity on bspR-dependent gene regulation. Taken together, we have identified bspR to be a regulator of T3SS3.
T3SS3 effector genes are regulated through the AraC family regulator bsaN and the putative chaperone bicA
To determine whether the regulation of T3SS3 genes by bspR could be through other T3SS regulators, a putative regulator bsaN was chosen for analysis. bsaN encodes an AraC family regulator with a 55% homology to InvF of Salmonella, which, together with chaperone protein SicA, is required for expression of effector genes in SPI-1 (Darwin and Miller, 2001). In the bsaN mutant (ΔbsaN::FRT), the expression of two known effector genes bopE (Stevens et al., 2003) and bopA (Stevens et al., 2002) was less than 1% of wild-type level (Fig. 2C). In comparison, expression of genes in the translocon operons (bicA, bipB, bprA, bipD) was 15–25% of wild-type level and the expression of genes encoding for structural components (bsaM) was not affected in the bsaN mutant (Fig. 2C). The expression level of remaining genes in the bsaN operon as represented by bsaQ was 67% of wild-type level. Furthermore, the translocon protein BipD was still produced and secreted by the bsaN mutant, but effector protein BopE was not present (Fig. 2F). These data are consistent with the predicted function of BsaN as a transcriptional activator for effector genes. In addition, genes adjacent to bopA and bopE in the T3SS3 locus such as bicP, putative regulator genes bprB and bprC, and hypothetical gene BPSS1521 (bprD) also had highly reduced expression (Fig. 2C). InvF requires a co-activator SicA in Salmonella (Darwin and Miller, 2001). The homologue of SicA in T3SS3 is BicA (78% homology). A bicA mutant (ΔbicA::FRT) showed an almost identical expression and secretion profile as the bsaN mutant (Fig. 2D and F). Thus, both bsaN and bicA are essential for expression of effector genes and several putative regulators but not for structural genes. However, expression of chaperone/translocon genes only partially depends on bsaN/bicA. Since the expression of bsaN and bicA was reduced to less than 0.2% in the bspR mutant compared with wild-type level (Fig. 2B), downregulation of effector genes in the bspR mutant is most likely due to the absence of bsaN and bicA.
bprA, bprB and bprC were predicted to encode regulators of T3SS3 (Stevens et al., 2002). BprD shares weak homology with a sigma-54-dependent transcriptional activator (Holden et al., 2004). The reduced expression of putative regulators bprB, bprD and bprC in the bsaN and bicA mutants could indicate a possible signal relay within the T3SS locus. Deletion of bprA (ΔbprA::FRT) led to 45% decrease of fliC, whereas expression of all the other genes examined was not significantly changed in the culture condition (Fig. 2E). Deletion of bprB and bprC together with bprD (ΔbprBDC::FRT) did not result in significant changes in expression of T3SS3 genes (Fig. 2E) nor secretion (Fig. 2F). Therefore, these putative regulator genes do not play a role in the regulation of T3SS3 expression.
Transmembrane regulator bprP regulates bsaN and all other T3SS3 genes
Besides the known T3SS3 genes, BPSS1553 and BPSS1554 were also downregulated in the 17B9 and bspR mutants (Fig. 2A and B, Table S1). We name BPSS1553 bprP and BPSS1554 bprQ. These genes do not show significant homology to any known T3SS genes. The N-terminal of BprP has a recognizable DNA-binding domain and a C-terminal putative transmembrane (TM) domain (Fig. 3A). This domain organization resembles that of the transmembrane regulator ToxR of Vibrio cholera (Miller et al., 1987). Given its proximity to the T3SS3 gene cluster (BPSS1520–1552) and its co-regulation with T3SS3 genes, it is possible that bprP is involved in the regulation of T3SS3 genes. All the T3SS3 genes examined showed similar degree of downregulation in the bprP mutant (ΔbprP::FRT) as in the bspR mutant (Figs 2B and 3B). For example, structural gene operons (bsaN, bsaM), translocon (bipB, bipD) and effector (bopA, bopE) genes all showed less than 0.5% of wild-type expression level (Fig. 3B). Western blot analysis confirmed that BopE and BipD were not secreted in the bprP mutant but secretion could be restored through complementation of the bprP gene in trans (Figs 2F and 3C). The bprP mutant also lost the ability to kill macrophages (Fig. 3D). The expression of bprP does not require bsaN or bicA because it was not affected in the bsaN and bicA mutants (Fig. 2C and D). Thus, bprP is required for the activation of all components of the T3SS3 secretion complex including translocons and secreted effectors, although activation of effectors is most likely through bsaN and bicA. bprQ expression was co-regulated with bprP in the bspR mutant (Fig. 2A) but the deletion of bprQ did not have a negative impact on the secretion function (Fig. 3C) nor on macrophage cytotoxicity of the bacteria (Fig. 3D), suggesting that bprQ is not essential for the expression and function of T3SS3 under our culture condition. In fact, the bprQ mutant appeared to have the opposite phenotype of the bprP mutant.
