Identification of genes encoding components of the swarmer cell flagellar motor and propeller and a sigma factor controlling differentiation of Vibrio parahaemolyticus

Vibrio parahaemolyticus possesses two distinct motility systems, the polar system used for swimming in liquid environments and the lateral system used for swarming over surfaces. Growth on surfaces induces swarmer cell differentiation and expression of the lateral motility system. Mutants, created by transposon mutagenesis of a clone expressing lateral flagellin and gene disruption in V. parahaemolyticus, were unable to swarm and failed to make lateral flagellin; therefore, unlike the case for the polar system, there is one gene (lafA) encoding lateral flagellin. In addition to lafA, other genes required for swarming but not for swimming were identified by gene replacement mutagenesis. The nucleotide sequence of the clone determined open reading frames (ORFs) and deduced amino acid sequences showed similarities to flagellar components of other bacteria: flagellin, hook-associated protein (HAP2), motor components, and flagellar sigma factor (sigma 28). Many sigma 28 factors have been shown to recognize cognate promoters; however, expression of lafA in Escherichia coli required LafS, and E. coli sigma 28 did not substitute. Also, there were no sequences preceding genes encoding flagellin or HAP2 resembling the sigma 28 consensus promoter. The product of the sigma-like gene seems to be a unique member of the sigma 28 cluster. It appears the result of requiring expression for immunodetection of flagellin clones was that the sigma locus was fortuitously cloned, since the sigma and lafA loci were not contiguous in the chromosome. This work initiates identification and placement of genes in a scheme of control for swarmer cell differentiation; three levels have been identified in the transcriptional hierarchy.

This work initiates identification and placement of genes in a scheme of control for swarmer cell differentiation; three levels have been identified in the transcriptional hierarchy.
The marine bacterium and human pathogen Vibrio parahaemolyticus possesses two distinct motility systems, each appropriate for movement under different circumstances (2,48). Translocation over surfaces, or swarming, results in radially spreading colonies and colonization of surfaces, while movement in liquid is called swimming (20,52). The polar flagellar system (Fla) propels the bacterium in liquid medium. Growth on surfaces induces the developmental program called swarmer cell differentiation, in which septation ceases, the cells elongate, and a new flagellar system, the lateral system (Laf), is synthesized. The polar system is expressed essentially constitutively, while the lateral flagellar system is expressed only when the bacterium is in contact with surfaces or in a viscous environment, i.e., under conditions in which the polar flagellum is not functional (38). Thus, the two gene systems interact: performance of the polar system is in some way coupled to transcription of the swarmer cell gene system. To elucidate the signal transduction event by which physical information is processed and converted to a signal that leads to swarmer cell differentiation, we are defining some of the key genes and the regulatory hierarchies of the two flagellar systems.
The gene systems that encode the two motility systems are large, each composed of 40 or more genes without overlap except for shared chemotaxis components (41,47). Flagellum-driven motility requires genes encoding components of a propeller and a motor (34). The flagellum is composed of a basal body which appears as a set of rings mounted on a rod that extends from the cytoplasm through the cell wall, a hook which couples the filament to the basal body, and the * Corresponding author. Electronic mail (Internet) address: mccar ter@psl.wisc.edu.
propeller, or filament, which is a hollow, semirigid helical tube that polymerizes at the tip. Proteins called HAPs (hook-associated proteins) are required for joining the filament to the hook and capping the distal tip of the filament. The mystery of flagellar assembly, in particular the nature of the flagellum-specific export pathway which is distinct from the signal sequence-dependent pathway utilized for most exported bacterial proteins, has not been solved (22,23,31). The motor is powered by the electrochemical membrane potential. It couples the movement of ions through a channel (which is formed by MotA in Escherichia coli) to rotation of the propeller. Another protein (called MotB in E. coli) is believed to function as a stator. Together, these two proteins are believed to act as force-generating units which transmit torque to the filament through the basal body. Additional proteins include components of the switch complex, which is important for modulation of direction of flagellar rotation and also required for both organelle assembly and energy conversion.
In this report, we begin to identify genes belonging to the lateral flagellar system and to order these genes in a hierarchical scheme of gene control. The genes encode structural components of the flagellum and the motor, as well as a sigma factor of RNA polymerase important for directing swarmer cell development. Sequence information on structural components of both motility systems should provide insight with respect to the signals for flagellar export and assembly, laf and fla promoter recognition and regulatory sites, and perhaps evolutionary clues for these two distinct systems. Defining the cascade of regulation should help us begin to understand how the lateral and polar gene systems interact. parahaemolyticus strain BB22 and strain LM1017 (lafX313::lux) have been described previously (5,38). XTnphoA and E. coli CC118 [araDl39 A(ara leu)7697 AlacX74 AphoA20galEgalK thi rpsE rpoB argE(Am) recAl] were obtained from Colin Manoil (17). E. coli MC1000 [araD139 A(ara, leu)7697 AlacX74 galU galK strA] and MC1000 (flhD::TnlO Kanr) were from Phil Matsumura (9), and E. coli YK410 (araD139 AlacU169 strA thipyrC46 nalA thyA his) and YK4181 (YK410fliA) were from May Macnab (28). Cosmid pLM1457 has been described elsewhere (39), and cosmid pLM1489 contains the V. parahaemolyticus polar flagellin genesflaC andflaD in the vector pRK415 (25).
