Regulated. bronchiseptica and B. pertussis. Deletion of bvgas or modulation. In this phase vag genes are not induced and vrg loci

JOURNAL OF BACTERIOLOGY, June 1993, p. 3468-3479 0021-9193/93/113468-12$02.00/0 Copyright 1993, American Society for Microbiology Vol. 175, No. 11 Flagellin Gene Transcription in Bordetella bronchiseptica Is Regulated by the BvgAS Virulence Control System BRIAN J. AKERLEY1 AND JEFF F. MILLER' 2* Department of Microbiology and Immunology, School of Medicine,`* and Molecular Biology Institute,2 University of California, Los Angeles, California 90024 Received 19 January 1993/Accepted 19 March 1993 The products of the bvgas locus activate expression of a majority of the known Bordetefla virulence factors but also exert negative control over a class of genes called vrg genes (bvg-repressed genes). BvgAS negatively controls the production of flagella and the phenotype of motility in Bordetefla bronchiseptica. In this studyjfa, the flagellin gene, was cloned and characterized to facilitate studies of this negative control pathway. An internalfaa probe detected hybridizing sequences on genomic Southern blots ofbordetefla pertussis, BordeteUa parapertussis, and Bordetefla avium, although B. pertussis and B. parapertussis are nonmotile. FlaA is similar to the FliC flagellins of Salmonella typhimurium and Escherichia coli, andflaa complemented an E. coli flagellin mutant. Insertional inactivation of the chromosomal flaa locus eliminated motility, which was restored by complementation with the wild-type locus. Analysis offlaa mrna production by Northern (RNA) blotting and primer extension indicated that negative regulation by BvgAS occurs at the level of transcription. The transcriptional start site offlaa mapped near a consensus site for the alternative sigma factor, cf, encoded by flia in E. coli and S. typhimurium. Consistent with a role for afll4 analog in B. bronchiseptica, transcriptional activation of ajfaa-lacz fusion in E. coli requiredfli4 and aflaa-linked locus designatedfrl.frl also efficiently complemented mutations in the flagellar master regulatory locus,flhdc, ofe. coli. Our analysis of the motility phenotype of B. bronchiseptica suggests that the Bordetefla virulence control system mediates transcriptional control offaa through a regulatory hierarchy that includes thefrl locus and an alternative sigma factor. The BvgAS signal transduction system mediates a morphogenic program involving coordinate regulation of virulence genes as well as alterations in cell shape, surface structures, and colony morphology (3, 43, 44, 73). This program is observed as a biphasic transition in members of the genus Bordetella (1, 11, 30, 36). These organisms are gram-negative aerobes which colonize primarily the upper respiratory tracts of humans and other animals. Bordetella pertussis produces whooping cough in humans (73), and Bordetella bronchiseptica is a common commensal that also causes disease in wild and domesticated animals (23). Genetic, molecular, and phenotypic analyses indicate that these two species are closely related (2, 29, 50) and that both species express functionally interchangeable alleles of bvgas (49). Environmental signals, including sulfate anion, nicotinic acid, and growth at low temperature, specifically modulate the activity of the BvgS protein (43, 45). Current models of Bvg function suggest that BvgS phosphorylates BvgA (3, 43, 65, 68), which then binds to specific DNA sequences to regulate transcription (55, 56). In the Bvg' phase the BvgAS system is active, resulting in the expression of vag genes (bvg-activated genes) and the repression of other genes designated vrg genes (bvg-repressed genes) (30). The vag loci encode an array of products, including potential adhesins such as filamentous hemagglutinin (FhaB) and pili (53, 74), an adenylate cyclase toxin with hemolytic activity (20, 24), pertussis toxin (25), dermonecrotic toxin (39), and pertactin, an adhesin (13, 31, 37). These factors, with the exception of pertussis toxin, are expressed in both B. * Corresponding author. 3468 bronchiseptica and B. pertussis. Deletion of bvgas or modulation of BvgAS by environmental signals induces the Bvgphase. In this phase vag genes are not induced and vrg loci are expressed. Loss of expression of certain vag loci abrogates virulence (72). The role of vrg loci in virulence is not as well defined. However, a TnphoA insertion in vrg-6 reduces colonization of mice by B. pertussis, suggesting that pathogenesis also requires factors that are negatively controlled by bvgas (10). Analysis of diverse Bvg-regulated phenotypes suggests that the Bordetella virulence regulon includes multiple regulatory inputs. BvgAS has been implicated as a direct activator of the flab and bvgpl promoters, while Bvgmediated activation of the pertussis toxin operon and the adenylate cyclase toxin gene appears to require additional factors (22, 24, 27, 45, 46, 57). Furthermore, Beattie et al. (8, 9) have proposed that bvgas regulates vrg-6 in B. pertussis by a mechanism involving sequences downstream of the promoter. We recently reported that Bvg negatively controls motility in B. bronchiseptica (1). Studies of systems that control motility in other organisms indicate that flagellar genes are usually transcribed as a hierarchical unit regulated by a global control locus. In Salmonella typhimurium and Escherichia coli (5, 7, 34, 62), a transcriptional control factor, cyclic AMP (camp) receptor protein, mediates a response to nutrient limitation and initiates a transcriptional cascade of flagellar genes. Caulobacter crescentus motility control involves a currently unknown coupling mechanism between spatially organized gene expression and cell cycle-dependent synthesis of flagella (21, 51). Control over a large number of structural genes by a small number of key regulatory factors represents a common feature of these systems.

VOL. 175, 1993 BvgAS REGULATES FLAGELLIN TRANSCRIPTION 3469 TABLE 1. Bacterial strains and plasmids used in this study Strain or plasmid Description Source or reference E. coli SMl0)pir DH5a MC4100 MC4101 YK410 YK4104 YK4131 YK4136 YK4146 thi thr leu tona lacysupe reca::rp4-2-tc::mu (Xpir R6K) Kmr F- hsdrl7 supe44 thi-i recal gyra rel4l A(argF-lac)U169 +80dlacZAM15 X- F- arad139 A(argF-lac)U169 rpsls50 relal flhd5301 deoci ptsf25 rbsr X- MC4100 recal aradl39 A(argF-lac)U169 rpsl thi nala thya pyrc46 his X- YK410fliA YK410 flhd YK410fhC YK410fliC 48 BRL 45 45 33 33 33 33 33 Bordetella spp. GP1SN GP1SND BB7865D DM107 DM106 BB7865 BB7870 F912934 15311 BP370 Plasmids pbr322 pbrx1 pbks+ prk290 pss1129 prs550 ptrc99a pba10 pba45 pba61 pba63 pba64 pba66 pba67 pba250 pjm650 pjmflaa-z pjm655 pjm656 Bacteriophages XRS45 kflaa-z Smr Nalr derivative of guinea pig lung isolate of B. bronchiseptica; Bvg' GPlSN::pBA67; Gmr BB7865::pBA67; Gm' AbvgAS derivative of GP1SN; Bvg- GP1SN containing bvgs-c3 constitutive mutation; Bvgc Smr Nalr derivative of human isolate of B. bronchiseptica; Bvg' Spontaneous Bvg- derivative of BB7865 B. avium isolate from a 3-week-old turkey B. parapertussis type strain B. pertussis Apr Tcr cloning vector Apr Tcs pbr322 with a deletion from EcoRI to EcoRV Apr high-copy-number cloning vector Broad-host-range vector; IncP1 Tcr OriTIncP Bordetella suicide vector; Apr Gmr Sms OriTIncP laczya fusion vector; ApT pbr322-derived expression vector containing the Ptrc promoter and lacjq; Apr 14-kb EcoRI fragment with the flaa and fri loci in pbr322; Fig. 1 pbks+ containing the 1.6-kb EcoRV-PstI fragment upstream of flaa BamHI-BglII subclone of pba10; Fig. 1 BamHI-BglII subclone of pba10 containing a deletion at the KpnI site inflaa; Fig. 1 BamHI-SnaBI subclone of pba10; Fig. 1 BamHI-AflhI subclone of pba10; Fig. 1 flaa-internal PstI-KpnI fragment in PstI-HindIII sites of pss1129 SacII-SnaBI fragment containing flaa in EcoRI of prk290; Fig. 1 14-kb EcoRI fragment from pba10 in prk290; Fig. 1 flaa' SacII-PstI fragment in prs550 ptrc99a containing the AflII-SnaBI fragment containing flaa pjm655 with flaa in the opposite orientation X laczya fusion vector for recombination with prs plasmids containing transcriptional fusions; bla-'laczya imm2l inde Ap5 KmS XRS45 x pjmflaa-z recombinant; Kmr addresses the mechanism of vrg gene regulation by BvgAS. Our approach involves the characterization of both positive and negative regulators of flagellin gene expression in order to determine the level at which Bvg exerts negative control. We have cloned, characterized, and investigated the distribution of the flagellin gene, flaa, in Bordetella spp. Analysis of the transcriptional regulation of flaa established bvgas as a negative regulatory locus. We investigated similarities and differences between motilityassociated genes of B. bronchiseptica and E. coli by using complementation of flagellar mutants. In E. coli the flagellin gene encodes the primary constituent of the filament structure. Flagellin represents a late component of the hierarchy and requires early genes for its expression (14, 32, 35, 52). The use of flaa as a reporter gene in E. coli has allowed the identification of two transcriptional activators of flaa that 1 1 1 49 49 1American Type Culture Collection represent potential targets of BvgAS-mediated negative control. 