ABSTRACT
The nine laf (lateral flagellum) genes of mesophilic aeromonads are in the Aeromonas salmonicida genome. The laf genes are functional, except for lafA (flagellin gene), which was inactivated by transposase 8 (IS3 family). A pathogenic characteristic of mesophilic aeromonads (lateral flagella) is abolished in this specialized pathogen with a narrow host range.
Aeromonas salmonicida is an important pathogen of salmonid fish, producing the systemic disease furunculosis (12). As in Vibrio parahaemolyticus (8, 9), two types of flagella are responsible for motility in aeromonads. A polar unsheathed flagellum is expressed constitutively that allows the bacteria to swim in liquid environments. In media where the polar flagellum is unable to propel the cell, aeromonads express peritrichous lateral flagella (21), a phenomenon associated with the colonization of surfaces, as such hyperflagellated cells demonstrate increased adherence. We have demonstrated the importance of the polar flagellum of A. hydrophila in the invasion of fish cell lines (11, 13). More recently, we have described a polar flagellar gene region in A. caviae that appears to be essential for adherence to human epithelial cells in vitro (19). Traditionally, the genus Aeromonas has been divided on the basis of motility, with A. salmonicida being the typical nonmotile species (17). However, a report by McIntosh and Austin (10) indicated that five A. salmonicida isolates were able to produce a pole-located sheathed flagellum when grown at supraoptimal incubation temperatures (30 to 37°C) and in the presence of 18% (wt/vol) Ficoll. However, incubation under these conditions never renders more than 1% of the population motile (twisting/tumbling movement) or flagellated. Genetic evidence demonstrated that A. salmonicida strains are capable of producing flagella, as two flagellin genes (flA and -B) were identified and characterized (24). Nine lateral flagellar genes, lafA to -U, for A. hydrophila and four A. caviae genes, lafA1, lafA2, lafB, and fliU, have been isolated (3). Mutant characterization and nucleotide and N-terminal sequencing demonstrated that the A. hydrophila and A. caviae lateral flagellins are almost identical but are distinct from their polar flagellum counterparts. Mutation of Aeromonas lafB or lafS or both A. caviae lateral flagellin genes caused the loss of lateral flagella and a reduction in adherence and biofilm formation. Mutation of lafA1, lafA2, fliU, or lafT resulted in strains that expressed lateral flagella but had reduced adherence levels. Mutation of the lateral flagellar loci did not affect polar flagellum synthesis, but the polarity of the transposon insertions on the A. hydrophila lafT and -U genes resulted in nonmotility (3). In a recent study of lateral flagella and swarming motility in different Aeromonas species, four different isolates of A. salmonicida reacted positively with a DNA probe for lateral flagella that was used to correlate lateral flagella and swarming motility in mesophilic Aeromonas spp. (7).
In this study, we provide the genetic basis to explain why A. salmonicida strains are able to hybridize with a DNA laf gene probe but are unable to produce lateral flagella.
Presence of laf genes and lateral flagella.
The nucleotide sequences of the lateral flagellins from A. hydrophila AH-3 (lafA) and A. caviae Sch3N (lafA1 and lafA2) were aligned by using ClustalW (22) in order to find common oligonucleotides able to amplify this DNA region. In the same alignment, the polar flagellins of A. caviae Sch3N (flaA and flaB) and A. salmonicida (flaA and flaB) were introduced in order to verify that the common oligonucleotides for lateral flagellins are unable to amplify the polar flagellins. From these data, we found two highly conserved domains (the C- and N-terminal domains) in which oligonucleotides could be designed to specifically amplify the lateral flagellins. The oligonucleotides designed were Laf1 (5′-GGTC TGCGCATCCAACTC-3′) and Laf2 (5′-GCTCCAGACGGTTGATG-3′). In a PCR with Laf1, Laf2, and genomic DNA from either A. hydrophila AH-3 or A. caviae Sch3N, we obtained a single band of approximately 550 bp that was confirmed to be the lateral flagellin gene by DNA sequencing (Fig. 1A).