Growth conditions and media. Vibrio strains were propagated at 30°C in heart infusion medium (25 g of Bacto Heart Infusion (Difco) plus 20 g of NaCl per liter). Swarming was examined on solidified heart infusion media with 1.5% Bacto Agar (Difco). The minimal medium of Broach et al. (8) supplemented with 0.4% galactose, 20 mM NH4Cl, and 2% NaCl was used for conjugation and semisolid motility plates with 1.5 and 0.36% Bacto Agar, respectively. LB, 2xYT, and NZCYM media for propagation of E. coli strains and lambda phage were prepared as described by Maniatis et al. (35). Antibiotics (from Sigma Chemical Co.) were used at the following concentrations: 125 ,ug/ml (gentamicin), 100 ,ug/ml (kanamycin), and 10 ,ug/ml (tetracycline). 5-Bromo-4-chloro-3-indolyl phosphate (XP; Sigma) was used at 40 ,ug/ml. TnphoA and gene replacement mutagenesis. Mutagenesis with ATnphoA was performed by infecting a culture of E. coli CC118 carrying cosmid pLM1457 (tetracycline resistant). Multiple 2-ml cultures of exponentially growing cells (optical density at 600 nm of 0.4) in NZCYM medium supplemented with tetracycline were incubated with 250 Rl of XTnphoA (at titers of >1011 PFU/ml) for 30 min to allow adsorption. Then, 2 ml of 2xYT plus tetracycline was added to each tube, and cultures were grown for 2 h at 37°C with shaking, at which time 2 ml of 2xYT plus tetracycline, kanamycin, and 0.4% glucose was added to each tube, and cells were grown overnight. Cultures were reinoculated into selective media (optical density at 600 nm of 0.1) and grown to saturation. Cosmid DNA was prepared and used to transform strain CC118, selecting for tetracycline and kanamycin resistance. The procedure for gene replacement in V. parahaemolyticus has been outlined elsewhere (49). For mobilization of the cosmids carrying the transposon mutations into V. parahaemolyticus, the tra donor was pRK2013 (from Gary Ditta [15]); for selecting gene replacement, the "kick-out" plasmid was pPHIJI (from Ron Taylor [6]). Counterselection against the E. coli parents was done on minimal medium with galactose as the carbon source and appropriate antibiotics for plasmid and cosmid maintenance. Strain constructions were confirmed by Southern blot analysis of restricted genomic DNA (37,40). Transposon insertions in the lafA locus did not perturb the lafS locus and vice versa.
Immunological techniques. Western immunoblot analysis with anti-LafA antiserum 127 and anti-Fla antiserum 129 was done as previously described (39)

RESULTS
Cloning and transposon mutagenesis. Cloning of the locus encoding lateral flagellin has been described elsewhere (39). Briefly, a lambda bank was screened for production of lateral flagellin antigen. This bank was constructed by ligating size-selected Sau3A partial digests of V. parahaemolyticus chromosomal DNA into the XEMBL3 vector cleaved with BamHI. One of the antigen-producing clones was subcloned into the broad-host-range vector pLAFRII for analysis. This cosmid, pLM1457, was mutagenized with TnphoA in order to localize the structural gene, lafA, for sequencing and to create lafA mutations to be used for gene replacement in V. parahaemolyticus. From this mutagenesis, cosmids that failed to produce LafA antigen of the appropriate molecular size (29 kDa) were obtained. Strains carrying some of these mutated cosmids (e.g., with transposon 1494 [Fig. 1, lane 2]) synthesized fusion proteins of a higher molecular weight that cross-reacted in Western analysis with both anti-LafA and anti-alkaline phosphatase antisera. Intriguingly, we obtained other cosmids with transposons (e.g., transposon 1694 in [Fig. 1, lane 7]) that also failed to produce LafA but which on restriction analysis mapped quite distal to the first class. Given the positions and orientations of the transposons (Fig.   2), it did not seem likely that mutants of the second class were simply polar upon lafA expression, and we postulated that there was a gene (or genes) on cosmid pLM1457 necessary for expression of lafA. One other interesting class of transposon insertions, represented by transposon 1692, resulted in overproduction of LafA (Fig. 1, lane 8). Lateral flagellin appeared to be unstable in E. coli (long exposures of Western blots showed degradation in all strains producing LafA); however, it was most apparent for strains carrying cosmids with transposons such as transposon 1692, which J. BACTERIOL. Overnight cultures of E. coli CC118 carrying the following cosmids, which are derivatives of pLM1457 with transposon insertions, were concentrated twofold in sample buffer for SDS-polyacrylamide gel electrophoresis. The reference sample was wild-type V. parahaemolyticus BB22 harvested from a