INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Environmentally responsive regulation of virulence gene expression in the human pathogen Bordetella pertussis is mediated by the bvgAS locus, which encodes a member of the two-component family of bacterial signal transduction systems (33). BvgS is an unorthodox sensor/transmitter protein which contains, in addition to the typical periplasmic, transmembrane, and histidine kinase domains, a response regulator domain and an additional domain termed the output domain. Through a phosphorelay mechanism, this protein is capable of transferring phosphate to the response regulator BvgA, resulting in increased affinity of BvgA for its binding sites in virulence gene promoters (1, 4, 5, 16, 37). In addition to regulating its own expression, the bvgAS locus governs the expression of all known protein virulence factors of this organism, including those encoding filamentous hemagglutinin (fha), pertussis toxin (ptx), and adenylate cyclase toxin/hemolysin (cya). Under standard culture conditions, BvgS activates BvgA to stimulate expression of these virulence genes. However, when grown in media containing MgSO₄ or nicotinic acid, or at reduced temperatures, bvg-regulated expression of virulence factors is represseda phenomenon known as antigenic, or phenotypic, modulation (19).

In vitro transcription experiments involving the ptx and fha promoters have demonstrated a dependence on phosphorylated BvgA (6, 7, 28). At both promoters, binding of BvgA-phosphate to consensus inverted heptameric repeats upstream of the 35 promoter motif has been seen, with additional sequences between these repeats and the 35 motif also being protected in DNase I footprinting experiments (5, 6, 42). However, the repeats in the ptx promoter region are further upstream than in the fha promoter, and the inverted heptamers are separated by 10 bp (33). These differences in promoter architecture may be responsible for functional differences in the two promoters observed in vivo, in that higher concentrations of BvgA-phosphate are required to activate ptx transcription than are needed to activate fha transcription. This phenomenon has been observed in Escherichia coli as well as in B. pertussis (25, 36).

Further demonstration of functional differences between these two promoters is provided by missense mutations affecting the extreme C terminus of BvgA which abolish ptx transcription but have little or no effect on fha transcription (30). The region of the BvgA protein affected by these mutations may be involved in interaction with RNA polymerase at regulated promoters. This is suggested by studies of other members of the subfamily of response regulators to which BvgA is related by sequence similarity in its C-terminal domain. In both the LuxR protein of Vibrio fischeri and the UhpA protein of E. coli, short C-terminal deletions affected the ability of these proteins to stimulate transcription at regulated promoters (10, 38).

We report here an analysis of the phenotype conferred by short C-terminal deletions of BvgA. Unexpectedly, when provided in trans in B. pertussis, these deletions conferred a profound growth inhibition which was dependent on the presence of a functional chromosomal bvg locus. Mutations which relieved this growth inhibition included mutations which affected the level of expression or the primary structure of the B. pertussis RNA polymerase  subunit. These observations provide additional genetic evidence for an interaction between the C-terminal portion of the BvgA protein and RNA polymerase.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, plasmids, and media. Bacterial strains and plasmids used in this study are described in Table 1. E. coli DH5, which was used as a transformation recipient for all cloning steps, was obtained from Bethesda Research Laboratories. E. coli strains were grown on L agar or in L broth (21) supplemented with antibiotics as appropriate. B. pertussis strains were grown on Bordet-Gengou agar (Difco) containing 1% proteose peptone (Difco) and 15% defibrinated sheep blood. Concentrations of antibiotics were 10 µg/ml for tetracycline, kanamycin, and gentamicin; 50 µg/ml for nalidixic acid; and 100 µg/ml for streptomycin. For growth of B. pertussis strains under modulating conditions, 50 mM MgSO₄ was added.