BprP binds directly to the bsaN promoter
ToxR binds to the promoter region ToxT and directly activates its expression (Krukonis et al., 2000). Since bprP is required for the expression of bsaN, BprP could be similarly binding to the promoter region of bsaN. Recombinant full-length histidine (His)-tagged BprP was present mainly in the membrane fraction of bacterial lysate at 37 kD (Fig. 4A, left panel). In contrast, a truncated BprP (BprP1–222) without the TM and the C-terminal periplasmic domains was present exclusively in the soluble but not the membrane fraction at 26 kD (Fig. 4A, right panel). This confirms that the putative TM domain is required for BprP membrane localization. Using the promoter region of bsaN as a probe, we tested the DNA binding activity of BprP using the soluble and membrane fractions in the electrophoretic mobility shift assay (EMSA). Membrane fractions containing full-length BprP bound the bsaN promoter probe and retained the labelled probe in the wells and this interaction could be disrupted by adding increasing concentrations of unlabelled probe (Fig. 4B, left panel). Membrane fractions from Escherichia coli expressing BprP1–222 or the E. coli BL21 expression host did not bind to the probe. The BprP1–222 present in the soluble fraction could bind to the bsaN promoter probe despite the absence of the TM and periplasmic domains (Fig. 4B, right panel). To determine whether the DNA binding activity is sufficient for activation of bsaN and bopE expression, we complemented the bprP mutant with plasmids expressing BprP1–222 or the full-length BprP. Only the full-length but not the truncated BprP restored the expression of bsaN and bopE in the bprP mutant (Fig. 4C).
bspR controls T3SS3 through bprP
bprP is required for the expression of all known genes in the bspR regulon and the expression of bprP requires bspR. This strongly suggests that the major role of bspR in the regulation of T3SS3 is to activate the expression of bprP. Introducing bprP or bprPQ in trans into the bspR mutant restored the expression of T3SS3 genes (Fig. 5A) and the secretion of BopE and BipD (Fig. 5B). The lowered expression of BipD and BopE in the bspR mutant complemented with bprPQ suggests that bprQ might play a regulatory function, supported by the enhanced secretion and cytotoxicity in the bprQ mutant (Fig. 3C and D).
bspR cascade regulates T6SS5 through bsaN
The T6SS5 genes of B. pseudomallei are highly homologous and arranged similarly to Burkholderia mallei T6SS locus (BMAA0751–BMAA0727) (Schell et al., 2007; Shalom et al., 2007) so we adopt the nomenclature of Schell et al. to ensure consistent naming of homologues. In B. mallei, overexpression of virAG (BMAA0745–6) or an AraC family regulator BMAA1517 led to increased expression of the T6SS locus (Schell et al., 2007). These genes are highly conserved in B. pseudomallei and corresponded to virAG (BPSS1494–5) and bprC (BPSS1520) respectively. In our microarray data, several genes in the T6SS5 cluster showed greater than 50% reduction in expression level in 17B9 mutant (Table S1). Since the expression of bprC was downregulated to 1–3% of wild-type level in bspR, bprP, bsaN and bicA mutants (Figs 2B–D and 3B), it is possible that the expression of T6SS5 genes is regulated by bspR-dependent signalling cascade.
