Analysis of BvgA Activation of the Pertactin Gene Promoter in Bordetella pertussis

Abstract

Bordetella pertussis, the causative agent of whooping cough, regulates expression of its virulence factors via a two-component signal transduction system encoded by the bvg regulatory locus. It has been shown by activation kinetics that several of the virulence factors are differentially regulated. fha is transcribed at 10 min following an inducing signal, while ptx is not transcribed until 2 to 4 h after the inducing signal. We present data indicating that prn is transcribed at 1 h, an intermediate time compared to those of fha and ptx. We have identified cis-acting sequences necessary for expression of prn in B. pertussis by using prn-lac fusions containing alterations in the sequence upstream of the prn open reading frame. In vitro transcription and DNase I footprinting analyses provided evidence to support our hypothesis that BvgA binds to this sequence upstream of prn to activate transcription from the promoter. Our genetic data indicate that the region critical for prn activation extends upstream to position −84. However, these data do not support the location of the prn transcription start site as previously published. We used a number of methods, including prn-lac fusions, reverse transcriptase PCR, and 5′ rapid amplification of cDNA ends, to localize and identify the bvg-dependent 5′ end of the prn transcript to the cytosine at −125 with respect to the published start site.

Bordetella pertussis, the causative agent of whooping cough, alternates between a virulent Bvg⁺ phase and an avirulent Bvg⁻ phase in response to environmental stimuli. A central regulatory locus called bvg encodes the two-component signal transduction system that mediates this phenotypic modulation. The two components involved in the regulatory system are BvgS, a transmembrane sensor protein, and BvgA, a transcriptional activator protein (1). BvgS autophosphorylates at histidine-729, and after a series of sequential intramolecular phosphorylations, the phosphate is relayed to BvgA (29). The phosphorylated form of BvgA has increased affinity for Bvg-activated promoters (4) and is responsible for the transcriptional activation of bvg-regulated genes (21, 26, 29). Expression of bvg-activated genes can be modulated in the laboratory by growth at a low temperature (25°C) or in the presence of magnesium sulfate or nicotinic acid (16).

The BvgAS system coordinately regulates expression of several virulence-associated factors of B. pertussis. These include adhesins such as filamentous hemagglutinin (Fha) and pertactin (Prn), as well as toxins such as pertussis toxin (Ptx) and adenylate cyclase-hemolysin toxin (Cya). Not only is there regulation between the phenotypic phases of B. pertussis, but a number of bvg-activated genes have been shown to be differentially regulated. A kinetic study has indicated that fha and bvg are transcribed just 10 min after an inducing signal (temperature shift) but that ptx and cya are not transcribed until 2 to 4 h after the inducing signal, as measured by S1 nuclease protection analysis (23). The induction of ptx and cya transcription correlated with the accumulation of intracellular BvgA (23). The phosphorylated form of BvgA has been shown to directly activate the expression of fha, bvg, ptx, and cya (3, 15, 22, 26, 30). ptx requires higher levels of phosphorylated BvgA for expression than fha does (26). Gel retardation and DNase I footprinting analyses have demonstrated that BvgA interacts directly with sequences upstream of the inducible promoters for fha, bvg, ptx, and cya (3, 15, 22, 26, 30). In these upstream sequences, pairs of heptanucleotide inverted repeats with the consensus sequence TTTC(C/T)TA have been identified and shown by genetic and biochemical means to be important cis-regulatory elements for promoter activity (14, 15, 19, 22). It is thought that the heptameric, inverted repeats are the initial binding site for the transcriptional activator BvgA (3). Genetic and DNase I protection data (3, 19) support a model of ptx activation in which phosphorylated BvgA binds to an upstream, high-affinity BvgA binding site. Cooperative binding of BvgA dimers along the sequence between the primary binding site and the promoter allows BvgA to interact with RNA polymerase (RNAP) (3).

