ABSTRACT
Expression of virulence determinants by Bordetella pertussis, the primary etiological agent of whooping cough, is regulated by the BvgAS two-component regulatory system. The role of a second two-component regulatory system, encoded by risAS, in this process is not defined. Here, we show that mutation of B. pertussis risA does not affect Bvg-activated genes or proteins. However, mutation of risA resulted in greatly diminished expression of Bvg-repressed antigens and decreased transcription of Bvg-repressed genes. In contrast, mutation of risS had no effect on the expression of Bvg-regulated molecules. Mutation of risA also resulted in decreased bacterial invasion in a HeLa cell model. However, decreased invasion could not be attributed to the decreased expression of Bvg-repressed products, suggesting that mutation of risA may affect the expression of a variety of genes. Unlike the risAS operons in B. parapertussis and B. bronchiseptica, B. pertussis risS is a pseudogene that encodes a truncated RisS sensor. Deletion of the intact part of the B. pertussis risS gene does not affect the expression of risA-dependent, Bvg-repressed genes. These observations suggest that RisA activation occurs through cross-regulation by a heterologous system.
Bordetellae are gram-negative coccobacilli that cause respiratory diseases in several host species. These include whooping cough in humans, caused by Bordetella pertussis and human strains of B. parapertussis. B. pertussis, B. parapertussis, and B. bronchiseptica are closely related and are considered to be subspecies within the B. bronchiseptica cluster (4, 24, 49, 63). In particular, B. pertussis and B. parapertussis appear to have evolved independently from a progenitor bacterium most similar to B. bronchiseptica through selective gene loss and inactivation (14, 49). These subspecies share many similarities but possess interesting differences, which are likely to account for their different host specificities and abilities to cause disease (27, 28). One clear similarity is the presence in all of these three bordetellae of the BvgAS two-component regulatory system, encoded by the bvg locus (for Bordetella virulence gene), which controls the expression of virulence determinants in all three subspecies (36, 53).
BvgAS is a member of the large family of two-component response regulators. These are comprised of a sensor protein (that is generally transmembrane) and a cytoplasmic transcriptional regulator protein. These systems respond to environmental signals and regulate target gene transcription (46). Specifically, the BvgS sensor can respond to signals such as temperature and high concentrations of β-vitamin derivatives (e.g., nicotinic acid) and sulfate anions (34, 40). Under permissive conditions, such as growth at 37°C in low sulfate and nicotinic acid concentrations, BvgS initiates a complex phosphorelay cascade that ultimately results in phosphorylation of BvgA (62). Phosphorylated BvgA then acts as a transcriptional activator at cis-acting sites for a number of genes known as Bvg-activated genes, or vags (52). An additional level of regulatory complexity is evident in that the vags can be separated into early- and late-activated genes whose temporal expression differs when the bacteria are switched between conditions (32, 54). When Bordetella spp. are grown in vitro at low temperatures or in the presence of high sulfate or nicotinic acid concentrations, the BvgAS system is inactive: there are low levels of phosphorylated BvgA transactivator, resulting in low, basal, or virtually absent transcription of the individual vags and low expression of their encoded virulence factors. This process has been called antigenic or phenotypic modulation, and the chemicals that induce it are called modulators (34, 40). Modulation also results in the up-regulation of the expression of another set of genes, known as Bvg-repressed genes, or vrgs (33).
The vags include the bvgAS locus itself (51), bvgR encoding a transcriptional repressor (41), and a broad set of other genes (9, 25, 35, 38, 50). The vag-encoded products include the majority of known virulence factors of these pathogens and are shared mostly between B. pertussis, B. parapertussis, and B. bronchiseptica. Notable exceptions include pertussis toxin, which is believed to be expressed only in B. pertussis (5); PagP, which is apparently not expressed in B. pertussis (50); and the Bcr type III secretion system, which, despite being transcriptionally active in B. pertussis, is nonfunctional (29, 38).
In contrast, the vrgs encode factors whose expression is more divergent between the broad-host-range pathogen B. bronchiseptica and the obligate human pathogen B. pertussis (14, 49). Their products include B. bronchiseptica-specific flagella (2) and urease (39) and B. pertussis-specific vrgs and antigens (Vras) with unknown functions (3, 29, 33, 57). The repression of the vrgs appears to be mediated principally through the vag-encoded transcriptional repressor BvgR, which is believed to bind at cis-repressive sites of the vrg genes (7, 41, 42). The expression of the vrgs is maximal late in the bacterial growth phase (55), and individual vrgs are differentially modulated by individual in vitro modulators (29). In addition to the vags and the vrgs, there is a third gene class, Bvg-intermediate, exemplified by the bipA gene (60) whose transcription is maximal at semimodulating conditions (16).
A second two-component regulatory system, RisAS, has been described and analyzed in B. bronchiseptica (30, 65). This system is related to the EnvZ-OmpR systems of other gram-negative bacteria including Escherichia coli, Salmonella enterica serovar Typhimurium, and Shigella flexneri (22) that reciprocally regulate the expression of porins OmpC and OmpF in response to osmolarity. Interestingly, they are also implicated in bacterial virulence. For example, Xenorhabdus nematophilus envZ mutants have decreased the production of a number of outer membrane proteins and are less pathogenic in host tobacco worms (23), S. enterica serovar Typhimurium ompR mutants are less virulent in mice (18), ompR is required for Yersinia enterocolitica survival in macrophages (10), and both ompR and envZ mutants of S. flexneri are less able to invade HeLa cells (8).
