A New Gene Locus of Bordetella pertussis Defines a Novel Family of ...

2 downloads 0 Views 388KB Size Report
By the introduction of a cosmid library containing genomic B. pertussis DNA into this suppressor strain, we isolated a cosmid which suppressed the phenotype of ...
JOURNAL OF BACTERIOLOGY, Aug. 1996, p. 4445–4452 0021-9193/96/$04.0010 Copyright q 1996, American Society for Microbiology

Vol. 178, No. 15

A New Gene Locus of Bordetella pertussis Defines a Novel Family of Prokaryotic Transcriptional Accessory Proteins THILO M. FUCHS,1 HEIKE DEPPISCH,1 VINCENZO SCARLATO,2

AND

ROY GROSS1*

Theodor-Boveri-Institut fu ¨r Biowissenschaften, Lehrstuhl fu ¨r Mikrobiologie, Universita ¨t Wu ¨rzburg, Am Hubland, D-97074 Wu ¨rzburg, Germany,1 and Istituto Ricerche Immunobiologiche, I-53100 Siena, Italy2 Received 5 May 1996/Accepted 21 May 1996

Recently, a novel type of regulatory mutation causing differential effects on the expression of virulence genes due to a slight overexpression of the RNA polymerase a subunit (RpoA) was found in Bordetella pertussis (N. H. Carbonetti, T. M. Fuchs, A. A. Patamawenu, T. J. Irish, H. Deppisch, and R. Gross, J. Bacteriol. 176:7267–7273, 1994). To gather information on the molecular events behind this phenomenon, we isolated suppressor mutants of the RpoA-overexpressing strains after random mutagenesis. Genetic characterization of these suppressor strains revealed the existence of at least three distinct groups of dominant alleles. Mutations occurred either in the rpoA locus itself, in the bvg locus, or in unknown gene loci. One mutant of the latter group was further characterized. By the introduction of a cosmid library containing genomic B. pertussis DNA into this suppressor strain, we isolated a cosmid which suppressed the phenotype of the suppressor strain, thus restoring the negative effect on transcription of the ptx and cya toxin genes. Mutagenesis of the cosmid with Tn5 led to the identification of the gene locus responsible for this phenomenon. Its DNA sequence revealed the presence of an open reading frame (ORF) consisting of 2,373 bp coding for a hypothetical 86-kDa protein with extensive sequence similarities to ORFs with not yet identified functions of Escherichia coli, Haemophilus influenzae, and Neisseria meningitidis. The new gene, termed tex, for toxin expression, seems to be an essential factor for B. pertussis, as it cannot be deleted from the bacterial chromosome. All members of this new protein family show significant sequence similarities with the mannitol repressor protein MtlR and with the presumptive RNAbinding domains of the Pnp and ribosomal S1 proteins of E. coli in their N- and C-terminal parts, respectively. These sequence similarities and the fact that the tex gene was isolated by virtue of its effects on gene expression in B. pertussis indicate that the members of this new protein family may play an important role in the transcription machinery of prokaryotic organisms. two groups. The upstream activating sequences present in these promoters are either close to the RNA polymerase binding region (bvg and fha promoters) or far upstream (ptx and cyaA promoters) (15, 17, 31). Recent work of several laboratories showed that the BvgA transcription factor directly interacts with the upstream activating sequences of both types of promoters, but transcription seems to be initiated differently in the various cases, depending on the concentration and phosphorylation status of the BvgA transcription factor (5, 31, 38). Nevertheless, additional factors such as the recently identified Baf protein (9) or differential effects of DNA topology (16) may still be involved in the activation of some of the promoters. Two types of mutants which showed differential effects in virulence gene expression in B. pertussis have been described. Both classes of mutants were characterized by a strong reduction in transcription of the toxin promoters, whereas transcription of the other factors was not affected. These mutations were localized either in the BvgA transcription factor itself (39) or in the a locus of B. pertussis (6, 7). In the latter case, the mutations caused an overexpression of the a subunit of RNA polymerase as a result of mutations in the Shine-Dalgarno region of the rpoA gene. The mechanism by which overexpressed a subunit causes a negative effect on toxin gene expression remains poorly understood. A direct interaction of a and BvgA was suggested to be required for the activation of the toxin promoters (7). Since the activation of the toxin promoters seems to occur only in the presence of a high BvgA concentration, the overexpression of RpoA may decrease the number of free BvgA molecules by titrating out this transcription factor.

Bordetella pertussis, the etiological agent of whooping cough (4), expresses a set of virulence-related factors such as several adhesins, e.g., filamentous hemagglutinin (FHA), pertactin (PRN), and fimbriae, and several toxins, e.g., pertussis toxin (PTX) and adenylate cyclase toxin (CyaA), which confers a hemolytic phenotype. The regulation of expression of these factors is coordinate and depends on the respective growth conditions; i.e., the virulence factors are expressed at body temperature but not at low temperature, a phenomenon termed phenotypic modulation (21, 43). The BvgAS two-component system is responsible for this coordinate regulation and controls virulence gene expression on the transcriptional level (1, 3, 17, 18, 27, 40, 41). The bvgAS locus is genetically unstable. Spontaneous mutations which inactivate the two-component system and thereby cause the lack of expression of the virulence genes can occur with a high incidence (25). Therefore, these mutations in the bvg locus lead to avirulent bacteria, the so-called phase variants (23). Despite the coordinate regulation mediated by this two-component system, differential regulatory phenomena such as different induction kinetics of transcription of the various factors or different expression patterns in the heterologous host Escherichia coli were noted, which led to the classification of the virulence factors into two groups, (i) the two toxins PTX and CyaA and (ii) the other factors (33, 34, 42). These differences in their expression are also reflected by differences in the promoter structures of the * Corresponding author. Mailing address: Lehrstuhl fu ¨r Mikrobiologie, Biozentrum, Universita¨t Wu ¨rzburg, Am Hubland, D-97074 Wu ¨rzburg, Germany. Phone: (931) 888 4403. Fax: (931) 888 4402. Electronic mail address: [email protected]. 4445

4446

FUCHS ET AL.

