Flagella are much more than organelles of locomotion and have multiple roles that contribute to pathogenesis. Bacteria, such as Vibrio parahaemolyticus and Aeromonas spp., that possess two distinct flagellar systems (a polar flagellum for swimming in liquid and lateral flagella for swarming over surfaces) are relatively uncommon and provide ideal models for the independent investigation of the contributions of these different types of motility and other flagellar functions to virulence and how they are controlled. Studies with the above organisms have already increased our understanding of how bacteria sense and colonize surfaces forming biofilms that enable them to survive and persist in hostile environments. These insights are helping to identify possible new targets for novel antimicrobials that will both prevent or disrupt these processes and enhance the effectiveness of existing antibiotics. Aeromonas lateral flagella, in addition to mediating swarming motility, appear to be adhesins in their own right, contribute to microcolony formation and efficient biofilm formation on surfaces, and possibly facilitate host cell invasion. It is, therefore, likely that the ability to express lateral flagella is a significant virulence determinant for the Aeromonas strains able to cause persistent and dysenteric infections in the gastrointestinal tract, but further work is needed to establish this.
Interest in bacterial flagella has a long history dating back to the late 17th Century when Antony van Leeuwenhoek was first fascinated by the ‘small living animals which moved themselves very extravagantly’ in the scurf of his teeth when he examined it through one of his single-lens microscopes . This interest has continued to grow in more recent times with the realization that not only are flagella the organelles of locomotion (swimming and swarming motilities), but they contain a sophisticated export apparatus that is closely related to type III secretion pathways for virulence factors. They also have additional roles in adhesion, surface colonization, biofilm formation, and invasion that contribute to pathogenesis. Research into these multifunctional roles and how they are regulated is helping to provide a platform for the development of much-needed novel anti-infective strategies.
The number and distribution of flagella on bacteria varies. Single (monotrichous) and multiple (peritrichous) arrangements are the ones most frequently found on pathogenic bacteria. Flagellar expression is, however, influenced by growth conditions. On solid media, many bacterial species express more flagella than they do when grown in liquid media. Some species, like Proteus mirabilis, increase the numbers of their existing flagella while others, like Vibrio parahaemolyticus, have a single polar flagellum in liquid media but on solid media show mixed flagellation consisting of the polar flagellum (Fla) and a completely different peritrichous (or lateral) flagellum (Laf) which may be present in large numbers [2,3]. As will be discussed (Section 4), such microorganisms have two distinct genetic systems encoding the Fla and Laf, respectively. The possession of flagella is a costly commitment for the bacterium in terms of resources and energy. Hence, bacterial species in the latter category are relatively uncommon and the expression of both Fla and Laf is highly regulated.
As this review will show, species that have two distinct flagellar systems present unique advantages for understanding how bacteria detect surfaces and for dissecting flagellar roles in pathogenesis.
2Bacterial species with two distinct flagellar systems
V. parahaemolyticus is the best-studied organism with distinct lateral flagella. This bacterium is common in marine and estuarine environments and causes gastroenteritis, wound infections, and septicaemia in man. It is a primary cause of seafood-associated food poisoning worldwide. Lateral flagella have also been described on about seven other Vibrio species (some of which have a similar disease spectrum to V. parahaemolyticus) , as well as only a small number of other bacterial species. These species are Rhodospirillum centenum (a purple photosynthetic bacterium) , Azospirillum spp. (nitrogen-fixing rhizobacteria that colonize plants) , Helicobacter mustelae (the causative agent of chronic gastritis and ulcer disease in ferrets) , and Plesiomonas shigelloides and Aeromonas spp. (opportunistic and gastroenteric pathogens of man) [8,9].
Genetic analysis of the flagellar systems of V. parahaemolyticus is well advanced , yet surprisingly little is known about the contributions of the different flagellar roles to virulence. Recent studies with Aeromonas spp. are helping to resolve this situation. These organisms are ubiquitous, aquatic bacteria that are also found in many foods. They comprise a diverse genus, but of the now more than 14 recognized DNA hybridization groups, just three (A. hydrophila HG1, A. veronii biovar sobria HG8/10, and A. caviae HG4) are responsible for >85% of human infections.