The RisAS system is considered to be an ortholog, not a homolog, of OmpR/EnvZ, as it does not respond to osmolarity as an environmental signal and apparently does not regulate the expression of porins in B. bronchiseptica (30). Rather, RisAS coordinately regulates a variety of factors, acting independently of Bvg-mediated regulation. Furthermore, Ris expression is induced when B. bronchiseptica is intracellular and is required for intracellular survival, resistance to oxidative stress, and persistence in a mouse infection model (30, 65). This implicates Ris as a second two-component regulatory system important in the pathogenesis of the bordetellae.
In this study, we describe a comprehensive analysis of the RisAS system in B. pertussis and show that risA affects the transcription and expression of vrgs but not the vags, whereas risS does not affect vrg or vag expression probably because it is a pseudogene in B. pertussis. risA mutants adhere just as well as wild-type bacteria but are less efficient at invasion of HeLa cells, indicating a role in the host-pathogen relationship for this system. RisA can function independent of RisS, suggesting the possibility of cross talk between regulatory systems and the integration of multiple regulatory signals in the expression of Bvg-regulated molecules in B. pertussis.
MATERIALS AND METHODS
Bacterial strains and plasmids.The strains and plasmids used in this study are listed in Table 1. Bordetella pertussis strains were grown on Bordet-Gengou agar (BGA; Difco, Detroit, MI) containing 15% sheep's blood (Dalynn, Calgary, Alberta, Canada) for 3 days at 37°C in a humidified atmosphere prior to experimental assays, unless otherwise noted. To induce modulation, B. pertussis strains were grown on BGA containing 5 mM nicotinic acid and 40 mM MgSO4 (7, 25). E. coli strains were grown in L broth or on L agar. When necessary, the following antibiotics at the indicated concentrations were added to the media: nalidixic acid, 30 μg/ml; gentamicin, 15 μg/ml (for maintenance of B. pertussis and E. coli strains) or 30 μg/ml (for selection of transconjugants); ampicillin, 100 μg/ml; kanamycin, 50 μg/ml; streptomycin, 200 μg/ml; and tetracycline, 10 μg/ml. PCR, DNA isolation, cloning, and sequencing were performed by standard methods. Clones were confirmed by sequence analysis, restriction analysis, or Southern hybridization as appropriate.
Strains and plasmids
Mutagenesis of the risA and risS genes in B. pertussis.The Bordetella pertussis risAS locus was identified and cloned by our own unpublished analysis and confirmed through analysis of published sequences (30) and the genomic sequence of B. pertussis (49). The genomic sequence data were produced by the B. pertussis Sequencing Group at the Sanger Centre and were obtained from ftp://ftp.sanger.ac.uk/pub/pathogens/bp/ . Allelic replacement of an insertionally disrupted risA in BP536 (59) was performed using the allelic exchange vector pRTP1 (58). Briefly, PCR analysis of a transfected B. pertussis DNA cosmid library identified one colony whose DNA acted as a template for the amplification of risA. The cosmid library was constructed in the vector pHC79 (Boehringer Ingelheim, Ridgefield, CT) from B. pertussis CN2992 (Wellcome collection) chromosomal DNA partially digested with Sau3A and transfected into E. coli XL1-Blue (Stratagene, La Jolla, CA). PCR was performed with primers bpORF (5′-CTGCTGGTTCTCGACCTG-3′) and bpORR (5′-GGGTTGAAGGGCTTGGAC-3′), which generated a 200-bp risA fragment. The risA-containing cosmid, pBP2, was digested with SalI-ClaI to release a 1.7-kb DNA fragment which was cloned into pBluescript II KS(+) (Stratagene) to create pBP3. A kanamycin resistance (neo) cassette was inserted at the EcoRI site of the risA fragment in pBP3 to create pBP15. Insertionally disrupted risA DNA was released as a 3.5-kb PvuII fragment and cloned into pRTP1 to create pBP16. Triparental matings were used to mobilize pBP16 from donor E. coli CC118λpir (15) into recipient B. pertussis BP536 using E. coli plasmid-mobilizing helper strain S17-1(pNJ5000) (12). Double-crossover events were selected on media containing streptomycin and kanamycin. Southern hybridization using genomic DNA from resultant colonies confirmed recombination and mutagenesis. One mutant, BPOR, was chosen for further analysis.
To address the possibility that insertion of the neo gene into risA created downstream effects on other genes, we created an in-frame deletion of risA by homologous recombination in B. pertussis. Allelic replacement of the deletion mutant into BP536 was performed using the allelic exchange vector pSS2141 (42). Briefly, PCR was performed with primers risArev (5′-AAGCTTAGCGGGAAGACGAAGTTTCGAAGGCAA-3′) and risAfor (5′-CTCGAGGGGGCACGAGACGGCGCTCCTG-3′), which generated a 948-bp risA-containing fragment. PCR was performed using B. pertussis strain BP338 (64) DNA as the template. The PCR product was cloned into vector pCR2.1 using the TOPO TA cloning kit (Invitrogen Life Technologies, Carlsbad, CA) to generate pTS198. A 204-bp in-frame deletion was generated in risA (ΔrisA) by restriction digestion with SstII followed by religation to create pTS210. The ΔrisA construct was cloned as an EcoRI fragment from pTS210 into pSS2141 to generate pTS219. Triparental matings were used to mobilize pTS219 from donor E. coli DH5α (Invitrogen) into recipient B. pertussis BP536 using E. coli plasmid-mobilizing helper strain MM294(pRK2013) (20). Integration of the plasmid into the chromosomal copy of risA was selected for by plating on media containing gentamicin. A second homologous recombination resulting in the excision of pSS2141 DNA was selected for by plating transconjugates on media containing streptomycin. Isolates were screened for the ΔrisA mutant by PCR using the risAfor and risArev primer pair and template prepared from bacterial colonies (13). The in-frame ΔrisA mutant was confirmed in one clone, BPΔOR, which was selected for further analysis.
