An Improved Recombination-Based In Vivo Expression Technology-Like Reporter System Reveals Differential cyaA Gene Activation in Bordetella Species

Abstract

Bordetella pertussis and Bordetella bronchiseptica rely on the global two-component regulatory system BvgAS to control expression of distinct phenotypic phases. In the Bvg⁻ phase, expression of vrg genes, including those required for motility in B. bronchiseptica, is activated and genes encoding virulence factors are not expressed. Conversely, in the Bvg⁺ phase, genes encoding virulence factors are highly expressed while genes necessary for motility are repressed. Although several genetic analyses have demonstrated the importance of the Bvg⁺ phase during respiratory infection, Bvg-regulated gene activation in B. bronchiseptica has not been investigated in vivo. To address this, we developed a plasmid, pGFLIP, that encodes a sensitive Flp recombinase-based fluorescent reporter system able to document gene activation both in vitro and in vivo. Using pGFLIP, we demonstrated that cyaA, considered to be a “late” Bvg⁺ phase gene, is activated substantially earlier in B. bronchiseptica than B. pertussis following a switch from Bvg⁻ to Bvg⁺ phase conditions. We show that the altered activation of cyaA is not due to differences in the cyaA promoter or in the bvgAS alleles of B. bronchiseptica compared to B. pertussis, but appears to be species specific. Finally, we used pGFLIP to show that flaA remains repressed during infection, confirming that B. bronchiseptica does not modulate to the Bvg⁻ phase in vivo.

INTRODUCTION

Bordetella species are Gram-negative bacterial respiratory pathogens. The genus includes Bordetella pertussis, an obligate human pathogen and the causative agent of whooping cough, and the closely related Bordetella bronchiseptica, which can infect a wide range of mammals, including several species that are commonly studied in the laboratory (1–3). These bacteria rely on the global two-component regulatory system BvgAS for virulence (1–3). The BvgAS phosphorelay regulates gene expression patterns according to environmental cues and controls at least three distinct phenotypic phases: Bvg minus (Bvg⁻), Bvg plus (Bvg⁺), and Bvg intermediate (Bvgⁱ) (4, 5). All evidence thus far suggests that the Bvg⁺ phase is necessary and sufficient for infection and that modulation to the Bvg⁻ or Bvgⁱ phase does not occur in vivo (6–8). Although it has been hypothesized that the Bvgⁱ phase and/or the Bvg⁻ phase is required for transmission or survival outside a host (7, 9, 10), a recent study provided evidence that neither of these phenotypic phases is required for B. bronchiseptica transmission in swine (8).

Genes that define these three phenotypic phases have been divided into four classes based on their expression profile. Class 1 (late Bvg⁺ phase) genes include cyaA (encoding the bifunctional hemolysin/adenylate cyclase toxin ACT) and ptxA (encoding the catalytic subunit of the AB₅-type pertussis toxin) (4). Class 2 (early Bvg⁺ phase) genes include those encoding filamentous hemagglutinin (fhaB), fimbriae (fim2 and fim3), and bvgAS itself (4). bipA is the only class 3 (Bvgⁱ phase) gene that has been characterized. Class 4 (Bvg⁻ phase) genes include those encoding proteins involved in motility (frlAB) and chemotaxis in B. bronchiseptica (4).

Expression of genes that define the various Bvg-dependent phenotypic phases is determined mechanistically by the location and affinity of BvgA binding sites near the transcription start site (4). Class 1 genes contain multiple low-affinity BvgA binding sites 5′ distal to the start of transcription (4, 11), while class 2 genes contain high-affinity BvgA binding sites proximal to the transcription start site (12–14). The promoter region of the class 3 gene bipA contains high-affinity BvgA binding sites 5′ proximal to the transcription start site and low-affinity sites 3′ to the transcription start site (10, 15). Although it has been hypothesized that BvgAS directly represses transcription of frlAB in B. bronchiseptica (16), BvgA binding to the frlAB promoter has not been demonstrated and BvgAS-mediated repression of at least some genes in B. pertussis is indirect (17).

In vitro transcription and DNA binding studies indicate that the phosphorylated form of BvgA (BvgA∼P) is required to activate transcription of Bvg⁺ phase genes and that a higher concentration of BvgA∼P is necessary to bind “late” Bvg⁺ phase promoters than “early” Bvg⁺ phase promoters (11, 18). Although the natural signals that activate the BvgAS system are unknown, it is possible to modulate Bordetella spp. to the Bvg⁻ phase in the laboratory by adding a chemical modulator (MgSO₄ or nicotinic acid) to the growth medium or by growing bacteria at 25°C (15, 19–21). When bacteria are grown under Bvg⁻ phase conditions, class 4 genes are expressed while class 1 to 3 genes are not expressed. Following a switch from Bvg⁻ phase to Bvg⁺ phase conditions, transcription of class 4 genes ceases and transcription of class 2 genes, along with the sole class 3 gene bipA, is immediately activated. After several hours, class 1 genes are expressed and class 3 (bipA) genes are repressed. These data are consistent with the model in which the concentration of BvgA∼P within the cell is nearly zero in the Bvg⁻ phase, low in the Bvgⁱ phase, and high in the Bvg⁺ phase (4, 5, 11, 15, 18, 22, 23).

In contrast with the extensive in vitro characterization of the steady-state expression patterns of BvgAS-regulated genes in both B. pertussis and B. bronchiseptica, as well as kinetic analyses of gene expression upon switching from modulating to nonmodulating conditions and vice versa (5, 10, 11, 15, 22–24), only one study has investigated BvgAS-dependent gene regulation in vivo (24). Veal-Carr et al. utilized recombination-based in vivo expression technology (RIVET) to analyze the kinetics of BvgAS-activated gene expression in B. pertussis both in vitro and following intranasal infection of mice (24, 25). They showed that the in vivo activation of Bvg⁺ phase genes, including ptxA, cyaA, fhaB, and prn, temporally recapitulated the activation pattern of these genes upon switching B. pertussis from Bvg⁻ to Bvg⁺ phase conditions in vitro; i.e., fhaB was activated early postinoculation (p.i.) (∼1 h), followed by prn (∼4 h), then later by cyaA (∼12 h) (24). Significantly, the fact that the patterns of gene activation were nearly identical in bacteria switched from Bvg⁻ to Bvg⁺ phase conditions in vitro and after inoculation of mice indicates that the mouse lung is a Bvg⁺ phase environment (24).

We constructed a plasmid, pGFLIP, that encodes a Flp recombinase-based fluorescent reporter system to assess the activation kinetics of genes in vivo (26). The region of pGFLIP delivered to the chromosome contains gfp and nptII genes, encoding green fluorescent protein (GFP) and neomycin phosphotransferase (conferring kanamycin resistance [Km^r]), respectively, flanked by Flp recombinase target (FRT) sites. The plasmid also contains a promoterless flp recombinase gene with a multiple cloning site (MCS) immediately 5′ to the start codon. Upon expression of flp under the control of a promoter of interest, the gfp and nptII genes are permanently excised from the bacterial chromosome.