J. BACTERIOL.

TABLE 1. Bacterial strains and plasmids used Strain or plasmid

Strains E. coli K-12 DH5a SM10 B. pertussis Tohama I BP347 BC75 RPV3 SUP1 SUP2 SUP3 SUP4 SUP5 SUP6 SUP7 B. bronchiseptica BB7865 Plasmids pBluescript pSS1129 pMMB208 pLAFR2 pLA-L1

pLA-L1-Tn5 pSS-tex pSS-tex-Kan

Source or reference(s)

Relevant feature(s)

High-efficiency transformation Mobilizing strain

GIBCO BRL 35

Wild type, but rpsL Derivative of Tohama I, bvgS::Tn5 Derivative of Tohama I, phase variant with mutation in the a locus Derivative of Tohama I, phase variant with mutation in the a locus Suppressor mutant of RPV3 Suppressor mutant of BC75 Suppressor mutant of BC75 Suppressor mutant of RPV3 Suppressor mutant of RPV3 Suppressor mutant of RPV3 Suppressor mutant of RPV3

43 43 6, 8

Wild type, but rpsL

25

High-copy-number cloning vector Bordetella suicide vector Broad-host-range vector Broad-host-range vector pLAFR2 containing a 25-kb B. pertussis DNA insert including the tex locus pLA-L1, but tex::Tn5 pSS1129 containing a 7-kb insert including the entire tex gene pSS1129 containing a kanamycin cassette flanked by tex sequences

Stratagene 40 26 13 This study

7 This This This This This This This

study study study study study study study

This study This study This study

In this study, we examined the molecular events leading to a lack of transcription of the toxin promoters after overexpression of RNA polymerase a subunit. For this purpose, we mutagenized the RpoA-overexpressing strains with methylating agents and screened for suppressor mutants, which again were able to express all virulence genes, including the toxin genes. The characterization of these mutants led to the identification of a new gene locus encoding a protein with interesting sequence similarities to proteins from various unrelated bacterial genera and to specific domains of proteins interacting with nucleic acids. MATERIALS AND METHODS Bacterial strains, plasmids, and growth media. The strains and plasmids used in this study are described in Table 1. E. coli strains were grown in LB medium (Gibco). Bordetella strains were grown at 378C on Bordet-Gengou (BG) agar plates supplemented with 15% sheep blood (4) or in Stainer-Scholte (SS) liquid medium (37). If necessary, antibiotics were added at the following concentrations: chloramphenicol, 20 mg/ml; kanamycin, 50 mg/ml; spectinomycin, 100 mg/ ml; streptomycin, 100 mg/ml; and tetracycline, 12,5 mg/ml. Molecular biology procedures. DNA cleavage with restriction enzymes, PCR, DNA ligation, preparation of plasmid and chromosomal DNA, gel electrophoresis on agarose and polyacrylamide, and transformation of E. coli were carried out according to standard procedures (32). For DNA sequencing, a T7 polymerase kit from Pharmacia with [a-35S]dATP was used according to the manufacturer’s protocol. For Southern hybridizations, an ECL direct nucleic acid labeling and detection kit (Amersham) was used. Western blotting (immunoblotting) was performed as described previously (7, 18). RNA analysis was performed as recently described (7, 18). Conjugations into B. pertussis were carried out as previously described (17), using E. coli SM10 as a donor (35). Isolation of suppressor mutants. Strains RPV3 and BC75 were inoculated in SS medium and grown to an optical density at 595 nm of 0.7. Then the bacteria were harvested and divided into two samples. One half of the culture was

resuspended in SS medium, and the other half was suspended in Tris-maleic acid buffer (pH 6.0). A freshly prepared solution of N-methyl-N9-nitro-N-nitrosoguanidine was added to a final concentration of 100 mg/ml to the latter sample. The bacteria were incubated without shaking at 378C for various periods of up to 30 min. After this treatment, the bacteria were centrifuged and resuspended in SS medium and plated on BG plates. A killing curve was established, and the mutagenesis was repeated under conditions which did not result in more than 50% killing. After plating on BG agar plates, screening for suppressor mutants was carried out to look for hemolytic colonies. Construction of a B. pertussis cosmid library. High-molecular-weight chromosomal DNA of B. pertussis Tohama I was prepared from 50 ml of a logarithmically growing culture in SS liquid medium (37). The cells were harvested, resuspended in 10 ml of TE buffer (50 mM EDTA, 50 mM Tris-HCl [pH 8.0]), and frozen at 2208C. Then 1 ml of a lysozyme solution (10 mg/ml) was added, and the samples were incubated on ice for 45 min. Finally, 6 ml of a sodium dodecyl sulfate (SDS) solution (0.5% SDS, 0.4 M EDTA, 50 mM Tris-HCl [pH 8.0]) and proteinase K (final concentration, 1 mg/ml) were added. The solution was incubated at 508C for 1 h, then extracted twice with phenol-chloroform, and dialyzed overnight against TEN buffer (0.1 M NaCl, 1 mM EDTA, 10 mM Tris-HCl [pH 8.0]). Then RNA was digested by incubation with DNase-free RNase followed by phenol-chloroform extraction and extensive dialysis. The chromosomal DNA prepared in this manner was partially digested with Sau3A, and fragments of 20 to 30 kb were isolated on a sucrose gradient. After dialysis, the fragments were ligated into BamHI-cleaved and dephosphorylated pLAFR2 cloning vector (13). The ligation mixture was in vitro packaged into l phage particles, using a Gigapack Gold packaging kit (Stratagene, San Diego, Calif.), and transfected into the E. coli SM10. From SM10, the library was then conjugated into the respective B. pertussis SUP strains as recently described (17), and transconjugants were plated out on BG agar plates, selecting for tetracycline and streptomycin. After 4 to 5 days of incubation at 378C, the plates were screened for nonhemolytic colonies. Tn5 mutagenesis. Transposon mutagenesis was carried out as described recently (24). Cosmids carrying a Tn5 insertion were isolated by conjugation into B. pertussis, selecting for transposon-mediated kanamycin resistance. The nucleotide sequence around the Tn5 insertion point was determined by using sequencing primers derived from the ends of Tn5 and thereafter by primer walking of about 2 kb of both DNA regions flanking the transposon. In addition, the sequence of the Tn5 insertion point was determined in the original clone without transposon insertion. Construction of pSS-tex-Kan. The 1,543-bp EcoRI fragment containing the 59 region of the tex gene (see Fig. 4A) was cloned in pBluescript and digested with SfiI, cutting the fragment in two. After a fill-in reaction, the blunt-ended kanamycin cassette was cloned into the linearized plasmid, thereby flanking the kanamycin cassette on both sides with tex gene sequences. The tex gene fragment containing the kanamycin cassette in its center was then isolated after EcoRI digestion and cloned into the allelic exchange vector pSS1129 (40). The resulting construct, pSS-tex-Kan, was then conjugated into B. pertussis, and selection for double-crossover events was carried out as described elsewhere (7, 40). Computer analysis. For the analysis of DNA and protein sequences, the programs of the Genetics Computer Group package were used (10). Protein homologies were detected by using the FASTA and TFASTA programs and were further elaborated by using the PILEUP program. The transcriptional terminator of the tex gene was predicted by using the TERMINATOR program, and the isoelectric points were calculated by using the ISOELECTRIC program. Nucleotide sequence accession number. The sequence data reported are deposited in the GenBank database under accession number X95386.