65 58 Stratagene 16 67 63 Pharmacia 63 MATERIALS AND METHODS Bacterial strains, plasmids, and growth conditions. Bacterial strains and plasmids are described in Table 1. Bordetella strains were grown on Bordet-Gengou agar plates (Becton Dickinson Microbiology Systems, Cockeysville, Md.) containing 15% sheep blood. Broth cultures of Bordetella spp. were grown in Stainer-Scholte medium (SSM) (64). E. coli strains were grown on LB agar or LB broth (1% tryptone, 0.5% yeast extract, and 1% NaCl) (Difco Laboratories, Detroit, Mich.). Additions to media after autoclaving where indicated were at the following concentrations: 40 mm MgSO4, 10 mm nicotinic acid, 25 ptg of gentamicin per ml, 20

3470 AKERLEY AND MILLER pug of chloramphenicol per ml, 10 pug of tetracycline per ml, 100 pug of ampicillin per ml, or 40 jig streptomycin per ml. 5-Bromo-4-chloro-3-indoyl-i-D-galactopyranoside (X-Gal) (40 [tg/ml) was added to agar media where indicated to detect 13-galactosidase production. Motility assays for B. bronchiseptica were conducted in SSM with 0.35% agar and antibiotics where indicated. Motility assays for E. coli were conducted in LB broth containing 0.35% agar and antibiotics where indicated. DNA methods. Standard methods were used for the isolation of plasmid and chromosomal DNA, restriction enzyme digestions, agarose gel electrophoresis, and DNA ligations (58). Restriction enzymes, calf intestinal alkaline phosphatase, Klenow fragment, T4 DNA polymerase, T4 DNA ligase, and T4 polynucleotide kinase were from Promega Corp. (Madison, Wis.), New England Biolabs (Beverly, Mass.), or Bethesda Research Laboratories (Gaithersburg, Md.) (BRL) and were used according to the manufacturer's instructions. Radiolabeled nucleotides were from Amersham (Arlington Heights, Ill.). Cloning steps were conducted in competent E. coli DH5ao prepared by standard techniques. E. coli YK410 derivatives were transformed by high-voltage electroporation with a Gene Pulser apparatus (Bio-Rad, Richmond, Calif.) as previously described (17). Colony and Southern hybridizations. DNA electrophoresed on agarose gels was transferred to SS-Nytran nylon membranes (Schleicher & Schuell, Keene, N.H.) by standard techniques (58) and immobilized by UV irradiation in a Stratalinker (Stratagene, La Jolla, Calif.). Colonies were lifted onto Hybond-N membranes (Amersham) and lysed, and DNA was then immobilized in a Stratalinker. Hybridizations with degenerate oligonucleotide probes labeled to an approximate specific activity of 3 x 107 cpm/lig with [y-32p]atp and T4 polynucleotide kinase were conducted according to the procedure of Wood et al. (75) except that 0.5% sodium dodecyl sulfate (SDS) was added to the hybridization buffer to reduce nonspecific binding to nylon membranes. Probes FN24 and FN27 were degenerate oligonucleotides based on the previously determined N-terminal amino acid sequences of flagellin in B. bronchiseptica (VAQN NLNK and MAAVINTNY, respectively). Hybridizations with oligonucleotides were conducted at 37 C. Washes were conducted at 45 to 55 C in the presence of 3 M tetramethylammonium chloride as described previously (75). Doublestranded DNA probes for hybridization analysis were isolated from low-melting-point agarose, denatured, and radiolabeled with [ot-32p]datp to specific activities of approximately 2 x 107 cpm/jig by using the Random Primers DNA Labeling System (BRL). Hybridizations with [a-32p] datp-labeled heterologous probes were conducted at 55 C as described previously (58). Membranes were washed at 55 C in 0.1x SSC with 0.1% SDS (0.1x SSC is 15 mm NaCl plus 1.5 mm sodium citrate). Determination of nucleotide sequence. DNA sequencing was performed by the dideoxy sequencing method of Sanger et al. (59), using a commercially available kit (Sequenase; U.S. Biochemical Corp., Cleveland, Ohio) essentially according to the manufacturer's instructions, with [a-35s] datp. Labeling reactions were conducted at 4 C, which seemed to improve polymerase processivity over GC-rich regions. Double-stranded DNA templates were denatured with NaOH. Acrylamide gels containing 4 to 20% formamide were used to resolve gel compression artifacts. Deletions were made in DNA subcloned from pba10 into pbks+ (Stratagene), using exonuclease III and mung bean nuclease (26). Subclones were sequenced with primers (Promega) J. BACTERIOL. complementary to the T7 and T3 sites in Bluescript. Custom primers were synthesized by an Applied Biosystems model 470B automated DNA synthesizer (Applied Biosystems, Foster City, Calif.), purified by oligonucleotide purification cartridge chromatography (Applied Biosystems), and used in addition to subclones to obtain sequences of both strands for the majority of the gene. Sequences were analyzed with the University of Wisconsin sequence analysis software (15). A potential terminator was identified by using the TERMINA- TOR program (12). RNA analysis. Total cellular RNA was isolated from mid-logarithmic-stage cultures of B. bronchiseptica by the hot-phenol method as described previously (70). Doublestranded probes for Northern (RNA) hybridizations were prepared as described above. Electrophoresis of RNA and Northern hybridizations were conducted as described previously (58). Oligonucleotides used for primer extension assays (Bvg-P, 5'-GAGGACTITGTTGTACATG; BAO10, 5'- CCAGCGACAAGTAGTTGG; BAO3, 5'-CAGGTTGTTCT GGGCAAC; and BAO5, 5'-GTGAAGTTCGGAAATGTG) were purified as described above and end labeled with T4 polynucleotide kinase. Reaction conditions were as follows. RNA (10 jig) was annealed to approximately 1 ng of each radiolabeled primer for 1 h at 440C in PE buffer (0.25 M KCl and 10 mm Tris [ph 8.0]), and extensions with Moloney murine leukemia virus H+ reverse transcriptase (BRL) were conducted at 440C after the addition of 0.8 volume of RT buffer (25 mm KCl, 50 mm Tris [ph 8.3], 10 mm dithiothreitol, 3.5 mm MgCl2, 0.5 mm each deoxynucleoside triphosphate, and 0.1 mg of bovine serum albumin per ml). The products were concentrated by ethanol precipitation and boiled in 30 mm NaOH prior to electrophoresis. The product from 3 jig of RNA was loaded in each lane. Detection of flagellin. Cell surface-associated flagellin was detected by immunoblotting with monoclonal antibody 15D8 as previously described (1, 18). Briefly, B. bronchiseptica grown overnight on Bordet-Gengou plates containing 10 mm nicotinic acid was suspended in phosphate-buffered saline at 4 C, and flagella were sheared from cells by vortexing for 30 s. Cells were removed by two sequential centrifugations at 10,000 x g. Proteins in the cell-free supernatants were concentrated by trichloroacetic acid precipitation, and amounts equivalent to an optical density of 1.5 at 600 nm were analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (10% polyacrylamide). To confirm equal loading, total protein was visualized by Coomassie staining as previously described (1). Construction of chromosomal fla-lacz fusions. A SacII- BamHI fragment containing the SacII-PstI 5' of the flagellin gene, including the first codon, was obtained from pba45. The cohesive ends were removed with T4 DNA polymerase prior to the ligation of phosphorylated BamHI linkers to the blunt ends. This fragment was cloned into the BamHI site of prs550 (Table 1) and screened for the orientation consistent with the direction offlaa transcription to create pjmflaa-z. The flaa-lacz fusion was then transferred onto XRS45 by in vivo homologous recombination in MC4100 (Table 1). The recombinant phage XflaA-Z was used to lysogenize E. coli YK4104 and YK410. These lysogens were then transformed with either pbrx1 or pba66. Cells were harvested in the mid-logarithmic stage, and 0-galactosidase assays were performed as described by Miller (47). Three independent samples were measured in duplicate experiments. Nucleotide sequence accession number. The DNA sequence data described in this paper have been deposited in GenBank with the accession number L13034.