Plasmid construction. Deletions of the bvgA gene were constructed by PCR. A fragment containing the entire bvgA gene and its promoter region was amplified by using an upstream primer including the EcoRI site at position 1 and a downstream primer including the SphI site at position 817, 6 bp downstream of the bvgA stop codon (coordinates as in GenBank accession no. M25401). Deletions removing codons immediately upstream of the stop codon were introduced by specifying the appropriate sequence in the downstream primer. In this way, genes coding for BvgA which lacked 1, 2, 3, 6, 9, 12, 15, or 18 C-terminal amino acids (bvgAC1-bvgAC18) were synthesized and subsequently cloned between the EcoRI and SphI sites of pSS1894. To permit introduction of these bvgA deletions into the B. pertussis chromosome, flanking sequences were restored and the resulting constructs were recloned into the allelic exchange vector pSS1129. In this way, plasmids pSS2189, pSS2190, pSS2191, pSS2192, pSS2193, pSS2194, pSS2647, and pSS2650 were created. To permit introduction of the mutant bvgA genes in trans, an EcoRI site was added at the SphI site by cleavage and ligation with the self-complementary oligonucleotide 5'-CATGCGAATTCG-3'. The resulting EcoRI fragments were recloned into pRK290 to create pSS2342, pSS2344, pSS2244, pSS2219, pSS2246, pSS2248, pSS2250, and pSS2252. In a similar way, pSS2068, containing the wild-type bvgA gene was created.

Strain construction. B. pertussis BP953 contains fhaB-lacZ and ptx-phoA transcriptional gene fusions to allow monitoring of virulence factor gene expression regulated by the bvgAS locus. The fhaB-lacZ fusion was created as follows. A BamHI-SalI fragment containing lacZ and derived from pRS1551 (27) was added between the most upstream BamHI site and the most downstream XhoI site in the 10.0-kb EcoRI fragment containing fhaB. The resulting EcoRI fragment containing the fusion was recloned into pSS1129 to create pSS1581. This plasmid was used to replace the chromosomal fhaB gene to create BP947, using previously described methodology (31). The ptx-phoA fusion was created as follows. Plasmid ptxA4-50-6, kindly provided by Cynthia Lee, consists of a 4.7-kb EcoRI fragment containing the pertussis toxin operon cloned into pBR325. The oligonucleotide linker 5'-pGATCTCGCGGCCGCGA-3' was added at the unique BglII site in this plasmid, resulting in the addition of a NotI site. A fragment containing the phoA gene of E. coli was created by PCR. The upstream primer sequence was 5'-ATTCCCGGGCGAGTACATTGCGAAAATAAAGTGAAACAAAGCACTA-3'. This primer includes the first 16 bp of the phoA gene, including the native GTG start codon, 9 bp upstream of the start codon, and an XmaI site at the 5' end. The downstream primer sequence was 5'-CGCAAGCTTGCGGCCGCATTTCAGCCCCAGAGCGGCTTTCATG-3', which contains the last 25 bp of the phoA gene prior to the termination codon and a NotI site near the 5' end. This PCR fragment was cloned by using its XmaI and NotI sites into the ptx operon between the naturally occurring XmaI site and the NotI site introduced at the BglII site. The resulting phoA open reading frame is predicted to code for 13 additional amino acids at its C terminus relative to wild-type phoA as a result of the lack of the native termination codon. This fusion was cloned as a BstBI fragment into the ClaI site of pSS1129 to create pSS1615, which was then used to replace the ptx locus of BP947 to create BP953 as previously described (31).

Selection of mutants surviving growth inhibition conferred by bvgAC3. Independent cultures of BP953[pS2244] or BP1079[pSS2244] were grown on Bordet-Gengou agar containing tetracycline and MgSO₄. After growth for 3 to 4 days, bacteria were resuspended in phosphate-buffered saline, the bacterial density was estimated by measurement of A₆₀₀, and dilutions were plated on Bordet-Gengou agar containing tetracycline but lacking MgSO₄. After incubation for 3 to 4 days at 37°C, colonies surviving selection were examined for expression of alkaline phosphatase and -galactosidase to differentiate Bvg phenotypes. Mutants were restreaked onto Bordet-Gengou medium lacking tetracycline, and colonies were screened for tetracycline sensitivity. In this way, the mutant strains were cured of plasmid pSS2244 prior to further analysis.

Visualization of alkaline phosphatase and -galactosidase expression by B. pertussis colonies. Colonies were allowed to adhere to nitrocellulose filters, which were then perfused by placing on 3MM filter paper (Whatman) saturated with Tris-HCl (pH 8.0) containing 200 µg of BCIP (5-bromo-4-chloro-3-indolylphosphate; Sigma) per ml and 500 µg of Magenta-Gal (Biosynth) per ml to visualize expression of alkaline phosphatase and -galactosidase, respectively.