To test our system, we cloned the promoter regions of several BvgAS-controlled genes, including the Bvg-activated genes cyaA, fhaB, and ptxA and the Bvg-repressed gene flaA (encoding flagellin), into pGFLIP and evaluated transcription activation in vitro and in vivo in B. bronchiseptica. Among other results, we found unexpectedly that the cyaA gene is expressed differently in B. pertussis and B. bronchiseptica.

MATERIALS AND METHODS

Strains, reagents, and growth conditions.

Escherichia coli was grown in lysogeny broth (LB; 10 g/liter tryptone, 5 g/liter yeast extract, 10 g/liter NaCl) or on LB plates with 1.5% agar at 37°C. Bordetella was grown in Stainer-Scholte (SS) broth (27) or on Bordet-Gengou (BG) plates with 1.5% agar (BD Biosciences, San Jose, CA) supplemented with 7.5% (B. bronchiseptica) or 15.0% (B. pertussis) defibrinated sheep's blood (Colorado Serum Company, Denver, CO) at 37°C (28). As required, culture media were supplemented with kanamycin (Km; 50 or 100 μg/ml), ampicillin (100 μg/ml), streptomycin (Sm; 25 μg/ml), MgSO₄ (20 mM or 50 mM), heptakis (1 mg/ml; Sigma), and diaminopimelic acid (DAP; 400 μg/ml) for the Δasd ΔaphA mobilizer strain RHO3 (29). Unless otherwise noted, all restriction enzymes and T4 DNA ligase were purchased from New England BioLabs.

Construction of pGFLIP and derivatives containing Bordetella promoters.

The Tn7 transposition plasmid pUC18T-mini-Tn7T-Km-FRT (30) was digested with BamHI, resulting in fragments of 3,636 and 1,299 bp in length containing the plasmid backbone and the nptII (Km^r) gene, respectively. Separately, a 797-bp fragment containing gfp driven by the constitutive Burkholderia pseudomallei rpsL promoter P_S12 was amplified by PCR from mini-Tn7-kan-gfp (31) using Pfu Ultra II (Agilent) and primers GFP_UP and GFP_DN. This fragment was blunt-end ligated into the cloning vector pJET1.2/blunt (Fermentas) and was transformed into E. coli DH5α according to the manufacturer's instructions. Using restriction sites introduced by PCR, the P_S12-gfp fragment was digested from pJET using BamHI and was ligated together with the BamHI-digested pUC18T-mini-Tn7T-Km-FRT backbone and nptII fragment. As the flp gene would be sensitive to transcription readthrough from either the P_S12 or the nptII promoter, primers specific to gfp (gfpseqR) and nptII (kanseqR) were used to confirm that both genes would be transcribed opposite the promoter of interest and would therefore not drive flp expression. Once the orientation of gfp and nptII was verified, the plasmid was digested with KpnI and StuI; a fragment containing promoterless flp amplified by PCR from pFLPe4 (30) using primers FLP_UP and FLP_DN was likewise digested with KpnI and StuI and was ligated into the digested vector. The resulting plasmid, pGFLIP, thus contained an MCS 5′ to flp, the flp recombinase gene, and constitutively expressed nptII and gfp genes flanked by FRT sites. pGFLIP was fully sequenced using a primer-walking approach with the primers listed in Table 1.

Table 1.

Primers used in this study

Primer	Sequence (5′–3′)a	Description
FLP_UP	ATCTACGGTACCATGAGCCAGTTCGATATCC	Forward and reverse primers to amplify flp from pFLPe4
FLP_DN	AGGTCCAGGCCTCTATATGCGTCTATTTATG
GFP_UP	ATATATGGATCCCAGCTGTTGACTCGCTTG	Forward and reverse primers to amplify gfp from mini-Tn7-kan-gfp
GFP_DN	ACCTGGGGATCCTTATTTGTATAGTTCATCC
Tn7seqF	GAGCGCTTTTGAAGCTGATGTGCT	Forward sequencing primer annealing within Tn7R
gfpseqR	GATGACGGGAACTACAAGACACGT	Reverse sequencing primer annealing within gfp
kanseqR	ATCGCCTTCTATCGCCTTCTTGAC	Reverse sequencing primer annealing within nptII
GFLseq1	ACGGTGAAAACCTCTGACACATGC	Forward sequencing primers for sequencing pGFLIP by primer walking
GFLseq2	CTGAAATCAGTCCAGTTATGCTGTG
GFLseq3	AAATCCGCCGCTAGGAGCTT
GFLseq4	GTCTGCCATGATGTATACATTGTGTG
GFLseq5	GGGACAACTCCAGTGAAAAGTTCTTC
GFLseq6	CTGATGCTCTTCGTCCAGATCATC
GFLseq7	CCTGCGTGCAATCCATCTTGTTCA
GFLseq8	CAGGGGATCTTGAAGTTCCTATTCCG
GFLseq9	GCAACAATTCTGGAAGCCTCATT
GFLseq10	TCTTTAGCGCAAGGGGTAGGATCG
GFLseq11	TCCAATTGAGGAGTGGCAGCAT
GFLseq12	TATCAGAGCTTATCGGCCAGCCT
GFLseq13	ATAAAGATACCAGGCGTTTCCCCC
GFLseq14	AAACAAACCACCGCTGGTAGC
GFLseq15	CGCAGAAGTGGTCCTGCAACTTTA
GFLseq16	CCGCGCCACATAGCAGAACTTTAA
gflpmcsF	GTTGACAAAGGGAATCAGGGGATC	Forward sequencing primer for inserts in the MCS
gflpmcsR	GAACTGGGTGTAGCGTCGTAAGCT	Reverse sequencing primer for inserts in the MCS
glmSF	CAGCTGCTGTCGTACCACACGG	Forward primer to confirm Tn7 insertion at glmS by PCR
cyaA_F	ATTATAGAGCTCTGCGAGCAGATGCAC	Forward and reverse primers to amplify PcyaA from RB50 and BPSM
cyaA_R	TATAATGGTACCGTGGATCTGTCGATAAGTAG
ptxprF	AGCTTCGAGCTCCAAGATAATCGTCCTGCTC	Forward and reverse primers to amplify PptxA from BPSM
ptxprR	ATATATGGGCCCTCCCGTCTTCCCCTCT
fhaprF2	AGGCCTGAGCTCGATAAGAAGAATATGCTT	Forward and reverse primers to amplify PfhaB from RB50
fhaprR2	ATATTCGGTACCATTCCGACCAGCGAAGTG
flaA_F	AATGAGCTCGCCGTGCTCAACGTCA	Forward and reverse primers to amplify PflaA from RB50
flaA_R	ATTATAGGTACCAGGCTCCCAAGAGAGAAA

^aThe underlined sequences represent restriction enzyme cut sites.