PCR amplification with several Aeromonas sp. genomic DNAs. (A) DNA amplification fragments (550 bp) obtained with oligonucleotides Laf1 and -2 and A. caviae Sch3N (lane 1), A. hydrophila AH-3 (lane 2), A. salmonicida A450 (lane 3), and four different A. salmonicida isolates (lanes 4 to 7). Lane MW contained molecular weight markers (1 Kb PLUS DNA Ladder; Invitrogen). (B) DNA amplification fragments obtained with oligonucleotides Laf 1 and -5 and A. caviae Sch3N (lane 1, 2,751 and 1,737 bp), A. hydrophila AH-3 (lane 2, 1,715 bp), A. salmonicida A450 (lane 3, 2,952 bp), and the same A. salmonicida isolates previously used in panel A (lanes 4 to 7, 2,952 bp). Lane St contained λ DNA digested with HindIII. (C) DNA amplification fragments obtained with oligonucleotides Laf 1 and −9 and A. hydrophila AH-3 (lane 1, 7,120 bp), A. salmonicida A450 (lane 2, 8,360 bp), and two of the A. salmonicida isolates used previously (lanes 3 and 4, 8,360 bp). Lane St contained λ DNA digested with HindIII.
Table 1 lists the strains used in this study. A. salmonicida strains were grown at 20°C, and mesophilic Aeromonas strains were grown at 30°C. The A. salmonicida strains were pathogenic isolates from moribund fish derived from a wide range of geographical locations (Canada, Japan, the United Kingdom, the United States, and Spain). All of the A. salmonicida strains used in this study were classified in accordance with Bergey's Manual of Systematic Bacteriology. All of the isolates were able to react with specific antibodies against the A layer and could be classified as typical pathogenic A. salmonicida strains (16).
Bacterial strains and plasmid used in this study
All 50 (100%) of the A. salmonicida strains showed a positive PCR resulting in a single band of approximately 550 bp, which was confirmed by sequencing in a number of cases (Fig. 1A). The strains were also tested by colony blot hybridization against a laf probe and for swarming by methods previously described (3, 19). There was a complete correlation among the PCR-positive and colony blot-positive strains. However, while laf-positive mesophilic Aeromonas sp. strains were able to swarm, none of the A. salmonicida strains tested were able to. Furthermore, when several A. salmonicida strains were examined by electron microscopy (EM) and a Western immunoblot assay for the presence of lateral flagella and production of lateral flagellin (3, 19), they were always negative.
As can be observed from the DNA sequences of A. hydrophila AH-3 (accession no. AY028400 ) and A. caviae Sch3N (accession no. AF348135 ), lafB is the highly conserved gene immediately downstream of the lateral flagellin gene(s) in the laf cluster. An oligonucleotide was designed within the lafB gene, Laf5 (5′-ATCGCTGGAGGTCATCTTG-3′). In a PCR with Laf1, Laf5, and genomic DNA from A. hydrophila AH-3 or A. caviae Sch3N, we obtained a single band (1,715 bp) for A. hydrophila AH-3 and two bands (2,751 and 1,737 bp) for A. caviae Sch3N (depending upon whether they have one or two flagellin genes). When we tested the A. salmonicida strains, we amplified a single band of 2,952 bp (Fig. 1B). The complete DNA sequence of the A. salmonicida band showed the presence of a putative transposase 8 from the IS3 family inserted within the A. salmonicida lafA gene. When other A. salmonicida strains were tested (n = 10) with the same oligonucleotides, the same lafA band (2,952 bp) containing the putative transposase was amplified. This putative transposase showed a high degree of identity and similarity (more than 60 and 75%, respectively) to TnpA from Pseudomonas putida or Erwinia carotovora and putative transposases from Pantoea agglomerans, Xanthomonas campestris, and Agrobacterium tumefaciens. Furthermore, it showed the typical transposase region at the beginning of the sequence and the integrase core domain at the end.
The complete A. salmonicida strain A450 laf gene cluster.