To determine the genes required for activation of T6SS5 genes in B. pseudomallei, we generated two additional deletion mutants of virAG and bprC and examined expression of three representative T6SS5 genes. Deletion of bspR, bprP or bsaN reduced the expression of T6SS5 genes to less than 8% of wild-type level. In contrast, the virAG deletion did not significantly affect the expression of tssA under the culture condition tested (Fig. 6), similar to the phenotype of virG mutant in B. mallei (Schell et al., 2007). In the bprC mutant, virA and virG expression was downregulated to 30% and 42%, respectively, whereas tssA expression decreased to 4% of wild-type level (Fig. 6).
Bacteria acquire T3SSs through horizontal gene transfer (Hueck, 1998). Once acquired, the expression of these genes needs to be integrated into the existing transcription regulatory network. Thus, similar T3SSs might be regulated through entirely different mechanisms in different genera of bacteria. The T3SS3 of B. pseudomallei plays an important role in its pathogenesis (Stevens et al., 2004) but its regulation is unknown. In this study, we identified a TetR family transcription regulator bspR that controls T3SS3. The TetR family of regulators, with the characteristic helix–turn–helix (HTH) DNA-binding motif, are well represented and widely distributed among Gram-positive and Gram-negative bacteria. They are typically repressors controlling genes whose products are most often involved in multidrug resistance, the biosynthesis of antibiotics, enzymes in catabolic pathways, quorum sensing and biofilm formation (Ramos et al., 2005). Thus, BspR represents a novel function for a TetR regulator that acts as a positive regulator of T3SS. It remains to be seen whether a similar mode of regulation may be employed by other pathogens to regulate their T3SSs.
Many T3SSs contain AraC family regulators that control the expression of the secreted T3SS effectors (Francis et al., 2002). We have shown that BsaN is such a regulator and required for the expression of two known T3SS3 effectors, BopA and BopE, and the putative chaperone, BicP. Similar to InvF of the Salmonella SPI-1 system, the chaperone protein BicA is required as a co-activator for BsaN. Furthermore, bsaN/bicA is required for the expression of three putative regulator genes, namely bprB, bprC and bprD. We have identified bprC to be the regulator of T6SS5 expression. Although bsaN activates the expression of the two-component system virA/virG, they do not activate T6SS5 genes. Thus, bprC but not virAG plays an important role in activating the expression T6SS5 genes downstream of T3SS3 gene activation in our culture condition. However, we could not rule out the possibility that the activation of the two-component system virA/virG requires environmental signals which are not present in our culture condition to function.
InvF is activated directly by transcriptional activator HilA (Lee et al., 1992) and many regulators of SPI-1 control the expression level of HilA directly or indirectly (Ellermeier and Slauch, 2007). However, there are no obvious homologues of HilA in B. pseudomallei. Our microarray data revealed that the bspR regulon includes a novel gene, bprP. BprP has an N-terminal DNA-binding domain belonging to the response regulator OmpR-type winged HTH family (Galperin, 2006) whose members also include the DNA-binding domains of HilA and ToxR. Despite the fact that BprP and ToxR have different C-terminal domains, the similarity in their DNA-binding domains prompted us to test whether BprP activates bsaN in a similar direct fashion as ToxR activates AraC regulator ToxT (DiRita and Mekalanos, 1991). Indeed, bprP is required for the expression of bsaN and BprP binds to the promoter region of bsaN. However, the DNA-binding domain of BprP is not sufficient for gene activation, suggesting that BprP needs to be localized to the membrane in order to activate transcription. Furthermore, bprP is required for the activation of all the genes in the bspR regulon as genes downregulated in the bspR mutant showed a similar degree of downregulation in the bprP mutant. Based on the identified role of bprP, we propose that the current delineation of T3SS3 be extended to include bprPQ. BprP may be able to respond to environmental signals by modulating the strength and specificity of its transcription activity. ToxR can activate ctxAB expression independent of ToxT in the presence of bile acids (Hung and Mekalanos, 2005). It is conceivable that host factors might be able to enhance the activity of the residual BprP expressed in the absence of bspR, consequently leading to partial restoration of T3SS3 in vivo. This scenario would explain the incomplete attenuation of virulence of the bspR mutant in our mouse model. Transmembrane protein ToxS enhances the expression of ToxR-dependent genes (DiRita and Mekalanos, 1991) and BprQ has a putative TM domain similar to ToxS. Thus BprQ is a potential candidate that could modify the activity of BprP in response to unidentified environmental cues.