Promoters for five Bordetella genes (cyaA, cyaA_Bp, fhaB, ptxA, and flaA) were cloned into the MCS of pGFLIP as follows. For cyaA and cyaA_Bp, 605- and 604-bp fragments containing the cyaA promoter were amplified by PCR from B. bronchiseptica RB50 and B. pertussis BPSM, respectively, using primers cyaA_F and cyaA_R. These fragments were digested with SacI and KpnI and ligated into pGFLIP. For fhaB, a 426-bp fragment containing the fhaB promoter was amplified by PCR from RB50 using primers fhaprF2 and fhaprR2, digested with SacI and KpnI, and ligated into pGFLIP. For ptxA, a 454-bp fragment containing the ptxA promoter was amplified by PCR from BPSM using primers ptxprF and ptxprR, digested with SacI and ApaI, and ligated into pGFLIP. For flaA, a 514-bp fragment containing the flaA promoter was amplified by PCR from RB50 using primers flaA_F and flaA_R, digested with SacI and KpnI, and ligated into pGFLIP. These constructs were delivered to the chromosome by transposase-mediated insertion as described below.

Transposase-mediated delivery of pGFLIP to the B. bronchiseptica attTn7 site.

The pGFLIP plasmid was delivered to Bordetella by triparental mating using a procedure modified from reference 28. B. bronchiseptica strains RB50 and RB52 were grown on BG agar for 48 h, and a portion of the cells was coincubated with conjugation-competent E. coli RHO3 cells (29) harboring pGFLIP and RHO3 cells containing the helper plasmid pTNS3 (30), which expresses tnsABCD from a constitutive promoter, on BG agar supplemented with DAP for 6 h at 37°C. Following incubation, cells were restreaked onto BG-Km agar containing 20 mM MgSO₄ (to maintain strains containing pGFLIP with Bvg⁺ phase promoters in the Bvg⁻ phase) or without MgSO₄ (to maintain RB50flaAFLP in the Bvg⁺ phase) and were incubated an additional 48 h at 37°C. Delivery of pGFLIP to BPSM followed the same procedure except that incubations required 4 days at 37°C and plates were supplemented with 50 mM MgSO₄. pGFLIP without a promoter driving flp was used as a positive control for GFP production and as a negative control for flp activation. The delivery of all constructs to the attTn7 site was confirmed by PCR using primers glmSF and gfpseqR (data not shown).

Evaluation of pGFLIP in vitro.

We determined the functionality of pGFLIP using a plate-based assay in which RB50cyaAFLP and RB50flaAFLP were modulated between promoter-inactive and promoter-active conditions. BG agar containing 20 mM MgSO₄ was used to derepress P_flaA and to deactivate P_cyaA, while BG agar without MgSO₄ was used to activate P_cyaA and to repress P_flaA (7). The pGFLIP strains were grown under promoter-inactive conditions for 48 h at 37°C and were determined to be GFP positive (GFP⁺) using a G:BOX Chemi imaging system with an UltraBright-LED blue transilluminator and an SW06 short-pass filter (495 to 600 nm; Syngene, Frederick, MD). A single GFP⁺ colony was resuspended in PBS and was diluted and plated on BG agar under promoter-active and promoter-inactive conditions. To demonstrate the loss of Km^r upon promoter activation, Km (100 μg/ml) was added to some plates. Plates were incubated for 48 h at 37°C, and white-light and fluorescent images were obtained using the G:BOX Chemi system.

For kinetic assays, strains containing pGFLIP were grown overnight with Km selection in SS under promoter-inactive conditions (20 mM MgSO₄ for RB50cyaAFLP, RB50cyaA_BpFLP, RB50fhaBFLP, and RB50ptxAFLP; no added MgSO₄ for RB50flaAFLP). Cells were washed twice in PBS with or without 5 mM MgSO₄ to prevent the premature activation of Bvg⁺ phase and flaA promoters, respectively, and were added to fresh SS medium under promoter-active conditions (20 mM MgSO₄ for RB50flaAFLP; no added MgSO₄ for RB50cyaAFLP, RB50cyaA_BpFLP, RB50fhaBFLP, and RB50ptxAFLP) at an optical density at 600 nm (OD₆₀₀) that was equivalent to 1 × 10⁹ CFU/ml. Tubes were incubated on a roller at 37°C for 8 h. At each time point, an aliquot of cells was removed, diluted in phosphate-buffered saline (PBS) under promoter-inactive conditions, and plated on BG agar under promoter-inactive conditions. Plates were incubated for 48 h at 37°C, and total CFU were enumerated. For kinetic assays with B. pertussis, cultures were inoculated at the CFU/ml corresponding to an OD₆₀₀ of ∼0.1, and cells were grown overnight in SS containing 50 mM MgSO₄, 100 μg/ml Km, and 1 mg/ml heptakis. Cells were washed in PBS containing 25 mM MgSO₄ to prevent premature flp activation. All other aspects of the assay were the same as for B. bronchiseptica, except that dilutions were plated on BG agar containing 50 mM MgSO₄ and plates were incubated 4 days at 37°C. GFP fluorescence was quantified using the G:BOX Chemi imaging system as described above. The percent resolution (the ratio of GFP⁻ colonies to the total number of colonies × 100%) was calculated for each strain at each time point.

Evaluation of pGFLIP in a murine model of infection.

Five- to 7-week-old BALB/c mice (Jackson Laboratories, Bar Harbor, ME) were inoculated intranasally with 1 × 10⁵ CFU of B. bronchiseptica pGFLIP strains in 50 μl of PBS. For infections with RB50cyaAFLP, RB50cyaA_BpFLP, RB50fhaBFLP, and RB50ptxAFLP, bacteria were grown overnight in SS medium containing Km (100 μg/ml) and 20 mM MgSO₄ to maintain cells in the Bvg⁻ phase. For infections with RB50flaAFLP, bacteria were grown overnight in SS media containing Km (100 μg/ml) to maintain cells in the Bvg⁺ phase. Lungs were harvested from infected mice at 1 h and 30 h postinoculation (p.i.). For RB50flaAFLP, right lungs were homogenized in 1 ml of PBS, diluted in PBS, and plated in duplicate on BG agar. For strains containing Bvg⁺ phase promoters in pGFLIP, homogenization, dilution, and plating were carried out in the presence of 20 mM MgSO₄. Percent resolution was calculated for each strain at each time point. This study was done in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Our protocol was approved by the University of North Carolina IACUC (protocol 12-307.0). All animals were properly anesthetized for inoculations, monitored regularly, and euthanized when moribund, and efforts were made to minimize suffering.

Statistical analyses.