RESULTS Isolation of suppressor mutants of B. pertussis BC75 and RPV3. The B. pertussis mutant strains BC75 and RPV3 show a strongly reduced amount of the ptx and cyaA mRNAs (6, 7). This reduced transcript accumulation has been associated with mutations in the rpoA gene that cause a slight overproduction of the a subunit of RNA polymerase (7). At the phenotypic level, bacteria overexpressing the a subunit of RNA polymerase do not show hemolysis on BG blood agar plates. We hypothesized that the use of strains BC75 and RPV3 and the hemolytic properties of B. pertussis on blood-containing agar plates may represent a useful tool for the isolation of new regulatory genes. Briefly, as depicted in Fig. 1, we performed a random mutagenesis of BC75 and RPV3 and isolated clones containing dominant mutations which suppress the a-overexpressing phenotype by screening for hemolytic colonies. Then, among these suppressor strains, we searched for clones carrying the dominant mutations outside the known regulatory genes (rpoA itself and bvgAS). Finally, in such strains, we

VOL. 178, 1996

FIG. 1. Schematic presentation of the strategy for isolating new genes involved in modulation of RNA polymerase a subunit expression and/or activity. The nonhemolytic (HLY2) phenotype of strains BC75 and RPV3 is caused by a slight overexpression of RNA polymerase a subunit (7). As a first step to identify new regulatory genes, mutants carrying suppressor mutations of the a-overexpressing phenotype of BC75 and RPV3 were isolated by random mutagenesis. In the second step, suppressor mutants were analyzed for the presence of suppressor mutations in the known regulatory rpoA (type I) and bvgA (type II) loci. In step 3, mutants with suppressor mutations in unknown gene loci x/y/z (type III) were used to screen a genomic library of a wild-type strain for antisuppressor functions counteracting the a-suppressing phenotype. By transposon mutagenesis, the gene locus responsible for the antisuppressor phenotype was then identified and characterized by sequence and mutation analysis.

looked for a dominant antisuppressor function encoded in the wild-type genome by screening for nonhemolytic clones after the introduction of a genomic library in trans. Using this scheme, we exposed the BC75 and RPV3 strains to nonlethal concentrations of N-methyl-N9-nitro-N-nitrosoguanidine and plated the bacteria on BG blood agar plates. On these plates, 7 of 25,000 screened colonies showed the desired hemolytic phenotype (strains SUP1 to SUP7). As expected, by Western blotting, strains BC75 and RPV3 and the SUP derivatives were shown to express FHA, PRN, and fimbriae (Fig. 2). In addition, the newly isolated strains SUP1 to SUP7 also expressed the two toxins PTX and CyaA (Fig. 2). Under modulating growth conditions (in the presence of 50 mM MgSO4), these strains did not show hemolysis on plates, thus indicating a correct function of the bvg regulatory system (data not shown). The expression levels of the toxins were estimated by densitometric scanning of the blots to reach a level of up to 100% of that of the wild-type strain. As outlined above, these strains were expected to carry mutations in loci which counteract the effect of a overexpression, i.e., either in the rpoA gene itself (compensatory mutations), in the bvg locus, or in new gene(s) involved in regulatory functions. To localize the suppressor mutations, the wild-type bvg or the RPV3 or BC75 rpoA locus, respectively, was introduced

TOXIN EXPRESSION IN B. PERTUSSIS

4447

into these strains. The SUP3 strain lost its ability to produce the toxins after allelic exchange with the BC75 rpoA locus, the SUP4 and SUP7 strains did so after allelic exchange with the RPV3 rpoA locus, and two clones (SUP2 and SUP5) did so after introduction of the bvg locus. Therefore, these strains carry mutations in these two loci. Interestingly, in two strains (SUP1 and SUP6), introduction of neither the bvg nor the rpoA locus caused lack of toxin expression. It is likely that in these strains, suppressor mutations occurred in new gene(s) involved in regulation of a functions. Identification of a new gene locus interfering negatively with toxin expression in B. pertussis. Since the hemolysin-negative phenotype of the parental strains BC75 and RPV3 was associated with a disequilibrium between the a subunit of RNA polymerase and other transcription factors, the equilibrium required for toxin expression has been at least partially regained in the suppressor strains SUP1 and SUP6. To isolate the dominant gene(s) involved in antisuppressor function and therefore in toxin expression, strains SUP1 and SUP6 were complemented with a genomic library of B. pertussis Tohama I present in the cosmid vector pLAFR2, and exconjugants were screened for a nonhemolytic phenotype. After introduction of the cosmid library, several nonhemolytic derivatives of the hemolytic SUP1 strain could be detected. From these nonhemolytic clones, the cosmids were isolated and compared with each other. Two of these cosmids contained unrelated inserts, as determined by comparison of the DNA fragment pattern after restriction enzyme digestion and Southern blot analysis (data not shown). For this study, one of the two different cosmid clones, pLA-L1, was further characterized. Curing of the SUP1(pLA-L1) strain from the cosmid resulted in a hemolytic phenotype, whereas reintroduction of the cosmid in the SUP1 strain restored the nonhemolytic phenotype (data not shown). This finding demonstrates that the negative effect on hemolysin expression is indeed due to the presence of the DNA fragment cloned in the cosmid. By immunoblot analysis, we then showed that pLA-L1 affects the production of both CyaA and PTX (Fig. 2). RNA slot blot analysis demonstrated that the negative effect on toxin expression is caused by a strong reduction in transcription of the two factors in the presence of the pLA-L1 cosmid (Fig. 3). Southern hybridization confirmed that the pLA-L1 cosmid did not contain the known B. pertussis regulatory rpoA or bvgAS loci (data not shown). The antisuppressor effect on toxin expression caused by pLA-L1 was observed only in strain SUP1. Introduction of pLA-L1 into strain SUP6, which very likely carries a mutation(s) different from that in strain SUP1, or into the wild-type strain Tohama I had no effect on virulence gene expression. This finding indicates that the expression of this new gene is