VOL. 175, 1993 BvgAS REGULATES FLAGELLIN TRANSCRIPTION 3471 E. coli complementation pbal 0 1 -'-r pjm650 E BH1 frli 500 bp pba61 F- -T~-..- flaa S2 2I a S B S2 A2 P P KliSBi B2 E +* + + i pba63 A i pba64 I l pba66 I- -I pba250 I w pjm655 Ptrci I pjm656 pba64 + pba250 I I1tr -iptrc pba66 + pba250 FIG. 1. Complementation of E. coli mutants by B. bronchiseptica sequences. The top line represents the structure of the 14-kb EcoRI fragment from GP1SN containing the flaa locus and an adjacent regulatory region designatedfrl. All plasmids are pbr322-derived replicons except for pjm650 and pba250, which are prk290 replicons. The locations of the flaa open reading frame and the transcription start site (arrow) are shown. The shaded PstI fragment hybridized to FN27 and was used as the probe for the Southern and Northern blots shown in Fig. 4 and 5, respectively. frl activity localizes to the BamHI-AflII restriction fragment present in pba66. Isogenic E. coli strains carrying mutations in the indicated loci are described in Table 1. Complementation offlhc and flhd mutations resulted in efficient swarming (+), and complementation of the E. coli flic mutant generates a bushy swarm phenotype (+ *). In all cases, vector plasmids did not complement motility mutations. Restriction sites: E, EcoRI; BH1, BamHI; S2, SacII; A2, AflII; P, PstI; K1, KpnI; SB1, SnaBI; B2, BglII. RESULTS Isolation of theflaa gene. To isolate the B. bronchiseptica flagellin gene, we designed two nonoverlapping degenerate oligonucleotides based on the previously determined N-terminal amino acid sequence of the protein (1). These probes, FN24 and FN27, detected a major common band and several weakly hybridizing bands unique to each probe on Southern blots of genomic DNA digested with BamHI, EcoRI, or BglII (data not shown). The common band in EcoRI-digested chromosomal DNA from strain GP1SN was approximately 14 kb. Restriction fragments in this size range were isolated and cloned into pbr322. Plasmid pba10 containing a single 14-kb EcoRI insert (Fig. 1) was isolated from this library on the basis of hybridization to FN27 and subsequent confirmation with FN24. FN27 hybridized to a single 260-bp PstI fragment on Southern blots of pba10 (shaded in Fig. 1). Southern hybridization of FN27 to pba10 digested with five additional restriction enzymes detected only one hybridizing fragment in each digest (data not shown). The 260-bp PstI fragment was subcloned and sequenced. The predicted amino acid sequence corresponded exactly to the N-terminal amino acid -I +* +* sequence of the flagellin protein expressed by GP1SN (1). The cloned gene is designated flaa. DNA sequence of JlaA. The complete DNA sequence of flaa was determined to facilitate an analysis of its regulation and role in pathogenesis. Sequence analysis (Fig. 2) revealed a 1,173-nucleotide open reading frame predicted to encode a protein of 391 amino acids with a molecular mass of 40.5 kda. This is consistent with the estimated size of the GP1SN flagellin (40 kda) previously determined by SDS-PAGE (1). As with other flagellins, no potential signal peptide-coding sequences were found. A potential transcriptional terminator was located and is shown in Fig. 2, downstream of the predicted translational stop codon. In keeping with the strong G+C base preference of other Bordetella genes (66), the predicted protein-coding region exhibits a G+C nucleotide composition of 64%, while the flanking DNA is 66% G+C. The predicted amino acid sequence of the B. bronchiseptica flagellin is remarkably similar to those of E. coli and S. typhimurium. Overall, FlaA displayed 74% amino acid similarity to S. typhimunium FliC, with 60.6% amino acid identity when compared by using BESTFIT (15). E. coli FliC

3472 AKERLEY AND MILLER 1 61 10 30 50 TGCGGGGACAGGCACCTGCCCCATCTCCCCCCGCCGCACGACGCCTGTCCGCAGGGGGAC 70 90 110 TGGCGCCTGCCCTATCCCGCCCGCGCCGCACGGACGCCTGTCCCCGCAGGAACATGCCCT 130 150 170 121 TTGCCGCCAGATTCCCCCGCACATTTCCGAACTTCACTTT=rlIp TCCGTCGCA 190.- r 210 230 181 AACCl CCGTAAlCAGGCAACAAAGGAAATCGCGGCGCTGTGCAAGCGAAAGTCCGATG 250 270 290 241 TTACAGATGGGCGGCCTAGCTGCCCGGTTTGAAGAAGCCTTTCTCTCTTGGGAGCCTCAA 310 330 350 301 M A A V I N T N Y L S L V A O N N L N K 370 390 410 361 GTCCCAATCGGCCCTGGGTAGCGCCATCGAGCGCCTGTCGTCGGGTCTGCGCATCAACAG S Q S A L G S A I E R L S S G L R I N S 430 450 470 421 A K D D A A G Q A I A N R F T A N V K G 490 510 530 481 CCTGACCCAGGCTGCCCGCAACGCCAACGACGGCATCTCGATCGCCCAGACGACCGAAGG L T Q A A R N A N D G I S I A Q T T E G 550 570 590 541 CGCGCTGAACGAAATCAACAACAACCTGCAGCGCATCCGCGAACTGACGGTTCAGGCCTC A L N E I N N N L Q R I R E L T V Q A S 610 630 650 601 CAACGGCACGAACTCGGCTTCGGACATCGACTCGATCCAGCAGGAAGTCAACCAGCGCCT N G T N S A S D I D S I Q Q E V N Q R L 670 690 710 661 GGAAGAAATCAACCGCATCGCCGAGCAGACCGACTTCAACGGCATCAAGGTCCTGAAGTC E E I N R I A E Q T D F N G I K V L K S 730 750 770 721 CAACGCCACCGACATGACCCTGTCGATCCAGGTCGGCGCCAAGGACAACGAAACGATCGA N A T D M T L S I Q V G A K D N E T I D 790 810 830 781 TATCAAGATCGATCGCAACTCGAACTGGAACCTGTATGACGCCGTGGGCACCGTCCCGGG I K I D R N S N W N L Y D A V G T V P G 850 870 890 841 CGGCACGGTCAACGGCGAGGCTCGCACCGTCAACGCGCTGGGCTTTGACGTGCTGTCGGC G T V N G E A R T V N A L G F D V L S A 910 930 950 901 CGTCACGACCACCATCGCTTCCGACACCGTGACCTTCGACGCCGCCGTGGCGGCCGCTGA V T T T I A S D T V T F D A A V A A A E 970 990 1010 961 ACAGGCCGCTGGCGCCGCCGTAGGCGACGGCAGCGTCGTCTCGTACGGCGATACCGCCAA Q A A G A A V G D G S V V S Y G D T A N 1030 1050 1070 1021 CCCGCAATACGCGGTCGTGGTCGACAATGCCGGCACGATGACCTCGTACGCCCTGACCTT P Q Y A V V V D N A G T M T S Y A L T F 1090 1110 1130 1081 CGACAAGGACGGCAAGGCCGCCCTGGGCGACCAGCTGGGTGCCGTCGCCTCGCAGGCTGC D K D G K A A L G D Q L G A V A S Q A A 1150 1170 1190 1141 GGAAGCCGCCGTCGGTACCAACGACGTCGCTGCCGGCGCCAACGTCACCGTGTCCGGCGG E A A V G T N D V A A G A N V T V S G G 1210 1230 1250 1201 CGCCGCCGACGCGCTGTCCAAGCTGGACGACGCCATGAAGGCCGTGGACGAACAGCGCAG A A D A L S K L D D A M K A V D E Q R S 1270 1290 1310 1261 CTCGCTGGGCGCGATCCAGAACCGTTTCGAATCGACCGTTGCCAACCTGAACAACACGAT S L G A I Q N R F E S T V A N L N N T I 1330 1350 1370 ---------------------------------------------- - - -- T N L S A A R S R I E D S D Y A T E V S 1390 1410 1430 1381 GAACATGACCAAGAACCAGATCCTGCAACAGGCTGGCACCTCGGTCCTGGCCCAGGCCAA N M T K N Q I L Q Q A G T S V L A Q A N 1450 1470 1490 1441 CCAAGTCCCGCAAAACGTCCTGTCGCTGCTGCGCTAAGCATCGGCACACCCCGGGCGGGC Q V P Q N V L S L L R * 1510 1530 1550 1501 ACAACCCGCCCGGGGCCATGGCAGTACGTACGCCCGGCCGCCTCACCGCGCCGGGCGTTT 1570 1561 TTGCGTCCGGCG FIG. 2. Nucleotide sequence of flaa from B. bronchiseptica GP1SN. The boxed sequences at positions 170 and 190 are u consensus sites, and the arrow indicates the transcriptional start site identified by primer extension. A putative ribosome-binding site preceding the start of translation is underlined. The 20 amino acids following the initiator methionine correspond to the N-terminal amino acid sequence determined from purified flagellin protein (1). A potential transcriptional terminator is indicated by dashed lines. J. BA=rRIOL. showed 70% similarity to FlaA, with 55.7% amino acid identity. A comparison of S. typhimurium and E. coli FliC proteins indicated 67.8% similarity and 57.2% identity. Therefore, the flagellin protein of S. typhimunium displays greater similarity to B. bronchiseptica FlaA than to E. coli FliC. As observed previously with other flagellins (28, 40), amino acids positioned at the amino and carboxyl termini of the protein are highly conserved, while the amino acid sequence diverges in the central region. The amino-terminal 95 amino acids and carboxy-terminal 46 amino acids of the GP1SN and S. typhimunium flagellins showed 92 and 93% identity, respectively. fla4 encodes flagellin and is required for motility. Our previous results demonstrated that two flagellin types, distinguished by apparent molecular mass (35 versus 40 kda), are produced by diverse clinical isolates of B. bronchiseptica, with only one form detectable in any particular strain (1). S. typhimurium alternates between expression of two flagellins (19, 69), the products offlic andfljb, both of which encode proteins similar to FlaA. To determine if the cloned flaa gene is required for B. bronchiseptica motility and to investigate the possibility that this organism expresses more than one flagellin gene, we constructed