bprP seems to be the most upstream regulator in the bspR regulon. It suggests that BspR activates the expression of bprP directly. However, we did not manage to demonstrate any interaction between BspR and the bprP promoter probe (data not shown). BspR could be activating bprP through another unknown protein or with the help of a co-factor not present in the expression host. It is also possible that a functional BspR requires post-translational modification. bspR itself may be regulated through as yet unidentified environmental cues and is likely to be coupled to other regulatory circuits in the bacteria. The mean survival time of bspR mutant-infected mice was almost three times that of mice infected with wild-type bacteria. Therefore, bspR is required for full virulence in vivo and it will be important to find out how bspR itself is regulated in vivo.
In summary, the bspR initiated regulatory cascade reveals a novel mechanism involving a TetR regulator located remotely from the T3SS but positively regulating its expression and the subsequent expression of T6SS5 (Fig. 7). Through this study, the major regulators controlling each step of the T3SS3 signalling cascade in B. pseudomallei has been deciphered. BspR, at the top of the cascade, activates BprP through an undefined mechanism. Transmembrane regulator BprP binds to the promoter region of BsaN and activates it similar to the way ToxR regulates ToxT. BprP also activates other genes in the T3SS3 cluster encoding components of secretion complex independent of BsaN and it remains to be seen if this also occurs by direct binding of BprP to their promoter regions. Together with chaperone BicA, BsaN transcriptionally activates the expression of secreted effectors such as BopE, the BprC regulator in T3SS3 and VirA/VirG in T6SS5. BprC in turn regulates the expression of T6SS5. It is still unclear whether the regulation by BsaN and BprC of their downstream genes is modified at the post-transcriptional level by other factors. For example in Shigella, OspD1 and Spa15 act as anti-activators of the T3SS by inhibiting the interaction between MxiE and IpgC (Parsot et al., 2005), which are homologues of BsaN and BicA respectively. However, the description of such a regulatory cascade in this important pathogen provides a framework for further understanding of the complex virulence regulation network in B. pseudomallei. Pharmaceutical intervention of BspR function and other components of this cascade may be an novel treatment strategy for a pathogen with very limited treatment options currently due to its inherent resistance to multiple antibiotics (Cheng et al., 2003).
Bacterial strains and culture conditions
Bacterial strains and plasmids used are listed in Table 1. Mutants created were derived from clinical strains of B. pseudomallei KHW (Liu et al., 2002) and K96243 (Holden et al., 2004). Antibiotics added to LB medium were as follows (in µg ml−1): for E. coli, ampicillin (Ap), 100; chloramphenicol (Cm), 25; kanamycin (Km), 25; tetracycline (Tc), 10; trimethroprim (Tm), 25; and zeocin (Zeo), 25; and for B. pseudomallei, Km, 250; gentamicin (Gm), 50; Tc, 50; Tm, 100; and Zeo, 1000. Plasmids were transferred conjugally from E. coli SM10 to B. pseudomallei on membrane filters with E. coli donors (∼108) and B. pseudomallei recipients (∼108) grown overnight at 37°C. Filters were incubated at 37°C on non-selective LB agar for 3 h before transferring filters onto selective media.
Table 1. List of strains and plasmids used for this study.