Statistical analyses were performed using Prism 5 (GraphPad Software, Inc.). Statistical significance was determined using the unpaired Student's t test or analysis of variance (ANOVA). Images were formatted using Adobe Photoshop CS5, and figures were generated using Adobe Illustrator CS5 (Adobe Systems, Inc.).

RESULTS

Design and construction of pGFLIP.

To study transcription activation of genes both in vitro and in vivo, we engineered a reporter system, pGFLIP, that provides both fluorescent and selectable markers. We designed our system for simplicity and ease of use. Therefore, all of the components necessary for pGFLIP function are contained on a single plasmid. The region of pGFLIP delivered to the chromosome includes gfp and nptII, each under the control of a strong constitutive promoter and together flanked by FRT sequences (Fig. 1A). The Flp recombinase gene flp is present in the opposite orientation from gfp and nptII with an immediate 5′ MCS to facilitate the insertion of a promoter of interest. In the absence of promoter activity, flp is not transcribed and cells remain GFP⁺ and Km^r. When the promoter of interest is activated, flp is transcribed and the gene product mediates recombination between FRT sites, permanently excising gfp and nptII.

Fig 1.

Design and mechanism of pGFLIP. (A) Diagram of pGFLIP. Tn7R and Tn7L, left and right ends of the Tn7 transposon, respectively; T₀ and T₁, bacteriophage λ and E. coli rrnB transcriptional terminators, respectively; FRT, Flp recombinase target; gfp, green fluorescent protein gene; nptII, neomycin phosphotransferase gene; MCS, multiple cloning site with restriction sites indicated; flp, Flp recombinase gene; oriT, origin of conjugative transfer; ori, ColE1 origin of replication; and bla, β-lactamase gene. (B) Schematic illustration of Tn7-mediated delivery and Flp-mediated excision of pGFLIP in RB50. The region of pGFLIP flanked by Tn7L and Tn7R sequences is delivered to the attTn7 site located between glmS and BB4801. While the promoter driving expression of the gene of interest (P_gene) remains inactive, gfp and nptII are expressed constitutively, resulting in fluorescent and Km^r bacteria. When P_gene is activated, flp is expressed and Flp recombinase mediates site-specific recombination between FRT sites, permanently excising gfp and nptII and yielding bacteria that are nonfluorescent and sensitive to Km.

To test the system, we cloned promoter regions from B. bronchiseptica and B. pertussis into the MCS to generate the strains described in Table 2. The region of pGFLIP between the Tn7L and Tn7R ends was delivered to the attTn7 site 3′ to the glmS gene in Bordetella spp. as described in Materials and Methods. Transcription terminators are present near the 5′ end of the transposon to prevent transcription readthrough from glmS. Delivery of promoterless pGFLIP to B. bronchiseptica RB50 and B. pertussis BPSM resulted in strains that were stably GFP⁺ and Km^r. These strains did not lose GFP fluorescence or Km^r when passaged multiple times in vitro in the absence of selection (data not shown).

Table 2.

Strains and plasmids used in this study

Strain or plasmid	Description	Source or reference
Strains
E. coli
DH5α	Molecular cloning strain	28
RHO3	Conjugation strain; Kms Δasd ΔaphA	29
Bordetella
RB50	Wild-type B. bronchiseptica strain; Smr	7
RB52	RB50 containing bvgAS from BP338; Smr	22
BPSM	Smr Tohama I derivative	29, 32
RB50FLP	RB50 with promoterless pGFLIP integrated at attTn7	This study
RB50cyaAFLP	RB50 with flp recombinase driven by PcyaA integrated at attTn7	This study
RB50cyaABpFLP	RB50 with flp recombinase driven by PcyaA from Tohama I integrated at attTn7	This study
RB52cyaABbFLP	RB52 with flp recombinase driven by PcyaA from RB50 integrated at attTn7	This study
RB52cyaABpFLP	RB52 with flp recombinase driven by PcyaA from Tohama I integrated at attTn7	This study
RB50fhaBFLP	RB50 with flp recombinase driven by PfhaB from RB50 integrated at attTn7	This study
RB50ptxAFLP	RB50 with flp recombinase driven by PptxA from Tohama I integrated at attTn7	This study
RB50flaAFLP	RB50 with flp recombinase driven by PflaA integrated at attTn7	This study
BPSMFLP	BPSM with promoterless pGFLIP integrated at attTn7	This study
BPSMcyaAFLP	BPSM with flp recombinase driven by PcyaA from RB50 integrated at attTn7	This study
BPSMfhaBFLP	BPSM with flp recombinase driven by PfhaB from RB50 integrated at attTn7	This study
BPSMptxAFLP	BPSM with flp recombinase driven by PptxA integrated at attTn7	This study
Plasmids
pUC18T-mini-Tn7T-Km-FRT	Mobilizable transposition vector; Apr Kmr	30
pFLPe4	Site-specific excision vector, source of Flp recombinase; Apr Kmr	30
mini-Tn7-kan-gfp	Mobilizable transposition vector, source of gfp driven by PS12; Kmr	31
pTNS3	Tn7 transposase expression vector containing tnsABCD; Apr	30
pGFLIP	pUC18-based vector with PS12-gfp and nptII flanked by FRT sequences and flp 3′ to the MCS; Apr Kmr	This study
pGFLIP-PcyaA	pGFLIP with flp recombinase driven by the RB50 cyaA promoter; Apr Kmra	This study
pGFLIP-PcyaABp	pGFLIP with flp recombinase driven by the BPSM cyaA promoter; Apr Kmra	This study
pGFLIP-PfhaB	pGFLIP with flp recombinase driven by the RB50 fhaB promoter; Apr Kmra	This study
pGFLIP-PptxA	pGFLIP with flp recombinase driven by the BPSM ptxA promoter; Apr Kmra	This study
pGFLIP-PflaA	pGFLIP with flp recombinase driven by the RB50 flaA promoter; Apr Kmra	This study

^aKm^r only under promoter-inactive conditions.

Functional evaluation of pGFLIP in vitro.

To evaluate the functionality of pGFLIP, we used two B. bronchiseptica promoters for which activity can readily be induced in vitro: P_cyaA and P_flaA. In B. bronchiseptica, the flaA promoter is highly transcribed in the Bvg⁻ phase, while the cyaA promoter is transcribed strongly in the Bvg⁺ phase (9, 10, 15, 22, 33). Therefore, RB50flaAFLP was constructed under Bvg⁺ phase conditions to prevent expression of flaA, while RB50cyaAFLP was maintained under Bvg⁻ phase conditions to prevent expression of cyaA. Km was added to the media during construction of these strains to select against any bacteria that had activated flp, thus ensuring that the population only contained GFP⁺ and Km^r cells. To test these strains for pGFLIP functionality, we suspended a single GFP⁺ colony of each strain in PBS and plated on BG, BG-Km, BG-MgSO₄, and BG-Km-MgSO₄ agar and observed colony formation and GFP fluorescence after two (B. bronchiseptica) or four (B. pertussis) days of incubation at 37°C.