flaa mutant strains by insertional mutagenesis. A DNA fragment internal to the deduced protein-coding region of flaa was used to create pba67, a Gmr suicide plasmid predicted to disrupt the flagellin gene by homologous recombination leading to cointegrate formation. pba67 was introduced into GP1SN and BB7865 from an E. coli donor. The genomic structure at the flaa locus of insertion mutants and subsequently complemented strains was determined by Southern blot analysis and was found to be consistent with a single chromosomal insertion as shown in Fig. 3A. Cointegrate strains GP1SND and BB7865D were nonmotile in soft-agar motility assays (Fig. 3C). These strains also failed to produce detectable levels of flagellin protein isolated from cell surfaces (Fig. 3B). Even with prolonged incubation in the presence of gentamicin, motile revertants were not observed, whereas motile revertants that had resolved the pba67 cointegrate were readily obtained in the absence of antibiotic selection. This result indicates that B. bronchiseptica, in contrast to S. typhimunium, does not contain another flagellin gene that can function in our flaa mutants under conditions that select for motility. Motility was restored by introducing pba250 (Fig. 1) into GP1SND or BB7865D, whereas the vector plasmid prk290 had no effect (Fig. 3C). pba250 is a prk290 derivative containing theflaa open reading frame, 900 bp of DNA upstream of the predicted translation start codon, and 50 bp downstream of the predicted termination codon. The swarm rate of the complemented strain shown in Fig. 3C is less than that of the wild type. This effect results from slower growth due to the presence of tetracycline, which was present during the assay to maintain the plasmid. These results show that flaa is required for motility and that the complete flaa gene is present on pba250, which is capable of restoring motility to flaa mutants. As observed in wild-type GP1SN, the complemented strain GP1SND/pBA250 expressed a major 40-kDa band (Fig. 3B) which reacts with the antiflagellin monoclonal antibody 15D8 (18). Consistent with our previous observations with partially purified flagella (1), the antibody detected an additional species migrating slightly more slowly than the 40-kDa band. This band represents either a 15D8-reactive protein which copurifies with FlaA, a modified derivative of

VOL. 175, 1993 BvgAS REGULATES FLAGELLIN TRANSCRIPTION 3473 A GP1 SN BB7865 Ap flaa x go pbd6r OnR OdR Gm OriT B 12 GP1SN FlaA -_ BB7865 -.: FMaA 2 3 4 5 6 7 8 7w Mr 44w.4. -58-41 -35-26 00* GP1 SN::BA67 BB7865::BA67 fi8a Gm OiR fla4 C 1 2 3 a h b 4 Downloaded from http://jb.asm.org/ FIG. 3. Disruption and complementation offlaa. (A) The chromosomalflaa loci in B. bronchiseptica GP1SN and BB7865 were disrupted by homologous recombination with pba67 to give the cointegrate structures shown. (B) Western blot (immunoblot) of flagellin preparations with monoclonal antibody 15D8. Lanes: 1, GP1SN; 2, GP1SND; 3, GPlSND/pBA250; 4, GPlSND/pRK290; 5, BB7865; 6, BB7865D; 7, BB7865D/pBA250; 8, BB7865D/pRK290. Molecular masses (kilodaltons) are indicated on the right. (C) Motility assays: la, GP1SN; 2a, GP1SND; 3a, GPlSND/pBA250; 4a, GPlSND/pRK290; lb, BB7865; 2b, BB7865D; 3b, BB7865D/pBA250; 4b, BB7865D/pRK290. All assays were in the presence of nicotinic acid. Preparations in 2a and 2b contain gentamicin to select for the cointegrate; those in 3a, 3b, 4a, and 4b contain gentamicin and tetracycline to select for both the cointegrate and either pba250 or prk290. FlaA, or a precursor which is cleaved to yield FlaA. As noted above, the amino acid sequence predicted from the flaa DNA sequence corresponds exactly to the previously determined N-terminal sequence of the major 40-kDa polypeptide. Several minor products which react with 15D8 migrate faster than the 40-kDa band in the SDS-PAGE gel. The sizes, expression patterns, and antibody reactivities of these species suggest that they are degradation products of FlaA. To further demonstrate the protein-coding capacity of flaa, we exploited the size difference between the flagellins of strains GP1SN (40 kda) and BB7865 (35 kda). Flagella were isolated from BB7865D/pBA250. pba250 conferred expression of the 40-kDa flagellin characteristic of GP1SN, while the vector alone had no effect. This result directly demonstrated that pba250 encodes a 40-kDa flagellin protein and that the 40-kDa flagellin can substitute for the 35-kDa species expressed by BB7865. Bordeteila pertussis, Bordetella avium, and BordeteUa paraperussis containflaa4-hybridizing sequences. Members of the Bordetella genus show a high degree of genetic similarity, yet they express species-specific as well as common phenotypes regulated by BvgAS. As shown in Fig. 4, a probe internal to flaa hybridized under high stringency to EcoRIdigested genomic DNA from each of the four recognized Bordetella species. A 14-kb fragment from B. bronchiseptica GP1SN and BB7865 hybridized to the probe as expected. The probe also hybridized to a 5.5-kb fragment from B. avium, a species which expresses motility and produces a flagellin that differs in estimated molecular weight from the two types found in B. bronchiseptica strains (1). In contrast, B. pertussis and B. parapertussis, which do not exhibit motility under any laboratory condition tested, contain either a 12- or a 14-kb DNA fragment which hybridized to the flaa probe. Furthermore, B. pertussis BP370 does not produce detectable levels offlaa mrna when tested by Northern hybridization with the flaa-specific probe or by primer extension (42). These results suggest that the genomes of B. pertussis and B. parapertussis contain inactive flagellin genes. pba10 complements fla loci in E. coli. The similarity between the B. bronchiseptica and E. coli flagellins suggested the possibility of functional conservation between motility-associated gene products. We therefore tested complementation of a set of E. colifla mutants by pba10 and its on November 12, 2018 by guest

3474 AKERLEY AND MILLER FIG. 4. 1 2 3 4 5 14 kb - 10 MUXgm 12 kb - 5.5 kb- Southern hybridization of EcoRI-digested genomic DNA from B. pertussis BP370 (lane 1), B. bronchiseptica GP1SN (lane 2), B. bronchiseptica BB7865 (lane 3), B. avium F912934 (lane 4), and B. parapertussis 15311 (lane 5). The probe used is the same as for Fig. 5A. derivatives (Fig. 1). pba1o and derivative pba61 complemented E. coli YK4146 containing a nonrevertible mutation in the flagellin gene, flic. This complementation, however, was inefficient, producing a swarm phenotype recognizable after 18 h as opposed to approximately 6 h for the wild-type strain. In addition, swarming was qualitatively different, leading to "bushy swarms" similar to those described by Komeda and lino for E. ccli hook-associated protein mutants (33). It is likely that this unusual motility in YK4146 complemented with pba1o is caused by a defect in assembly of heterologous flagellar components (see below). pba63 contains a deletion predicted to create a nonsense mutation in the coding region of flaa and does not complement flic. Truncation or deletion of flaa also abrogated flic complementation (pba64 and pba66, respectively). In E. coli the flia gene, encoding e, is required for expression of late genes in the flagellar regulon (14) and is closely linked to the fiagellin locus. The cloned 14-kb DNA region containing faa, however, failed to complement motility in a flia4 mutant of E. ccli (Fig. 1). The possibility remains, however, that B. bronchiseptica contains a flia analog that is unlinked tofaa or is not functional in E. ccli. Surprisingly, sequences adjacent to flaa efficiently complemented flhc and fihd mutations in E. ccli (Fig. 1). flhdc constitutes a master regulatory locus which is required for expression of the entire fiagellar regulon in E. ccli (5). pba1o derivatives were used to map the flhdc-complementing region, which we designate fri (flagellin regulatory locus). pba66 represents the smallest clone containing fri. It is likely that fri plays a key role in BvgAS-mediated regulation of B. brcnchiseptica motility. Although pba2so contains sequences sufficient for complementation of flaa mutations in B. bronchiseptica, it was not sufficient for flig complementation in E. ccii (Fig. 