Plasmid or strain
Source or reference
Source of the inverted terminal repeats (ITR) of Himar1, Apr
KHW bspR::Ωkm-2, mini-Himar1 cassette inserted at codon 52, Kmr
KHW ΔbsaN::FRT, codon 14–236 of BPSS1546 was deleted
KHW ΔbicA:FRT, codon 1–168 of BPSS1533 was deleted
KHW ΔbprBDC::FRT, codon 67 of BPSS1522-codon 209 of BPSS1520 was deleted
KHW ΔbprQ::FRT, codon 4–121 of BPSS1554 was deleted
KHW ΔbpsR::zeo, codon 13–205 of BPSS1105 was replaced by zeo cassette, Zeor
KHW ΔbprP::FRT, codon 11–311 of BPSS1553 was deleted
KHW ΔbprA::FRT, codon 2–83 of BPSS1530 was deleted
KHW ΔbprC::zeo, codon 209–353 of BPSS1520 was replaced by zeo cassette, Zeor
KHW ΔvirAG::FRT, codon 1–233 of BPSS1494 and 56–614 codon of BPSS1495 were deleted
K96243 contains pKO1105 integrated on the chromosome, Tcr
Construction of transposon and generation of transposon mutants
Himar1-based mini-transposon was constructed by cloning the Wkm-2 cassette from pUTE618 (Chen et al., 2004) into the unique BglII site in pMMOrf (Lampe et al., 1999). A hyperactive Himar1 transposase (tnp) gene from pET29C9 (Lampe et al., 1999) was fused to a promoter (P4325) from Bacillus cereus (Dunn and Handelsman, 1999). The mini-Himar1 cassette and P4325-Himar1-tnp were cloned into pUTE583 (Chen et al., 2004) to generate transposon plasmid pUTE664. The Gram-positive replication origin and erythromycin-resistant gene of pUTE664 were replaced with oriT from pOT182 (Merriman and Lamont, 1993) to generate pUTE664-oriT. A library of transposon insertion mutants was generated by mobilizing pUT664-oriT from the E. coli SM10 into KHW. After mating for 3 h at 37°C, transconjugants were selected for growth on LB medium containing gentamicin and kanamycin.
Screening of cytotoxicity mutants and determining the mini-transposon insertion sites
Transposon mutants were cultured in LB in 96-well plates at 37°C overnight. Mutants were diluted 10 times and subcultured for 1 h before being inoculated onto RAW264.7 cells. Cells were examined under microscope at 6–8 and 18–20 h after infection. Mutants inducing significantly less oncotic cell death (< 10% per field) were subjected to quantitative cytotoxicity assay by measuring LDH release. A mutant was considered for further characterization only if it showed significant attenuation in cytotoxicity to mouse peritoneal macrophages and human THP-1 cells (Sun et al., 2005). To determine insertion sites, the transposon-genomic junctions were sequenced from mutants. PCR was performed on Sau3AI-cut circularized genomic DNA using primer pair Himar1-2 and Himar1-3 (Table S2) extended outward from within the mini-Himar1 cassette. PCR products were sequenced using primer Himar1-4.
Gene deletion by allelic exchange
Deletion mutants were generated by replacement of the respective gene or gene clusters with antibiotic-resistant cassettes. Approximately 1 kb fragments upstream and downstream of the genes were amplified from genomic DNA and cloned into pK18mobsacB (Schäfer et al., 1994). The FRT-flanked tet cassette from pFRTT1 or zeocin resistance cassette (zeo) from pUC18T-mini-Tn7T-Zeo-lox (Choi et al., 2008) was inserted between the gene fragments. The plasmids were electroporated into E. coli SM10 and conjugated into strain KHW. Homologous recombination was selected for retention of tet or zeo markers and loss of the plasmid marker (Km). bspR and bprC were replaced by zeo cassette. bprP, bprQ, bsaN, bicA, bprA, bprBDC and virAG were replaced by tet cassette. Deletion of the chromosomally integrated tet marker was achieved by introduction of Flp recombinase on a curable plasmid as described (Hoang et al., 1998).