Prn is an outer membrane protein that is synthesized as a 93-kDa precursor and then processed to 60- and 30-kDa forms. Prn contains an Arg-Gly-Asp (RGD) domain and has putative roles in attachment to and invasion of epithelial cells (11, 17). Prn has been shown to be bvg regulated (7), but the cis-acting DNA sequences necessary for prn expression have not been characterized. Although a transcription start site was identified by primer extension analysis (12), there is no ς⁷⁰ consensus promoter at the appropriate region upstream of this site in the prn sequence. Therefore, we wanted to examine the region necessary for bvg activation of prn. Our hypothesis was that BvgA binds to the sequence upstream of prn and activates transcription from the promoter, similar to other bvg-activated genes. In this study, we attempt to characterize the important cis-regulatory sequences upstream of and within the prn promoter by both genetic and biochemical means and we identify the true bvg-dependent transcription start site of prn.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains and plasmids used or created in this study are shown in Table 1. Escherichia coli strains were grown on Luria-Bertani (LB) agar. B. pertussis strains were grown on Bordet-Gengou (BG) agar (Difco) supplemented with 15% sheep blood or in Stainer-Scholte liquid medium (25). The following antibiotics were added to the concentrations indicated (micrograms per milliliter) when necessary: ampicillin, 100; chloramphenicol, 20; tetracycline, 10; gentamicin sulfate, 10; kanamycin, 50; streptomycin, 400; and nalidixic acid, 20. Bacterial conjugations were performed as described previously (13) with E. coli S17.1 as the donor strain (24).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant feature(s)	Reference or source
Strains
E. coli S17.1	IncP plasmid Tra functions integrated into the chromosome	24
B. pertussis
Tohama I	Wild type	15a
Tohama I Δbvg	Tohama I with deletions of 2.6- and 2.5-kb EcoRI fragments containing the bvg operon	This work
NMD615	Recipient strain in which the prn promoter is replaced by Kanr	This work
NMD series	prn-lac fusions with alterations of the prn promoter and upstream region, as described in the text	This work
Plasmids
pNMD603	Variation of pLAFR2 in which the polylinker sequence was replaced by new cloning sites	This work
pJHC1	Mobilizable suicide vector derivative of pSS1129 containing polylinker and lac sequence from pBluescript	This work
pNMD601	prn upstream sequence between ClaI and SmaI in pBluescript	This work
pNMD633	prn −153 to +147 cloned for DNase I protection analysis	This work
pTE-PRN	prn −153 to +147 in pTE-103 transcription vector	This work

DNA manipulations and allelic exchange.

DNA manipulations were carried out by standard molecular methods. Restriction sites were engineered by using overlap extension PCR (5) and subcloned into the appropriate plasmid. Sequence additions or replacements were achieved by introduction of complementary oligonucleotides containing the appropriate overhanging sticky ends. Constructs were introduced into B. pertussis as plasmids by conjugation with the mobilizable vector pNMD603, a variation of pLAFR2 (6), or into the B. pertussis genome by allelic exchange using the pJHC1 derivative of pSS1129 (27, 28), a mobilizable suicide vector. To aid in the screening of conjugants after allelic exchange, we constructed a recipient strain in which a kanamycin resistance gene cassette replaced the prn region between the EcoRI and NcoI sites (NMD615). Kanamycin sensitivity was then used as an indicator of successful integration of our engineered sequences, which were then confirmed by PCR analysis.

RT-PCR Analysis.

Total RNA was prepared from the Tohama I and Tohama I Δbvg strains by extraction with Trizol LS reagent (Gibco BRL) and then treatment with RNase-free DNase I (Boehringer Mannheim) to remove any contaminating DNA. Total RNA (2 μg for time course or 3 μg for prn start site mapping) was used in reverse transcriptase (RT) reactions (all components from Gibco, BRL) with SuperScript II RT and primed with random hexamers to synthesize first-strand cDNA. Samples without the addition of RT were also run to verify the absence of DNA contamination. After treatment with RNase H (Gibco BRL), 10% (time course) or 5% (prn start site mapping) of the first-strand reaction product was used as the template in subsequent PCRs. Time course analysis PCR mixtures (50 μl) contained 100 pmol of the primers listed below, 1× PCR buffer, 1.5 mM MgCl₂, 0.4 mM deoxynucleoside triphosphates, and 0.5 μl of Taq DNA polymerase (all components from Gibco BRL). The reactions were run for 20 cycles of 94°C denaturation, 52°C annealing, and 72°C extension in a thermal cycler. One hundred picomoles of the primers shown below (see Fig. 6) paired with primer 651 (5′GGTCGGAGCCCTGGATA3′) was used in the detection of bvg-dependent prn transcripts. The PCR mixtures also contained 42 μl of PCR Supermix (Gibco BRL) and 5% dimethyl sulfoxide (DMSO) (Fisher Scientific), and the reactions were run for 25 cycles of 94°C denaturation, 65°C annealing, and 72°C extension in a thermal cycler. The products were electrophoresed on a 2% agarose gel, stained with ethidium bromide, and visualized with UV light.

FIG. 6.