FIG. 2. Western blot showing expression of PTX (17), PRN (22), and FHA (30) in whole cell lysates of the B. pertussis wild-type strain Tohama I and regulatory mutants. Lanes: a and b, Tohama I; c, BP347; d, RPV3; e, SUP1; f, SUP6; g and h, SUP1/pLA-L1.

4448

FUCHS ET AL.

FIG. 3. RNA slot blot analysis of ptx transcripts in the B. pertussis wild-type strain Tohama I and in regulatory mutants demonstrating the negative effect of pLA-L1 on toxin expression in the SUP1 strain. Bands: a, Tohama I; b, RPV3; c, SUP1; d, SUP6; e, SUP1/pLA-L1.

dominant in the mutant strain and acts as an antisuppressor of the a-suppressor mutation in strain SUP1. Thus, the new gene is likely to be involved in regulation of a expression and/or function. To identify the gene responsible for the negative effect on toxin expression encoded on the 25-kb DNA insert of pLA-L1, the cosmid was mutagenized with Tn5 in E. coli. After conjugation of the pLA-L1 derivatives carrying random Tn5 insertions, clones which did not lose their hemolytic properties were isolated. One cosmid conferring the desired phenotype, carrying a single Tn5 insertion in a 1.6-kb EcoRI fragment, was then used for subcloning and subsequent sequence analysis of the respective gene locus, which was termed tex, for toxin expression. Nucleotide sequence of the new gene locus tex and identification of a highly conserved family of proteins present in evolutionary distant genera. The DNA sequence around the Tn5 insertion point in pLA-L1 revealed the presence of an open reading frame (ORF) consisting of 2,373 bp. This putative ORF has the typical codon usage of B. pertussis and, accordingly, the Genetics Computer Group program CODON PREFERENCE indicates a high coding probability throughout the entire ORF (data not shown). The transposon was inserted at the beginning of the tex gene 210 bp downstream of the putative ATG start codon. In addition, two incomplete ORFs with high coding probability are found about 300 bp upstream and downstream of the tex gene. Both ORFs are oriented in the opposite direction of the tex gene. The upstream ORF encodes a hypothetical protein with significant similarity to an ORF of unknown function from Photobacterium strain SS9 (ORF1; GenBank accession number X67094) (2). The protein encoded by the downstream ORF shows highly significant sequence similarities with the tryptophanyltRNA synthetase (TrpS) of E. coli and Haemophilus influenzae (data not shown) (12, 19). Between the end of the tex gene and the presumptive trpS gene, we found a stem-loop structure which is likely the rho-independent transcription terminator of the tex gene. As both ORFs surrounding the tex gene are oriented in the opposite direction of tex and as a result of the presence of the putative transcription terminator at its 39 end, the tex gene is likely to be transcribed as a monocistronic transcript (Fig. 4). Therefore, the effect of the transposon insertion in the tex gene is not due to polar effects on a proximal gene(s). The proposed ATG start codon of the tex gene is not preceded by a well-conserved canonical ribosomal binding site; however, as indicated in Fig. 4, a so-called downstream box is found near the start codon (36). This sequence is homologous to sequence positions 1422 to 1436 of the 16S rRNA of B.

J. BACTERIOL.

pertussis and, similarly to several cases described in E. coli, may permit efficient translation of the tex mRNA (36). The amino acid sequence deduced from the nucleotide sequence was used for a homology search in the GenEmbl/GenBank databases, using the TFASTA program. Interestingly, various highly homologous sequences deriving from E. coli ORF_o622 (accession number U18997) and from H. influenzae ORF HI0568 (19) could be detected. In addition, further strong sequence homologies are located in the vicinity of the above-mentioned E. coli and H. influenzae loci, but in different reading frames of the analyzed DNA sequences. Indeed, in the respective GenEmbl/GenBank annotations, frameshift sequencing errors were anticipated for both loci joining various putative ORFs with high coding probability in these regions. Confirmation of the B. pertussis sequence came from recent results obtained with Neisseria meningitidis, in which a highly related ORF has been identified in the neighborhood of the capsule locus (the so-called region E) (14, 28). Figure 5 shows the sequence of the B. pertussis protein and its astonishing similarity to the hypothetical proteins from the other organisms. The sequence conservation is very high: the proteins of this family show about 60% amino acid identity and between 74 and 78% sequence similarity. The protein sequences of E. coli and H. influenzae shown in Fig. 5 were tentatively assembled from various pieces of different reading frames according to the highest similarity to the B. pertussis sequence. One frameshift (position 54) in the case of E. coli and three frameshifts (positions 418, 689, and 763) in the case of the H. influenzae were introduced. As the positions of the potential sequencing errors resulting in frame shifts in the E. coli and H. influenzae ORFs are located in very highly conserved parts of the proteins, we propose that the true primary structures of these proteins very likely resemble the sequences presented here, but, the loci have to be resequenced

FIG. 4. Schematic presentation of the tex locus and its promoter region. (A) The tex locus and the surrounding region containing additional ORFs is shown. The deduced amino acid sequence of the downstream ORF indicates that the respective gene codes for the tryptophanyl-tRNA synthetase TrpS of B. pertussis. The arrows above the ORFs indicate their transcriptional polarity. The insertion point of Tn5 in the construct pLA-L1-Tn5 and the insertion point of the kanamycin cassette in the pSS-tex-Kan constructs are shown. TRM, predicted rhoindependent terminator. (B) The tex gene coding region is indicated by italics and by the translated protein sequence below the nucleotide sequence. Grey stars above the nucleotide sequence indicate nucleotides complementary to the 39 end of the B. pertussis 16S rRNA. The black stars indicate those nucleotides which may form a so-called downstream box (DB-box) and which are likely to be involved in the initiation of translation (36).