Representatives of Aeromonas spp. are found in most aquatic environments, but A. hydrophila may predominate in spring water, A. veronii biovar sobria in recreational lakes and river water, while A. caviae is often more common in alkaline and marine waters. All three species are found in sewage-contaminated water. These mesophilic aeromonads (particularly A. hydrophila) colonize water-dwelling plants and animals (fish, leeches, frogs), and cause diseases in reptiles, fish, shellfish, and snails. They are isolated as part of the faecal flora of a wide variety of animals, including ones destined for human consumption (pigs, cows, sheep and poultry). In retail foods, they are most common in seafoods and shellfish, but all three species are found in most foods. A. hydrophila and A. veronii biovar sobria predominate in meat and poultry, while A. caviae is often the most common Aeromonas species found in salads and raw vegetables [11,12].
Human infections with the above three species include water-associated wound infections (particularly by A. hydrophila and A. veronii biovar sobria), and serious opportunistic infections such as septicaemia and meningitis in immunocompromised individuals. Some strains of these three species are also primary gastrointestinal pathogens whose combinations of virulence determinants (not yet fully characterized) result in diarrhoea and/or dysenteric and persistent infections, most notably in children <5 years of age and the elderly in the summer months. A. hydrophila and A. veronii biovar sobria are more likely to produce the haemolytic, cytotoxic enterotoxin (Act/aerolysin) than are strains of A. caviae. A. caviae is particularly reported as an enteropathogen in infants <1 month of age, and in travellers where persistent infections of >17 months have been reported. Some strains of all three species can invade cultured human epithelial cells (HEp-2) and intestinal cells (Caco-2), and it is thought that invasion may be the mechanism that results in the dysenteric presentation, although this link remains to be confirmed [11–14].
Early light microscopic observations, subsequently confirmed by genetic and functional analyses, suggested that not all Aeromonas strains are able to express lateral flagella, thus increasing interest in the Laf as a possible virulence determinant that might contribute to the persistent and invasive clinical presentations described above [15,16].
3Flagellar structure and type III secretion apparatus
Flagella (Fla and Laf) are essentially helical propellers. Details of their structure and assembly are reviewed more extensively elsewhere [17–19]. In brief, a flagellum consists of the filament made up of polymerized flagellin subunits (of which there may be more than one antigenic type), attached by a hook structure to the basal body. The latter membrane-embedded, rotary device is composed of several ring structures (cytoplasmic (C), membrane-supramembrane (MS), peptidoglycan and lipopolysaccharide layers), and the torque-generating machinery, MotA and MotB (two proton-conducting or sodium ion-driven proteins depending on the organism), several of which are arranged around the MS and C rings. The flagellum has a specialized type III export apparatus that translocates protein subunits in an ATP-dependent manner across the plane of the cytoplasmic membrane and delivers them to a central channel in the basal body-hook-filament structure where they eventually reach their assembly point at the distal end. Here, there is a capping protein that is required for the addition of flagellin subunits to the filament. Subunits in the channel insert between the end of the filament and the cap that prevents them escaping into the cell's surroundings. The assembly of the hook also requires a capping protein.
The flagellar type III secretion apparatus shows homology with the contact-dependent type III secretion apparatus of Gram-negative bacteria which exports toxins and other virulence factors and may itself act as another general mechanism for the transport of proteins that influence bacterial–host interactions in pathogenesis . Several extracellular proteins, including a virulence-associated phospholipase, have been shown to be exported by the flagellum export apparatus of Yersinia enterocolitica. Type III secretion proteins themselves are, therefore, of interest as possible targets for therapeutic interventions .
The flagellar structure outlined above not surprisingly has a complex genetic basis. Studies of the flagella of organisms such as Escherichia coli and Salmonella enterica have shown that in excess of 50 genes are involved in motility and chemotaxis [23,24]. These genes are arranged in >14 flagellar operons that are distributed in three (or more) major clusters that operate in a regulatory cascade. Thus, there is a hierarchy of gene expression. In E. coli, the class 1 operon encodes the transcriptional activator of class 2 operons. Class 2 genes include structural components of the basal body-hook structure, and the transcriptional activator for the class 3 operons. The class 3 operon includes flagellar filament structural genes and the chemotaxis signal transduction system that directs the cell's motion. A checkpoint mechanism ensures that class 3 genes are not transcribed before functional basal body-hook structures are completed [24,25].