With either genomic DNA from A. hydrophila AH-3 or the cosmid pCOS-LAF (3) in a PCR (annealing temperature of 56°C, extension time of 7 min 30 s, and PlatinumTaq polymerase high fidelity from Invitrogen) with oligonucleotides Laf1 and Laf9 (5′-CCAGATTCTTTTCCGCCTG-3′), a single DNA band of 7,120 bp was obtained (Fig. 1C). Oligonucleotide Laf9 is located in the last gene of the laf cluster, lafU; this allowed complete amplification of the nine laf genes. By applying the same amplification procedure to A. caviae Sch3N or A. veronii bv. sobria AH-1 genomic DNA, we obtained a single DNA band with a size similar to that of the AH-3 band. Sequence analysis of both edges confirmed the presence of the lafA and lafU genes in these strains. However, when A. salmonicida strain A450 was tested, we obtained a single band of 8,360 bp (Fig. 1C). Sequencing of the complete DNA band amplified from A. salmonicida strain A450 (accession no. AY1295578 ) revealed nine open reading frames (ORFs): lafA, -B, -C, -X, -E, -F, -S, -T, and -U (Fig. 2). All of the ORFs were transcribed in the same direction, and putative Shine-Dalgarno sequences were found upstream of all of the ORF start codons. The only putative transcriptional terminator sequence was found between lafA and lafB, suggesting that lafB- to U form a single transcriptional unit. The coding sequence was preceded by a putative σ54 promoter sequence. From lafB to lafU, the DNA sequence homology with the same region of A. hydrophila AH-3 was always greater than 90% (3).
Schematic representation of the A. salmonicida A450 lateral flagellin loci. Flagellar genes and ORFs are indicated by arrows, as is the direction of transcription. ORFs are named after their homologues in the A. hydrophila AH-3 laf gene cluster. The putative transposase 8 of the IS3 family is indicated by a shaded arrow along with its direction of transcription. RF (+1, +2, +3, −1, −2, and −3) indicates the reading frame. The lollipop structure depicts the approximate position of the putative transcriptional terminator. PstI, SalI, EcoRV, BamHI, and BglII restriction sites are shown.
Complementation analysis of mesophilic Aeromonas Laf− mutants with A. salmonicida laf genes.
The AH-1982 (lafB) and AH-1983 (lafS) mutant strains of A. hydrophila and mutant strain AAR9 (lafB) or tandem flagellin mutant strain AAR6 (lafA1 lafA2) of A. caviae produced polar flagella but not lateral flagella (3). Mutant AH-1984 (lafT) of A. hydrophila produced polar and lateral flagella but was nonmotile (3). Mutant AAR6 was unable to be complemented by plasmid pINA1 (lafA of A. salmonicida with the putative transposase inserted), as lateral flagella were not detected by EM.
However, mutants AAR9 and AH-1982 were fully complemented when plasmid pINA2 (lafB of A. salmonicida) was introduced into them, as lateral flagella were detected on the complemented strains by EM. A similar situation was observed for mutant AH-1983 and plasmid pINA3 (lafS of A. salmonicida). Furthermore, plasmid pINA4 (lafT of A. salmonicida) was fully able to complement the nonmotile phenotype when introduced into mutant AH-1984. None of these strains were complemented when we introduced the plasmid vector alone (pACYC184) into the mutants.
Lateral flagella, as well as the polar flagellum, in mesophilic Aeromonas strains seem to be involved in bacterial adhesion to eukaryotic cells and the ability to form biofilms (3, 5, 19). In order to see if the A. salmonicida laf genes were fully functional, we tested them for the ability to complement the mesophilic Aeromonas laf mutants for adhesion to HEp-2 cells and biofilm formation in vitro by using previously described assays (18, 23). As shown in Table 2, A. hydrophila mutants AH-1982 (lafB) and AH-1983 (lafS) and A. caviae mutants AAR6 (lafA1 lafA2) and AAR9 (lafB) showed an approximately 85% reduction in adherence to HEp-2 cells in comparison with the respective wild-type strains whereas mutant AH-1984 showed only a 50% reduction (P = <0.0005). Furthermore, the same mutants showed a drastic reduction in the ability to form biofilms (Table 2). The results obtained with the complemented strains (Table 2) are in agreement with the phenotypic complementation mentioned above. Briefly, lafB, -S, and -T mutants recovered values of adhesion to HEp-2 cells and biofilm formation similar to those of the corresponding wild-type strains. However, again, the AAR6 mutant of A. caviae (lafA1 lafA2) was not complemented by plasmid pINA1 (lafA of A. salmonicida with the putative transposase inserted). Again, no changes in adhesion to HEp-2 cells or biofilm formation were found in the mutants when the plasmid vector alone (pACYC184) was introduced.