VOL. 178, 1996

TOXIN EXPRESSION IN B. PERTUSSIS

4449

carefully. In any case, a potential function of these genes for the various organisms is not known. A careful evaluation of further sequence similarities of this new protein family to proteins present in the SwissProt database revealed two additional interesting sequence motifs: (i) in the N termini of the proteins, a significant similarity to the mannitol repressor protein MtlR of E. coli (11) (Fig. 6A); and (ii) in the C termini, a strong homology to the putative RNA binding domains of the E. coli Pnp and RpsA (ribosomal S1) proteins (29) (Fig. 6B). In the latter sequence motif, several glycines and hydrophobic amino acids were proposed to be important for the interaction of Pnp and S1 with ribonucleic acids. Interestingly, these amino acids are highly conserved also in the Tex protein family (Fig. 6B). The presence of several less conserved regions among the four proteins suggests that the proteins of the Tex family consist of several functional domains which may be separated by nonspecific linker regions. At least two such putative linkers can be detected which, in Tex, are located from sequence positions 250 to 300 and 600 to 620. These linkers may divide the protein at least into three functional regions, an N-terminal domain (amino acids [aa] 1 to 250), a central part (aa 300 to 600), and a C-terminal domain (aa 620 to 791). The calculation of the isoelectric point (pI) of the whole Tex protein gives a value of 6.48. However, the pI values of the individual protein domains reveal a very uneven distribution of acidic and basic properties. The N-terminal domain (aa 1 to 250) has a pI of about 5.0 and is nearly as acidic as the homologous MtlR protein, with a calculated pI of 4.5 (11). In contrast, the Cterminal domain (aa 600 to 791) containing the putative RNA binding domain has a calculated pI of 10.67. This finding is in agreement with the assumption that the C-terminal domain of Tex may have nucleic acid binding properties. The identification of sequence motifs in the Tex protein family potentially involved in interactions with nucleic acids seems to be especially relevant, because the tex gene was originally identified by

FIG. 5. Multiple sequence alignment of the Tex protein and the products of three ORFs from E. coli (Coli), H. influenzae (Haem), and N. meningitidis (Neiss). The sequence alignment was made using the Genetics Computer Group PILEUP program. Only amino acids identical or homologous in all four proteins are boxed. The stars within the E. coli and H. influenzae sequences indicate frameshifts, which were introduced in the respective DNA sequences to obtain maximal similarity among the various proteins (see the text for details). The bars below the sequences in the N- and C-terminal parts of the proteins indicate the regions with homology to the MtlR and Pnp or S1 proteins, respectively (Fig. 6). Groups of similar amino acids are as follows: D,E,Q,N; A,S,G,P,T; F,Y,W; K,R,H; V,L,I,M; C.

FIG. 6. Sequence similarities of the Tex protein family to several E. coli proteins. (A) Similarity to the mannitol repressor protein, MtlR. Identical or similar amino acids present in at least four of the five sequences are boxed. (B) Similarity to the RNA binding domain of the Pnp and S1 (RpsA) proteins. Identical or similar amino acids present in at least five of the six sequences are boxed. The conserved glycine residues previously proposed to be important for the interaction with RNAs are marked by black arrows. Groups of similar amino acids are as follows: D,E,Q,N; A,S,G,P,T; F,Y,W; K,R,H; V,L,I,M; C.

4450

FUCHS ET AL.

virtue of its negative effect on transcription of the toxin genes in B. pertussis. It is therefore possible that these proteins constitute a new family of transcription-associated factors. Mutation analysis of the tex locus and effects of its overexpression. A direct relation of the tex locus to the suppressor mutation present in strain SUP1 was investigated by the exchange of the chromosomal tex locus of the SUP1 strain with the wild-type gene. For this purpose, the entire tex locus was cloned in the pSS1129 suicide vector (40). The resulting construct, pSS-tex, was introduced into the hemolytic SUP1 strain, and double-crossover events were selected. In the case of the presence of suppressor mutation(s) in the tex locus of the SUP1 strain, the repair with the wild-type locus should again cause the nonhemolytic phenotype of the parental strain RPV3. However, although several hundreds of recombinants were screened, no nonhemolytic derivatives could be detected on the blood agar plates. This observation indicates that the suppressor mutation(s) occurred not in the tex locus of the SUP1 strain but in another, still unknown locus. To investigate the function of the tex locus, we tried to delete the tex gene from the B. pertussis chromosome. For this purpose, a kanamycin cassette flanked by tex gene sequences was cloned into the allelic exchange vector pSS1129 and introduced into the wild-type strain, selecting for double-crossover events. However, despite repeated attempts, it was not possible to knock out the gene from the chromosome. Only in a strain carrying the pLA-L1 cosmid containing the tex locus in trans was it possible to select mutants which after double-crossover events had the kanamycin cassette integrated into the chromosomal copy of the tex gene (data not shown). The inability to inactivate the tex locus when present in a single copy clearly indicates an essential role of the tex gene for cellular viability. To further prove this hypothesis, we tried to cure the strain harboring the kanamycin cassette in the chromosomal copy of the tex gene from the pLA-L1 cosmid. However, whereas the pLA-L1 cosmid was lost in a wild-type B. pertussis strain, we were not able to identify strains carrying the inactivated chromosomal tex gene which had lost the cosmid, confirming the assumption that the tex gene is an essential factor. The presence of the tex gene on the low-copy-number vector pLAFR2 abrogates transcription of the toxin promoters in the SUP1 strain (Fig. 3). To analyze whether the lack of an effect of pLA-L1 in the wild type may be due to the relatively low overexpression of Tex caused by this construct, we cloned the tex gene in the pMMB208 vector under the control of the inducible tac promoter. However, attempts to introduce the resulting construct into the wild-type strain were not successful. Transconjugants harboring pMMB208 with the tex gene insert were strongly impaired in viability and grew only to very tiny colonies on BG agar plates (data not shown). Obviously, under noninducing conditions, there already occurs significant expression of the tex gene, which must be harmful for the bacteria. Therefore, the tex gene can be neither deleted from the chromosome nor strongly overexpressed in B. pertussis. DISCUSSION In this study, we tried to gain more insight into the mechanism of differential activation of virulence promoters of B. pertussis. The basis for this study was the recent finding that in two spontaneous B. pertussis mutants (BC75 and RPV3), a slight perturbation of the equilibrium of transcription factors, a less than threefold overexpression of the RNA polymerase a subunit, caused the lack of transcription of some factors belonging to the Bvg regulon of B. pertussis (6, 7). In these mutants, the ptx and cyaA loci showed a strongly reduced