Schematic of prn promoter sequence and oligonucleotides used in RT-PCR analysis. Sequential oligonucleotide primers were paired with an antisense primer in the prn ORF in PCRs with total B. pertussis cDNA used as a template to determine the 5′ extent of the prn transcript. PCR product results are shown to the right of the respective oligonucleotide primers. +, strong band; +/−, faint band; −, absence of PCR product. The putative −35 and −10 sequences and +1 are indicated in bold.

Time course analysis.

Strain Tohama I was grown on nitrocellulose filters on BG agar plates containing 50 mM MgSO₄ to modulate bvg activity. At time zero, the filters were transferred to medium without MgSO₄ to induce bvg activity. Total RNA was prepared as described above from cells at times 0, 30, 60, 240, and 480 min after induction. RT-PCR, with primers specific for sodB (5′CTGCCTTACGCTCTGGATG3′, antisense 5′GGACGGGCATTGCGGTAAT3′), fha (5′CCTAAAACGAGCAGGCCG3′, antisense 5′GAACTTGTTGTGCGAGAC3′), ptx (5′ACCGCAAGAACAGGCTG3′, antisense 5′GTCGATCGGCATGCTGTTC3′), and prn (5′GCACCACGCTGGCCATG3′, antisense 5′GACGACGTGACACTGCC3′), was used as described above to determine promoter activation. To analyze the RT-PCR data, portions of the RT-PCR samples were run on an agarose gel and stained with the fluorescent dye Vistra Green (Amersham) and band intensities were quantified by analysis on a FluorImager SI system using ImageQuant software (Molecular Dynamics). The band intensities were normalized to the sodB standard, a bvg-independent superoxide dismutase of B. pertussis (9).

DNA sequence and database analysis.

A 900-bp fragment between the ClaI and SmaI sites upstream of the prn open reading frame (ORF) was cloned into pSK-Bluescript (Stratagene). The resulting clone, pNMD601, was used to sequence both strands of the prn sequence using a Sequenase kit (U.S. Biochemicals). The sequence was analyzed by use of a Blast search of the GenBank database. All of the additional constructs were sequenced with an ABI automated sequencer to confirm deletions and replacements.

β-Galactosidase (Lac) assays.

Filter Lac assays were used as a quick screen for the plasmid-borne prn-lac fusion strains as described previously (19). For quantitative measure of β-galactosidase levels, strains were grown for 2 days on agar medium and assayed three independent times as described previously (19). β-Galactosidase levels were determined by the method of Miller (20), and Lac units were determined by the formula [(OD₄₂₀ − 1.5 × OD₅₅₀)/OD₆₀₀] × 1,000, where OD₄₂₀, e.g., is the optical density at 420 nm (19). Statistical differences were determined by Student’s t test.

DNase I footprinting and in vitro transcription analyses.

A 300-bp DNA fragment of the upstream prn sequence (−153 to +147) was generated by PCR with oligonucleotides containing BamHI and SalI restriction sites. This fragment was cloned into either the pKS-Bluescript vector (Stratagene) to yield pNMD633 or into the pTE103 transcription vector (10) to yield pTE-PRN. Footprinting analysis was carried out as described previously (2) with fragments generated from pNMD633. Reaction mixtures contained 150 nM E. coli ς⁷⁰-saturated RNA polymerase (Pharmacia), 15 mM acetyl phosphate (Ac∼P), and BvgA at 0.58 or 1.2 μM. Reaction mixtures were electrophoresed on a 6% polyacrylamide sequencing gel and exposed for autoradiography in a PhosphorImager cassette. Transcription assays were carried out as described previously (2). Reaction mixtures contained 0.5 pmol of super-coiled pTE-PRN plasmid, 150 nM E. coli ς⁷⁰-saturated RNA polymerase (Pharmacia), or 1.4 μM purified Bordetella bronchiseptica RNA polymerase and between 0 and 0.78 μM BvgA. Where indicated, Ac∼P was added at a final concentration of 15 mM. Reaction mixtures were electrophoresed on a 6% sequencing gel and exposed to autoradiography in a PhosphorImager cassette.

5′ RACE analysis.

