INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Aeromonas species, which have recently been placed in a new family, Aeromonadaceae, are responsible for causing a variety of human infections, including septicemia, wound infections, meningitis, pneumonia, and gastroenteritis (5). Among various virulence factors produced by Aeromonas species, enterotoxins are by far the most important in causing Aeromonas-mediated infections (1, 13, 28, 32). The cytotoxic enterotoxin gene (act) from the human diarrheal isolate SSU of Aeromonas hydrophila has been cloned, sequenced, and hyperexpressed in our laboratory (14). Four biological activities, namely, hemolytic, cytotoxic, and enterotoxic activities as well as lethality, have been shown in mice to be associated with cytotoxic enterotoxin (Act) (39). Act is a single-chain polypeptide with an estimated molecular mass of 52 kDa (40). The toxin protein is secreted as an inactive precursor (54 kDa), which is converted into the active form by proteolytic processing near the C terminus (14). Act is an aerolysin-related toxin which exhibited approximately 90% homology with an aerolysin from a fish isolate of Aeromonas bestiarum (previously designated A. hydrophila) (1, 23, 24). In contrast, comparison of Act with an aerolysin from a diarrheal isolate of Aeromonas trota revealed approximately 75% homology (1, 9, 26). Recently, an aerolysin-related toxin also was isolated from a gram-positive organism, Clostridium septicum (7).

We identified regions on Act involved in the biological functions of the toxin by deletion analysis, generation of antipeptide antibodies, and site-directed mutagenesis (16). Our data indicated that although Act had significant homology with aerolysin, there are enough differences that differential folding of these two protein molecules could occur (16, 17, 19). Further, our data suggested that there may be different loci coding for specific biological activities of Act.

Mechanism-of-action studies revealed that Act operated by creating pores, estimated to be 1.14 to 2.8 nm in diameter, in the erythrocyte membranes (17). The toxin appeared to undergo aggregation when preincubated with cholesterol, which resulted in a loss of Act's hemolytic activity (17), indicating cholesterol to be one of the receptors for Act (17). Recently, Nelson et al. (34) reported that Thy-1, a major surface glycoprotein of T lymphocytes, is a high-affinity receptor for aerolysin from A. bestiarum.

In the present study, we have constructed transposon and isogenic mutants that were defective in the production of Act from wild-type A. hydrophila SSU to determine Act's precise role in the overall virulence of Aeromonas. These mutants not only were devoid of Act-associated biological activities but were significantly less virulent in mice than wild-type Aeromonas, proving unequivocally the role of Act in Aeromonas-mediated infections.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. A. hydrophila SSU, a diarrheal isolate, was obtained from the Centers for Disease Control and Prevention, Atlanta, Ga. The identity of this culture as A. hydrophila was confirmed by DNA-DNA hybridization and ribotyping (5). Isolate A52 of an Aeromonas species was provided by M. Kai, Tokai University, Kanagawa, Japan. A strain of Escherichia coli harboring plasmid pME9 with transposon Tn5-751 was obtained from S. P. Howard, University of Regina, Regina, Saskatchewan, Canada. The transposon Tn5-751 had two antibiotic resistance genes coding for kanamycin and trimethoprim. Rifampin- and streptomycin-resistant spontaneous mutants of A. hydrophila were prepared during these studies. Suicide vector pJQ200KS, which contained a P15A origin of replication, a sacB gene from Bacillus subtilis, and a gentamicin resistance gene, was obtained from M. K. Hynes, The University of Calgary, Calgary, Alberta, Canada (36). E. coli S17-1, with streptomycin and trimethoprim resistance and lysogenized with pir (20, 36), was from S. J. Libby, North Carolina State University, Raleigh, N.C. Plasmid pMW1823, another suicide vector, with a chloramphenicol resistance gene from pACYC184, an origin of replication from plasmid pSC101, and the mob region from plasmid pJM703.1, was provided to us by V. L. Miller, Washington University School of Medicine, St. Louis, Mo. Plasmid pXHC95 contained a 2.8-kb BamHI DNA fragment from A. hydrophila chromosomal DNA and harbored the act gene. This plasmid had an ampicillin resistance gene and was propagated in E. coli XL1-Blue cells. Plasmid pUC4K contained a 1.2-kb kanamycin resistance gene cassette, which represented a portion of the transposon Tn903 (Pharmacia Biotech Inc., Piscataway, N.J.). The E. coli clones with recombinant plasmids, as well as Aeromonas cultures, were stored in Luria-Bertani (LB) medium containing 25% (vol/vol) glycerol at 70°C. The concentrations of antibiotics used to grow cultures were as follows: 50 µg of ampicillin per ml, 40 µg of rifampin per ml for transposon mutants and 300 µg of rifampin per ml for isogenic mutants, 25 µg of streptomycin per ml, 25 µg of trimethoprim per ml, 50 µg of kanamycin per ml, 15 µg of gentamicin per ml, and 20 µg of chloramphenicol per ml.