A high-density tiling array was custom-fabricated using Nimblegen's photolithographic Maskless Array Synthesis platform. One hundred per cent of the 5728 true protein-coding genes of strain K96243 (Holden et al., 2004) with 50mers tiling across the entire genome are represented. The 127 coding sequences that have been annotated as pseudogenes, gene remnants or partial coding sequences (Holden et al., 2004) are not included in the microarray. Total bacterial RNA was isolated using Trizol and RNA extraction kit (Invitrogen). RNA was treated with TURBO DNase I (Ambion). cDNA was synthesized from mRNA using the SuperScript II reverse transcriptase (Invitrogen). The purified cDNAs from both wild-type strain and mutant were labelled with Cy5 and Cy3 respectively (Cy5-ULS, Cy3-ULS, Kreatech Diagnostics). Hybridization of labelled cDNA to the array was performed and images acquired from array slides as previously described (Ong et al., 2004). Data obtained from hybridizations of two independent RNA preparations of each bacterial strain were used in each analysis. Raw microarray data were LOWESS (Locally Weighted Scatter Plot Smoother) normalized using GeneSpring GX (Agilent) and the median ratio values of probes corresponding to Sanger's 5728 genes comparing mutant with wild type were computed. The signal intensities were visualized using SignalMap (NimbleGen Systems).
Measurement of gene expression by real-time PCR
Real-time PCR primers are listed in Table S2. Bacteria were grown in LB broth overnight and subcultured in RPMI for 3 h at 37°C. Total RNA was isolated as described above. Reverse transcription was performed using High Capacity cDNA synthesis kit (ABI). Transcripts were quantified using SYBRgreener qPCR Supermix for iCycler (Invitrogen) in a Bio-Rad iQ5 machine. Relative RNA level of a particular gene in mutant strains was normalized to that of wild type using the 2−ΔΔCt method (Livak and Schmittgen, 2001) with 16S rRNA as reference gene. Results were reported as mean with standard deviation of triplicate samples. All results were analysed with Student's t test. Differences were considered statistically significant when P < 0.01.
Detection of BopE and BipD by Western blot
Bacteria were subcultured for 3 h in LB at 37°C with constant agitation. Bacterial lysate and culture supernatant were prepared as described (Hii et al., 2008). Same amount of protein from wild-type and mutant strains were separated by 12% SDS-PAGE. BopE and BipD were detected with specific rabbit polyclonal antiserum followed by detection with HRP-conjugated anti-rabbit IgG (Stevens et al., 2003).
Electrophoretic mobility shift assay (EMSA)
Membrane fractions of bacterial lysate were prepared from E. coli expressing bprP in pET-28 vector (Novagen) and EMSA was performed according to the methods described (Darwin and Miller, 2001). A biotin-labelled primer bsaN4-4 and an unlabelled primer bsaN4-1 (Table S2) were used to generate biotin-labelled PCR fragments from bsaN promoter region as probe (136 bp). Membrane proteins were mixed with 0.5 mg ml−1 of bsaN promoter probe in 20 mM Tris with 1.25 mM EDTA, 5 mM NaCl, 50 mM KCl, 5% glycerol, 5% polyethylene glycol 6000, 1 mg ml−1 acetylated BSA and 100 mg ml−1 poly (dI·dC). The mixture was incubated at 30°C for 30 min before being separated on a 5% PAGE. Probes were blotted onto nylon membrane and detected using Chemiluminescent Nucleic Acid Detection Module (Pierce).
Female, 6- to 8-week BALB/c mice were purchased from Laboratory Animals Centre (National University of Singapore). Infection of mice was carried out in animal BSL3 facility with protocol approved by institution IACUC committees. Mice were infected with ∼100 cfu intranasally (Liu et al., 2002). Animals were then monitored daily for survival.
We thank M.P. Stevens (Institute for Animal Health, UK) for the BipD and BopE polyclonal serum, H.P. Schweizer (Colorado State University) for the zeocin resistance cassette and pFLP2 plasmid and T.M. Koehler (University of Texas Health Science Center) for providing experimental materials. This work is supported by Grants T208A3105 from the Ministry of Education and NMRC/1221/2009 from the National Medical Research Council.