Total RNA was synthesized as described above from the Tohama I and Tohama I Δbvg strains of B. pertussis. Using the 5′ rapid amplification of cDNA ends (RACE) system, version 2.0 (Gibco BRL), we synthesized first-strand cDNA by using 50 ng of prn-specific antisense oligonucleotide 654 (5′CCTTGATGGTGGTTCCGCTG3′) to prime the reactions. The higher temperature and volume protocol for first-strand cDNA synthesis of transcripts with high GC content was followed. After purification of the first-strand product, terminal deoxynucleotidyl transferase was used to add homopolymeric C tails to the 3′ ends of the cDNA. This reaction was carried out on ice for 1 h, and the mixture contained 20% DMSO. The tail provided an anchor for subsequent nested priming. PCR amplification using prn-specific primer 651 (5′GGTCGGAGCCCTGGATA3′) and the kit’s abridged anchor primer was performed for 35 cycles of 94°C denaturation, 60°C annealing, and 72°C extension. The reaction mixtures contained 100 pmol of each primer, 5 μl of dC-tailed cDNA, 5% DMSO, 5 μl of 10× PCR buffer, 3 μl of 25 mM MgCl₂, 2 μl of 10 mM deoxynucleoside triphosphate mix, 0.5 μl of Taq DNA polymerase (Gibco, BRL), and distilled water to 50 μl. After the PCR products were purified by using Wizard PCR purification preps (Promega), 1 μl was used as a template in a second PCR using prn-specific primer 652 (5′GCCTCGAGCTGGCGCTCACCGGTCTTGA3′) containing an XhoI site and the kit’s abridged universal amplification primer. The products were examined on a 2% agarose gel stained with ethidium bromide. The same 5′ RACE product was obtained from two separate pools of total RNA. The product was then gel purified, cut with XhoI and SpeI, and cloned into both pBluescript SK and KS (Stratagene). Several clones were then sequenced with an ABI automated sequencer.

RESULTS

Activation kinetics of prn.

It has been shown that several bvg-activated genes are differentially regulated (23). Therefore, we wanted to examine the activation kinetics of prn to begin to characterize the effect of bvg regulation on prn expression. To examine the activation kinetics, we used an RT-PCR assay to detect transcripts of several bvg-activated genes over the course of time after induction of the Bvg system. B. pertussis Tohama I cells were modulated on plates containing 50 mM MgSO₄. At time zero, cells were removed from the MgSO₄-containing plates to induce the Bvg system. Total RNA was prepared from the cells at 0, 30, 60, 240, and 480 min. RT-PCR, with primers specific for sodB, fha, ptx, and prn, was used to determine promoter activation (Fig. 1). The absence of transcription at time zero and the increase in the level of transcription at later time points strongly suggest the bvg dependence of these genes, although fha transcription was routinely difficult to reduce to zero by MgSO₄ modulation. The same kinetic patterns were obtained when the band intensities were normalized to the sodB standard and plotted. As shown in Fig. 1, fha was activated early (30 min) after induction and ptx was activated late (240 min), results similar to those described previously (23). Transcription of prn was reproducibly observed after 60 min, indicating a third, intermediate class of promoter activation (Fig. 1). prn RT-PCR products from a variety of prn-specific primers were consistently weaker than RT-PCR products from other bvg-activated genes. This indicates that prn is weakly transcribed and is consistent with the reduced levels of β-galactosidase activity we obtain from prn-lac fusions in comparison to those obtained from fha- or ptx-lac fusions.

FIG. 1.

RT-PCR analysis of promoter activation kinetics. RT-PCR was used to detect transcripts of sodB, fha, ptx, and prn after induction of the Bvg system. The time course of induction is shown on ethidium bromide-stained agarose gels at 0, 30, 60, 240, and 480 min for the bvg-dependent fha, ptx, and prn promoters and the bvg-independent standard, sodB. Results from a typical experiment are shown.

Characterization of the prn promoter-activating sequence.

We hypothesized that the transcriptional activator BvgA would interact with sequence upstream of prn to initiate transcription, but we did not know the extent of sequence necessary for prn activation. Initially, to identify the region upstream of prn important for bvg activation, we made a series of plasmid-borne prn-lac transcriptional fusions that were introduced into B. pertussis Tohama I by conjugation. Results from β-galactosidase assays (data not shown) indicated that the region 220 bp upstream of the prn ORF is important for bvg-dependent promoter activation but may not be sufficient for full promoter activity.

Published sequence extends only as far as a ClaI site 147 bp upstream of the prn ORF; therefore, we sequenced both strands of the DNA upstream of the ClaI site to further examine the region of importance for prn activation. The sequence revealed part of an ORF with codon usage typical of B. pertussis. No strong homologies to the predicted protein of this ORF were identified by a search of the databases; the strongest was a segment of a Moraxella bovis pilin gene inverting protein (recombinase) with 41% identity and 63% similarity over 63 residues. The 335-bp intergenic sequence revealed long inverted repeats just downstream of the upstream ORF, a likely transcription terminator. An unusually AT-rich (53% compared to the 35% average for B. pertussis) region of about 200 bp, from positions −107 to +108, was recognized to contain numerous potential BvgA binding half sites based on homology to other bvg-activated promoters. The sequence data in combination with the plasmid-borne lac fusion data strongly suggested that the bvg-activating region for prn lies in this sequence upstream of the prn ORF.