plate. Lanes: 0, BB22; 1, pLM1457 (parent, no insertion); 2, pLM 1494 (lafA::TnphoA [note high-molecular-weight fusion product]); 3, pLM1687 (intergenic lafA and lafB); 4, pLM1688 (lafB::TnphoA); 5, pLM1498 (lafU::TnphoA); 6, pLM1690 (lafT::TnphoA); 7, pLM1694 (lafS::TnphoA); 8, pLM1692 (lafL::TnphoA); 9, pLM1457 (parent). The blot was probed with anti-lateral flagellin antiserum and subsequently with radioiodinated protein A. overproduced LafA. Since there appeared to be more functions pertinent to lateral flagellar gene expression encoded by this clone than simply the flagellin structural gene, the entire region was sequenced. A restriction map, with positions of some of the transposons indicated, and the sequencing strategy are shown in Fig. 2.
The kafA gene. The DNA sequence and the translated products of the proposed reading frames presented in Fig. 3 correspond to bp 1 to 3768 on the restriction map in Fig. 2. The open reading frame (ORF) encoding lateral flagellin, LafA, starts at position 739 with an ATG. It is preceded by sequences with features that resemble a site for translational initiation in E. coli, including a Shine-Dalgarno domain which has appropriate spacing preceding the initiation codon (16). The ORF is also preceded by two regions of inverted, dyad symmetry that show some sequence similarity to each other (indicated by the open circles in Fig. 3). There are no sequences resembling consensus flagellar promoters which have been identified preceding flagellar genes in a variety of organisms (and there are no sequences resembling u70 or a" consensus promoters [18]). The translated protein contains 284 amino acids with a calculated molecular size of 29,727 Da, which correlates well with the observed migration of LafA in sodium dodecyl sulfate (SDS)-polyacrylamide gels (39). LafA appears similar to other flagellins (24) in that it is quite rich in alanine (44 residues) and has only one proline, two histidines, and no cysteine or tryptophan residues. Most of the similarity occurs at the amino and carboxyl termini, with internal variability. Some of the flagellins with outstanding similarity scores include PSEFLAA from Pseudomonas aeruginosa, STYH1A from Salmonella paratyphi-A, STYFLGH1I from Salmonella typhimurium, ECOHAG from E. coli, BACFLAGA from Bacillus subtilis, SMAHAG from Serratia marcescens, TRPFLABP from Treponema phagedenis, BORFLA2 from Borrelia burgdorferi, and CAJ FLAAB from Campylobacter jejuni. For example, GCG BestFit comparison of LafA with PSEFLAA yielded the highest homology scores, with 69.4% similarity (S)/52% identity (I) (quality = 245 and quality ratio = 0.862).
Gene replacement experiments in the wild-type strain BB22 that substituted lafA::TnphoA fusions for the wildtype allele yielded mutant bacteria that were unable to swarm, failed to produce lateral flagellin subunit, and produced fusion proteins (i.e., proteins that cross-reacted with both alkaline phosphatase and LafA antisera). Polar motility (i.e., swimming in semisolid motility medium) was not affected. The apparent molecular weights on SDS-polyacrylamide gels of the fusion proteins were larger than that of either alkaline phosphatase or LafA. The fusions were regulated by growth on a surface; no fusion proteins were produced when mutant bacteria were grown in liquid me-  I I I I I I I I I I I I I I I IlI III I I I I I I I I   I   1000  2000 3000 4000 5000 6200 bp Restriction map and sequencing strategy. Restriction sites shown are the ones used for cloning from cosmid pLM1457 into M13 mp18 or mpl9. Abbreviations: B, BamHI; P, PstI; S, SalI; X, XIaI. The following restriction fragments were used for construction of the sequencing templates: 1, B-X (1.2 kb); 2, P-P (0.6 kb); 3, P-P (2.3 kb); and 4, S-S (2.4 kb). In addition to these templates, deletions of these clones were prepared for sequencing by using the International Biotechnologies Cyclone I Biosystem. The joints (*) between templates 2 and 3 and templates 3 and 4 were crossed by priming the parental full-length clone (pLM1457, double stranded) with synthetic oligonucleotides homologous to sequences at one end of template 2 or 3. Overlapping sequences were obtained for both strands of DNA and are indicated by the arrows. ORFs corresponding to the genes indicated proceed from left to right for lafA and lafB and right to left for lafL, lafS, lafT, and lafU. Only parts of the ORFs corresponding to lafL and lafU were cloned and sequenced. Representative TnphoA insertions are shown on the restriction map; filled circles indicate those insertions that form alkaline phosphatase fusion proteins, and arrows indicate the direction of transcription of phoA. 3363 VOL. 175, 1993   