J. BACTERIOL.

transcript accumulation, whereas other members of the Bvg regulon were not affected. After random mutagenesis of these mutants with chemical agents, we succeeded in isolating suppressor mutants which obviously had at least partially regained the balance of factors required for transcription of the entire Bvg regulon (Fig. 1). By the introduction of the respective loci into these mutants, several of these suppressor mutations could be mapped to either the bvg or the rpoA locus. Recent results obtained demonstrated that the BvgA transcription factor is able to interact directly with all virulence promoters analyzed so far, the ptx, cyaA, fha, and bvg loci (5, 31, 38, 42). The localization of suppressor mutations within the bvg and rpoA loci is in agreement with these results and further strengthens the hypothesis that BvgA and the RNA polymerase a subunit may interact physically during the transcription initiation process at the toxin promoters (7). The molecular characterization of the mutations in the bvg and rpoA loci of these suppressor strains is in progress and may provide further insights into the activation mechanism at the various virulence promoters and the proposed interaction of the RpoA and BvgA proteins. Two suppressor mutants (SUP1 and SUP6) could not be repaired to the parental phenotype of the BC75 and RPV3 strains by introduction of the bvg or rpoA locus. Obviously, these strains carry mutations in unknown loci which suppress the a-overexpressing phenotype by modulating either expression of the a subunit or its activity. As shown in Fig. 1, these strains were used as a tool to select for new gene loci which counteract the activity of the suppressor mutations present in SUP1 and SUP6. We expected that this selection procedure would allow the identification of gene loci, which, because of their interference with the suppressor phenotype, are likely to be involved in the control of a activity. For the identification of potential antisuppressor loci, we introduced a genomic library of B. pertussis in the two strains and screened for nonhemolytic colonies on blood agar plates. In the case of the SUP1 strain, we were able to identify two cosmids with unrelated inserts, which exerted a dominant negative effect on transcription of the two toxin promoters, whereas the other Bvg-dependent factors were not affected (Fig. 2). One of them, pLA-L1, was further characterized, and the gene locus responsible for the antisuppressor effect was identified by Tn5 mutagenesis. The associated gene, termed tex, encodes a putative 85.73-kDa protein. As expected, the antisuppressor effect of the tex locus on toxin transcription can be observed only in the SUP1 suppressor strain, not in the other suppressor strains or in the wild type. Since the suppressor mutation does not map within the tex gene of strain SUP1, the relationship of the tex gene to the gene locus containing the suppressor mutation in the suppressor strain SUP1 remains to be established. Experiments are under way to identify the gene locus in the SUP1 strain responsible for suppression of the phenotype caused by a overexpression. As determined by RNA slot blot analysis, the presence of the tex gene on the low-copy-number vector pLAFR2 causes only a slight overexpression of its transcript (data not shown). Attempts to overexpress the tex gene more strongly by using an expression vector failed because of the impairment of cellular viability after introduction of the appropriate construct into the bacteria. This finding indicates that raising the Tex concentration over a certain limit may be deleterious for the bacteria. On the other hand, it was impossible to inactivate the tex gene, which indicates its essential character for the bacteria. Therefore, Tex concentration seems to be a critical parameter, since Tex cannot be deleted and only a mild overexpression can be tolerated by the bacteria. So far there is no evidence re-

VOL. 178, 1996

garding how the tex locus exerts a negative effect on toxin expression in the suppressor strain. However, the analysis of the primary structure of the predicted Tex protein revealed some clues, which are likely to be important for a future understanding of this phenomenon. First, the Tex protein was found to be highly homologous to other hypothetical proteins from various organisms such as E. coli, H. influenzae, and N. meningitidis. The degree of sequence identity is extremely high, reaching up to 75% (Fig. 5). Such a high degree of sequence conservation among organisms, which in evolutionary terms are only very distantly related, is expected for proteins with important housekeeping functions. Indeed, our unsuccessful attempts to delete the tex gene suggest that Tex is an essential factor for B. pertussis. Second, two sequence motifs present in several proteins interacting with nucleic acids were identified in all four members of the novel protein family. In their N-terminal parts, a sequence similarity is present with the mannitol repressor protein MtlR of E. coli (Fig. 6A), which is a poorly understood transcriptional regulator and does not possess typical motifs of other transcription factors such as a helix-turn-helix motif (11). The similarity of Tex to a transcriptional repressor protein is consistent with the fact that the tex gene was isolated by virtue of its negative effects on transcription from several promoters when present in higher copy number. However, it is unlikely that Tex is a specific repressor protein of the toxin promoters, because this effect is seen only in the SUP1 strain, not in the wild-type strain. In the C-terminal parts of the members of the new protein family, the potential RNA binding motif of the polynucleotide phosphorylase Pnp and the ribosomal S1 (RpsA) is conserved (29) (Fig. 6B). In the case of the S1 protein, this sequence motif is present four times in its primary structure. The only known common property of these two proteins is the fact that both interact with RNA. In the Tex protein family, within this motif all structural motifs which were previously proposed to be structurally relevant, e.g., six glycines and several hydrophobic amino acids, are conserved (Fig. 6B). Therefore, the proteins of the Tex family contain two sequence motifs present in proteins interacting with nucleic acids. This finding suggests that the proteins of the Tex family also interact with nucleic acids. The fact that the C-terminal part of Tex containing the S1/Pnp motif has very basic properties is in agreement with this assumption. The essential character of the tex gene for B. pertussis, the strong conservation of the Tex protein family, and the presence of the sequence motifs possibly involved in interactions with nucleic acids suggest that the members of the novel protein family may be part of the transcription machinery. They may not have been identified as such previously because mutations in these genes or overexpression due to cloning in multicopy systems may severely impair cellular viability. Experiments are under way with the aim of elucidating the functions of these factors and how the tex locus interferes with the a-overexpressing phenotype of the SUP1 strain, thereby affecting transcription of the toxin genes in B. pertussis. In addition, as the suppressor mutation(s) in the B. pertussis SUP1 strain is not located in the tex locus, it is important to identify the gene locus carrying the suppressor mutations which counteracts the overexpression of the a subunit and which confers Tex sensitivity to the toxin promoters in this way. The system outlined in Fig. 1 allowed the identification of a new gene locus which very likely codes for a factor involved in the bacterial transcription machinery. This system is based on perturbations in the equilibrium of transcription factors, in our case the RNA polymerase a subunit. This result is reminiscent of the recent finding by Kitten and Willis (20), who showed