To further characterize the sequence necessary for prn activation, allelic exchange was used to introduce the prn upstream region with sequence alterations into the chromosome of B. pertussis Tohama I. Additional EcoRI and ClaI restriction sites were introduced into a BglI site upstream of the AT-rich region (Fig. 2). A series of deletions and replacements with unrelated sequence were constructed and then fused to a promoterless lac gene at the NcoI site 210 bp into the prn ORF (Fig. 2). The fusion constructs were then introduced into the B. pertussis chromosome by allelic exchange, and prn promoter activity was measured by quantitative β-galactosidase assays.

FIG. 2.

Effect of deletions and replacements upstream of the chromosomal prn-lac transcriptional fusion on prn promoter activity. A restriction map of the prn upstream region (orf, upstream open reading frame) is shown at the top (not drawn to scale). Constructs are shown on the left, and promoter activities (10³ Lac units) with standard deviation bars are shown on the right. The gray bars represent β-galactosidase levels when the strains were grown under nonmodulating conditions, and the black bars represent β-galactosidase levels when the strains were grown in the presence of 50 mM MgSO₄. (A) NMD616, wild type; (B) NMD618, strain containing additional EcoRI and ClaI cloning sites inserted at the BglI site; (C) NMD623, strain with deletion of sequence between engineered and wild-type EcoRI sites; (D) NMD625, strain with replacement of the EcoRI fragment deletion with unrelated sequence of same size; (E) NMD630, strain with replacement of EcoRI fragment deletion with a 16-bp wild-type sequence upstream from the EcoRI site; (F) NMD631, strain with replacement of EcoRI deletion with a 26-bp wild-type sequence upstream from the EcoRI site.

Introduction of the additional cloning sites (in NMD618) did not significantly alter the wild-type transcriptional activity of the prn-lac fusion (Fig. 2A and B). However, transcriptional activity was completely eliminated by deletion of the sequence between the EcoRI sites (NMD623) (Fig. 2C) as well as by the deletion of the sequence between the ClaI sites (data not shown). In addition, replacement of the EcoRI (NMD625) (Fig. 2D) and the ClaI (data not shown) deleted fragments with unrelated sequences of the same length eliminated promoter activity. These data indicate that full prn transcriptional activity requires a specific sequence upstream of the EcoRI site.

To identify the 5′ extent of the activating sequence for prn, we annealed complementary olignucleotides of increasing length corresponding to the wild-type sequence and cloned them into the EcoRI deletion construct, pNMD623 (Fig. 2E and F). The changes were introduced into the chromosome of B. pertussis by allelic exchange, and prn promoter activity was again determined by β-galactosidase assays. Transcriptional activity was restored to near-wild-type levels by the reintroduction of only 16 bp upstream of the EcoRI site (NMD630) (Fig. 2E and F). These data suggest that the cis-regulatory region necessary for prn promoter activation extends from the ClaI site to just upstream of the EcoRI site upstream of the prn ORF (−68 to +6) (see Fig. 7) and that sequences around the EcoRI site are crucial for activation.

FIG. 7.

Schematic diagram of the sequence upstream of prn. The end of an ORF with codon usage typical of B. pertussis is followed by 335 bp of intergenic sequence and the start of the prn ORF. A putative transcription terminator for the upstream ORF is depicted by inverted arrows. Relevant restriction sites are indicated in italics, and the reported transcription start site (12) is indicated in bold and marked by #. The region of primary DNase I protection by BvgA (from positions −94 to −52) is outlined with a solid line, while the region of weak, secondary protection by BvgA (from positions −51 to +22) is outlined with a dashed line. DNase I-hypersensitive sites in the presence of BvgA are noted with an asterisk. The transcription start site identified in this paper is indicated in bold and marked as +1, and the numbering is in reference to this start site. Putative −10 and −35 promoter sequences are in boldface type and labeled. A putative primary BvgA binding site is indicated in lowercase letters.

Biochemical analyses of BvgA interaction with the prn promoter region.