dium, but fusion proteins were synthesized when the mutant strains were grown on solidified medium. These are the first mutants known to have lesions in lafA. Since they fail to produce lateral flagellin and are unable to swarm, we conclude that, in contrast to the multiple genes encoding the polar flagellins, there appears to be only a single gene encoding lateral flagellin. The IqIR gene. The region between lafA and the next ORF extends from nucleotides 1596 to 2369 in Fig. 3. This ORF, designated lafBl, starts with a GTG codon preceded by sequences resembling the Shine-Dalgarno sequence at a distance of 7 bp. The lafA and lafB genes do not appear to be organized in a transcriptional unit. There are sequences resembling an E. coli rho-independent transcriptional terminator following the lafA coding region (positions 1631 to 1661). Moreover, disruption of the intergenic region did not affect swarmer cell gene expression, since a strain with transposon 1697 introduced by gene replacement was able to swarm. The lafB gene is preceded by one region of inverted dyad symmetry that shows sequence similarity (indicated by the open circles in Fig. 3) to the two dyad regions preceding lafA. The deduced protein sequence is 445 amino acids in length and has a calculated molecular weight of 48,234. When the lafB gene was disrupted by gene replacement with lafB::TnphoA insertions, the mutants failed to swarm. A TBLASTN search revealed homology to STYFLID, the S. typhimurium fliD gene product (the distal capping protein called HAP2). Like other flagellar axial proteins which polymerize to form the flagellar basal body and filament, this protein lacks cysteine, has a 1.8% proline content, and possesses heptad repeats of hydrophobic amino acids at the carboxyl terminus (21). The best identity between the Salmonella and Vibrio proteins occurs within the carboxylterminal half, consistent with the putative function and location of the protein in the flagellar structure. As the distal capping protein, one end of the molecule interacts with flagellum and the other is exposed to the environment (21).
The lateral flagellar sigma gene (laSJ). The DNA sequence presented in Fig. 4 corresponds to the reverse complement of the restriction map in Fig. 2, correlating with positions 6218 to 3762. Within this sequence is an ORF, designated lafS, that starts at nucleotide 240 with an ATG codon. It is preceded at a distance of 7 bases by sequences resembling the Shine-Dalgarno consensus. This ORF codes for a potential polypeptide of 242 amino acids with a calculated molecular size of 27,8i7 Da. Introduction of transposons, such as transposon 1694, into the V. parahaemolyticus chromosome affected swarming: lafS::TnphoA mutants were unable to swarm and failed to synthesize lateral flagellin. Polar motility was unaffected. When the deduced protein sequence was used to search the GenBank data base, the following similarities to proteins approximately 28 kDa in size were found: STYFLIA, the S. typhimurium flagellar sigma factor (58% S/33% I by BestFit analysis); PSEFLIA, the P. aeruginosa flagellar sigma factor (56% S/31% I); BACSIGD, the B. subtilis flagellar sigma factor (48% S/28% I); STMSWHIG, the Streptomyces coelicolor developmental sigma factor (50% S/27% I); and STMPROZ, a homolog of STMSWHIG and a minor sigma factor of Streptomyces aureofaciens (52% S/28% I). Three of these proteins are alternative sigma subunits of RNA polymerase that are required for flagellin synthesis (19,45,51). The fourth, STMSWHIG (and probably STMPROZ), is a factor required for initiation of spore development (10). In multiple sequence alignments, the other sigma factors in the c8 cluster appear more closely related to each other than to V. parahaemolyticus.
The sigma subunit of RNA polymerase confers the specificity of promoter selection. Inspection of the homology between the above sigma factors reveals two regions with greater than 70% similarity (Fig. 5). These two regions of similarity map in the consensus regions of alignment, designated 2.4 and 4.2, that have been identified upon comparison of many sigma factors (33). Domain 2.4, which encompasses region I in Fig. 5, is believed to be important in -10 recognition of promoter sequences by RNA polymerase, while domain 4.2, which encompasses region II, has been implicated as being important for -35 hexamer recognition. The V. parahaemolyticus sequence seem significantly different. In region I, the V. parahaemolyticus sequence varies at two positions (marked with asterisks) which were conserved among all of the other sigma factors shown. In region II, V. parahaemolyticus varies at six positions where the others remain identical.