Enzyme assays. For measurement of -galactosidase, bacteria to be assayed were recovered by sterile swab from Bordet-Gengou agar and suspended in Z buffer (21). The A₆₀₀ was measured, and 0.05 ml was diluted with 1 ml of Z buffer. Cells were permeabilized by the addition of 30 µl of 0.1% sodium dodecyl sulfate and 30 µl of chloroform followed by vortexing, and the assay was completed as described by Miller (21). For measurement of alkaline phosphatase, bacteria were suspended in 1.0 M Tris-HCl (pH 8.0). The A₆₀₀ was measured, and 0.5 ml of cell suspension was diluted with 0.5 ml of 1.0 M Tris-HCl (pH 8.0). The cells were permeabilized as described above, and the assay was completed as described by Brickman and Beckwith (8). Units in both cases were defined by the following equation: units = [1,000 × A₄₂₀  (1.75 × A₅₅₀)]/(T × V × A₆₀₀), where T is the incubation time in minutes and V is the volume of permeabilized cells in milliliters.

Mapping of mutations in B. pertussis. Donors used for Hfr mapping in this study contained selectable markers in the form of insertions of Tn2048, a derivative of the mini-Tn5 transposon delivered by pUT-Kan (11, 20). From a set of 12 such insertions spaced evenly around the B. pertussis Tohama I chromosome, three markers near the bvg locus, and three markers near the rpoA locus were used to localize mutations to these regions. The genetic background of the donor strains was Tohama I, str rif ptx-phoA. Chromosomal sequences were mobilized by a library of random B. pertussis chromosomal fragments cloned into pSS1853 by using generalized conjugation as described previously (32) into the mutant strains, with selection on Bordet-Gengou agar containing kanamycin and nalidixic acid. Linkage of a mutation to the different Tn2048 insertions was determined by visualization of the expression of ptx-phoA and fhaB-lacZ by exconjugant colonies as described above.

Recovery of mutations from B. pertussis. Plasmids pSS2320, pSS2321, pSS2322, pSS2324, pSS2325, and pSS2326 were recovered from B. pertussis mutant strains by purification of DNA by the alkaline lysis method (3) and transformation into E. coli DH5. Plasmids containing the chromosomal bvgA gene from B. pertussis mutant strains were isolated by using the allelic recovery plasmid pSS2197 previously described for this purpose (34). Plasmids containing the rpoA gene from B. pertussis mutant strains RPV5 and BP1187 were isolated in a similar manner. Briefly, pSS2306 was transferred to the mutant strains with selection for gentamicin resistance and counterselection for nalidixic acid resistance. Strains containing pSS2306 thus integrated at the rpoA locus by a single crossover in the 7-kb BamHI fragment containing the rpoA gene were subsequently mated with SS1840 to recover the plasmid liberated by a second crossover. Plasmids so isolated were screened for the ability to confer the differential phenotype following reintroduction into BP953. In this way, pSS2316 and pSS2318 were isolated.

Western analysis. Western blot analyses were performed as previously described (29). Samples were prepared from suspensions of B. pertussis strains grown on Bordet-Gengou agar which were adjusted to have the same optical density. The monoclonal antibody 4RA2, recognizing epitopes on the  subunit conserved between E. coli and B. pertussis, was kindly provided by Nancy Thompson and Richard Burgess (35). To detect BvgS, the murine monoclonal antibody 23/5 isolated in this laboratory was used (40).