Adhesion to HEp-2 cells and biofilm formation of different mesophilic Aeromonas wild-type and mutant strainsa
We clearly demonstrated by mutant complementation of mesophilic Aeromonas spp. that A. salmonicida genes lafB to -U are fully functional either by their phenotypic traits (presence or absence of lateral flagella or swarming motility) or by their ability to adhere to HEp-2 cells and form biofilms in vitro. The only A. salmonicida laf gene that is nonfunctional is lafA, as assessed by mutant complementation of mesophilic Aeromonas spp. It was shown that in all of the A. salmonicida isolates tested, there was a putative transposase 8 from the IS3 family inserted in the middle of the lafA gene. This gene is responsible for lateral flagellin production, which in some cases is present as a single gene, as in A. hydrophila AH-3 (3), or as two genes, as in A. caviae Sch3N (3). A. salmonicida appears to have only a single lafA gene, which was inactivated in all of the strains tested.
We tried repeatedly to introduce plasmid pCOS-LAF (3) into different A. salmonicida strains. Although cosmid vector pLA2917 alone (1) is able to enter and replicate in A. salmonicida strains, no transconjugants were obtained when pCOS-LAF was introduced. Moreover, if a plasmid construct (pACYC184) with the A. hydrophila AH-3 lafA gene alone was introduced into A. salmonicida strains, we obtained the same negative results.
The lack of transconjugants of A. salmonicida strains with pCOS-LAF or a plasmid construct with the A. hydrophila AH-3 lafA gene alone does not allow us to state that A. salmonicida strains just require a intact copy of the lafA gene for production of lateral flagella. Inactivation of lafA was enough to abolish production of lateral flagella in A. salmonicida, but maybe other genes affecting global (polar and lateral) flagellar synthesis are lacking in A. salmonicida strains, thus explaining the lack of production of polar and lateral flagella. For instance, Vibrio parahaemolyticus strains use more than 60 genes to produce polar flagella (8). It is tempting to speculate that supraoptimal incubation temperatures plus high-osmolarity (Ficoll) conditions may enhance excision of the transposase from the A. salmonicida lafA gene, rendering a small percentage of the population able to produce lateral flagella. In this respect, A. salmonicida appear to be similar to Shigella spp., which are also classified as nonmotile, as IS elements have been detected in the flagellar genes of various strains (2). However, flagellate and motile strains of Shigella spp. have been reported, although at a low frequency (4). As with Shigella spp., A. salmonicida does not show any degeneracy in its flagellar genes. This suggests that the insertion of the IS element was a recent event or that keeping these genes and having a small amount the A. salmonicida population able to become motile are somehow important for the survival and biology of the organism.
Production of lateral flagella by mesophilic Aeromonas spp. is a pathogenic factor, as it enhances adhesion to eukaryotic cells and the ability to form biofilms (3). A. salmonicida strains are pathogens with a narrow host range, salmonid fish. In this bacterium, this pathogenic factor is not expressed even though the genes appear to be present in its genome. This is, to our knowledge, the first genetically well-documented case in which a pathogenic character is abolished in a pathogen with a high level of specialization because of its narrow host range.
ACKNOWLEDGMENTS
This work was supported by a Plan Nacional de I + D grant (Ministerio de Ciencia y Tecnología, Spain) and by Generalitat de Catalunya. R.G. is the recipient of a predoctoral fellowship from University of Barcelona.
We thank Maite Polo for technical assistance.
FOOTNOTES
- Received 23 July 2002.
- Accepted 21 October 2002.
- Copyright © 2003 American Society for Microbiology