An in-frame deletion of the risS gene was performed by first amplifying a 2,140-bp fragment of DNA containing the risAS locus by PCR using the primers risASfor (5′-AAGCTTCACCGCCTCATGCGACAC-3′) and risASrev (5′-AAGCTTGCCAGCACCGACGAAATGTC-3′), which generated flanking HindIII sites (underlined). The risAS PCR product was cloned into vector pCR2.1 to generate pTS264. The risAS fragment was then cloned from pTS264 into pUC9 (44) as a HindIII fragment to generate pTS269. A 432-bp in-frame deletion was generated in risS (ΔrisS) by restriction digestion with SstI followed by religation to create pTS270. The ΔrisS-containing fragment was amplified by PCR from pTS270 using primers risASfor and risASrev and was cloned into vector pCR2.1 to generate pTS272. The ΔrisS-containing fragment was then cloned into pSS2141 from pTS272 as a BamHI fragment to generate pTS273. Allelic replacement of risS with ΔrisS in B. pertussis strain BP536 was performed using pTS273 and the mating and selection procedures described for the generation of the ΔrisA mutant. Clones were screened for the ΔrisS mutant by PCR using the risASfor and risASrev primer pair. The in-frame ΔrisS mutant was confirmed in one clone, BPΔRS, which was selected for further analysis.
A mutant with constitutive expression of the vags and insensitivity to modulation was generated as a control strain in the analysis of bvg-regulated genes and phenotypic analysis of the ΔrisA mutant BPΔOR. This was achieved by allelic exchange of the constitutive bvgS mutant bvgS-C3 into the chromosome of parental strain BP536 using E. coli SM10(pJM503) (45) as previously described. Screening for hemolysis identified constitutive bvgS strains that remained hemolytic on BGA containing the modulators 5 mM nicotinic acid and 40 mM MgSO4. One strain, BPSC, was chosen for further analysis. To assess the affect of a mutation in risA in a strain with insensitivity to modulation, we constructed B. pertussis strain SCOR by allelic exchange of ΔrisA as described using BPSC as the parental strain.
Comparisons of the B. pertussis risAS locus with the risAS locus of other bordetellae was done with genome sequences (49) made available from the Sanger Centre (these sequence data were produced by the B. bronchiseptica and B. parapertussis Sequencing Groups at the Sanger Centre and were obtained from ftp://ftp.sanger.ac.uk/pub/pathogens/bp/ , from our own unpublished sequence data, and by BLAST analysis of published sequences available in the public databases [GenBank nonredundant nucleotide and protein databases, last accessed January 2005]).
Complementation of the ΔrisA mutant.To confirm that the phenotype described in ΔrisA bacteria is a result of the disruption of risA only, we constructed a plasmid containing risA to use for complementation using the low-copy-number vector pRK310 (17). PCR was performed using primers risASfor (5′-AAGCTTCACCGCCTCATGCGACAC-3′) and risACrev (5′-AAGCTTAGGCTGACCAGCATCAGTGC-3′) to generate a 1,314-bp piece of DNA with flanking HindIII sites (underlined). This DNA contains risA starting 438 bp upstream of DNA of the risA open reading frame and ending 138 bp into the risS open reading frame. The PCR product was cloned into vector pCR2.1 to generate plasmid pTS263, which contains the risA gene flanked by BamHI sites. The risA gene was cloned into pRK310 as a HindIII fragment to create pRisA. Plasmids pRK310 and pRisA were introduced into B. pertussis strains from E. coli DH5α by triparental matings using E. coli MM294(pRK2013) and selecting for tetracycline-resistant B. pertussis.
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting.Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting were performed as previously described (13). Proteins were resolved using 15% (37.5:1) acrylamide/bis-acrylamide separating gels. The S1 subunit of pertussis toxin, pertactin, Vra-a, or Vra-a and Vra-b were detected with monoclonal antibodies X2X5 (31), BBO5 (11), and 1G7 or monoclonal antibody 7H1 (57), respectively. Peroxidase-conjugated goat anti-mouse immunoglobulin G (whole-molecule) secondary antibody was purchased from Pierce (Rockford, IL). Antibody binding was visualized by chemiluminescence using enhanced chemiluminescence Western blotting detecting reagents (Amersham, Piscataway, NJ). Apparent molecular weights were determined by comparison with the BenchMark prestained molecular weight ladder (Invitrogen).