1). This result was unexpected, and we hypothesized that the lack of complemnentation could be due a requirement for one or more genes in addition toflaa that either play a structural role in assembling the fiagellar filament or are required for flaa expression in E. coli YK4146. To differentiate between these possibilities, the flaa structural gene was inserted downstream from an IPTG (isopropyl-p-d-thiogalactopyra- promoter to create pjm655. pjm655 noside)-inducible Ptrc weakly complemented the flic mutation in the absence of IPTG, and addition of IPTG greatly increased the motility phenotype. Although the extent and kinetics of swarming in the pjm655-complemented mutant were similar to those of the wild type, the bushy swarm phenotype was still observed. The vector ptrc99a, and pjm656 containing flaa in the orientation opposite to that of pjm655, did not complement the flic mutation. These results indicate that expression of flaa alone is sufficient for flic complementation and that the inability of pba250 to induce swarming in flic mutants results from insufficient flaa expression. The addition of pba64 or pba66 to flic mutants containing pba250 restored complementation of the motility defect. Both of these plasmids contain fri, and it is therefore likely that the product(s) of this locus acts in trans to increase flagellin production by pba250 (see below). Transcriptional control of faa by bvgas. To begin an analysis of the mechanism of negative control of motility, we analyzed transcription of the flaa gene. A 260-bp PstI fragment internal to the coding region of flaa (Fig. 1) was hybridized to total cellular RNA isolated from GP1SN and a Bvg- derivative, DM107, in which the bvgas locus is deleted. A 1.4-kb RNA species was detected in DM107 but not in GP1SN grown under conditions permissive for Bvg activity (Fig. 5A). The distance between the start site and the potential transcriptional terminator, located from nucleotide positions 1227 to 1260, predicts a monocistronic mrna molecule approximately 1.36 kb in length, in agreement with the Northern blot data. The role of bvgas in the transcriptional regulation offlaa was confirmed by primer extension (Fig. SB and C). Total cellular RNA from B. bronchiseptica was reverse transcribed from kinase end-labeled primers complementary to the 5' end of either flaa or bvga. B. bronchiseptica strains were grown to mid-log phase in broth cultures in the presence or absence of environmental signals that modulate BvgAS activity. The flaa primer yielded a single extension product in RNA samples from GP1SN grown in the presence of nicotinic acid or MgSO4. The same product was detected in the bvg deletion mutant DM107 grown at 370C without the addition of modulators. No product was detected with RNA from GP1SN grown in the absence of modulating signals or with strain DM106 grown in the presence of nicotinic acid. DM106 contains the bvgs-c3 allele, which encodes a constitutively active BvgS protein that is not inactivated by modulators (43). These results indicate that flaa is negatively regulated by bvg at the transcriptional level. As a control for RNA preparation and to verify the phase (Bvg' or Bvg-) of the B. bronchiseptica cultures used for Fig. SB, primer extension reactions were performed with a primer hybridizing to the 5' J. BACT1ERIOL. end of the bvgas operon (Fig. SC). bvgas transcription is autoregulated in Bordetella spp. by a mechanism involving a switch between two distinct promoters (57, 60). In the presence of modulating signals, the bvgp2 promoter is active. Conditions which are permissive for activation of bvg-dependent genes induce the bvgpl promoter and repress bvgp2. As shown in Fig. SC, the transcript corresponding to the P1 promoter is expressed in GP1SN grown in SSM at 370C, and the P2 extension product is not detected. In the presence of Bvg-modulating signals, the P2 promoter is expressed, while P1 is not detected. The bvgs-c3 strain, DM106, expresses the P1 transcript, and not

VOL. 175, 1993 A1 2 ~~~~~~~~~~~~~~~~Ii B AGCT 1 ece 4.il 7, F : BvgAS REGULATES FLAGELLIN TRANSCRIPTION 3475 2 3 4 5 C AGCT 1 2 3 4 5 7-X~~~~~il -1ḃII bvg P2 D B. br AdF --No ~ ~~~~~~ ft 00 * a* -.A -32-10 +1 onch. flaa TTGC TTAA GTCCGTCGCAAACCT GCCGTAATICCAGGC S.t. flgk TTGC GTCCACGTAGTCGCT GCCGGAATICAACGAGTATTGAAG H.c. flic ATTC TAAA GGTTGTTTTACGACA GACGATAA CAGG jittgacggcga S.t. fljb ATAG TAAA GTTTATGCCTCAACTIGTCGATAA CCTGGATGACACAGG Consensus TAAA N15 1GCCGATAAI FIG. 5. Transcriptional analysis of flaa. (A) Northern blot of total cellular RNA isolated from mid-log-phase cultures of GP1SN (Bvg'; lane 1) and DM107 (Bvg-; lane 2) probed with a 260-bp Pst fragment internal toflaa. (B) Primer extension of theflaa transcript, using primer BAO3. RNA was prepared from mid-log-phase cultures of GP1SN grown in SSM (lane 1), GP1SN grown in SSM plus nicotinic acid (lane 2), GP1SN grown in SSM plus MgSO4 (lane 3), DM107 grown in SSM (lane 4), and DM106 grown in SSM plus MgSO4 (lane 5). (C) The same RNA samples analyzed in panel B were used for primer extension with a primer (Bvg-P) complementary to the bvgas transcript. The positions of the bvgp2 and bvgpl start sites are shown. (D) The B. bronchiseptica flaa promoter region is compared with three oa'-dependent promoters and the a' consensus sequence. P2, even in the presence of modulating conditions. The positions of the bvg-specific extension products agree with analyses of these promoters in B. pertussis (57, 60) and B. parapertussis (61). Figure SC, lane 4, serves as a control for nonspecific products, since DM107 does not contain the DNA region containing bvg. An additional band (running above the P2 extension product) was present in all lanes, indicating that it represents an artifact unrelated to bvgas regulation. Results with the bvg-specific primer extend the observation of bvg autoregulation to B. bronchiseptica. Together, the flaa and bvg results demonstrate that flaa transcription is coordinately regulated with the state of bvg expression. The start site of flaa transcription maps to a sequence closely resembling the consensus recognition site for the alternative sigma factor, I, encoded by the fli4 genes of E. coli and S. typhimurium (Fig. SD) (14, 52). The position of the flaa start site at nucleotide -102 relative to the translational start codon was confirmed by primer extension with bvg Pl two additional primers complementary to sequences upstream and downstream of the site recognized by BAO3. The primer upstream of the putative start site, BAO5, did not detect a transcript. The downstream primer, BAO10, yielded a product corresponding to the same start site observed with BAO3 (data not shown). The presence of a putative o-f consensus site near the transcriptional start site suggests that transcription of theflaa gene may require a FliA analog in B. bronchiseptica. Role offli4 andfrl in the transcriptional activation offlaa. To analyze the requirements for transcriptional activation of the flaa gene, we exploited the functional homology between elements of the motility regulons of B. bronchiseptica and E. coli. AflaA-lacZ fusion was constructed and crossed onto XRS45. The fusion immediately follows the flaa ATG and includes no other protein-coding sequences offlaa. The resulting recombinant phage, XflaA-Z, was used to produce chromosomal transcriptional reporter fusions in the isogenic wild-type andflia E. coli backgrounds, YK410 and YK4104,