Expression of the lateral flagellar gene lafA in E. coli is independent of the E. coli flagellar sigma factor. Many of the alternative sigma factors in the a" cluster have been shown to recognize similar (and, it has been demonstrated in some cases, interchangeable) promoters (10,18). Therefore, expression of lafA was examined in various E. coli flagellar mutant strains. Western blot analysis suggested that the E. coli flagellar genesfliA, encoding the flagellar sigma, and the master regulatory genes (flhCD) that control expression of fli4, were not required for expression of lafA on clone pLM1457 in E. coli. Production of LafA was compared in the following pairs of wild-type and mutant strains: MC1000/ MC1000 flhD::TnlO (FlhD-FlhC-) and YK410/YK4181 (YK410 fli4). The level of expression of lafA from cosmid pLM1457 was similar in E. coli wild-type and flagellar regulatory mutant strains (Fig. 6, lanes 1 to 4). The data suggest that no E. coli flagellar proteins were required for expression of lafA. In contrast, expression of V. parahaemolyticus polar flagellin genes flaC and flaD, which are encoded by cosmid pLM1489, appeared to require E. coli flagellar genes for expression (lanes 5 to 8). It is unlikely that expression of lafA in E. coli is under the control of a cosmid promoter, since mutations in the lateral flagellar sigma gene (e.g., transposon 1694), which map on cosmid pLM1457 downstream of the direction of transcription of the lafA gene (Fig. 2), resulted in loss of expression of lafA (Fig. 1, lane 7).
Moreover, other transposon insertions resulted in overproduction of LafA in E. coli (an example of which is shown in lane 8 of Fig. 1). These transposons, which all have the same orientation of insertion mapped upstream of the lafS gene (see transposon 1692 in Fig. 2). It seems likely that increased production of lafA was the result of increased transcription of lafS promoted by the transposon insertion. We conclude that expression of the gene encoding lateral flagellin in E. coli is dependent on expression of another Vibrio gene, lafS, which encodes a protein that resembles a flagellar sigma factor of RNA polymerase, and that the E. coli flagellar sigma cannot substitute for function of LafS.
The lateral flagellar motor genes (lafT and laJV). Located 11 bases downstream of the lafS gene is an ORF, designated lafT, that initiates with an ATG preceded at a distance of 7 bases by sequences that resemble the Shine-Dalgamo sequence. The predicted protein encoded for by the gene sequence contains 285 amino acids with an estimated molecular size of 31,304 Da. Its overall hydrophobicity is consistent with it being a membrane protein (51% Ala, Leu, Ile, Val, Phe, Met, and Pro). The protein is homologous to the E. coli motA gene product: BestFit analysis yields 57.5% S/35.8% I (quality = 210 and quality ratio = 0.74). More-VOL. 175, 1993 Sites that resemble E. coil ribosome binding sites (RBS) are indicated by underlined sequences which exhibit homology with the 3' end of E. ccli 16S rRNA. Selected restriction sites are shown to facilitate orientation to the physical map. The sequence presented is in reverse orientation to the restriction map presented in Fig. 2 and corresponds to bp 6218 to 3762. The coding sequences extend from nucleotides 1 to 227 for lafL, from 240 to 968 for lafS, from 979 to 1836 for lafT, and from 1836 to 2438 for lafU. Only parts of the sequences corresponding to lafL and lafU were obtained. over, upon comparison with the E. coli protein, six domains seem conserved (12). There are four regions, each composed of 20 to 25 mainly hydrophobic amino acids (with 1 or 2 charged amino acids per domain), that could form membrane-spanning segments according to calculations of Eisenberg et al. (14) and Kyte and Doolittle (32). Internal to the molecule are two hydrophilic, highly charged regions. The first charged region (including amino acids 52 to 73) contains four acidic, four basic, and two glutamine residues. The second charged region (amino acids 142 to 157) contains six acidic, one basic, and two glutamine residues.
The termination codon, TAA, of lafT overlaps the initiation codon, ATG, of another ORF, designated lafU. This ATG is preceded at a distance of 4 bases by sequences resembling the Shine-Dalgamno sequence. The deduced pro-tein sequence revealed homology to the E. ccli motB gene product (BestFit analysis: 53.7% S/31.1% I; quality = 111 and quality ratio = 0.59). Three domains show conservation to domains noted for MotB (11,50), while the central region of the molecule appears more variable. At the amino terminus, there is a hydrophilic region (7 basic, 3 acidic, and 2 glutamine residues in the first 23 amino acids) followed by a hydrophobic domain (including amino acids 26 to 49). This domain has characteristics of a membrane-spanning domain in terms of hydrophobicity, length, and potential a-helical structure. There is a second hydrophobic region located near the carboxyl terminus (corresponding to region C in reference 11). This region is 22 amino acids in length, long enough to span a membrane; however, it has 4 charged residues. As was observed for motB::TnphoA fusion strains and consis-  tent with LafU being a membrane protein, some of the E. coli strains with cosmids carrying TnphoA insertions that mapped in lafU were blue on plates with XP, a chromogenic substrate of alkaline phosphatase (11,36). Gene replacement with lafU::TnphoA and lafT::TnphoA yielded mutant bacteria that were able to swim but unable to swarm and failed to synthesize lateral flagellin.