When plated under promoter-active conditions (20 mM MgSO₄ for RB50flaAFLP; no added MgSO₄ for RB50cyaAFLP) in the absence of Km, all RB50flaAFLP and RB50cyaAFLP colonies examined had lost fluorescence, and when tested subsequently for growth on agar containing Km, all colonies had lost Km^r (Fig. 2A and D). As expected, plating either strain under promoter-active conditions and in the presence of Km resulted in a lack of growth due to loss of nptII from the chromosome (Fig. 2E and H). Conversely, when plated under promoter-inactive conditions (no added MgSO₄ for RB50flaAFLP; 20 mM MgSO₄ for RB50cyaAFLP) in the presence of Km, all RB50flaAFLP and RB50cyaAFLP colonies maintained fluorescence (Fig. 2F and G). Likewise, under promoter-inactive conditions in the absence of Km, all RB50flaAFLP colonies maintained fluorescence (Fig. 2B), indicating that flaA is not expressed under these conditions. Unexpectedly, however, when RB50cyaAFLP was plated under promoter-inactive conditions in the absence of Km selection, approximately 14% of the colonies had lost fluorescence at the time of imaging (Fig. 2C). These colonies did not grow when restreaked onto BG-Km-MgSO₄, indicating that cells in these colonies were not gfp mutants but had lost fluorescence and Km^r due to Flp-mediated recombination. These results suggest that the transcription activity of P_cyaA under Bvg⁻ phase conditions in vitro is close to the threshold level of flp transcription required for Flp-mediated recombination between FRT sites. These plate-based assays for fluorescence and Km^r demonstrated that pGFLIP is indeed functional and that both markers (gfp and nptII) can be used to document promoter activation under active and inactive conditions.

Fig 2.

Plate-based validation of the pGFLIP reporter system using P_cyaA and P_flaA. Left column, images of RB50cyaAFLP plated under Bvg⁺ phase (A) and Bvg⁻ (C) phase conditions (achieved by supplementing BG agar with 20 mM MgSO₄) in the presence (E) or absence (G) of 100 μg/ml Km. Right column, images of RB50flaAFLP plated under Bvg⁺ phase (B) and Bvg⁻ (D) phase conditions in the presence (F) or absence (H) of 100 μg/ml Km. White light photographs are on the left and fluorescent images are on the right in each pair of images.

Kinetic analysis of Bordetella gene activation using pGFLIP.

We next assessed the ability of this system to monitor gene activation over time in bacteria grown in liquid culture. Strains containing pGFLIP plasmids with cyaA, fhaB, ptxA, and flaA promoters were grown overnight under promoter-inactive conditions and switched to promoter-active conditions, and the percent resolution (i.e., loss of GFP fluorescence and Km^r) was calculated over 8 h. For the first 4 h after switching from Bvg⁺ phase to Bvg⁻ phase conditions, the percentage of RB50flaAFLP colonies that had lost fluorescence remained below 20% (Fig. 3A), indicating that less than 20% of the bacteria had reached or surpassed the threshold of flp expression to result in recombination during this time. Approximately 75% of cells had lost fluorescence at 6 h, and this proportion was maintained at 8 h (Fig. 3A). When grown under Bvg⁻ phase conditions for 24 h, more than 85% of cells demonstrated a loss of fluorescence and Km^r (data not shown). The fact that resolution did not reach 100% suggests that the maximum level of flaA expression under in vitro Bvg⁻ phase conditions (SS medium containing 20 mM MgSO₄) is just at or above the threshold of flp expression needed for Flp-mediated recombination between FRT sites. The relatively long amount of time required for the majority of RB50flaAFLP cells to cross the threshold of flp expression is in agreement with a previous study that examined the expression kinetics of the BvgAS-regulated operon frlAB, which encodes the E. coli FlhDC homologs FrlAB that are at the top of the flagellum transcription cascade in B. bronchiseptica (16). Upon shifting from Bvg⁺ phase to Bvg⁻ phase conditions, a frlAB-lacZ fusion strain exhibited a gradual increase in β-galactosidase activity over time, reaching approximately 40% of the maximum overnight expression level by 7.5 h (15).

Fig 3.

Kinetic analysis of Bordetella gene activation in vitro. B. bronchiseptica strains grown under promoter-off conditions and in the presence of 100 μg/ml Km were washed and placed in fresh SS medium under promoter-on conditions (A) or promoter-off conditions (B). Colonies arising from aliquots plated over 8 h were monitored for loss of fluorescence, and the percent resolution was calculated. Results are the means ± standard errors of the means (SEM) for experiments performed in duplicate or triplicate.

In contrast, when RB50fhaBFLP cells were switched from Bvg⁻ phase to Bvg⁺ phase conditions, the fhaB promoter was apparently activated as early as 1 h, at which time approximately 75% of cells had lost fluorescence (Fig. 3A). At 2 h and later time points, nearly 100% of RB50fhaBFLP cells had lost fluorescence, consistent with the classification of fhaB as a class 2 gene requiring a relatively low level of BvgA∼P for activation. As a canonical class 1 (or “late”) gene, ptxA had only been activated in approximately 8% of cells at 1 h (Fig. 3A). Approximately 20% of cells had activated ptxA by 4 h, increasing to 90% of cells by 8 h. Surprisingly, cyaA, which is also considered to be a class 1 gene, was activated in over 85% of cells at 1 h, and in more than 95% of cells at 2 h and later time points (Fig. 3A). These data suggest that at early time points after switching to promoter-active conditions, the cyaA promoter is activated to a level sufficient to drive flp-mediated recombination. This result was unexpected based on cyaA transcription results in studies of B. pertussis (5, 11, 18, 23, 24).

At the initiation of each kinetic experiment, we observed a consistent background resolution (i.e., the percent resolution at 0 h) for each Bvg⁺ phase promoter in pGFLIP. RB50cyaAFLP displayed the highest background resolution at 8.9%, while the background resolution for fhaB and that for ptxA were significantly lower at 1.54% and 1.99%, respectively (P < 0.01). The background resolution for RB50flaAFLP was the lowest of the pGFLIP constructs at 0.23% (Fig. 3A). The fact that a portion of RB50cyaAFLP cells had lost fluorescence at 0 h suggests that low-level expression of cyaA might be occurring in a BvgAS-independent manner, which could interfere with measuring BvgAS-dependent P_cyaA activation over time. Therefore, to test the sensitivity of pGFLIP to background resolution, we grew RB50cyaAFLP, RB50fhaBFLP, and RB50ptxAFLP strains overnight as shown in Fig. 3A, washed the cells to remove Km, and maintained the cultures under Bvg⁻ conditions in the absence of selection for 8 h. At 0, 1, 4, and 8 h, the percent resolution was determined for each strain. We hypothesized that if P_cyaA was indeed active under Bvg⁻ conditions, we would observe a steady increase in resolution over time for RB50cyaAFLP but not for RB50fhaBFLP or RB50ptxAFLP. As expected, RB50cyaAFLP displayed the greatest background resolution at 11.9%, increasing to 18.1% at 8 h (Fig. 3B). Loss of fluorescence for RB50fhaBFLP and RB50ptxAFLP remained essentially unchanged over 8 h, averaging 0.22% and 5.4%, respectively (Fig. 3B). Although cyaA had been activated in approximately one-fifth of the population at 8 h, these data do not account for the >85% resolution observed for RB50cyaAFLP as early as 1 h following the switch to Bvg⁺ phase conditions.