Our genetic analyses identified the DNA sequence necessary for bvg-dependent transcriptional activation from the prn promoter. To test the hypothesis that BvgA is directly involved as the transcriptional activator of prn, we cloned a 300-bp DNA fragment containing prn-specific sequence from positions −153 to +147. The resulting clone, pTE-PRN, was used for in vitro transcription assays that were carried out as described in Materials and Methods. No detectable transcript was present with the vector alone (Fig. 3, lane 1) or with unphosphorylated BvgA (Fig. 3, lanes 2 and 6), providing evidence that transcription from the prn promoter is directly BvgA dependent. An increasing level of transcription driven from the prn promoter in pTE-PRN was detected with increasing levels of phosphorylated BvgA when both E. coli RNAP (Fig. 3, lanes 3 to 5) and B. bronchiseptica RNAP (Fig. 3, lanes 7 to 9) were used. Transcription from the prn promoter required the phosphorylation of BvgA, similar to in vitro transcription results from other bvg-activated promoters (26). Based on the relative migration of the transcript compared to an RNA ladder, the location of the transcription start site was considerably farther upstream than that of the previously published site (12).

FIG. 3.

BvgA-mediated in vitro transcription analysis of the prn promoter. Transcription reaction mixtures contained 0.5 pmol of supercoiled pTE-PRN plasmid, 150 nM E. coli ς⁷⁰-saturated RNAP or 1.4 μM purified B. bronchiseptica RNAP, and between 0 and 0.78 μM BvgA. Where indicated, Ac∼P was added at a final concentration of 15 mM. Lanes: 3 and 7, 0.20 μM BvgA; 4 and 8, 0.39 μM BvgA; 2, 5, 6, and 9, 0.78 μM BvgA. Reaction mixtures were electrophoresed on a 6% polyacrylamide sequencing gel and exposed for autoradiography in a PhosphorImager cassette. The arrow indicates the prn transcript.

In light of the requirement for phosphorylated BvgA for transcription from the prn promoter, we hypothesized that the DNA sequence that we had identified as necessary for prn activation by our genetic analyses would correspond to the BvgA binding sequence. To test this, we used DNase I footprinting analysis with prn promoter fragments and purified BvgA as described in Materials and Methods. The footprinting was performed on DNA fragments encompassing the same 300 bp upstream of the prn ORF as in the in vitro transcription analysis, including the entire region that had been identified as important by the genetic analyses. Clear protection of a region encompassing nucleotides −94 to −52 was observed, while a region of weaker protection from nucleotides −51 to +22 was identified (Fig. 4). Protection was dependent on phosphorylation of BvgA (Fig. 4, lane 2 versus lane 3), and the level of protection increased progressively upon the addition of a higher concentration of phosphorylated BvgA (Fig. 4, lane 4) and of E. coli RNAP (Fig. 4, lane 5) in the reaction mixtures. The region of primary protection surrounds the aforementioned EcoRI site at nucleotide −68 (see Fig. 7) and is thus consistent with our genetic analysis. The area of secondary protection extended to just downstream of the ClaI site, approximately 100 bp upstream of the published transcription start site.

FIG. 4.

DNase I footprinting analysis of the prn promoter. Protection reaction mixtures, where indicated, contained 150 nM E. coli ς⁷⁰-saturated RNAP, 15 mM Ac∼P, and BvgA at 0.58 (lane 3) or 1.2 (lanes 2, 4, and 5) μM. Reaction mixtures were electrophoresed on a 6% acrylamide sequencing gel and exposed for autoradiography in a PhosphorImager cassette. The black rectangle represents the region of primary protection by BvgA from positions −94 to −52, and the open rectangle represents the region of secondary protection by BvgA from −51 to +22.

Identification of the bvg-dependent transcription start site of prn.

The results of our genetic and biochemical analyses brought into question the accuracy of the published transcription start site. In addition, our attempts to repeat the primer extension analysis resulted in the same extension product as that published previously (12), but this product was obtained not only from the wild-type but also from the Tohama I Δbvg and modulated B. pertussis strains (data not shown). This would indicate that transcription initiating at this site is not bvg dependent, which is not consistent with it being the relevant prn transcription start site.