Cosmid 1457 is a chimera, since lafA and lafS loci are not contiguous on the chromosome. In contrast to the E. coli motB gene product, which is 309 amino acids in length, the Western blot analysis, overnight cultures of strains were concentrated twofold in sample buffer for SDS-polyacrylamide gel electrophoresis. Control lane 0 contains wild-type V. parahaemolyticus BB22 harvested from a plate. Both lateral (Laf) and polar (Fla) flagellins are produced. Three of the four polar flagellins (FlaA, FlaC, and FlaD) comigrate, and the fourth (FlaB) appears slightly smaller. Lanes: 0, strain BB22; 1, MC4100/pLM1457 (lafA'); 2, MC4100 flhD::TnJO/pLM1457; 3, YK41O/pLM1457; 4, YK4181 (YK41O fliA4)/pLM1457; 5, MC4100/pLM1489 (flaC' flaD'); 6, MC4100 flhD::TnlO/pLM1489; 7, YK41O/pLM 1489; 8, YK4181 (YK41OfliA4)/pLM1489. Cross-reacting E. ccli proteins in the background serve as controls for total amounts of cell protein loaded in each lane. potential protein encoded for by this ORF consists of 202 amino acids. There is no recognizable terminator in the nucleotide sequence; however, there is a Sau3A site at the very end of the sequence (GATC at position 2401 in Fig. 4). Southern blot analysis comparing restricted wild-type chromosomal and cosmid pLM1457 DNAs suggested that pLM1457 is a chimeric cosmid. For example, as shown in Fig. 7, when pLM1457 was used as a probe, the 0.6-kb PstI fragment 1 was present in the cosmid and the chromosome (lanes 4 and 5), and this fragment was perturbed (lane 6) in a mutant strain carrying transposon 1494, which maps in the 0.6-kb PstI fragment. The 1.0-kb BglII fragment was also present in the cosmid and chromosome (lanes 2 and 3), and this fragment was unperturbed in strains with insertions mapping in the lafA or lafS locus (lanes 9 through 14).  13 and 14). Neither the 2.3-kb PstI fragment 2 (lanes 4 and 5) nor the 2.4-kb Sall fragment (lanes 7 and 8) of the cosmid was found in the chromosome. Both of these fragments span the putative chimeric joint. Thus, either two unrelated pieces of DNA were fortuitously joined together or an original, larger clone was deleted to produce pLM1457 and the other positive clones identified. This was an unexpected result given that XEMBL3 accepts 22-kb inserts and that the lambda bank was made with 18to 22-kb fragments of DNA purified by agarose gel electrophoresis; however, it is a likely result because the antibody screen was based on selection for expression of lafA. It seems that lafA expression requires the lafS locus, and thus, in effect, the screen was for the two regions to be contained on the same vector.
The iaJL gene. The lafS gene is preceded by a small ORF, with the intercistronic region being 12 bases. This ORF, designated lafL, appears to encode 74 residues at the carboxyl terminus of a protein that shows homology to the FliL proteins of other bacteria; for example, BestFit analysis reveals approximately 52% S/30% I with the proteins of E. coli and S. typhimurium, 49% S/19% I with Caulobacter crescentus protein, and 45% S/12% I with B. subtilis protein (1,26,29,53). These fliL genes encode membrane proteins approximately 150 amino acids in length. Although the function offliL is unknown, in E. coli and S. typhimurium, it is found in an operon that encodes other proteins that are members of the flagellar switch complex which is involved in energizing and switching the direction of rotation of the flagellum. Introduction of transposon mutation 1692 into the V. parahaemolyticus chromosome affected swarming: the lafL::TnphoA mutant was unable to swarm and failed to synthesize lateral flagellin. Polar motility was unaffected.