Construction of transcriptional reporters. β-Galactosidase (β-Gal) transcriptional reporters were generated using the suicide vector pFUS2 (3) and PCR-amplified B. pertussis DNA. The following primers were designed to amplify an internal fragment of bvgR, kpsM, and vrg-6, with flanking HindIII and KpnI restriction sites (underlined): 5′-AAGCTTTGTTCATCCAGCCAGAGCCC-3′ (bvgRF) and 5′-GGTACCCATGGTCGCCCAGCAGATACC-3′ (bvgRR), 5′-AAGCTTCGTTCTTCCTGTTCAGCCACAC-3′ (kpsMF) and 5′-GGTACCAATAAAGGTGCTCCCACGACTTG-3′ (kpsMR), and 5′-AAGCTTGCCGGACTCATGCTCTCCAGCG-3′ (vrg-6F) and 5′-GGTACCGCGGTCATAGTGCCGATGGC-3′ (vrg-6R). The sizes of the cloned fragments were 441 bp, 309 bp, and 234 bp for bvgR, kpsM, and vrg-6, respectively. PCR was performed using B. pertussis strain BP338 DNA as the template, and the products were cloned into pCR2.1. The resulting plasmids were digested with HindIII-KpnI and cloned in-frame with the groES gene of pFUS2, resulting in a transcriptional fusion with the lacZ gene. These fusion constructs were introduced into the chromosomes of strains BPOR, BPΔOR, BPSC, and BPΔRS and their isogenic parent, BP536, from E. coli DH5α in matings using E. coli MM294(pRK2013). Gentamicin selects for insertional duplication mutations in the chromosomal copy of the genes and creates a β-Gal transcriptional reporter under the control of the chromosomal promoters. Strains containing reporters were maintained under gentamicin selection to select against potential excision of the plasmids. To complement the ΔrisA mutant, pRisA (or control pRK310) was mated into BPΔOR reporter strains from DH5α in matings using E. coli MM294(pRK2013). Unique clones were generated for each transcription assay.
β-Galactosidase assay.Transcriptional analysis was done using bacteria grown on solid media as described previously (3). Transcription was monitored using β-Gal as a reporter using a standard colorimetric assay (40). Data were collected from four independent assays. Student's t test was used to analyze the data.
HeLa cell adherence assay.The HeLa cell adherence assay was performed as a modification of the B. pertussis HeLa cell invasion assay (see below). Bacteria were incubated with HeLa cell monolayers for 90 min in Eagle minimal essential medium (MEM; Invitrogen) supplemented with 3% fetal bovine serum (FBS; Flow Laboratories, McLean, VA) at 37°C in 5% CO2. HeLa monolayers were then washed five times for 5 min each with rotation at 120 rpm at room temperature using 500 μl MEM-3% FBS. Monolayers were then rinsed with trypsin buffer and treated with 0.25% trypsin solution (Invitrogen) to lift the monolayers and dislodge adherent bacteria. Cells were diluted in MEM-3% FBS and plated onto BGA in duplicate for determination of total CFU. Data were collected from three independent assays. Student's t test was used to analyze the data.
HeLa cell invasion assay.The gentamicin invasion assay was performed as previously described (19). Separate control experiments demonstrated that the risA mutant did not result in an appreciably greater sensitivity to gentamicin than in parental strains containing wild-type risA (data not shown). Colony counts were collected from at least three independent assays. Student's t test was used to analyze the data.
RESULTS
Sequence analysis of the risAS locus of B. pertussis.The ompR/envZ orthologs risA and risS were compared between Bordetella spp. using the published genome sequences of B. pertussis, B. parapertussis, and B. bronchiseptica; published risAS gene sequences; and our own sequence analysis. The DNA sequences of risAS are nearly identical, with a few, mostly conservative, nucleotide changes between alleles (30). The strains studied were B. bronchiseptica BB7866, B. pertussis Tohama I, B. parapertussis 15311 and ML/180, and B. avium 35086. The genomic sequences of B. pertussis Tohama I, B. parapertussis 12822, and B. bronchiseptica RB50 also reveal fundamentally identical risAS alleles, with the notable exception of B. pertussis Tohama I (49). The Tohama I risAS sequence differs by only a few bases but most notably contains an additional C at position 1848. This is within the coding region of risS and introduces a frameshift in the risS gene. This results in a downstream opal stop codon and a deduced RisS protein that is truncated at residue 323 with a predicted molecular mass of 36 kDa instead of the 52 kDa predicted for the RisS protein of the other bordetellae. EnvZ of Xenorhabdus nematophilus is only 342 amino acids long but can complement E. coli envZ-null mutants in an osmodependent manner; however it is truncated in its N-terminal perplasmic region and still contains all of the domains important for signal transduction (61). In contrast, B. pertussis RisS is predicted to lack an EnvZ-like transmitter domain, which includes the autophosphorylation site as well as other invariant amino acids and conserved domains. It is thus highly unlikely that the EnvZ ortholog RisS is functional as an environmental sensor in B. pertussis. The domain structures of EnvZ and RisS variants are illustrated in Fig. 1.
Comparison of domain structures of EnvZ and RisS proteins. (A) E. coli EnvZ; (B) Xenorhabdus nematophilus AN6/1 EnvZ; (C) B. bronchiseptica RB50 RisS; (D) B. pertussis Tohama I RisS. HAMP, receptor histidine kinase domain; HisKA, dimerization/phosphoacceptor, HATPase_c; histidine kinase-like ATPases.
It is possible that this single base change is peculiar to the strains examined. We therefore used PCR to amplify risS from other Bordetella isolates and performed sequence analysis to compare risS alleles. The strains tested were CN2992 and six clinical isolates from the United Kingdom Pertussis Reference Laboratory which had been passaged in vitro a maximum of two times: BP-K72, BP-K75, BP-K76, BP-K79, BP-K82, and BP-K83. All of these strains contained the extra cytosine resulting in the frameshift mutation in risS. Control strains of B. parapertussis (CN2591 and three ovine strains, BPP-HI, BPP-C, and BPP-k2) and B. bronchiseptica (CN 7635E and six rabbit isolates, BBr-4 through BBr-9) did not contain the frameshift mutation and had a full-length risS gene. Hence, this mutation appears to be common in B. pertussis strains but rare or absent in B. parapertussis and B. bronchiseptica.