DNA sequence analysis. DNA sequence analysis was performed by the dideoxy-chain termination method, using a Sequenase kit with deazaGTP to reduce artifacts due to G:C compression (U.S. Biochemical). Single-stranded templates for sequencing were isolated following cloning into mp18 and mp19 (41).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Short C-terminal deletions of BvgA confer a bvg-dependent growth inhibition in trans. To assess the phenotype conferred by deletion mutations affecting the C terminus of BvgA, deletions removing 1, 2, 3, 6, 9, 12, 15, and 18 amino acids were constructed and are illustrated in Fig. 1. These deleted bvgA genes were used to replace the wild-type bvgA gene of B. pertussis BP953. This strain contains transcriptional gene fusions of the fhaB gene, encoding filamentous hemagglutinin, to the lacZ gene of E. coli, encoding -galactosidase, and of the ptx operon, encoding pertussis toxin, to the phoA gene of E. coli, which encodes alkaline phosphatase. Deletions removing more than one C-terminal amino acid conferred a Bvg phenotype in that colonies were Lac and Pho as assessed by colony lifts and perfusion with colorimetric substrates BCIP and Magenta-Gal (data not shown). In addition, these colonies were nonhemolytic, indicating a lack of expression of the cya operon, encoding adenylate cyclase toxin/hemolysin (data not shown). The bvgAC1 mutation conferred a mild regulatory phenotype in that ptx-phoA expression was reduced approximately twofold, while fha-lacZ expression remained at wild-type levels (data not shown). To assess the effects of these mutations in trans, the deleted bvgA genes were cloned into pRK290, which is capable of replication in B. pertussis with a copy number of 5 to 7 (14). These plasmids were transferred by conjugation to BP953. As indicated in Fig. 1, deletion of one amino acid had no effect, but deletion of two or three amino acids of the bvgA gene in trans resulted in a complete inhibition of detectable growth on Bordet-Gengou agar. Deletions removing 6, 9, 12, or 15 amino acids had an intermediate effect, as revealed by a reduction in colony size but not number, and deletion of 18 residues relieved the inhibition. Expression of the growth-inhibitory phenotype conferred by the bvgAC3 gene of plasmid pSS2244 was dependent on the activity of the wild-type chromosomal bvgAS locus, as no inhibition was observed with a strain containing a chromosomal bvgA null mutation, or of BP953[pSS2244] grown in the presence of 50 mM MgSO₄, which represses synthesis of bvgAS-activated genes, including the bvgAS locus itself (data not shown).

View larger version (37K): [in this window] [in a new window]   FIG. 1.   BvgA proteins expressed by plasmid-borne bvgA alleles in trans. The top line depicts DNA sequence features of the EcoRI-SphI fragment containing bvgA. P1 and P2 are Bvg-inducible and constitutive promoters, respectively (23, 24). BBS is a BvgA binding site (23). The coding sequence of wild-type bvgA (bvgAw.t.) is depicted by an open arrow. Nucleotide coordinates for the EcoRI and SphI sites refer to the first nucleotide of those sites with reference to GenBank accession no. M25401. The BvgA proteins are depicted as open boxes below the DNA map, with the number of residues in each protein to the right. The shaded areas correspond to the region of amino acid sequence similarity with other two-component response regulator proteins, and the stippled areas correspond to the putative helix-turn-helix (HTH) DNA binding motif. The positions of the bvgA1056 and bvgA1060 mutations which affect ptx transcription but not fha transcription (30) are depicted by arrowheads on the BvgA wild-type protein. Plasmid names and their encoded bvgA alleles are shown at the left; the growth phenotype conferred upon BP953 is shown at the right.

Selection for mutations which relieve growth inhibition conferred by BvgAC3. Cultures of BP953[pSS2244] or BP1079[pSS2244] grown on Bordet-Gengou agar containing MgSO₄ (permissive conditions) were plated on Bordet-Gengou agar lacking MgSO₄ (nonpermissive conditions). This permitted the isolation of mutants which relieved the growth-inhibitory effect of the BvgAC3 protein. As shown in Fig. 2, when colonies surviving this selection were examined for expression of fhaB-lacZ and ptx-phoA, a variety of colonial phenotypes were observed. Dark blue colonies expressed both fusions normally and were Cya⁺ by hemolysis (data not shown), a Bvg⁺ phenotype. Colorless colonies lacked expression of either gene fusion and were nonhemolytic, a Bvg phenotype. Pink colonies had reduced hemolysis and expression of ptx-phoA but normal or nearly normal expression of fha-lacZ, a differential Bvg phenotype. Each of these classes were further characterized as follows.