4.1Flagellar genetics of V. parahaemolyticus
The polar flagellum (Fla) of V. parahaemolyticus has around 60 genes for its biogenesis and function, many of which are homologues of E. coli flagellar genes . Homologous genes and a similar genetic organization are also found in other Vibrio species and Pseudomonas aeruginosa. It is a complex organelle as there are six polar flagellin genes, the structure is sheathed, and it is powered by a sodium ion gradient. The regulatory hierarchy of V. parahaemolyticus is thought to be similar to that reported for V. cholerae with genes in the hierarchy spread over five clusters situated in four regulatory classes .
An entirely different gene set encodes the V. parahaemolyticus lateral flagellum (Laf). An operon of eight genes includes those that encode a single flagellin (lafA), the capping protein (lafB), the lateral flagellar specific sigma factor (lafS), and the lateral flagellar proton-powered motor proteins (lafT/lafU) . In addition, there is another region of lateral flagellar genes (designated lfg) that appear to be the equivalent of the polar flagella flg genes that encode some of the basal body, hook and rod proteins .
4.2Flagellar genetics of Aeromonas spp.
The gene sets for the Aeromonas polar and lateral systems are also distinct [9,28]. (Recent studies, discussed further in Section 7, have shown that mutations in the polar fla genes do not prevent strains from expressing lateral flagella as was first reported .) The Aeromonas flagellin loci are shown in Fig. 1.
The polar flagellin locus (cloned from A. caviae strain Sch3) has two flagellin genes (flaA and flaB), a flagellar gene of unknown function (flaG), and genes encoding the filament capping protein (flaH), and a putative flagellin chaperone (flaJ) . Other Aeromonas polar flagellin specific genes that have more recently been identified (transposon mutagenesis) in this strain include the flg operons (encoding basal body, chemotaxis, hook and rod proteins), fliM (encoding the switch protein), and cheA (encoding a chemotaxis protein, discussed in Section 6), all of which are thought to be in distinct gene clusters (J.G. Shaw, unpublished observations).
The lateral flagellin locus of A. caviae strain Sch3 consists of two flagellin subunit genes (lafA1 and lafA2) which show 87% identity at the nucleotide level, and lafB and fliU, which encode a capping protein and an N-lysine methylase involved in flagellar biosynthesis, respectively [9,16]. An operon of nine lateral flagellar genes isolated from an A. hydrophila strain has only a single lateral flagellin gene (lafA), which is the more common finding in the mesophilic aeromonads. In addition to lafA and lafB, the operon includes genes for the lateral flagellar specific sigma factor (lafS), and the chemotaxis motor protein genes, lafT and lafU. These latter genes have all subsequently also been found in the A. caviae strain, as shown in Fig. 1. (J.G. Shaw, unpublished observations). The laf locus of the A. hydrophila strain has the same organization as the laf locus of V. parahaemolyticus, with the exception of an extra open reading frame (lafX) between lafC and lafE in A. hydrophila. The Aeromonas lateral flagellins, LafA1 and LafA2, have around 60% identity and 75% similarity with the V. parahaemolyticus LafA . It is likely that as in V. parahaemolyticus, Aeromonas spp. possess a lfg-like operon containing additional lateral flagellar specific genes. Both the polar and lateral Aeromonas flagella are unsheathed . The Aeromonas flagellar systems, therefore, appear to parallel those of V. parahaemolyticus, but are slightly less complex. Southern hybridization studies with an approximately 1-kb laf gene probe (Pst1 fragment) to the A. caviae flagellin subunits (Fig. 1) showed that only around 58% of mesophilic Aeromonas strains possessed laf genes. These were also functional in mesophilic strains .