The risA gene is upstream of risS, and the two genes are separated by only 4 bp, as found in ompR-envZ systems. They are cotranscribed in B. bronchiseptica (30). The deduced B. pertussis RisA protein comprises 244 amino acid residues with a molecular mass of 27.8 kDa and shares a 65.7% identity with the E. coli OmpR protein. This response regulator family of proteins is defined by a conserved domain of approximately 100 amino acids that extends over the N terminus of each of these proteins. In particular, the residues that correspond to D12, D55, and K105 in the E. coli OmpR protein tend to be conserved among the response regulator protein family. These are found at amino acid positions D17, D60, and K110 in the deduced B. pertussis RisA. Residue D60 of RisA lies in a domain between V58 and R76 that is completely conserved between B. pertussis RisA and E. coli OmpR and is likely to be the site of RisA phosphorylation. In addition, the region between P83 and I121 that accommodates the invariant K110 is almost completely conserved between the two proteins (30).
Expression of Bvg-regulated proteins.We have previously noted that two B. pertussis Vras are not expressed in a risA transposon mutant (our unpublished observations). Furthermore, mutation of the risAS locus is known to have pleiotropic effects on B. bronchiseptica gene expression (30). We thus analyzed protein expression in risAS mutants to investigate their patterns of Bvg-regulated protein expression. Blots were probed for expression of the Bvg-activated proteins pertussis toxin and pertactin and the Bvg-repressed antigens Vra-a (Fig. 2 and Fig. 3, lower-molecular-mass bands) and Vra-b under both nonmodulating and modulating conditions. B. pertussis strain BPOR, which has risA insertionally disrupted by a neo cassette, demonstrated decreased expression of the Vras even when the bacteria were grown under modulating conditions but did not appear to have altered expression of the vag products pertactin (Prn) or pertussis toxin subunit S1 (data not shown). To confirm that the decrease in expression of the Vras was due specifically to the mutation in risA and not polar effects due to the neo insertion, we constructed mutants with single in-frame deletions in risA (BPΔOR) or risS (BPΔRS). Levels of Vra-a and Vra-b were greatly diminished in the ΔrisA mutant compared to the wild type under both nonmodulating and modulating conditions, while the overall expression of the Vras was still increased under modulating versus nonmodulating conditions (Fig. 2, lanes 1 through 4). In contrast, the ΔrisS mutant did not show any alteration in the expression of the Vras compared to the wild type (Fig. 2, lanes, 1 and 2 and 5 and 6).
Expression of Bvg-repressed antigens in risA and risS mutants. Lanes: 1, parental Bordetella pertussis strain BP536 grown on BGA without modulators; 2, B. pertussis strain BP536 grown on BGA with the modulators 5 mM nicotinic acid and 40 mM MgSO4; 3, ΔrisA mutant (BPΔOR) grown on BGA; 4, ΔrisA mutant grown on BGA with modulators; 5, ΔrisS mutant (BPΔRS) grown on BGA; 6, ΔrisS mutant grown on BGA with modulators. Relative protein levels were assessed by Western blotting as described in Materials and Methods. (A) Vra-a (VraA); (B) Vra-b (VraB). Migrations of molecular weight (MW) markers are indicated on the right.
Expression of Bvg-regulated proteins. Lanes: 1, parental Bordetella pertussis strain BP536 grown on BGA without modulators; 2, B. pertussis strain BP536 grown on BGA with the modulators 5 mM nicotinic acid and 40 mM MgSO4; 3, bvgS-C (BPSC) mutant grown on BGA; 4, bvgS-C mutant grown on BGA with modulators; 5, ΔrisA mutant (BPΔOR) grown on BGA; 6, ΔrisA mutant grown on BGA with modulators; 7, ΔrisA mutant with low-copy-number vector [BPΔOR(pRK310)] grown on BGA; 8, BPΔOR(pRK310) grown on BGA with modulators; 9, ΔrisA mutant with complementing risA plasmid [BPΔOR(pRisA)] grown on BGA; 10, BPΔOR(pRisA) grown on BGA with modulators. Relative protein levels were assessed by Western blotting as described in Materials and Methods. (A) pertactin (Prn); (B) the S1 subunit of pertussis toxin (S1); (C) Vra-a (VraA); (D) Vra-b (VraB). Migrations of molecular weight (MW) markers are indicated on the right.