View larger version (110K): [in this window] [in a new window]   FIG. 2.   Selection for mutations which suppress the growth-inhibitory phenotype conferred by pSS2244. BP953[pSS2244] was grown under permissive conditions and plated under nonpermissive conditions at an approximate density of 1 × 107 or 3 × 107 per plate (8.5-cm diameter). Colonies were allowed to adhere to nitrocellulose filters and developed with BCIP and Magenta-Gal as described in Materials and Methods. Dark blue colonies express ptx-phoA and fha-lacZ, pink colonies express primarily fha-lacZ, and white colonies express neither fusion.

bvgAC3

bvgAC3

Plasmid-borne bvgA mutations conferring a differential Bvg phenotype. The differential phenotypes conferred by six of the plasmid-borne bvgA alleles were analyzed by -galactosidase and alkaline phosphatase assays in order to quantitate the differential expression of ptx-phoA and fhaB-lacZ. These data are presented in Table 2 (plasmid-borne bvgA mutations) and demonstrate, in comparison to BP953 containing the pRK290 vector alone, that ptx-phoA expression is repressed upon introduction of these alleles in trans.

View this table: [in this window] [in a new window]   TABLE 2.   -Galactosidase and alkaline phosphatase assays of mutant strains displaying a differential phenotype

View larger version (32K): [in this window] [in a new window]   FIG. 3.   Plasmid-borne mutations in the bvgAC3 gene which relieve growth inhibition of pSS2244 and confer a differential phenotype. The positions and nature of the mutations are shown above the map of the EcoRI-SphI bvgAC3-containing fragment. Other features are as described for Fig. 1, with the exception of the hatched boxes, which represent predicted nonsense (n.s.) peptide introduced into the bvgA open reading frame as a result of frameshift (f.s.) mutation or IS insertion.

View larger version (14K): [in this window] [in a new window]   FIG. 4.   MgSO4 modulation of the bvg-regulated gene fusions fha-lacZ and ptx-phoA. BP953 was grown on Bordet-Gengou agar containing different concentrations of MgSO4. Cells were recovered and assayed as described in Materials and Methods. Each point represents the mean of three separate assays.  -Gal, -galactosidase; Alk. PhoS., alkaline phosphatase.

Chromosomal bvg-linked mutations conferring a differential Bvg phenotype. In all but 1 of the 15 strains harboring chromosomal bvg-linked mutations, the location of the mutation conferring the differential phenotype could be unequivocally localized to the 816-bp EcoRI-SphI fragment containing the bvgA gene and upstream region. This was accomplished by using the allelic recovery plasmid pSS2197 to recover this fragment from the mutant strains and reintroduce it into the BP953 genetic background according to procedures described elsewhere (34). Except in the case of bvg-1247, which appeared to map to the region encoding the periplasmic domain of BvgS, the mutation was associated with the EcoRI-SphI bvgA fragment in all cases. Six mutants of this class were selected for further characterization. The results of enzyme assays performed to quantitate the differential expression of ptx-phoA and fhaB-lacZ in these strains are presented in Table 2 (chromosomal bvgAS mutations). The expression of the ptx locus in all six strains is reduced to very low levels, while expression of the fha locus is reduced at most twofold. In all cases, the expression of fha-lacZ was repressed by addition of MgSO₄ to the growth media, demonstrating that these mutations did not affect modulation mediated by the bvgAS locus (data not shown).

Chromosomal rpoA-linked mutations conferring a differential Bvg phenotype. Mutations affecting RpoA have been shown previously to cause a differential Bvg phenotype (9). One of these previously described rpoA alleles was examined and was also found to relieve the growth-inhibitory phenotype conferred by pSS2244 but apparently not to the same degree, as colony size was reduced. This rpoA allele was recovered from its parental strain RPV5 and transferred by allelic exchange to the BP953 chromosome to create BP1242. A quantitative comparison of the regulatory phenotypes caused by rpoA-linked mutations isolated in this study with that from RPV5 is shown in Table 2 (rpoA-linked mutations). It can be seen from these data that the effects of these mutations were similar, although BP1187, BP1196, and BP1246 showed a somewhat more pronounced differential effect.