Hierarchy of control of lateral flagellar gene expression. Genes in the swarmer cell system are expressed in response to growth on a surface. Expression of the lafA gene, which encodes the flagellar filament protein, requires expression of the laJLSTUV locus encoding the sigma subunit and motor parts. Introduction of mutations in any of the identified genes in this second locus resulted in loss of swarming and loss of production of lateral flagellin in V. parahaemolyticus. Thus, lafA is placed in a scheme of regulation under the control of the sigma locus (Fig. 8). One other lateral flagellar locus has been studied (5,38); it encodes a gene, that we now designate lafX, which shows surface-dependent gene expression. Strain LM1017 contains an operon fusion between the lafX gene and indicator luminescence genes (lux). It is VOL. 175, 1993 Southern blot analysis of restricted cosmid and chromosomal DNA prepared from V. parahaemolyticus wild-type and transposon mutant strains probed with labeled pLM1457, showing that cosmid pLM1457 is a chimera (the lafA and lafS loci are not contiguous on the chromosome). Arrows at the left indicate approximate sizes in kilobase pairs; arrows at the right indicate fragments in restricted wild-type DNA that are perturbed in DNA prepared from transposon mutant strains. Lanes: 1, lambda DNA markers (the 3.5-kb band hybridizes to the probe); 2, cosmid pLM1457 restricted with BglII; 3, wild-type strain BB22 restricted with BglII; 4, cosmid pLM1457 restricted with PstI; 5, BB22 restricted with PstI; 6, mutant strain LM4066 (with transposon 1494) restricted with PstI; 7, cosmid pLM1457 restricted with Sall; 8, BB22 restricted with SalI; 9, cosmid pLM1457 restricted with BglII; 10, BB22 restricted with BgIII; 11, mutant strain LM4066 (with transposon 1494) restricted with BglII; 12, mutant strain LM4247 (with transposon 1687) restricted with BglII; 13, mutant strain LM4251 (with transposon 1690) restricted with BglII; 14, mutant strain LM4242 (with transposon 1694) restricted with BglII. The diagram shows the chimeric joint and selected restriction fragments from pLM1457. The asterisks indicate fragments that do not correspond with the chromosome. Locations of the transposons used to created specific mutant strains are shown. The chimeric joint is indicated by the wavy line.
Abbreviations: P, PstI; G, BglII; S, SalI; A, Sau3A. unable to swarm, fails to make lateral flagellin, and produces bioluminescence only when grown on surfaces and not in liquid media (38). Introduction of sigma mutations, such as mutation 1694, into strain LM1017 did not affect expression or surface-dependent regulation of the laf::lux fusion. It is placed at the highest position as yet identified because it retains its normal, surface-dependent regulation in the presence of mutations placed lower in the scheme (i.e., lafS). The lafS gene encodes a sigma factor, and the lafT and lafU genes encode motor parts. They appear to be part of a large operon, designated by the arrow. appears that regulation of the lafX gene occurs at a point higher in the cascade of gene control; therefore, it is placed above the sigma locus in the hierarchy.

DISCUSSION
V. parahaemolyticus has two alternative cell types, the swimmer cell and the swarmer cell. The swimmer cell is adapted for survival in liquid environments; the rod-shaped bacterium is propelled by a single sheathed, polar flagellum. When a swimming cell contacts a surface, a series of changes occur that transforms the swimmer to a swarmer cell. The swarmer cell is adapted for colonization of surfaces; these highly elongated, superflagellated cells can rapidly move over and colonize surfaces. The two propulsive systems are distinct, having no motor or propeller components in common. Somehow the bacterium is able to assemble two distinct appendages simultaneously, and we are interested in the signals that direct export and morphogenesis of the polar and lateral organelles. Previous work has shown that the composition of the polar filament is complex; i.e., there are multiple subunits in the polymer, and the filament is sheathed by an extension of the cell outer membrane (2,38). The polar flagellum is powered by the sodium motive force (4). The lateral flagella show more resemblance to E. coli flagella in that they are unsheathed, arranged peritrichously, and powered by the proton motive force (1, 4). Purified lateral flagellar filament appears to be composed of only a single flagellin subunit (39). Purified polar flagellar filament appears to be composed of two flagellin subunits; however, disruption of the genes encoding polar flagellins revealed that there were four genes (38). In this work, we have shown that gene disruption of lafA created a mutant that is unable to swarm and fails to synthesize lateral flagella. Thus, there is only a single gene encoding the subunit that polymerizes to form the lateral flagellar filament.
Flagellar morphogenesis is a complex cascade of events whereby gene expression is carefully coupled to assembly of the organelle (34). This hierarchy of gene regulation was first unraveled for E. coli by using gene fusions to lacZ (27) and has now been worked out for S. typhimurium (30) and C. crescentus (13,43). For these bacteria, expression of the many genes involved in motility and chemotaxis is grouped into at least three transcriptional classes (early, middle, and late). All of the genes of a particular class must be expressed before expression of the next, or later, class. This leads to sequential assembly of the basal body structure, addition of the hook via the HAPs, which are important for filament addition, and finally polymerization of flagellin monomers at the growing tip. Failure at any point of this assembly completely shuts down later gene expression. As depicted in Fig. 8, we have begun to identify genes in the motility system of V. parahaemolyticus that is used for surface translocation and to place these genes into a scheme of control. Three levels have been identified in the transcriptional hierarchy.