To ensure that mutation of risA was specifically responsible for the decreased expression of the Vras, we introduced either low-copy-number control vector pRK310 (17) or this vector containing the wild-type risA gene (pRisA) into the ΔrisA mutant BPΔOR and then analyzed the plasmid-containing strains in complementation experiments. In addition, we constructed and analyzed an isogenic mutant containing the bvgS-C3 mutant (BPSC) (45), which has a Bvg-constitutive phenotype, to act as a Bvg-activated phenotypic control. Immunoblots were probed for pertussis toxin (subunit S1) and pertactin and the Bvg-repressed antigens Vra-a and Vra-b (Fig. 3). Pertussis toxin and pertactin were expressed only in the absence of modulators in the wild-type strain and ΔrisA mutant regardless of the presence of pRK310 or pRisA, and their expression levels did not differ appreciably between the ΔrisA stains and their wild-type isogenic parent (Fig. 3A and B). In addition, standard zones of hemolysis were observed when risA mutants were grown on blood agar (data not shown), indicating normal levels of the vag-encoded bifunctional adenylate cyclase-hemolysin. As expected, the Bvg-constitutive mutant BPSC constitutively expressed pertussis toxin, pertactin, and the hemolytic phenotype regardless of the presence of modulators. Conversely, Vra-a and Vra-b were highly induced under modulating conditions in wild-type bacteria but were constitutively repressed in BPSC under the conditions tested (Fig. 3C and D).
Expression of these Bvg-repressed products was greatly diminished in the ΔrisA mutant compared to the wild type even in the presence of modulators, with levels of the Vras in the modulated ΔrisA mutant comparable to nonmodulated levels in the wild-type strain. Introduction of vector control pRK310 did not restore wild-type expression levels of the Vras under either nonmodulating or modulating conditions (Fig. 3). However, introduction of pRisA complemented the ΔrisA mutant, restoring strong expression of the Vras to a level severalfold higher than that produced by the wild-type strain in the presence of modulators (Fig. 3C, lane 10). These data indicate that risA appears to be required for the expression of Bvg-repressed products, perhaps in a specific manner that does not affect the expression of Bvg-activated products. In addition, the increased copy number of risA when supplied even on a low-copy-number plasmid appears to augment expression of the Vras, suggesting that increasing levels of the RisA activator may increase expression of the Bvg-repressed antigens.
Transcriptional analysis of Bvg-regulated genes.To determine if the reduction of expression of Bvg-repressed products in a ΔrisA background was due to reduced transcription, we analyzed insertional lacZ fusion reporters (Table 2). As expected, transcription of the Bvg-repressed genes kpsM (which is predicted to encode a capsule transport protein) (3, 29) and vrg-6 (which is predicted to encode a small protein of unknown function) (7) were greatly increased under modulating conditions in the wild-type strain, while their transcription was constitutively repressed in BPSC. Transcription of these vrgs was greatly reduced in the ΔrisA mutant BPΔOR, even under modulating conditions. However, introduction of pRisA restored transcription of the vrgs to a level that was about double that of the wild type. In contrast, the Bvg-activated gene bvgR was transcribed at a much higher level under nonmodulating conditions than under modulating conditions in wild-type bacteria, while it had constitutively high transcription in the control strain BPSC. As in the wild type, transcription of bvgR was still high under nonmodulating conditions in the risA mutant and again was greatly reduced in the presence of modulators. Introduction of pRisA did not affect the expression of bvgR in any of the strains tested.
β-Galactosidase activities of B. pertussis pFus2 integrants
To test the statistical significance of the transcriptional fusion data, Student's t test was used. When wild-type BP536 and the ΔrisA mutant BPΔOR were compared under modulating conditions, transcription of the vrgs kspM and vrg-6 was significantly decreased in BPΔOR (P < 0.01), demonstrating that vrg transcription is significantly lower when risA is mutated. Transcription of the vag bvgR was not significantly different between the wild-type bacteria and the risA mutant, confirming that risA has no apparent role in the transcription of the vags. Analysis of transcription data for pRisA-complemented ΔrisA mutants demonstrated significantly increased transcription of vrg-6 under both modulating and nonmodulating conditions (P < 0.01). Significantly increased transcription of kpsM after pRisA complementation was seen only under modulating conditions; however, kspM transcription was very low under nonmodulating conditions and near the limit of detection for the assay used. These data strongly suggest that the presence of multiple copies of risA significantly increases transcription of the vrgs regardless of the affect of Bvg regulation. In summary, these data indicate that mutation of risA greatly diminishes the expression of Bvg-repressed products at the level of transcription, while not apparently affecting the transcription of the vags.
Analysis of risA mutants for adherence and invasion in HeLa cell models.Transcription of risA occurs when B. bronchiseptica is inside eukaryotic cells, and risA mutants have reduced ability to survive within these cells (30, 65). We tested B. pertussis risA mutant BPOR, the constitutive mutant BPSC, and their isogenic wild-type parent for adherence to, and survival in, HeLa cells (Fig. 4).
HeLa cell adherence or invasion by a risA mutant. (A) adherence assay; (B) invasion assay. BP536, parental B. pertussis strain; BPSC, bvgS-C mutant; BPOR, risA mutant. Values represent 103 CFU recovered from gentamicin-treated (panel B only) and washed monolayers. Error bars represent standard errors of the means for three replicates. *, P < 0.01, compared to BP536.
A reduction in survival in the HeLa cell invasion assay could be due to reduced adherence, invasion, or survival. To distinguish between reduced adherence and reduced invasion or intracellular survival, we performed an adherence assay using HeLa cells. After 90-min incubation and washing to remove nonadherent bacteria, there was no difference in adherence to HeLa cell monolayers between BPOR and BPSC (P > 0.1) relative to the wild-type control (Fig. 4A).
The invasion of wild-type bacteria and BPSC was not significantly different (P > 0.1). However, there was approximately fivefold less survival of BPOR relative to the wild type (P < 0.01) (Fig. 4B). The total numbers of invasive wild-type B. pertussis were in close agreement with those previously published (19), and the strains tested did not have any appreciable differences in sensitivity to gentamicin (data not shown).