TOXIN EXPRESSION IN B. PERTUSSIS

4451

that overexpression of certain ribosomal proteins can suppress mutations in the lemA gene, coding for a two-component sensor protein in Pseudomonas syringae. Interestingly, the members of the Tex protein family also show significant homology to a ribosomal protein. Therefore, similar systems exploiting phenotypical effects of disequilibria between interacting factors may be more generally applicable for the cloning of genes which are difficult to isolate by classical cloning or mutation strategies. ACKNOWLEDGMENTS We thank D. Beier, J. Daniels, W. Goebel, J. Gross, J. Hacker, and J. Kreft for careful reading of the manuscript and many discussions. M. Frosch is thanked for sharing unpublished Neisseria sequence data. Antisera against Bordetella virulence factors were obtained from F. Mooi, R. Rappuoli, and A. Ullmann. Part of this work was carried out during a short term stay of T.M.F. at the IRIS in Siena. The project was supported by the Deutsche Forschungsgemeinschaft (GR1243/2-2) and the Human Frontier Science Program Organization. REFERENCES 1. Arico, B., J. F. Miller, C. Roy, S. Stibitz, D. Monack, S. Falkow, R. Gross, and R. Rappuoli. 1989. Sequences required for Bordetella pertussis virulence factors share homology with prokaryotic signal transduction proteins. Proc. Natl. Acad. Sci. USA 86:6671–6675. 2. Bartlett, D. H., E. Chi, and M. E. Wright. Sequence of the ompH gene from the deep-sea bacterium Photobacterium SS9. Gene 131:125–128. 3. Beier, D., B. Schwarz, T. M. Fuchs, and R. Gross. 1995. In vivo characterization of the unorthodox BvgS two-component sensor protein of Bordetella pertussis. J. Mol. Biol. 248:596–610. 4. Bordet, J., and O. Gengou. 1909. L’endotoxine coquelucheuse. Ann. Inst. Pasteur (Paris) 23:415–419. 5. Boucher, P. E., and S. Stibitz. 1995. Synergistic binding of RNA polymerase and BvgA phosphate to the pertussis toxin promoter of Bordetella pertussis. J. Bacteriol. 177:6486–6491. 6. Carbonetti, N. H., T. M. Fuchs, A. A. Patamawenu, T. J. Irish, H. Deppisch, and R. Gross. 1994. Effect of mutations causing overexpression of RNA polymerase a subunit on regulation of virulence factors in Bordetella pertussis. J. Bacteriol. 176:7267–7273. 7. Carbonetti, N. H., N. Khelef, N. Guiso, and R. Gross. 1993. A phase variant of Bordetella pertussis with a mutation in a new locus involved in the regulation of pertussis toxin and adenylate cyclase toxin expression. J. Bacteriol. 175:6679–6688. 8. Cookson, B. T., D. E. Berg, and W. E. Goldman. 1990. Mutagenesis of Bordetella pertussis with transposon Tn5tac1: conditional expression of virulence-associated genes. J. Bacteriol. 172:1681–1687. 9. DeShazer, D., G. E. Wood, and R. L. Friedman. 1995. Identification of a Bordetella pertussis regulatory factor required for transcription of the pertussis toxin operon in Escherichia coli. J. Bacteriol. 177:3801–3807. 10. Devereux, J., P. Haeberli, and O. Smithies. 1984. A comprehensive set of sequence analysis programs for the Vax. Nucleic Acids Res. 12:387–395. 11. Figge, R. M., T. M. Ramseier, and M. H. Saier. 1994. The mannitol repressor (MtlR) of Escherichia coli. J. Bacteriol. 176:840–847. 12. Fleischmann, R. D., M. D. Adams, O. White, R. A. Clayton, E. F. Kirkness, A. R. Kerlavage, C. J. Bult, J. F. Tomb, B. A. Dougherty, J. M. Merrick, K. McKenney, G. Sutton, W. FithHugh, C. A. Fields, J. D. Gocayne, J. D. Scott, R. Shirley, L. I. Liu, A. Glodek, J. M. Kelley, J. F. Weidman, C. A. Phillips, T. Spriggs, E. Hedblom, M. D. Cotton, T. R. Utterback, M. C. Hanna, D. T. Nguyen, D. M. Saudek, R. C. Brandon, L. D. Fine, J. L. Fritchman, J. I. Fuhrmann, N. S. Geoghagen, C. L. Gnehm, L. A. McDonald, K. V. Small, C. M. Fraser, H. O. Smith, and J. C. Venter. 1995. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269:496– 512. 13. Friedman, A. M., S. R. Long, S. E. Brown, W. J. Buikema, and F. M. Ausubel. 1982. Construction of a broad host range cosmid cloning vector and its use in the genetic analysis of Rhizobium mutants. Gene 18:289–296. 14. Frosch, M., U. Edwards, K. Bousset, B. Krausse, and C. Weissgerber. 1991. Evidence for a common molecular origin of the capsule gene loci in gram negative bacteria expressing group II capsular polysaccharides. Mol. Microbiol. 5:1251–1263. 15. Goyard, S., and A. Ullmann. 1993. Functional analysis of the cya promoter of Bordetella pertussis. Mol. Microbiol. 7:693–704. 16. Graeff-Wohlleben, H., H. Deppisch, and R. Gross. 1995. Global regulatory mechanisms affect virulence gene expression in Bordetella pertussis. Mol. Gen. Genet. 247:86–94. 17. Gross, R., and R. Rappuoli. 1988. Positive regulation of pertussis toxin expression. Proc. Natl. Acad. Sci. USA 85:3913–3917.