5Regulation of Laf expression
For V. parahaemolyticus, the Fla appears to act as a sensor inducing Laf expression in conditions that slow its rotation. Thus, in addition to growth on solid surfaces or in increased viscosity, antibodies that inhibit Fla rotation, sodium channel-blocking drugs (e.g. phenamil) that slow the Fla motor, and mutations that affect swimming in liquid, all lead to constitutive Laf expression . A novel three-gene operon (scrABC) that appears to be involved in the control of Laf expression in V. parahaemolyticus has recently been described . It potentially encodes a pyridoxal-phosphate-dependent enzyme, an extracellular solute-binding protein, and a membrane-bound GG-DEF- and EAL-motif sensory protein. Mutants with defects in any of these three genes show decreased swarming motility (Section 6) and decreased laf gene expression .
The Fla does not appear to function as a sensor for Aeromonas spp., however. Mutations that result in the loss of the Aeromonas Fla (e.g. flaA:flaB, flaH and flaJ mutants, Fig. 1) do not cause the constitutive expression of Laf . In fact, such fla mutations appear to affect Laf expression negatively in some as yet unexplained way (Section 7). How Aeromonas spp. sense changing environments to regulate Laf expression is yet to be determined.
6Flagellar motilities: their roles in colonization of surfaces and biofilm formation
Two quite different types of motilities are mediated by bacterial flagella. In addition to swimming motility in liquid, flagella mediate swarming motility on solid surfaces or in viscous conditions. The latter motility is associated with cell differentiation to the ‘swarmer phenotype’ (discussed below). Swarming motility has been demonstrated in many different bacterial genera, both Gram-negative and Gram-positive . Like swimming motility, it is thought to contribute to bacterial colonization of host cell surfaces and play a significant role in biofilm formation.
Biofilms (complex communities of adherent bacteria encased in exopolysaccharide and with a distinctive architecture characterized by pillars and channels) are the preferred mode of existence of bacteria on environmental surfaces and host tissues . They are important environmental sources of infection and lie at the root of many persistent and indwelling device-related infections [32,33]. The biofilm mode of growth allows survival in a hostile environment. Bacteria in biofilms are generally more resistant to host defences and antimicrobial agents and are more likely to express more virulent phenotypes .
Fig. 2 summarizes what are believed to be the essential steps in biofilm formation.
Swimming motility and chemotaxis are important in the initial approach and attachment to surfaces (Fig. 2). In swimming motility, the flagella rotate counter-clockwise and the bacterium is propelled forward (a ‘run’). When the flagellar rotation is clockwise, the cell randomly tumbles in place (a ‘tumble’). Runs are more frequent if the bacterium approaches a chemoattractant (e.g. amino acids, sugars, oligopeptides). Conversely, decreasing concentrations of attractants or increasing concentrations of repellents (extremes of pH, some metal ions, hydrophobic amino acids) cause bacteria to increase their tumbling frequencies. Chemotaxis has been most extensively studied in organisms that are able to express only one flagellar type, such as E. coli. Transmembrane receptors, methyl-accepting chemotaxis proteins (MCPs), initiate the signal cascade when specific compounds bind to them. They transmit the chemotactic signal to the cytoplasm where cellular proteins (CheA, CheW, CheY, CheZ, CheR, and CheB) are responsible for the transduction of the chemical signal to the flagellar switch. CheA is the first of the Che proteins to receive the chemotaxis signal inside the cytoplasm. It is a histidine kinase that, in association with the coupling protein, CheW, interacts with the MCPs, autophosphorylates, and then transfers the phosphate group to other Che proteins, CheY and CheB. Phosphorylated CheY then interacts with the flagellar switch to increase the tumble frequency . By contrast, very little is known of chemotaxis in other families, including Vibrionaceae, where it appears to be much more complex. The V. cholerae genome contains 43 potential MCPs and three cheA genes have been described [10,36]. For V. parahaemolyticus, at least some of the chemotaxis components are shared by the two flagellar systems as defects in cheA or cheB affect not only swimming but also swarming motility (Laf mediated) . The situation appears similar for Aeromonas spp. (S.M. Kirov and J.G. Shaw, unpublished observations).