Analysis of Bvg-constitutive mutants in the HeLa cell invasion model.The fact that BPSC and wild-type bacteria survived equally well suggests that the reduction in survival of the risA mutant is not due to the diminished expression of Bvg-repressed molecules. We therefore constructed a BvgS-constitutive-ΔrisA double mutant and tested it for survival in the HeLa cell invasion assay to determine if risA mutation affected survival in the Bvg-constitutive background. The ΔrisA mutant BPΔOR and the BvgS-constitutive/ΔrisA double mutant SCOR both survived less well than the wild type in the HeLa cell invasion assay (Fig. 5) (P < 0.01), even though the Bvg-constitutive strain BPSC did not differ in its survival capability in the assay (P > 0.1). Interestingly, SCOR survived better than BPΔOR (P < 0.01), suggesting that an element of the decreased survival in risA mutants may be due to alteration in the expression of factors not regulated by Bvg.
HeLa cell invasion by Bvg-constitutive strains. BP536, parental B. pertussis strain; BPSC, bvgS-C mutant; BPΔOR, ΔrisA mutant; SCOR; bvgS-C/ΔrisA double mutant. Values represent 103 CFU recovered from gentamicin-treated and washed monolayers. Error bars represent standard errors of the means for four replicates. *, P < 0.01, compared to BP536.
DISCUSSION
The BvgAS two-component regulatory system is at the apex in the hierarchical control of virulence factor expression required for the pathogenesis of the bordetellae. However, in other pathogens, multiple regulatory networks contribute to virulence and this is likely to be the case here, too. The OmpR/EnvZ ortholog RisAS is the second two-component regulatory system to be described in the bordetellae through experimentation with B. bronchiseptica strains (30, 65). In particular, ris-dependent products have been implicated in resistance within macrophages to acid and oxidative stresses by B. bronchiseptica (30, 65) and ris is required for bacterial persistence in a mouse model of infection (65). Expression of ris is bvg independent, and mutation of ris produces pleiotropic effects on protein expression. We investigated potential roles for the RisAS system of B. pertussis in the regulation of virulence, using risAS mutants.
Our results suggest that RisA is required for the expression of Bvg-repressed products in B. pertussis. Vra-a and Vra-b are surface-exposed proteins that are tightly regulated by the BvgAS system in B. pertussis (43, 57). Mutation of risA results in greatly decreased expression of Vra-a and Vra-b with levels produced by modulated risA mutants only about as high as those produced by unmodulated wild-type B. pertussis. However, Vra-a and Vra-b were still induced by modulation, suggesting that their expression is still repressed by the BvgASR regulatory cascade but that a functional risA gene is required for high-level expression of the Vras. Complementation of the risA mutant with risA on a low-copy-number plasmid restored the high-level, BvgAS-regulated expression of the Vras to a level a severalfold higher than that found in wild-type B. pertussis. This suggests that an increased level of RisA transactivator increases expression of the Vras.
Transcriptional analysis using gene fusions demonstrated a dependence on risA for transcription of the vrgs. Transcription of vrg-6 and kpsM is virtually absent when risA is disrupted. When risA is supplied in trans, the levels of vrg expression in modulated bacteria are about twice those of modulated wild-type bacteria. These results suggest that risA affects vrg expression and Vra production at the level of transcription. Furthermore, the increased expression of these Bvg-repressed molecules as a result of there being multiple copies of risA suggests that increasing concentrations of the RisA transactivator induces higher levels of expression of the vrgs. It will be interesting to determine whether RisA acts directly to bind the cis-activating sites of the vrgs or indirectly by altering the expression of other regulatory molecules.
The Bvg-activated products appear not to be dependent on RisA for expression, as evidenced by equivalent levels of hemolysis on blood agar (a result of adenylate cyclase production) and comparable levels of pertussis toxin and pertactin expression between the wild type and risA mutants under nonmodulating conditions. Similarly, ectopic or unregulated production of Bvg-regulated products was not observed in risA mutants. For instance, bvgR has a normal bvg-activated transcription pattern when risA is mutated and pertussis toxin and pertactin are also not detected under nonmodulating conditions. Thus, it would seem that RisA could play a specific function in gene regulation as an activator of expression of the vrgs. However, it will be important to look at an expanded set of genes and proteins, though genomic and proteomic techniques, to address the likelihood that RisA affects genes other than just the vrgs.
It is perhaps surprising that a phenotype was detected in B. pertussis risA mutants, considering that the gene encoding the cognate sensor RisS is a pseudogene. All the B. pertussis strains tested have the frameshift mutation in risS, which would result in a prematurely truncated RisS sensor lacking an EnvZ-like transmitter domain, and the putative site of RisS autophosphorylation. This mutation was not found in risS of B. bronchiseptica or B. parapertussis. When B. pertussis risS was mutated to contain a large deletion upstream of the frameshift, it resulted in no obvious phenotypic effect on gene expression. Thus, maximal expression of Bvg-repressed products is dependent on functional risA but is independent of risS. It is interesting to speculate that RisA of B. pertussis may be activated by phosphorylation by a kinase other than RisS via cross talk between regulatory systems. Interestingly, heterologous phosphotransfer has been demonstrated between the purified components of two-component regulatory systems in vitro (48). In addition, phosphorylation of OmpR has been observed in envZ-null mutants of E. coli, suggesting that OmpR can be phosphorylated by a kinase that is functionally homologous to EnvZ (21). Cross-regulation has been observed in vivo between the AcrB sensor of anaerobiosis and the OmpR regulator of porin expression in E. coli, suggesting interplay between phosphorelay systems (37). Therefore, it is possible that another regulatory system, other than BvgAS or RisAS, may regulate the expression of the vrgs in B. pertussis via RisA activation.