Although we used a number of different primers, repeated attempts at primer extension to identify an additional prn transcript were unsuccessful, possibly due to the low level of prn transcript and the high GC content of B. pertussis nucleic acids. Therefore, we used a variety of alternative methods in an attempt to identify the bvg-dependent prn transcription start site. We first used a series of prn-lac transcriptional fusions (Fig. 5) to localize the region that resulted in bvg-dependent transcriptional activity from the prn promoter. We created constructs in which sequences were deleted from the NcoI site at the prn-lac fusion to a number of NcoI or NsiI sites engineered in the upstream sequence. By determining which deletions abolished transcriptional activity, we would be able to localize the prn promoter. The deletions were introduced to the chromosome of B. pertussis Tohama I, and the resulting strains were used in β-galactosidase assays to determine prn promoter activity. NMD637 and NMD638 retained bvg-dependent transcriptional activity (Fig. 5B and C). The maintenance of activity in these strains provides genetic evidence that the published start site (12) is incorrect and that the true transcription start site is located further upstream. Sequence was also deleted up to nucleotides −9 (NMD643) and −21 (NMD642). Both deletions abolished prn transcriptional activity (Fig. 5D and E) and, in combination with the aforementioned deletions, localized the prn transcription start site to a region of 30 bp flanking the ClaI site.

FIG. 5.

Effect of sequential deletion of the sequence upstream of the prn ORF on prn promoter activity. Alterations are shown on the left, and promoter activities (10³ Lac units) with standard deviation bars are shown on the right. The gray bars represent β-galactosidase levels when the strains were grown under nonmodulating conditions, and the black bars represent β-galactosidase levels when the strains were grown in the presence of 50 mM MgSO₄. (A) NMD616, wild type; (B) NMD637, strain with deletion from lac fusion junction to position +61; (C) NMD638, strain with deletion from lac fusion junction to position +22; (D) NMD643, strain with deletion from lac fusion junction to position −8; (E) NMD642, strain with deletion from lac fusion junction to position −23. The putative −10 sequence and +1 are indicated in bold, and the ClaI site is italicized.

To identify the 5′ end of the prn transcript, we used a series of sequential 5′ primers corresponding to sequence in this region in an RT-PCR analysis to establish the point at which we were no longer able to generate a PCR product. The 5′ primers are shown in Fig. 6 and were all paired for PCRs with an oligonucleotide complementary to the sequence 172 bp into the prn ORF. Total RNA was isolated from B. pertussis Tohama I and used in first-strand cDNA synthesis reactions with random hexamers. The cDNA was used as a template for PCR, and the results are shown in Fig. 6. This analysis localized the start of the transcript to the sequence between primers 669 and 666 or between bp −12 and +8 (Fig. 6). This result is consistent with our previous genetic and biochemical data and further supports the conclusion that the published transcription start site is erroneous.

Finally, we used a variation of the 5′ RACE technique to identify the precise prn transcription start site. This technique allowed us to use the power of PCR amplification to overcome the difficulties associated with the low level of prn transcript. We were also able to adapt the protocol to overcome problems associated with the high GC content of B. pertussis. The same 5′ RACE products were obtained from two separate pools of total RNA isolated from strain Tohama I, but the product was not evident when RNA from the Tohama I Δbvg strain, a Bvg⁻ strain of B. pertussis, was used. The use of primers containing XhoI and SpeI sites facilitated the cloning of the PCR products, and six clones were sequenced to determine the terminal nucleotide. The sequence indicated that the transcription start site for prn is the cytosine located 125 bp upstream of the published transcription start site (Fig. 7). This result is consistent with both the RT-PCR analysis and the chromosomal prn-lac fusion data. A putative −10 promoter sequence of GAGAAT is located 7 bp upstream of this +1. Twenty base pairs upstream of this −10 sequence there is a potential −35 sequence of TTGCTT (Fig. 7).

DISCUSSION

In this study, we have provided preliminary evidence that the kinetics of prn promoter activation in B. pertussis are intermediate between the early fha promoter and late ptx promoter. We have also identified cis-acting sequences that contribute to prn promoter activation in B. pertussis and have identified a bvg-dependent transcription start site. We have provided in vivo genetic data and in vitro biochemical data that together support the hypothesis that BvgA binds to the sequence upstream of prn and activates transcription from the promoter, similar to its activity at other bvg-activated genes.