Swarming motility, as mentioned above, involves bacterial cell differentiation and depends on the increased flagellar expression induced on surfaces and by high viscosity. It is a social phenomenon involving highly co-ordinated movement and contact between neighboring bacteria. The hyper-flagellated, differentiated bacteria assemble into multicellular rafts of cells and migrate away from the colony . Swarming has been best characterized in Proteus spp. and V. parahaemolyticus[3,27,30,38]. In Proteus spp., the hyper-flagellation results from the increased expression of the same flagellar type used for swimming, while in V. parahaemolyticus it occurs because of the induction of lateral flagella (Sections 1 and 5).
Proteus swarming on agar has a striking ‘bull's-eye’ appearance. The initial colony of short (2–4μ), vegetative rods with 6–10 peritrichous flagella differentiates into the hyper-flagellated swarm cells (20–80μ) which migrate in unison away from the colony, halt and revert to vegetative cells before a second phase of swarm cell differentiation. This results in the concentric ring pattern . Similar concentric ring patterns are also seen with V. parahaemolyticus on some media (rich ones) .
We have shown conclusively that, as for V. parahaemolyticus, lateral flagella also mediate swarming motility in Aeromonas spp. Only Aeromonas laf-positive strains were able to swarm, and flagellar mutants unable to express lateral flagella and/or negative for Laf on whole cell immunoblots grew, but showed no movement, in swarm agar assays (Fig. 3, panels 1b–3b).
Swarming motility allows rapid colonization of the surface where the bacteria then replicate to form microcolonies (Fig. 2). Besides movement, lateral flagella have additional functions in the colonization of surfaces. For V. parahaemolyticus, lateral flagella form linkages between bacteria on the agar surface that contribute to the microcolony formation . They have also been shown to be important in the adherence of bacteria to chitin. Kinetic and inhibition studies showed that such adhesion is probably by a mechanism different from that used by polarly flagellated bacteria. The production of lateral flagella is thought to increase the forces holding the bacterium to the surface while allowing for movement of the bacterium on the substrate before irreversible adsorption and the formation of the polysaccharide glycocalyx and biofilm  (Fig. 2).
Aeromonas lateral flagella, like those of V. parahaemolyticus, also form linkages between bacteria. This has been shown both on agar and on cell surfaces . Their role as adhesins, distinct from polar flagellar adhesion, has been demonstrated not just on inert surfaces but also on intestinal cell lines, and their contribution to biofilm formation has been established using selected fla and laf flagellar gene deletion mutant strains (Section 7). Such studies illustrate the value of organisms with the two flagellar systems for studying the contributions of the different motilities (and other functions associated with swimmer and swarmer phenotypes) to virulence as these can be inactivated independently of each other.
While flagella have clearly been shown to contribute to the initial approach and also attachment (along with other filamentous structures, such as pili) for several bacterial species, most notably P. aeruginosa, there have been few studies that definitively link swarming motility to biofilm formation and virulence. The strongest evidence until now has been for P. mirabilis, where the differentiated swarmer cell appears to be the virulent form. This organism causes urinary tract infections in hospitalized patients, especially those with urinary catheters. Infections develop by the ascending route. Bacteria first colonize the urethra and then migrate to the bladder, ureter and kidney. Proteus swarmer cell differentiation is also accompanied by a marked increase in the expression of several putative virulence factors (cell-bound haemolysin, urease, metalloprotease, ability to invade urothelial cells) [42,43]. A relationship between secretion of virulence factors and swarming motility has also been found in a number of other bacterial genera . However, results from animal experiments with P. mirabilis are equivocal as the difficulty with this organism (and others, such as P. aeruginosa and E. coli which have a single flagellar system for both swimming and swarming) is in uncoupling the contributions of the two flagellar motilities . With bacteria that have two distinct flagellar systems, such as with Aeromonas spp., this difficulty can be overcome (Section 7).
Development of the mature biofilm is controlled, at least in part, through ‘quorum sensing’ (QS), a chemical signalling system by which bacteria poll cell density and regulate the expression of many genes at high cell densities (Fig. 2). Understanding this has already led to the use of competitive antagonists (such as furanones) that can interfere with the signals that trigger biofilm differentiation and decrease P. aeruginosa biofilm formation, for example [46–48]. For some bacteria, for example Serratia liquefaciens (which has one flagellar type for both swimming and swarming), QS has been shown to also influence the production of surfactant required for swarming motility on surfaces .