4452

FUCHS ET AL.

18. Gross, R., and R. Rappuoli. 1989. Pertussis toxin promoter sequences involved in modulation. J. Bacteriol. 171:4026–4030. 19. Hall, C. V., M. vanCleemput, K. H. Muench, and C. Yanofsky. 1982. The nucleotide sequence of the structural gene for Escherichia coli tryptophanyltRNA synthetase. J. Biol. Chem. 257:6132–6136. 20. Kitten, T., and D. K. Willis. 1996. Suppression of a sensor kinase-dependent phenotype in Pseudomonas syringae by ribosomal proteins L35 and L20. J. Bacteriol. 178:1548–1555. 21. Lacey, B. W. 1960. Antigenic modulation of Bordetella pertussis. J. Hyg. 58:57–93. 22. Leininger, E., M. Roberts, J. G. Kenimer, C. I. G. Fairweather, N. Novotny, and M. J. Brennan. 1991. Pertactin, an Arg-Gly-Asp containing Bordetella pertussis surface protein that promotes adherence of mammalian cells. Proc. Natl. Acad. Sci. USA 88:345–349. 23. Leslie, P. H., and A. D. Gardner. 1931. The phases of Haemophilus pertussis. J. Hyg. 31:423–455. 24. Manoil, C. 1990. Analysis of protein localization by use of gene fusions with complementary properties. J. Bacteriol. 172:1035–1042. 25. Monack, D., B. Arico, R. Rappuoli, and S. Falkow. 1989. Phase variants of Bordetella bronchiseptica arise by spontaneous deletions in the vir locus. Mol. Microbiol. 3:1719–1728. 26. Morales, V. M., A. Backman, and M. Bagdasarian. 1991. A series of widehost-range low-copy-number vectors that allow direct screening for recombinants. Gene 97:39–47. 27. Parkinson, J. S., and E. C. Kofoid. 1992. Communication modules in bacterial signaling proteins. Annu. Rev. Genet. 26:71–112. 28. Petering, H., S. Hammerschmidt, M. Frosch, J. van Putten, C. A. Ison, and B. D. Robertson. 1996. Genes associated with the meningococcal capsule complex are also found in Neisseria gonorrhoeae. J. Bacteriol. 178:3342–3345. 29. Re´gnier, P., M. Grunberg-Manago, and C. Portier. 1987. Nucleotide sequence of the pnp gene of Escherichia coli encoding polynucleotide phosphorylase. J. Biol. Chem. 262:63–68. 30. Relman, D. A., M. Domenighini, E. Tuomanen, R. Rappuoli, and S. Falkow. 1989. Filamentous hemagglutinin of Bordetella pertussis: nucleotide sequence and crucial role in adherence. Proc. Natl. Acad. Sci. USA 86:2637–2641. 31. Roy, C. R., and S. Falkow. 1991. Identification of Bordetella pertussis regu-

J. BACTERIOL.

32. 33.

34.

35.

36.

37.

38.

39.

40.

41.

42. 43.

latory sequences required for transcriptional activation of the fhaB gene and autoregulation of the bvgAS operon. J. Bacteriol. 173:2385–2392. Sambrook, J., E. F. Fritsch, and T. Maniatis. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Scarlato, V., B. Arico, A. Prugnola, and R. Rappuoli. 1991. Sequential activation and environmental regulation of virulence genes in Bordetella pertussis. EMBO J. 10:3971–3975. Scarlato, V., and R. Rappuoli. 1991. Differential response of the bvg virulence regulon of Bordetella pertussis to MgSO4 modulation. J. Bacteriol. 173:7401–7404. Simon, R., U. Priefer, and A. Pu ¨hler. 1983. A broad host range mobilization system for in vivo genetic engeneering: transposon mutagenesis in gram negative bacteria. Bio/Technology 1:784–791. Sprengart, M. L., E. Fuchs, and A. G. Porter. 1996. The downstream box: an efficient and independent translation initiation signal in Escherichia coli. EMBO J. 15:665–674. Stainer, D. W., and M. J. Scholte. 1970. A simple chemically defined medium for the production of phase I Bordetella pertussis. J. Gen. Microbiol. 63:7501– 7510. Steffen, P., S. Goyard, and A. Ullmann. 1996. Phosphorylated BvgA is sufficient for transcriptional activation of virulence-regulated genes in Bordetella pertussis. EMBO J. 15:102–109. Stibitz, S. 1994. Mutations in the bvgA gene of Bordetella pertussis that differentially affect regulation of virulence determinants. J. Bacteriol. 176: 5615–5621. Stibitz, S., and M. S. Yang. 1991. Subcellular localization and immunological detection of proteins encoded by the vir locus of Bordetella pertussis. J. Bacteriol. 174:4288–4296. Uhl, M. A., and J. F. Miller. 1994. Autophosphorylation and phosphotransfer in the Bordetella pertussis BvgAS signal transduction cascade. Proc. Natl. Acad. Sci. USA 91:1163–1167. Uhl, M. A., and J. F. Miller. 1995. BvgAS is sufficient for activation of the Bordetella pertussis ptx locus in Escherichia coli. J. Bacteriol. 177:6477–6485. Weiss, A. A., and S. Falkow. 1984. Genetic analysis of phase change in Bordetella pertussis. Infect. Immun. 43:263–269.