3476 AKERLEY AND MILLER respectively. Only background levels of P-galactosidase were detected in strains containing the fusions alone, suggesting that additional Bordetella factors may be required for flaa expression in E. coli. The XflaA-Z lysogens were transformed with either the vector control pbrx1 or pba66, which contains the fri locus, and expression of the fusions was measured in P-galactosidase assays. As in the initial lysogens, strains containing pbrx1 without Bordetella sequences expressed only low levels of 3-galactosidase activity (12.3 + 0.8 and 9.7 ± 4.2 U for YK410 and YK4104 lysogens, respectively). In the wild-type strain containing the fusion (YK410/XflaA-Z), pba66 induced an approximately fivefold increase in P-galactosidase expression (59.6 ± 6.6 U) compared with the same strain transformed with pbrx1. The increasedflaa-lacz expression resulted in blue colonies on agar containing X-Gal. pba66 did not potentiate this induction in the flia mutant, in which expression remained at a background level (10.5 ± 5.6 U). Consistent with complementation results, these data suggest that both thefrl locus from B. bronchiseptica and the flia gene from E. coli are required for expression of flaa in E. coli. Therefore, frl and flhdc differ in their abilities to activate flaa, although both are capable of activating E. coli flic. DISCUSSION The role of the Bordetella bvgas locus in the positive control of virulence factor expression has been recognized for a decade (41, 72, 73). More recent studies have highlighted the biphasic nature of the virulence regulon, and the identification of Bvg-repressed loci suggests that the regulatory program controlling infection may be quite complex (1, 30). An investigation of the B. bronchiseptica motility regulon was initiated to study the mechanism of negative regulation by BvgAS. The flagellin gene (flaa) from strain GP1SN was cloned and sequenced, and both its functional role and its transcriptional regulation were characterized in B. bronchiseptica and E. coli. The predicted N-terminal amino acid sequence offlaa is in exact agreement with previous results obtained by directly sequencing the 40-kDa GP1SN flagellin protein. Comparisons between FlaA and the FliC proteins of S. typhimurium and E. coli showed high levels of amino acid identity, especially at the amino and carboxyl termini. Surprisingly, the amino acid similarity between the flagellins of B. bronchiseptica and S. typhimurium is greater than the similarity between E. coli and S. typhimurium FliC proteins. The striking correspondence between the B. bronchiseptica flagellin and the flic products of S. typhimurium and E. coli suggested that additional features of the motility systems of these organisms may also be conserved. The chromosomal flaa gene was insertionally inactivated to examine the function and coding capacity of the cloned sequence and to determine whether the observed molecular weight difference between the flagellins of BB7865 and GP1SN is caused by flagellar phase variation or allelic heterogeneity. Both the GP1SN and the BB7865 flaa loci were disrupted by recombination with a homologous internal fragment cloned into a suicide vector, creating the cointegrate strains GP1SND and BB7865D. Both cointegrate strains were nonmotile, and no surface-associated flagellin was detectable. Complementation of the motility defects in the mutant strains by using a plasmid containing the flaa open reading frame indicated that abrogation of motility was not caused by polar effects on downstream genes. Introduction of the plasmid containing flaa into BB7865D, whose J. BACTERIOL. parent strain expressed the 35-kDa flagellin, restored motility, and the complemented strain produced the 40-kDa flaa gene product characteristic of GP1SN. This demonstrated both the coding capacity of the GP1SN flaa gene and the ability of the 40-kDa flagellin to substitute for the 35-kDa form. Motile revertants of GP1SND did not arise under conditions which selected for motility as well as maintenance of the cointegrate, although motile derivatives that had resolved the cointegrate were readily obtained in the absence of antibiotic selection. These results are consistent with the lack of additional flagellin genes observed by Southern hybridization of genomic DNA. We found no evidence for flagellar phase variation in B. bronchiseptica, and it is likely that flaa represents the sole flagellin gene encoded by this species. We measured flaa mrna production to determine if BvgAS controls flagellin synthesis at the transcriptional level. A flaa-specific probe detected a 1.4-kb transcript in total RNA from Bvg- phase cells that was absent in RNA from Bvg' cells. Primer extension experiments demonstrated a single initiation site 102 bp upstream of the flaa coding sequence. Interestingly, this start site is 7 bp downstream from a sequence resembling the consensus for an alternative sigma factor, o-. Modulation by environmental signals or deletion of the bvgas operon led to transcription from the flaa promoter, and a mutant with a gene encoding the constitutively active BvgS-C3 protein did not express the flaa transcript despite the presence of modulating signals. The flaa promoter exhibited coordinate regulation with bvg promoters analyzed in the same RNA samples. The flaa promoter was coinduced with bvgp2 and inversely regulated with bvgpl. We conclude that flaa is controlled at the level of transcription by a mechanism that involves negative regulation by Bvg. The exact position in the flagellar regulon at which Bvg exerts direct control remains to be determined. Beattie et al. (8, 9) have presented evidence that vig regulation in B. pertussis may involve a mechanism acting after transcriptional initiation. In contrast, transcriptional control represents the primary mechanism regulating flagellin expression in B. bronchiseptica and in previously observed pathways of flagellar morphogenesis in other bacteria. Our analysis of flaa expression in B. bronchiseptica and E. coli suggests a model for Bvg-mediated negative control of motility (Fig. 6). Flagellar synthesis in E. coli proceeds as a cascade of transcriptional control events leading to the ordered expression of numerous structural genes (40). Glucose starvation leads to increased production of camp, which functions as a coactivator with camp receptor protein to promote expression of the flhdc locus (34, 62). flhd and flhc are early genes of the flagellar hierarchy that encode products required for transcription of middle and late genes (5, 7, 35). flia is a middle gene that encodes the alternative sigma factor of, which is required for transcription of late gene promoters, including the promoter for flagellin (flic) (6, 52). We have detected a locus in B. bronchiseptica, designated fri, which is genomically linked to flaa. fri efficiently complemented mutations in the flhdc loci of E. coli and was required in trans to flaa for complementation of motility in an E. coli flic mutant. Expression of a chromosomal flaalacz transcriptional fusion in E. coli required bothfrl and the E. coli flia gene in trans. These results, in addition to the observation that the flaa transcriptional start site maps to a consensus recognition sequence for crf, suggest that transcription from the flaa promoter requires the flia gene product and one or more products of the fri locus. Although the fri locus of B. bronchiseptica efficiently

VOL. 175, 1993 E. coli Glucose camp/crp flhdc ]- middle genes flia (af) late genes flic ]- B. bronchiseptica S2, Nic, T<25 C 1 bvgas fri flaa FIG. 6. Model for the regulation of flaa. Arrows represent positive control, and horizontal lines indicate negative control. The circle surrounding fla indicates that its existence in B. bronchiseptica is predicted. The JiW locus and the flia gene are required for expression of flaa in E. coli. The dotted line indicates a potential pathway of firl-mediated regulation. Correspondence between B. bronchiseptica loci and E. coli loci on the basis of complementation is indicated by lines with brackets. See Discussion for details. complemented aflhd orflhc defect in E. coli, the functional flhdc locus in wild-type E. coli appeared to be insufficient for flaa transcriptional activation. In the presence of fil, however, wild-type E. coli transcribed the flaa promoter. Consistent with this finding, functional complementation of an E. coli flagellin mutant byflaa required thefl locus. It is possible that the E. coli flia gene product inefficiently interacts with the flaa promoter and that the frl locus catalyzes this interaction by increasing the expression levels offlia. Alternatively, gene products offrl may act in concert with FliA to directly activate transcription from the flaa promoter. The mechanism offri-mediated activation offlaa and the number offil-encoded proteins which are responsible for the complementation activities that we detect remain to be determined. The construction of nonpolar mutations in firl will provide a means to address these questions. We propose a model in which Bvg coordinately regulates