Thus far, two classes of bvg-activated promoters, based on their temporal activation after an inducing signal, had been identified: the early class represented by fha and bvg and the late class represented by ptx and cya (23). Our activation kinetics data indicate that prn promoter activation occurs at a time between that of fha and ptx and therefore would suggest that a third, intermediate class of bvg-activated promoter exists. Our classification is not to be confused with a Bvg-intermediate (BvgI) phase of B. bronchiseptica, which was shown to occur in response to semimodulating conditions (8). Although differences in the promoters of these genes are thought to be responsible for the differences in activation kinetics, the role that differential regulation of virulence factors plays in B. pertussis pathogenesis is unknown. It is possible that the differential regulation that we are characterizing in vitro may reflect the necessity of the Bvg regulon to be sensitive to small changes in microenvironments to provide modulating signals that control gene expression during the course of infection. Our data from both genetic and biochemical analyses did not support the location of the prn transcription start site as published. Our analysis identifies the bvg-dependent prn transcription start site as the cytosine located 125 bp upstream from the previously reported start site. A putative −10 promoter sequence based on similarity to ς⁷⁰ promoters is located 7 bp upstream of this start site, in good agreement with the consensus spacing of 6 to 8 bp (18). The putative −10 and −35 sequences of this prn promoter have 4 of 6 and 3 of 6 nucleotides, respectively, matching the consensus ς⁷⁰ promoter sequences (18). However, these sequences are separated by 20 bp, which is longer than the consensus 17-bp distance (18) but similar to the 21-bp distance of the bvg-activated ptx promoter (19). This may contribute to the requirement of this promoter for BvgA activation. The presence of these putative promoter sequences at the appropriate position upstream of the transcription start site provides additional support for our experimental data. The location of the prn transcription start site reveals a relatively long (159-nucleotide) leader sequence in the prn transcript, a feature that has not been observed in other well-characterized bvg-activated promoters. Partial deletion of the leader sequence did not appear to have an effect on transcription levels of our prn-lac fusion (Fig. 5), but the deletion of almost the entire leader sequence reduced transcriptional activity by almost 45% (Fig. 5C). This would indicate that the leader sequence, or at least part of it, is necessary for full prn transcriptional activity. It is possible that the leader sequence plays a role in the differential regulation of the prn promoter. We are currently examining the effect that leader sequence alterations and other changes have on the activation kinetics of prn.

Our data from the genetic analyses of the sequence upstream of prn strongly suggest that the sequence critical for prn activation extends upstream to position −84. Our initial prn-lac fusions indicated that the sequence up to −68 was important, but possibly not sufficient, for full prn transcription. Our subsequent transcriptional fusions determined that the additional sequence from −84 to −68 is crucial for transcriptional activity.

Our data from the biochemical analyses of the prn promoter corroborate this genetic data. In vitro transcription analysis from the prn promoter demonstrated that prn transcription is directly BvgA dependent. DNase I protection analysis demonstrated that BvgA binds to the sequence upstream of prn, in the region identified by the genetic data as necessary for transcription from the prn promoter. Both assays indicate that phosphorylation of BvgA is required for its binding to the prn sequence and subsequent transcription activation. This result is similar to that seen with other bvg-activated promoters (2, 3, 26). Within the region of strongest protection by BvgA, a possible tandem inverted heptanucleotide repeat sequence with homology to similar sequences implicated as primary BvgA binding sites at other BvgA-activated promoters is present between −74 and −60, surrounding the EcoRI site (Fig. 7). The hypothesis that this represents the primary BvgA binding site at the prn promoter is supported by our deletion and replacement data (Fig. 2).

The potential inverted repeat BvgA binding site of prn has 9 of 14 bp of the consensus BvgA binding site. The primary binding sites of fha and ptx have 14 of 14 and 10 of 14 bp of the consensus BvgA binding site, respectively. However, the ptx half sites are separated by 10 bp and the sequence is located more than 50 bp further upstream of the transcription start site than it is in the fha-activating region. The distance of the putative primary BvgA binding site upstream from the prn promoter is similar to that at the fha promoter. The affinity of BvgA for the primary binding sites may therefore be a major contributing factor to the kinetics of activation of these promoters. Our footprinting data indicate the presence of a secondary BvgA binding region downstream of the primary site upon addition of higher concentrations of phosphorylated BvgA (Fig. 4). However, there is overlap of this secondary BvgA binding (in the absence of RNAP) with the core promoter sequences, although addition of RNAP to the footprinting reaction mixtures resulted in a visible increase in protection in this region (Fig. 4). The possible competition between BvgA and RNAP binding to this region at higher BvgA concentrations may be an artifact of the in vitro analysis, or it may contribute to the relatively low level of prn transcription in B. pertussis. DNase I protection analysis and genetic data support a model of cooperative BvgA binding at the ptx promoter (3, 19). The data from this study suggest the possibility that the same phenomenon occurs at the prn promoter but with consequent transcription inhibition rather than activation as at the ptx promoter. We are currently investigating these and other aspects of prn promoter activation.