View larger version (37K): [in this window] [in a new window]   FIG. 5.   Western analysis of RNA polymerase  subunit expression. Samples were whole-cell extracts of bacterial suspensions normalized by optical density to contain the same numbers of cells. The blot was probed with monoclonal antibody 4RA2, which recognizes an epitope on the  subunit conserved between E. coli and B. pertussis, kindly provided by Nancy Thompson and Richard Burgess. Sizes are indicated in kilodaltons. Lanes: a, BP953; b, BP1187; c, BP1196; d, BP1246.

View larger version (23K): [in this window] [in a new window]   FIG. 6.   Amino acid sequence alignment of the  subunits of B. pertussis and E. coli. Alignment was performed at the National Center for Biotechnology Information by using the LFASTA network service. The substitution in BP1187 is indicated above the B. pertussis RpoA (BpeRpoA) sequence, and the position of a mutation in the E. coli  subunit affecting activation of class II promoters (22) is marked by an asterisk below the E. coli RpoA (EcoRpoA) sequence.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

An unexpected phenotype results from the expression, in trans, of deletions of two or three amino acid residues from the C terminus of the response regulator protein BvgA of B. pertussis. This phenotype, a profound inhibition of growth, is proposed to be the result of an inappropriate interaction of the mutant BvgA protein with the RNA polymerase enzyme, specifically the  subunit. Experimental support for this hypothesis comes from the observation that mutations which alter the amino acid sequence of the  subunit or which increase levels of the  subunit can relieve this inhibition of growth.

The portion of the BvgA molecule associated with this growth inhibition, and by inference the interaction with the  subunit, is the C terminus, downstream of the putative helix-turn-helix DNA binding motif. Experimental support for this conclusion comes from the observation that deletion of this region in BvgAC18 relieves the growth inhibition, as do larger truncations isolated as spontaneous mutations of the plasmid-borne bvgAC3 gene. It is suggested that this portion of the BvgA molecule normally interacts with the  subunit but in the BvgC2 and BvgC3 deletions does so in an uncontrolled or otherwise inappropriate way. Results from studies on other response regulators, with sequence similarity to BvgA in the C-terminal domain, are consistent with a role for the C-terminal sequence in interaction with RNA polymerase. For example, short C-terminal truncations result in an inability of LuxR to activate transcription, although it appears to be able to bind DNA normally (10). Similarly, UhpA C-terminal deletions are inactive in transcriptional activation of the uhpT promoter (38). The Bvg phenotypes observed here with chromosomally encoded BvgA C-terminal deletions are consistent with these observations in that removal of as few as two amino acids abolishes BvgA activity. Interestingly, this region of the BvgA molecule includes the sites of amino acid substitutions in two mutants which no longer activate ptx transcription but which activate fha transcription normally (30), consistent with the hypothesis that the differential effect of those mutations involves a change in the interaction of BvgA with RNA polymerase at the ptx promoter.

The growth-inhibitory phenotype observed with bvgAC3 in trans can also be overcome by mutations which affect the wild-type chromosomal bvg locus, either by eliminating expression (bvg knockouts) or by reducing expression and/or activity, through the insertion of IS481 or IS1002 just upstream of the bvgS gene. In the latter case, the differential phenotype observed in terms of ptx and fha transcription is likely to be due to differences in the amount of BvgA-phosphate required for maximal expression at these two loci. In a similar fashion, the truncated BvgA molecules presented in Fig. 3 may result in intermediate levels of activity, thus explaining the differential phenotype observed under these conditions. One explanation for this finding would involve the formation of mixed dimers of BvgA. Several different experimental approaches have suggested that BvgA forms dimers (2, 4, 24). Alternatively, these truncated derivatives may compete with wild-type BvgA for phosphorylation by BvgS. Genetic analysis of a similar trans-dominant phenotype conferred by truncated UhpA molecules would suggest that the latter possibility is most likely correct (38). The fact that several of the mutant strains harboring chromosomal bvg mutations have reduced transcriptional activity of fha-lacZ is consistent with their very low levels of ptx-phoA expression and the observation that the ranges of responsiveness of these two fusions to modulating signals overlap somewhat (Fig. 4). This finding suggests that these strains have even lower levels of Bvg activity than the plasmid-bearing strains shown above them in Table 2.