flagellar genes through negative control of positive regulators such asfrl and a Bordetella analog offlia. Since genetic and biochemical data provide evidence that negative regulation of the bvgp2 promoter involves a direct interaction between bvgp2 and BvgA (55), it is possible that BvgA also binds and directly represses one or more promoters in the flagellar hierarchy. Inspection of the flaa regulatory region did not identify sequences resembling the BvgA-binding sites located upstream from the fhab and bvga genes, and studies in E. coli failed to detect direct repression offlaa by BvgAS (42). The scheme shown in Fig. 6 makes several testable predictions. A flia analog should be present in B. bronchiseptica, and its inactivation, as well as inactivation offir, should eliminate flaa expression. In addition, fri,flia, or both should be subject to negative regulation at the transcriptional or posttranscriptional level by BvgAS. It is likely that a large number of motility-associated structural genes are coregulated with flaa. It is also possible that regulatory factors initially identified as controlling flagellin expression could regulate vrg loci that are required for phenotypes other than motility. The interaction of the virulence control system of B. BvgAS REGULATES FLAGELLIN TRANSCRIPTION 3477 bronchiseptica with the flagellar hierarchy suggests a possible relationship between virulence and motility. Numerous studies have addressed the role of motility in bacterial pathogenesis. A requirement for motility has been implicated for infection and/or disease production by Campylobacterjejuni (71), Vibrio cholerae (54), and Salmonella typhi (38), each of which inhabits mucosal surfaces during infection. An enigmatic observation is that B. bronchiseptica is motile whereas B. pertussis is not. These are highly related organisms that colonize analogous sites in their mammalian hosts, share a number of BvgAS-regulated factors, and contain nearly identical bvgas loci. Since motility is a complex and energetically costly phenotype, it is unlikely to be maintained if it does not confer a significant selective advantage. An interesting finding of this study is that B. pertussis chromosomal DNA hybridized to a flaa-specific probe under conditions of high stringency. The B. pertussis strain examined does not express detectable flaa mrna, suggesting that it contains an inactive flaa gene. Similarly, B. bronchiseptica contains an inactive pertussis toxin allele (4). It is therefore tempting to speculate that the absence of motility in B. pertussis may reflect a difference in the pathogenesis or life cycles of the two species. Alternatively, B. pertussis may express adaptations that replace motility in response to the same conditions that necessitate motility in B. bronchiseptica. We are currently examining the functional role of motility during respiratory infection by B. bronchiseptica and assessing the extent to which the flagellar regulatory elements identified in B. bronchiseptica are involved in negative regulation of BvgAS-controlled genes in B. pertussis. ACKNOWLEDGMENTS We thank Virginia Miller, Olaf Schneewind, and the members of the J. F. Miller laboratory for critically reviewing the manuscript, Sandy Wong for technical advice, Andrew Uhl for useful discussions, and Michael Starnbach for bacterial strains. This work was supported by Public Health Service grant AI-31548 (to J.F.M.) from the National Institute of Allergy and Infectious Diseases. B.J.A. was supported by Microbial Pathogenesis training grant AI-07323 from the National Institutes of Health. J.F.M. is a Pew Scholar in the Biomedical Sciences. REFERENCES 1. Akerley, B. J., D. M. Monack, S. Falkow, and J. F. Miller. 1992. The bvg4s locus negatively controls motility and the synthesis of flagella in Bordetella bronchiseptica. J. Bacteriol. 174:980-990. 2. Arico, B., R. Gross, J. Smida, and R. Rappuoli. 1987. Evolutionary relationships in the genus Bordetella. Mol. Microbiol. 1:301-308. 3. Arico, B., J. F. Miller, C. Roy, S. Stibitz, D. Monack, S. Falkow, R. Gross, and R. Rappuoli. 1989. Sequences required for expression of Bordetella pertussis virulence factors share homology with prokaryotic signal transduction proteins. Proc. Natl. Acad. Sci. USA 86:6671-6675. 4. Arico, B., and R. Rappuoli. 1987. Bordetella parapertussis and Bordetella bronchiseptica contain transcriptionally silent pertussis toxin genes. J. Bacteriol. 169:2847-2853. 5. Arnosti, D. 1990. Regulation of Escherichia coli o' RNA polymerase byflhd and flhc flagellar regulatory genes. J. Bacteriol. 172:4106-4108. 6. Arnosti, D. N., and M. J. Chamberlin. 1989. Secondary sigma factor controls transcription of flagellar and chemotaxis genes in Escherichia coli. Proc. Natl. Acad. Sci. USA 86:830-834. 7. Bartlett, D. H., B. B. Frantz, and P. Matsumura. 1988. Flagellar transcriptional activators FlbB and FlaI: gene sequences and 5' consensus sequences of operons under FlbB and FlaI control. J. Bacteriol. 170:1575-1581.

3478 AKERLEY AND MILLER J. BACTERIOL. 8. Beattie, D. T., S. Knapp, and J. J. Mekalanos. 1990. Evidence that modulation requires sequences downstream of the promoters of two vir-repressed genes of Bordetella pertussis. J. Bacteriol. 172:6997-7004. 9. Beattie, D. T., M. J. Mahan, and J. J. Mekalanos. 1993. Repressor binding to a regulatory site in the DNA coding sequence is sufficient to confer transcriptional regulation of the vir-repressed genes (vrg genes) in Bordetella pertussis. J. Bacteriol. 175:519-527. 10. Beattie, D. T., R. Shahin, and J. J. Mekalanos. 1992. A vir repressed gene of Bordetella pertussis is required for virulence. Infect. Immun. 60:571-577. 11. Bemis, D. A., H. A. Griesen, and M. J. G. Appel. 1977. Bacteriological variation among Bordetella bronchiseptica isolates from dogs and other species. J. Clin. Microbiol. 5:471-480. 12. Brendel, V., and E. N. Trifonov. 1984. A computer algorithm for testing potential prokaryotic terminators. Nucleic Acids Res. 12:4411-4427. 13. Charles, I. G., G. Dougan, D. Pickard, S. Chatfield, M. Smith, P. Novotny, P. Morrissey, and N. F. Fairweather. 1989. Molecular cloning and characterization of protective outer membrane protein P.69 from Bordetella pertussis. Proc. Natl. Acad. Sci. USA 86:3554-3558. 14. Chen, Y., and J. D. Helman. 1992. Restoration of motility to an Escherichia coli flia flagellar mutant by a Bacillus subtilis a factor. Proc. Natl. Acad. Sci. USA 89:5123-5127. 15. Devereux, J., P. Haeberli, and 0. Smithies. 1984. A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Res. 12:387-395. 16. Ditta, G., T. Schmidhauser, E. Yakobsen, P. Lu, X. Liang, D. R. Finlay, D. Guiney, and D. R. Helinski. 1985. Plasmids related to the broad host range vector, prk290, useful for gene cloning and for monitoring gene expression. Plasmid 13:149-153. 17. Dower, W., J. Miller, and C. Ragsdale. 1988. High efficiency transformation of E. coli by high efficiency electroporation. Nucleic Acids Res. 16:6127-6145. 18. Feng, P., R. J. Sugasawara, and A. Schantz. 1990. Identification of a common enterobacterial flagellin epitope with a monoclonal antibody. J. Gen. Microbiol. 136:337-342. 19. Gillen, K. L., and K. T. Hughes. 1991. Negative regulatory loci coupling flagellin synthesis to flagellar assembly in Salmonella typhimurium. J. Bacteriol. 173:2301-2310. 20. Glaser, P., D. Ladant, 0. Sezer, F. Pichot, A. Ullmann, and A. Danchin. 1988. The calmodulin-sensitive adenylate cyclase of Bordetella pertussis: cloning and expression in E. coli. Mol. Microbiol. 2:19-30. 21. Gober, J. W., R. Champer, S. Reuter, and L. Shapiro. 1991. Expression of positional differentiation during cell differentiation in Caulobacter. Cell 64:381-391. 22. Goldman, S. E. Hanski, and F. Fish. 1984. Spontaneous phase variation in Bordetella pertussis is a multistep non-random process. EMBO J. 3:1353-1356. 23. Goodnow, R. A. 1980. Biology of Bordetella bronchiseptica. Microbiol. Rev. 44:722-738. 24. Goyard, S., and A. Ullmann. 1991. Analysis of Bordetella pertussis cya operon regulation by use of cya-lac fusions. FEMS Microbiol. Lett. 77:251-256. 25. Gross, R., B. Arico, and R. Rappuoli. 1989. Genetics of pertussis toxin. Mol. Microbiol. 3:119-124. 26. Henikoff, S. 1984. Unidirectional digestion with exonuclease III creates targeted breakpoints for nucleotide sequencing. Gene 28:351-359. 