Neither the B. bronchiseptica bvgAS allele nor P_cyaA accounts for the unexpectedly early activation of cyaA.

As cyaA expression in RB50 was activated unexpectedly early compared to what has been observed for B. pertussis (5, 11, 18, 23, 24), we sought to determine whether differences in the cyaA promoter or in the bvgAS allele between B. bronchiseptica and B. pertussis accounted for this difference. There are five single nucleotide changes and one nucleotide insertion in the sequence 5′ to cyaA in B. bronchiseptica RB50 compared to B. pertussis BPSM. We hypothesized that replacing the cyaA promoter of RB50 with that of BPSM would delay the activation of P_cyaA relative to P_fhaB, similar to what was observed by Veal-Carr et al. (24). We cloned the cyaA promoter from B. pertussis BPSM into pGFLIP and introduced this plasmid into RB50, generating strain RB50cyaA_BpFLP. When evaluated in the kinetic assay, there was no difference in the rate of resolution between RB50cyaAFLP and RB50cyaA_BpFLP, suggesting that differences in the cyaA promoter do not account for the rapid resolution observed in B. bronchiseptica (Fig. 4).

Fig 4.

Effect of P_cyaA and bvgAS alleles on cyaA activation in B. bronchiseptica. RB50 and RB52, an RB50 derivative carrying the bvgAS allele from B. pertussis BP338 in place of the native bvgAS allele, each with pGFLIP containing RB50 and BPSM P_cyaA, were grown as shown in Fig. 3 and were switched to promoter-on conditions for 8 h. Loss of fluorescence was calculated for each strain as described in Materials and Methods. Results are the means ± SEM for experiments performed in duplicate.

Some strains of B. pertussis, such as BP338 and BPSM, exhibit decreased sensitivity to chemical modulation compared to B. bronchiseptica RB50 due to amino acid differences in BvgS, which causes these strains to remain in the Bvg⁺ phase at a lower concentration of modulator than RB50 (22). We hypothesized, therefore, that the bvgAS allele from BP338 would permit activation of cyaA more quickly and in a greater percentage of cells than the B. bronchiseptica RB50 bvgAS allele. To assess this, we utilized strain RB52, which contains the entire bvgAS locus and bvgA promoter from B. pertussis BP338 in place of bvgAS in RB50 (22). RB52 recapitulates the decreased sensitivity to modulation characteristic of both BP338 and BPSM, requiring ≥40 mM MgSO₄ to modulate to the Bvg⁻ phase, in contrast to RB50, which requires ≥10 mM MgSO₄ to fully modulate (15, 22). We introduced pGFLIP containing the RB50 cyaA promoter driving flp expression into RB52, producing strain RB52cyaA_BbFLP. We evaluated the kinetics of cyaA activation in RB52cyaA_BbFLP compared to RB50cyaAFLP and did not observe any difference. Likewise, when we evaluated an RB52 derivative containing the BPSM cyaA promoter driving flp expression (RB52cyaA_BpFLP) in the kinetic assay, there was no impact on cyaA activation compared to RB50cyaAFLP. Together, these data suggest that neither the cyaA promoter nor the bvgAS allele significantly affects the kinetics of cyaA activation in B. bronchiseptica as reported by the P_cyaA-flp promoter fusion in pGFLIP (Fig. 4).

Evaluation of Bordetella promoter activation in vivo.

Although the kinetics of Bvg⁺ phase gene activation in vivo have been examined for B. pertussis (24), these experiments have not been done in B. bronchiseptica or in the context of a natural bacterium-host interaction. To address BvgAS-regulated gene activation in B. bronchiseptica in vivo, we infected BALB/c mice intranasally with 1 × 10⁵ CFU of the RB50 pGFLIP strains shown in Fig. 3A, grown under promoter-inactive conditions (100 μg/ml Km, 20 mM MgSO₄). Mice were sacrificed and lungs were harvested at 1 h and 30 h p.i. Lungs were homogenized and dilutions were plated on BG agar containing 20 mM MgSO₄ to prevent further recombination. Total CFU were enumerated, and the percentage of colonies that had lost fluorescence (percent resolution) was calculated for each strain and time point. The percentage of RB50fhaBFLP that had lost fluorescence at 1 h was 95.3% ± 0.60%, while only 5.4% ± 0.69% of RB50ptxAFLP had lost fluorescence at this time (Fig. 5). Similar to the in vitro kinetic assay (Fig. 3A), the majority (85.3% ± 3.3%) of RB50cyaAFLP cells had lost fluorescence at 1 h. At 30 h p.i., essentially all cells had lost fluorescence, indicating that the environment in the mouse induces the expression of Bvg⁺ phase genes to a level at or above that required to activate flp expression (Fig. 5). The fact that the pGFLIP system functions in vivo much the same as in vitro when switched from Bvg⁻ to Bvg⁺ phase conditions suggests that the pattern of Bvg⁺ phase gene activation is similar between these two conditions and that the unexpectedly early expression from the cyaA promoter in vitro was not due to an artifact of that assay.

Fig 5.

Analysis of Bordetella gene activation in vivo using pGFLIP. RB50cyaAFLP, RB50fhaBFLP, RB50ptxAFLP, and RB50flaAFLP were grown as shown in Fig. 3, and 1 × 10⁵ CFU was inoculated intranasally into mice in a total volume of 50 μl. Mice were sacrificed at 0, 1, and 30 h p.i., with additional time points at 3, 5, and 7 days for RB50flaAFLP, and lungs were homogenized and plated on BG agar supplemented with 20 mM MgSO₄ and Sm (for RB50cyaAFLP, RB50fhaBFLP, and RB50ptxAFLP) or on BG agar supplemented only with Sm (for RB50flaAFLP). Loss of fluorescence was calculated for each strain as described in Materials and Methods. Results are the means ± SEM for experiments performed in duplicate with three mice per time point.