A key factor in flagellar regulatory cascades is an alternative sigma subunit for RNA polymerase (18). We have fortuitously cloned a lateral flagellar gene that appears to encode such a sigma factor while cloning the single structural gene encoding lateral flagellin. LafS, the lateral flagellar sigma factor, is related to the alternative sigma factors (&f2s) of the flagellar cluster, which includes not only the sigma subunits that direct synthesis of flagellar genes from a variety of organisms but also a sigma factor from Streptomyces coelicolor that controls morphological differentiation, i.e., the developmental fate of hyphae (10). However, the V.
parahaemolyticus sigma factor appears to be different; although it is highly related to these sigma factors, it is more distantly related than any of the other members of the cluster are to each other. In particular, this difference is apparent in for regions of the protein thought to be important in conferring promoter specificity.
This observation is consistent with other facts. Some of these sigma factors have been shown to recognize similar and functionally interchangeable promoters (i.e., those of B. subtilis, E. coli, and Streptomyces coelicolor [10,18]), and o28 consensus promoter sequences have been identified preceding flagellar genes from a variety of organisms. We have determined the nucleotide sequences of two genes encoding components important for formation of the lateral filament, the flagellin monomer (lafA) and a HAP-like protein probably important for polymerization of the filament (lafB). Sequences preceding both the lafA and lafB genes show no similarity to the &2 promoter consensus sequence. In contrast, there are sequences preceding the genes encoding polar flagellins that resemble the flagellar consensus sequences (37a), and expression of these genes is dependent on E. coli 2'H (FliA). Genetic analysis has shown that expression of lafA requires the product of the sigma-like gene lafS. Furthermore, this requirement is manifested in E. coli. The E. coli flagellar sigma factor (FliA) does not recognize the promoter for the lateral flagellin gene, and expression of lafA in E. coli is dependent on the lafS gene. In fact, because of this requirement, the original clone expressing lafA in E. coli is a chimera containing two distinct regions of the chromosome, the LafA and sigma loci. On the basis of promoter selectivity, it appears that the V. parahaemolyticus sigma factor is a unique member of the flagellar cluster, and analysis of the differences in protein sequence and promoter specificity may prove insightful for determining what is important for conferring specificity of promoter selection.
In E. coli, the motA and motB gene products are required for torque generation. Both of these proteins are found in the cytoplasmic membrane. It is thought that MotA functions as a proton channel (7) and that MotB serves as a stator connecting MotA to a stationary component of the cell wall (11). The amino terminus of MotB is thought to span the cytoplasmic membrane, with the remainder of the molecule localized to the periplasm. How these proteins function to generate torque is not clear; however, the conservation of deduced protein sequence between the proton-driven lateral flagellar motor components (LafT and LafU) and MotA and MotB of E. coli supports the present topological picture of the torque generators. There is only one region of significant difference. LatT has four membrane-spanning domains similar to those detected for MotA (12). It also has two regions of high charge which may be important for the handling of protons. The difference in these charged regions of V. parahaemolyticus is that they are not separated into an acidic and a basic domain as was observed for E. coli; both regions contain many acidic, basic, and polar amino acids.
The arrangement of genes around lafS suggests a transcriptional unit, and it will be interesting to determine how large this operon or gene cluster is. Positioning of the lateral motor genes in the sigma locus is curious. In E. coli, the mot genes are in an operon with two chemotaxis genes and require the flagellar sigma factor for expression. Also, the appearance of a gene resembling fliL in the sigma operon is intriguing. Although function of FliL is not known, it is usually found preceding a second gene, fliM, which is believed to be part of the motor's switch complex involved in modulating direction of flagellar rotation. Again, the gene organization appears unusual for V. parahaemolyticus. Mutants in the fliM locus of C. crescentus show abnormal cell division, and it has been postulated thatfliM orfliL may help couple flagellar regulation to the cell cycle (53). Flagellar morphogenesis for E. coli as well as C. crescentus appears to be coupled to the cell cycle (43,44). In contrast to these bacteria, synthesis of lateral flagella seems to be released from cell division events. When V. parahaemolyticus differentiates to a swarmer cell type, cell division ceases, the cell elongates, and hundreds of lateral flagella are synthesized. Perhaps the unique transcriptional organization of lateral flagellar genes is significant in terms of the particular timing and coordination of gene expression required for swarmer cell differentiation.