27. Huh, Y. J., and A. A. Weiss. 1991. A 23-kilodalton protein, distinct from BvgA, expressed by virulent Bordetella pertussis binds to the promoter region of vir-regulated toxin genes. Infect. Immun. 59:2389-2395. 28. Joys, T. M. 1988. The flagellar filament protein. Can. J. Microbiol. 34:452-458. 29. Kloos, W. E., N. Mohapatra, W. J. Dobrogosz, J. W. Ezzel, and C. R. Manclark. 1981. Deoxyribonucleotide sequence relationships among Bordetella species. Int. J. Syst. Bacteriol. 31:173-176. 30. Knapp, S., and J. J. Mekalanos. 1988. Two trans-acting regulatory genes (vir and mod) control antigenic modulation in Bordetella pertussis. J. Bacteriol. 170:5059-5066. 31. Kobisch, M., and P. Novotny. 1990. Identification of a 68- kilodalton outer membrane protein as the major protective antigen of Bordetella bronchiseptica by using specific-pathogenfree piglets. Infect. Immun. 58:352-357. 32. Komeda, Y. 1982. Fusions of flagellar operons to lactose genes on a Mu lac bacteriophage. J. Bacteriol. 150:16-26. 33. Komeda, Y., and T. Iino. 1979. Regulation and expression of the flagellin gene (hag) in Escherichia coli K-12: analysis of hag-lac gene fusions. J. Bacteriol. 139:721-723. 34. Komeda, Y., H. Suzuki, J. Ishidsu, and T. Iino. 1975. The role of camp in flagellation of Salmonella typhimurium. J. Bacteriol. 142:289-298. 35. Kutsukake, K., Y. Ohya, and T. Iino. 1990. Transcriptional analysis of the flagellar regulon of Salmonella typhimurium. J. Bacteriol. 172:741-747. 36. Lacey, B. W. 1960. Antigenic modulation of Bordetella pertussis. J. Hyg. 58:57-93. 37. Leininger, E., M. Roberts, J. G. Kenimer, I. G. Charles, N. Fairweather, P. Novotny, and M. J. Brennan. 1991. Pertactin, an Arg-Gly-Asp-containing Bordetella pertussis surface protein that promotes adherence of mammalian cells. Biochemistry 88:345-349. 38. Liu, S., T. Ezaki, H. Miura, K. Matsui, and E. Yabuuchi. 1988. Intact motility as a Salmonella typhi invasion-related factor. Infect. Immun. 56:1967-1973. 39. Livey, I., and A. C. Wardlaw. 1984. Production and properties of Bordetella pertussis heat-labile toxin. J. Med. Microbiol. 17:91-103. 40. Macnab, R. M. 1987. Flagella, p. 70-83. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, D.C. 41. Melton, A. R., and A. A. Weiss. 1989. Environmental regulation of expression of virulence determinants in Bordetella pertussis. J. Bacteriol. 171:6206-6212. 42. Miller, J. F., and B. J. Akerley. 1992. Unpublished data. 43. Miller, J. F., S. A. Johnson, W. J. Black, D. T. Beattie, J. J. Mekalanos, and S. Falkow. 1992. Isolation and analysis of constitutive sensory transduction mutations in the Bordetella pertussis bvgs gene. J. Bacteriol. 174:970-979. 44. Miller, J. F., J. J. Mekalanos, and S. Falkow. 1989. Coordinate regulation and sensory transduction in the control of bacterial virulence. Science 243:916-922. 45. Miller, J. F., C. R. Roy, and S. Falkow. 1989. Analysis of Bordetella pertussis virulence gene regulation by use of transcriptional fusions in Escherichia coli. J. Bacteriol. 171:6345-6348. 46. Miller, J. F., C. R. Roy, and S. Falkow. 1991. Regulation of fhab, bvg, and pix transcription in E. coli: a comparative analysis, p. 217-224. In C. R. Manclark (ed.), Proceedings of the Sixth International Symposium on Pertussis. DHHS publication no. (FDA) 90-1164. U.S. Public Health Service, Bethesda, Md. 47. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 48. Miller, V. M., and J. J. Mekalanos. 1988. A novel suicide vector and its use in construction of insertion mutations: osmoregulation of outer membrane proteins and virulence determinants of Vibrio cholerae requires toxr. J. Bacteriol. 170:2575-2583. 49. Monack, D. M., 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. 50. Musser, J. M., E. L. Hewlett, M. S. Peppler, and R. K. Selander. 1986. Genetic diversity and relationships in populations of Bordetella spp. J. Bacteriol. 166:230-237. 51. Newton, A., and N. Ohta. 1990. Regulation of the cell cycle division cycle and differentiation in bacteria. Annu. Rev. Microbiol. 44:689-719. 52. Ohnishi, K., K. Kutsukake, H. Suzuki, and T. Iino. 1990. Gene flia encodes an alternative sigma factor specific for flagellar

VOL. 175, 1993 operons in Salmonella typhimurium. Mol. Gen. Genet. 221:139-147. 53. 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. Nati. Acad. Sci. USA 86:2637-2641. 54. Richardson, K. 1991. Roles of motility and flagellar structure in pathogenicity of Vibrio cholerae: analysis of motility mutants in three animal models. Infect. Immun. 59:2727-2736. 55. Roy, C. R., and S. Falkow. 1991. Identification of Bordetella pertussis regulatory sequences required for transcriptional activation of thefhab gene and autoregulation of the bvgas operon. J. Bacteriol. 173:2385-2392. 56. Roy, C. R., J. F. Miller, and S. Falkow. 1989. The bvga gene of Bordetella pertussis encodes a transcriptional activator required for the coordinate regulation of several virulence genes. J. Bacteriol. 171:6338-6344. 57. Roy, C. R., J. F. Miller, and S. Falkow. 1990. Autogenous regulation of the Bordetella pertussis bvgabc operon. Proc. Natl. Acad. Sci. USA 87:3763-3767. 58. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 59. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463-5467. 60. Scarlato, V., A. Prugnola, B. Arico, and R. Rappuoli. 1990. Positive transcriptional feedback at the bvg locus controls expression of virulence factors in Bordetella pertussis. Proc. Natl. Acad. Sci. USA 87:6753-6757. 61. Scarlato, V., A. Prugnola, B. Arico, and R. Rappuoli. 1991. The bvg-dependant promoters show similar behaviour in different Bordetella species and share sequence homologies. Mol. Microbiol. 5:2493-2498. 62. Silverman, M., and M. Simon. 1977. The characterization of Escherichia coli mutants that are insensitive to catabolite repression. J. Bacteriol. 120:1196-1203. 63. Simons, R. W., F. Houman, and N. Kleckner. 1987. Improved single and multicopy lac-based cloning vectors for protein and operon fusions. Gene 53:85-96. 64. Stainer, D., and M. Scholte. 1971. A simple chemically defined BvgAS REGULATES FLAGELLIN TRANSCRIPTION 3479 medium for the production of phase I Bordetella pertussis. J. Gen. Microbiol. 63:211-220. 65. Stibitz, S., W. Aaronson, D. Mohack, and S. Falkow. 1989. Phase variation in Bordetella pertussis by a frameshift in a gene for a novel two component system. Nature (London) 338:266-269. 66. Stibitz, S., and T. L. Garletts. 1992. Derivation of a physical map of the chromosome of Bordetella pertussis Tohama I. J. Bacteriol. 174:7770-7777. 67. Stibitz, S., and M. S. Yang. 1991. Subcellular localization and immunological detection of proteins encoded by the vir locus of Bordetella pertussis. J. Bacteriol. 173:4288-4296. 68. Stock, J., A. J. Ninfa, and A. M. Stock. 1989. Protein phosphorylation and regulation of adaptive responses in bacteria. Microbiol. Rev. 53:450-490. 69. Stocker, B. A. D. 1949. Measurements of rate of mutation of flagellar antigenic phase in Salmonella typhimurium. J. Hyg. 47:398-413. 70. von Gabian, A., J. G. Belasco, J. L. Schottel, A. C. Chang, and S. N. Cohen. 1983. Decay of mrna in Escherichia coli: investigation of the fate of specific segments of transcripts. Proc. Natl. Acad. Sci. USA 80:653-657. 71. Wassenaar, T. M., N. M. C. Bleumink-Pluym, and B. A. M. van der Zeist. 1991. Inactivation of Campylobacterjejuni flagellin genes by homologous recombination demonstrates thatflaa but not flab is required for invasion. EMBO J. 10:2055-2061. 72. Weiss, A., E. L. Hewlett, G. A. Myers, and S. Falkow. 1983. TnS-induced mutations affecting virulence factors of Bordetella pertussis. Infect. Immun. 42:33-41. 73. Weiss, A. A., and E. L. Hewlett. 1986. Virulence factors of Bordetella pertussis. Annu. Rev. Microbiol. 40:661-686. 74. Willems, R., A. Paul, H. G. van der Heide, A. R. ter Avest, and F. R. Mooi. 1990. Fimbrial phase variation in Bordetella pertussis: a novel mechanism for transcriptional regulation. EMBO J. 9:2803-2809. 75. Wood, W. I., J. Gitschier, L. A. Lasky, and R. M. Lawn. 1985. Base composition-independent hybridization in tetramethylammonium chloride: a method for oligonucleotide screening of highly complex gene libraries. Proc. Natl. Acad. Sci. USA 82:1585-1588.