The dependence of the growth-inhibitory phenotype on expression of an intact bvg locus can be explained in several ways. BvgAS may simply be required for the expression of bvgAC3, itself a bvg-regulated gene. Alternatively, wild-type BvgA protein may play a direct role in the growth inhibition, perhaps through interaction with BvgAC3 to form a heteromultimer, which would then be the toxic form of the protein. To date, our efforts to distinguish between these two possibilities by expressing the bvgAC3 gene from a non-bvg-regulated promoter have been hampered by the unavailability of a promoter that is inducible in B. pertussis and that is of sufficient strength to achieve the growth-inhibitory phenotype, even in the presence of a wild-type chromosomal copy of bvgAS (data not shown). A third possibility is that manifestation of toxicity requires phosphorylation of BvgAC3 by BvgS. This hypothesis is currently being addressed by assessing the toxicity of BvgAC3 into which a D54N mutation has been incorporated, thus rendering it incapable of phosphorylation.

It has been previously reported that mutations affecting the level of the  subunit in B. pertussis affect bvg-regulated gene expression (9). In this case, the effect seen was a differential phenotype (Fha⁺ Ptx Cya). This has been interpreted to possibly indicate (i) an interaction between the  subunit and BvgA and (ii) titration of active BvgA-phosphate by excess  subunit. Biochemical data provide strong evidence of such an interaction as well, in that the C-terminal portion of the  subunit is required for bvg-activated transcription of the fha gene (6). The results reported here provide additional genetic data supporting such an interaction in that mutations which increase  subunit levels suppress the phenotype associated with a mutant BvgA protein.

Interestingly, a mutation affecting the primary structure of the  subunit was also found to overcome the growth inhibition conferred by BvgAC3. Typically, mutations in rpoA which suppress mutations in positive regulatory proteins have been found in the C-terminal third of this protein, a region believed to make up a separate conformational domain which is attached to the N-terminal portion of  by a flexible peptide linker (15). Two possible explanations for the effect of this mutation are proposed. One is that this mutation defines a site on the N-terminal domain of  which contacts the BvgA molecule in the process of transcriptional activation of the ptx promoter. However, this would not be the only contact required, because in vitro, BvgA-activated transcription of the ptx promoter is dependent on the C-terminal domain of the  subunit as well (7). A precedent for interaction of the N-terminal portion of  in transcriptional activation comes from recent reports of catabolite gene activator protein-activated transcription of the synthetic class II promoter CC(41.5) in E. coli. In this case, a mutation affecting the glutamic acid residue at position 165 of the E. coli  subunit had a significant negative effect (22). Interestingly, as shown in Fig. 5, the aspartic acid residue that was identified in the B. pertussis  subunit, at position 171, is nine amino acids away from the B. pertussis  residue corresponding to the E. coli  glutamic acid 165 and may represent an analogous mutation. Another possibility is that the glycine 171 mutation affects the assembly of  into the RNA polymerase core enzyme, either in terms of the efficiency of assembly or in terms of the configuration adopted by  in the assembled enzyme. If efficiency of assembly was low, unincorporated  subunit could accumulate with a differential effect on transcription of the ptx and fha genes as previously suggested (9). However, inefficient assembly either would be expected to result in a reduced growth rate due to reduced levels of RNA polymerase or would require higher levels of  to maintain the same level of RNA polymerase. However, neither of these conditions was observed. Fortuitously, the contribution of the corresponding residue in the E. coli subunit on assembly of RNA polymerase in vitro has been previously examined. It was found that a change of this aspartic acid at position 174 to an alanine negatively affected dimerization of the  subunit, the initial step in assembly, but subsequent assembly of ₂ and ₂' was normal (18). Inferring from this study and the observation of normal growth of the mutant B. pertussis strain, we may still ask whether the configuration and presentation of the mutant  subunit in the assembled RNA polymerase might not be altered and thus affect interaction with the BvgA protein at the ptx promoter. Clearly more study is required to distinguish between these models and to explain the behavior of this interesting mutant.