Like P. aeruginosa, Aeromonas spp. have a QS system that has been characterized and shown to be necessary for the development of the mature biofilm architecture . It is not yet known whether QS influences the early stages of biofilm formation for Aeromonas spp., but these organisms provide an ideal model for learning how flagellar activities contribute to them and are regulated. Such studies are likely to lead to the identification of additional targets for the prevention and treatment of biofilm infections. As yet, there have been few published functional studies with Aeromonas species but results to date are briefly summarized below.
7Aeromonas spp.: a model for dissecting flagellar roles in virulence
It has been established that both flagellar types function as adhesins (Fig. 3, panels 1a–3a). In in vitro cell adhesion assays, numbers of adherent bacteria are stained and counted after 90 min exposure to semi-confluent cell monolayers. The mean number of adherent bacteria per cell is estimated on three or more replicate coverslips [28,51]. Polar flagellar mutants of A. caviae strain Sch3 (e.g. AAR31/flaA:flaB, AAR59/flaH, and AAR 8/flaJ; Fig. 1), unable to express Fla or to swim, are virtually non-adherent to human epithelial (HEp-2) and intestinal (Caco-2, INT 407) cell lines (>95% decrease in adhesion compared to the wild-type strain) in these assays (figure 2, panels 1a–3a, in ). As mentioned (Section 5), these strains were originally thought to be Laf-negative as lateral flagellin was not detected on whole cell immunoblots and the mutants are unable to swarm on Eiken agar [9,16]. However, they have been designated here as Laf± (Fig. 3), as more recent experiments (Western blots of sheared rather than whole cell bacterial preparations) have suggested that Laf are expressed. The Laf are still under their natural control on surfaces, not constitutively as in V. parahaemolyticus polar mutant strains (J.G. Shaw, unpublished results). There is clearly something unusual about the Laf of these strains (functionally Laf-negative) and further studies are required to elucidate the regulation of Fla and Laf expression in Aeromonas spp. Removal of the Fla by shearing or agglutination with anti-flagellin antibodies also greatly reduced bacterial adherence by this A. caviae strain .
Centrifugation of these aflagellate mutants (AAR31/flaA:flaB, AAR59/flaH, and AAR8/flaJ) onto the cell monolayers only increased adherence slightly (<20%), whereas the adherence of Fla+ flagellar mutants (AAR269/flaA, AAR27/flaB, and AAR150/flaG, Fig. 1) was not affected . Taken together, the above results clearly establish that the Aeromonas Fla functions as a cellular adhesin. Motility is nonetheless also an important contributor to this initial adhesion to surfaces, as has been shown for other bacterial species. An Aeromonas CheA mutant strain (Fla+Laf+, but unable to swim (or swarm) in agar assays) showed a significantly reduced, random adherence pattern to all cell lines (S.M. Kirov and M. Castrisios, manuscript in preparation).
Lateral flagellar mutants (Fla+Laf–) of A. caviae strain Sch3 (AAR6/lafA1:lafA2 and AAR9/lafB; Fig. 1), and of a variety of other Aeromonas isolates of pathogenic species tested showed decreased (>50%) adhesion to HEp-2 cells . Wider testing of the A. caviae strain Sch3 mutants on intestinal cell lines (Henle INT 407 and Caco-2) confirmed similar decreased adherence (Fig. 3, panel 2a). Complementation of mutant strains to reinstate Laf expression and/or swarming motility restores adhesion to wild-type levels (, S.M. Kirov and M. Castrisios, manuscript in preparation). Transfer of the complete A. hydrophila laf locus on the plasmid pCOS-LAF into A. caviae strain Sch3 lafA1:lafA2 mutants restored Laf expression and swarming motility to wild-type levels in the microtiter plate assay .