B. bronchiseptica risAS expression is maximal at 37°C in the absence of magnesium cations, when the bacteria are intracellular (30). It will be important to determine which signals regulate RisA expression and activation in B. pertussis. This is of particular interest given that risS is a pseudogene in B. pertussis. It would not be surprising if RisA were to be activated by different regulatory sensors in B. pertussis and thus also different environmental signals. It will be interesting to determine if shuttling a wild-type B. bronchiseptica risS gene to B. pertussis alters the expression patterns of Bvg-repressed molecules in response to the presence of particular environmental signals.
It is apparent that the Bvg-repressed arm of the Bvg regulon has diverged during evolution of the subspecies of the B. bronchiseptica cluster (55). B. pertussis in particular expresses a more limited, but somewhat distinct, set of Bvg-repressed products compared to B. bronchiseptica. In addition, the requirement for RisA in B. pertussis for maximal expression of the vrgs contrasts with the pleiotropic nature of the Ris-regulated products in B. bronchiseptica wherein ris is required for expression of the B. bronchiseptica-specific acid phosphatase, but not bvg-repressed motility or urease (30). Our results indicate that the genes regulated by risAS and the mechanisms of RisA-mediated gene regulation have diverged during evolution of the bordetellae. It is possible that differences in risA gene regulation and differences in vrg expression in the bordetellae have resulted from the same selective pressure that resulted in the evolution of Bordetella subspecies with different host ranges, physiology, and disease presentations.
B. bronchiseptica ris mutants have reduced survival in macrophage invasion models (30, 65). B. pertussis can invade epithelial-like cells, and the expression of the Bvg-activated phase is required for adhesion and invasion in these models (6, 19). Our results in the HeLa cell model suggest that B. pertussis invasion is risA dependent, while adherence is risA independent. Since vrg expression is greatly diminished in risA mutants, it could be hypothesized that the vrgs are required for this intracellular survival or invasion. However, we found that Bvg-activated phase-locked bacteria, which constitutively repress the vrgs, are as able as wild-type bacteria to survive in HeLa cells. This suggests that the observed phenotype is probably due to effects on other genes that are dependent on risA for expression but that are not Bvg repressed and argues against modulation of the bacteria to affect increased survival in this model system. Although risA mutants of B. pertussis enter or survive less well in HeLa cells, the relevance of epithelial cell invasion by B. pertussis to the disease state is uncertain, as an intracellular niche has not been demonstrated for this bacterium. It is possible that invasion occurs in low numbers in humans in a location in which the bacteria are hard to detect. Alternatively, the readout in the invasion models could reflect another biologically relevant aspect of B. pertussis ecology, even if the bacterium does not invade eukaryotic cells in vivo, and it is conceivable that RisA has a biologically important function in B. pertussis in vivo through an influence on the expression of vrgs or other genes. Further work will be required to determine if the HeLa cell invasion data reflect a biologically relevant interaction of the bacterium with host cells or if they reflect a more general phenotype related to the ability of B. pertussis to survive stressful factors in its environment.
In B. bronchiseptica, biologically relevant phenotypes have been attributed to vrg-encoded products such as motility for flagella (1, 2) and urea degradation for urease (39). In B. pertussis, a relevant vrg-encoded biological phenotype is uncertain. Although a Bvg-repressed capsule has not been described for B. pertussis, kpsM, a capsule biosynthesis gene, has been described as a vrg in B. pertussis (3, 29). This is intriguing, as polysaccharide capsules are known to be important virulence factors for a number of bacteria (47). We have confirmed that kpsM is a B. pertussis vrg; however, it is unknown whether the KpsM protein is expressed in B. pertussis. Furthermore, in B. pertussis, several other genes of the capsule biosynthesis locus have been inactivated or have been shuttled to alternate positions in the chromosome (49). It is possible that the altered subset of capsule synthesis genes could have a biological function in B. pertussis.
The smaller number of vrgs expressed in B. pertussis compared with B. bronchiseptica may indicate that the B. pertussis vrgs are decaying relative to B. bronchiseptica. However, many vrgs are still expressed in B. pertussis, and moreover, the expression patterns of the vrgs differ between the two subspecies. It is hoped that further investigation of Ris- and Bvg-mediated regulation will elucidate a role (either current or ancestral) for both ris and the vrgs in the life cycle of the human pathogen B. pertussis.
ACKNOWLEDGMENTS
We thank F. Jacob-Dubuisson for providing plasmid pFUS2, J. Miller for providing E. coli strain SM10(pJM503), S. Stibitz for providing plasmids pSS2141 and pRK310, N. Guiso for providing B. bronchiseptica clinical isolates, and A. Weiss for providing monoclonal antibodies X2X5 and BBO5. We also gratefully acknowledge the Sanger Centre for allowing public access to Bordetella genome data prior to publication.
This work was supported by a Canadian Bacterial Diseases Network grant (through N.C.E. Canada) to M.S.P. and a Wellcome Trust Programme grant to D.M.
FOOTNOTES
- Received 11 January 2005.
- Returned for modification 29 March 2005.
- Accepted 26 April 2005.
- Copyright © 2005 American Society for Microbiology