Although all evidence thus far suggests that the Bvg⁺ phase is necessary and sufficient for Bordetella spp. to cause respiratory infection in rats and mice (6, 7, 22, 34), it is possible that rare in vivo environments exist that induce modulation to the Bvg⁻ phase. Therefore, using the activation of the flaA promoter as an indicator of the Bvg⁻ phase, we infected mice as described above and evaluated the loss of fluorescence at 1 h and 30 h p.i. We did not observe activation of flaA in any RB50flaAFLP cells at 1 h, 30 h, 3 days, 5 days, or 7 days (Fig. 5). These data suggest that B. bronchiseptica does not enter the Bvg⁻ phase in the mouse during the time period that we examined.

Kinetic analysis of Bordetella gene activation in B. pertussis reveals delayed activation of cyaA compared to B. bronchiseptica.

Due to the observation that neither the cyaA promoter nor the bvgAS allele affects the kinetics of cyaA activation in B. bronchiseptica, we hypothesized that additional, species-specific factors account for the differences in the activation of cyaA. To evaluate this possibility, we delivered the same pGFLIP plasmids as those used in B. bronchiseptica to the chromosome of B. pertussis BPSM and evaluated loss of fluorescence in the kinetic assay. We reasoned that if the activation of cyaA, fhaB, and ptxA in B. pertussis resembled the pattern seen in B. bronchiseptica, then the discrepancy with the study by Veal-Carr et al. (24) was likely due to differences in recombinase expression sensitivity between pGFLIP and RIVET (i.e., Flp recombinase is activated at a lower threshold of P_cyaA activity than TnpR, making cyaA appear to be activated sooner in the Flp-FRT system). Alternatively, if activation of cyaA, fhaB, and ptxA was canonical as in Veal-Carr et al. (24), then the difference in cyaA activation would be attributed to species-specific gene regulation.

In contrast to what we observed in B. bronchiseptica, the proportion of BPSMcyaAFLP that had lost fluorescence at 1 h was less than 10%, essentially equal to the level of background resolution (Fig. 6). There was a steady loss of fluorescence in BPSMcyaAFLP beginning at 2 h postswitch and continuing until 6 h, at which time more than 90% of colonies had lost fluorescence. The pattern of P_cyaA activation in B. pertussis matched that of P_ptxA, which displayed similar but somewhat earlier activation in B. pertussis compared to B. bronchiseptica (Fig. 6). Interestingly, the loss of fluorescence in BPSMfhaBFLP occurred more slowly than in RB50fhaBFLP, with approximately 55% of cells having lost fluorescence at 1 h. At 2 h postswitch, the proportion of GFP⁻ BPSMfhaBFLP equaled that of RB50fhaBFLP at 1 h, indicating that the activation of P_fhaB may be delayed in B. pertussis compared to B. bronchiseptica; however, the background resolution for BPSMfhaBFLP (29.5% ± 11.3%) was substantially higher than RB50fhaBFLP. At 1.99%, the background resolution for BPSMptxAFLP was identical to that seen for RB50ptxAFLP, while the background for BPSMcyaAFLP (4.8% ± 2.2%) was not significantly different compared to RB50cyaAFLP (Fig. 6).

Fig 6.

Kinetic analysis of Bordetella gene activation in vitro in B. pertussis. Strains grown under promoter-off conditions and in the presence of 100 μg/ml Km were washed and placed in fresh SS medium under promoter-on conditions. Colonies arising from aliquots plated over 8 h were monitored for loss of fluorescence, and the percent resolution was calculated. Results are the means ± SEM for experiments performed in triplicate.

DISCUSSION

Evaluating bacterial gene expression within the host is critical for understanding the complex host-pathogen interactions that result in disease, clearance, or asymptomatic colonization. One approach that has been developed to evaluate gene expression in the host is in vivo expression technology (IVET) (35). In IVET, promoter sequences are cloned 5′ to a gene that either confers resistance to an antibiotic or complements a specific auxotrophy (35–37). Under conditions in which the promoter is active, these genes are expressed and permit the survival of bacteria in hosts that have either been dosed with the relevant antibiotic (for antibiotic resistance-based selection) or naturally lack the ability to complement the auxotrophy (for auxotrophy-based selection) (35). A drawback of using IVET is that promoters that are activated transiently or at a low level may not be identified due to insufficient production of the missing survival factor (35). In RIVET, a modification of IVET, a promoter of interest drives the expression of a site-specific recombinase that irreversibly excises a genetic marker, often an antibiotic resistance gene (24, 35, 38). By selective plating, it is possible to determine if the promoter of interest was active at some time during infection. Unlike IVET, RIVET permits the detection of transiently or weakly expressed genes because the recombinase-mediated loss of a marker need only occur once to document promoter activation (35).

pGFLIP is a pUC18-based plasmid that, while conceptually similar to RIVET, possesses several advantages over this well-characterized genetic tool. The Tn7 transposon system specifically delivers sequences to the attTn7 site located 3′ to the highly conserved glmS gene, which provides pGFLIP with an especially broad host range that includes many Gram-negative bacteria (39). As a result of this specific recombination at the attTn7 site, only one integration event is required for all components of pGFLIP to be delivered in single copy to the bacterial chromosome. Single-copy delivery eliminates potential gene dosage issues inherent to multicopy plasmid systems, and integration at the attTn7 site does not disturb the native locus of the gene to be tested. Once delivered, a Tn7 transposon is stable in the absence of selection, unlike suicide plasmids that have been integrated into the chromosome via single-crossover homologous recombination, which can spontaneously resolve in the absence of selection. In contrast with other published systems, pGFLIP also possesses two markers—gfp and nptII—allowing either the loss of fluorescence or Km^r to indicate promoter activation.

In this study, we used pGFLIP to analyze the transcription activation of Bvg-regulated genes in both B. bronchiseptica RB50 and B. pertussis BPSM. Our results showed that, in vitro, fhaB and ptxA promoters were activated early and late, respectively, following a switch from Bvg⁻ to Bvg⁺ phase conditions, which is in agreement with previous reports for both Bordetella species (15, 23, 24). However, the cyaA promoter was activated unexpectedly early in our assay; these results appear to stand in contradiction to the established view that cyaA is transcribed solely as a late gene. Using B. bronchiseptica RB50 in a BALB/c mouse model, we found that the pattern of gene activation for cyaA, fhaB, and ptxA was nearly identical to that observed in vitro at early time points. Veal-Carr et al. reported similar agreement between in vitro and in vivo gene activation patterns using RIVET (24), although in that study, maximal activation of fhaB, cyaA, and prn occurred over a much greater time scale (approximately 24 h to full activation), likely a result of differences in sensitivity between the TnpR-res and Flp-FRT systems. Our use of pGFLIP to evaluate gene activation in B. bronchiseptica is both the first kinetic analysis of P_cyaA and the first in vivo kinetic analysis of any promoter to be reported for this organism.