Studies with Aeromonas flagellar mutant strains have also found that fully functional Fla and Laf are required for efficient biofilm formation on abiotic surfaces, such as plastic (microtiter plates)  and glass (borosilicate glass tubes) (Fig. 3, panels 1c–3c). In general the ability of the range of flagellar mutants to form biofilms in these assays paralleled the cell line adhesion results. In these quantitative biofilm assays, after the addition of bacteria and incubation, the surfaces are washed vigorously and the bacterial biofilm is stained with crystal violet, solubilized, and quantitated by measuring the absorbance read at 570 nm. The polar flagellin double mutant (AAR31, Fla−Laf±) consistently formed biofilms at only around 40% of the wild-type level, while the lateral flagellin double mutant strain (AAR6, Fla+Laf−) formed them at approximately 50% of the wild-type level. When the other flagellar mutants were tested in this model all had somewhat decreased abilities to form biofilms, irrespective of their phenotypes, but biofilm formation of the fla double mutant was restored virtually to wild-type levels by complementation of the mutant strain (S.M. Kirov and M. Castrisios, manuscript in preparation). Biofilm formation by lateral flagellar mutants has similarly been complemented in the microtiter plate assay . The contribution of lateral flagella to biofilm formation still requires further study, however, in more sophisticated biofilm models.
It should be noted that Fla and Laf are not the only filamentous structures to contribute to Aeromonas attachment and biofilm formation. As for P. aeruginosa, Type IV pili would also appear to play a role .
The CheA mutant of A. caviae strain Sch3 formed biofilms at <20% of the wild-type level. Testing of motility mutants of this strain, Mot− (Fla motor proteins) that will still express Fla but not swim, and LafT/U− (Laf motor proteins), able to swim but not swarm, will ultimately allow further dissection of the adhesive and motility contributions of flagella to biofilm formation. These studies will need careful interpretation however. How motility is affected, i.e. the effects of particular mutations on flagellar rotation and subsequent physical orientation of flagella, may impact adhesion and invasion capabilities . Also, flagellar expression (Fla and/or Laf) is tied to the regulation of other bacterial properties, for example exopolysacccharide production, involved in colonization and biofilm formation. The lesions in scrABC (Section 5) that decreased swarming in V. parahaemolyticus also increased capsular polysaccharide production . In V. cholerae, exopolysaccharide production and Fla expression have also been shown in several studies to be inversely related [54,55].
The above range of mutant strains will also help to elucidate flagellar roles in invasion. As mentioned (Section 2), invasion may be the virulence mechanism that results in Aeromonas-associated dysentery. Merino et al.  reported that for Aeromonas spp. polar flagella are essential for invasion of fish cell lines, but their studies did not determine whether it was motility, adhesion or a combination of both that was the significant function . A. caviae strain Sch3 has invasion efficiencies of 0.9% in HEp-2 cells and 4.8% in 14-day-old Caco-2 cells, compared to non-invasive Aeromonas strains and the invasive Y. enterocolitica strain 8081c which have values of ≪0.2% and 7.2%, respectively, in Caco-2 cells . It is, therefore, of some interest to determine whether Laf contribute to cellular invasion in this strain and other Laf-positive strains.
Although the advantages offered by Aeromonas spp. (and other bacteria with two distinct flagellar systems) for elucidating flagellar roles in pathogenesis have only recently been appreciated, it has already been established definitively with this bacterial model that the swarmer cell phenotype contributes to surface adhesion and biofilm formation independently of the initial swimming and attachment to surfaces. Invasion and in vivo colonization studies are now particularly required to establish the significance of the Laf as a virulence determinant for Aeromonas spp. and the swarmer phenotype for bacterial species more generally. Further investigations into the regulation and interrelationships of the two separate flagellar systems and other cellular components will ultimately have wide implications for the control of environmental biofilms and many biofilm infections.
I thank the University of Tasmania Institutional Research Grants Scheme and the Royal Hobart Hospital Research Foundation, Tasmania, Australia, for funding research from my laboratory quoted in this review. I am grateful to Dr Jon Shaw, University of Sheffield, UK, for his helpful feedback on the manuscript, and thank him and Miss Marika Castrisios, Discipline of Pathology, University of Tasmania, Australia, for allowing me to cite their unpublished work. Recent genetic analyses of Aeromonas flagellar genes conducted in Dr Shaw's laboratory were done in collaboration with Dr J.M. Tomás and colleagues, Department of Microbiology, University of Barcelona, Spain.