Although we expected P_cyaA to behave like a class 1 promoter in B. bronchiseptica, based on gene activation and expression data obtained by us and others for B. pertussis (5, 11, 18, 23, 24), our data nevertheless suggest that P_cyaA acts more like a class 2 promoter in this organism. However, it is not necessarily the case that cyaA reaches maximal expression at 1 to 2 h following a switch from Bvg⁻ phase to Bvg⁺ phase; it is possible that cyaA activation occurs in a stepwise manner, i.e., P_cyaA may rapidly reach a level of expression necessary to activate flp transcription in our system, but may not reach maximal expression until much later. This scenario would account for the apparently rapid activation of P_cyaA without requiring a bacterium to be producing and secreting a significant amount of ACT immediately upon switching to nonmodulating conditions. Evidence exists to support cyaA activation and ACT production, albeit at a reduced level, in the Bvgⁱ phase, as strains RB50i and RB53i are slightly hemolytic on BG agar (a consequence of the hemolysin activity of ACT), and RB53i produces measurable levels of cyaA transcript (9). We were able to determine that neither differences in the cyaA promoter nor differences in the bvgAS allele between the two species accounted for the difference in cyaA activation (Fig. 4), suggesting that other (potentially Bvg-independent) factors may be influencing cyaA activation in B. bronchiseptica compared to B. pertussis.

In B. pertussis BPSM, P_cyaA demonstrated an activation pattern consistent with both indirect (RIVET) and direct (RNA hybridization) assessments, reinforcing the conclusion that cyaA activation is indeed different between RB50 and BPSM (23, 24). Given the differences in host range and the ability of B. bronchiseptica to survive outside a host (7, 9, 10, 34), it is possible that the relatively early activation of cyaA in B. bronchiseptica is advantageous during the establishment of infection in a mammalian host or in transmission between hosts. Previous studies have shown that RB50 cyaA deletion mutants are more susceptible to clearance from the mouse respiratory tract, presumably as a result of neutrophil-mediated killing, and that ACT may interact with FHA (an “early” Bvg⁺ gene) on the cell surface and modulate cytokine production by the host (40, 41). All strains of B. bronchiseptica that have been tested show the same modulation characteristics in vitro (i.e., the same relatively low concentration of modulator is required for inducing the Bvg⁻ phase), while B. pertussis isolates exhibit variable resistance to modulation at concentrations of modulator equal to or greater than for B. bronchiseptica (22). These observations suggest that early expression of cyaA may not be detrimental to human-adapted B. pertussis but may be necessary for B. bronchiseptica to establish an infection.

For both B. bronchiseptica and B. pertussis, it has been shown that the Bvg⁺ phase is necessary and sufficient for infection (6, 7, 9, 16, 22). Our results do not contradict these observations for B. bronchiseptica, but they demonstrate that once inside the host, bacteria begin to transcribe Bvg⁺ phase genes within 1 h p.i., with essentially every cell having activated fhaB, cyaA, and ptxA by 30 h p.i. The transition of B. bronchiseptica from Bvg⁻ phase to Bvg⁺ phase upon inoculation also indicates that the mouse lung is a Bvg⁺ phase environment. We likewise provided evidence that B. bronchiseptica does not modulate to the Bvg⁻ phase in vivo, supporting studies conducted using Bvg⁻ phase-locked mutants that were unable to establish an infection and were quickly cleared from the respiratory tract and those using Bvg⁺ phase-locked mutants that displayed no colonization defect (7, 8). Moreover, ectopic expression of flagellin in Bvg⁺ phase B. bronchiseptica results in impaired persistence in the rat respiratory tract, possibly due to an immune response to this antigen (16). Therefore, modulation to the Bvg⁻ phase likely does not occur in vivo, as it would be disadvantageous to bacterial survival. We cannot rule out the possibility, however, that some Bvg⁺ RB50flaAFLP bacteria did modulate to the Bvg⁻ phase and were quickly eliminated by the immune response or were present in a niche other than the lung (such as the trachea or nasopharynx) and were not represented in the lung homogenates that we analyzed. It is also conceivable that B. bronchiseptica is able to partially modulate, perhaps to the Bvgⁱ phase, which would not be documented using RB50flaAFLP. We are currently constructing additional strains to test this possibility.

The pGFLIP plasmid has proven to be useful in understanding the regulation of gene activation in Bordetella. However, there remain caveats for the use of this system in other applications. As with other IVET and RIVET systems, pGFLIP requires that strains be manipulated under strict promoter-inactive conditions to prevent unwanted Flp-mediated resolution. This requirement poses problems for studying genes for which conditions of repression or lack of activation are unknown, genes that are essential for growth, and genes that may not be fully transcriptionally inactive in vitro. Additionally, like other systems that have been developed to monitor transcription, pGFLIP cannot provide information about posttranscriptional or posttranslational regulation of target genes. Finally, as was shown by the variable background resolution of Bvg-regulated promoters in our study, pGFLIP appears to be sensitive to low-level promoter activation even under “repressed” conditions for certain genes. Lee et al. were able to modulate the sensitivity of RIVET by mutating the ribosome binding site (RBS) of tnpR, effectively raising the threshold of promoter activity required for resolution (42). pGFLIP does not possess an RBS 5′ to flp, instead relying on the RBS of the promoter of interest, though it would be feasible to develop a modified pGFLIP plasmid that contains an RBS with reduced sensitivity to permit the study of genes that are not fully inactive or are constitutively active at a low level.

In this study, we used pGFLIP to detect the activation of BvgAS-regulated genes in Bordetella, but there are additional uses for this system to measure transcription activation at the population or single cell level. Using pGFLIP, fluorescence can be used to quickly differentiate cells that have activated the promoter of interest from those that have not. Over time, stochastic and/or transient promoter activation can result in sectoring of fluorescent colonies, permitting spatiotemporal observation of gene activation within a single colony (M. S. Byrd and E. Mason, unpublished observation). The addition of a second, non-gfp-encoded fluorescent label (e.g., a constitutively expressed fluorescent protein or a fluorescently labeled antibody) to cells already containing pGFLIP would allow cells that had activated the promoter of interest to be differentiated from cells that had not. Cells labeled using such an approach could be visualized using fluorescence-activated cell sorting or by microscopy. We are currently developing an improved pGFLIP plasmid that contains a constitutively expressed fluorescent protein gene not flanked by FRT sites that will provide a two-color to one-color readout upon activation of flp by the promoter of interest. The development of pGFLIP has resulted in a sensitive genetic tool that can be used to document promoter activation in a broad range of Gram-negative bacteria both in vitro and in vivo. Our use of pGFLIP to document the activation of Bvg-regulated promoters revealed unexpectedly early activation of cyaA in B. bronchiseptica, suggesting a possible explanation for the less restrictive host range of this organism compared to B. pertussis, and is the first in vivo use of a recombination-based genetic reporter of B. bronchiseptica gene activation.