Transposon mutagenesis. The transposon Tn5-751 from plasmid pME9 in E. coli was delivered to Aeromonas by conjugation as previously described (20, 38). Briefly, both E. coli(pME9) and streptomycin-resistant A. hydrophila SSU were grown under static conditions at 37°C overnight. The cultures were mixed (5 ml each) at a concentration of 8 × 10⁶ cells/ml, centrifuged (4,000 × g for 10 min), resuspended in 200 µl of LB medium, and plated on LB plates without any antibiotic pressure. After 4 h of incubation at 37°C, the culture was removed from the plate and various dilutions (10⁴ to 10⁹) of the sample were plated on LB plates with streptomycin, kanamycin, and trimethoprim. The cultures were identified as Aeromonas by a positive oxidase test to differentiate them from E. coli and by an automated identification system (Vitech, St. Louis, Mo.) in the Clinical Microbiology Laboratory, The University of Texas Medical Branch, Galveston. Further, dot blot hybridization (6) was utilized to differentiate toxin-bearing Aeromonas from nontoxigenic E. coli used in the conjugation experiment. The denatured total DNA samples were applied to a nitrocellulose membrane under vacuum in a dot blot apparatus (Bio-Rad, Hercules, Calif.). The filters were dried and baked at 80°C for 2 h. The blots were prehybridized and hybridized by using Quikhyb (Stratagene, La Jolla, Calif.) at 68°C as described by the manufacturer. The probes used included a 1.4-kb DNA fragment containing the full-length act gene and a 439-bp DNA fragment representing the 5' end of the toxin gene (14) and were labeled with [-³²P]dCTP (ICN, Irvine, Calif.) by using a random primer kit (GibcoBRL, Gaithersburg, Md.). The filters were washed at 68°C in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) (pH 7.0) plus 0.1% sodium dodecyl sulfate (SDS) for 1 h and then in 1× SSC plus 0.1% SDS for 30 min at 68°C. The blots were exposed to the X-ray film at 70°C for 2 to 12 h.

Construction of isogenic mutants of Aeromonas via double-crossover recombination. Recombinant plasmid pXHC95 (14) was used to construct isogenic mutants (Fig. 1). The BstEII restriction enzyme was used to digest this plasmid and to linearize it, since there was only one BstEII site within the 2.8-kb BamHI DNA insert (Fig. 1). There were no BstEII sites in the pBluescript expression vector used to construct recombinant plasmid pXHC95. The ends of the 2.8-kb DNA fragment were made blunt by using a PCR polishing kit (Stratagene). Subsequently, a 1.2-kb kanamycin gene cartridge was isolated from plasmid pUCK4 by restriction with enzyme PstI, which bordered the kanamycin gene cassette (Fig. 1). An appropriate-sized DNA fragment was excised from a 0.8% agarose gel, extracted with phenol-chloroform, ethanol precipitated (6), and finally purified with a GeneClean II kit (Bio 101, Vista, Calif.). The ends of the kanamycin gene cassette were made blunt and ligated to the blunt-ended 2.8-kb BstEII DNA fragment by using T4 DNA ligase (Promega, Madison, Wis.).

View larger version (30K): [in this window] [in a new window]   FIG. 1.   Flow diagram showing construction of various recombinant plasmids. Recombinant plasmid pXHC95 contained a 2.8-kb BamHI DNA fragment from chromosomal DNA of A. hydrophila SSU with the act gene (II). A kanamycin resistance gene cartridge from plasmid pUCK4 was introduced within the act gene to generate recombinant plasmid pXHC97.1 before ligation of the truncated act gene in the suicide vector pJQ200 to generate plasmid pXHC97.2 (III). A 2.8-kb BamHI DNA fragment containing the act gene also was subcloned in another suicide vector, pMW1823, to generate plasmid pXHC97.3 (I) for the purpose of generating a revertant of A. hydrophila with parental biological activity of Act. MCS, multiple cloning site.

Conjugation between A. hydrophila and E. coli. The recombinant E. coli S17-1(pXHC97.2) (Fig. 1) was conjugated with rifampin-resistant Aeromonas, as described previously for development of transposon mutants (20, 38), and the transconjugants were selected on LB agar plates with rifampin, kanamycin, and gentamicin to select for single-crossover transconjugants or with rifampin, kanamycin, and 5% sucrose to select for double-crossover transconjugants. The integration of the recombinant suicide vector into the chromosome of Aeromonas was confirmed by Southern blot analysis.

Northern blot analysis. Total RNA was isolated from various transposon mutants and wild-type A. hydrophila by using the total RNA isolation kit from Qiagen. The RNA samples (6 µg) were subjected to electrophoresis on a 1.2% formaldehyde-agarose gel with 1× MOPS buffer (0.2 M MOPS [morpholinepropanesulfonic acid] [pH 7.0], 0.05 M sodium acetate, 0.01 M EDTA, pH 8.0) (6). The RNA was transferred to a nylon membrane (GibcoBRL), and after baking, the filters were prehybridized, hybridized, and washed as described for dot blot analysis. The amount of RNA in each lane was quantitated by densitometer scanning (Applied Imaging, Pittsburgh, Pa.) of 23S and 16S rRNA bands after ethidium bromide staining of the gel. All of the reagents used for Northern blot analysis were treated with diethylpyrocarbonate.

Measurement of biological activities. The biological activity of Act was determined by cytotoxic, hemolytic, and enterotoxic assays.

(i) Cytotoxic assay. The culture filtrates, cell lysates, and cell membranes from various Aeromonas cultures (grown for 18 h in LB medium at 37°C) were diluted twofold with PBS and added to Chinese hamster ovary (CHO) cells. After 18 to 20 h of incubation at 37°C in the presence of 5% CO₂, the cytotoxic activity was recorded. The cytotoxic unit was defined as the reciprocal of the highest dilution of the toxin demonstrating 50% destruction of CHO cells (39).

(ii) Hemolytic assay. A volume of 100 µl of PBS was added to each of the wells of a 96-well microtiter plate. Next, 100 µl of twofold-diluted toxin preparation was added, followed by 100 µl of 2% rabbit erythrocytes (39). The plate was incubated at 37°C for 1 h and observed for hemolytic activity. The hemolytic unit was defined as the reciprocal dilution of Act demonstrating 50% lysis of rabbit erythrocytes. To demonstrate that the residual cytotoxic and hemolytic activities in various transposon mutants were indeed due to Act, antibodies specific for Act were used to neutralize Act's biological effects. The toxin preparations were mixed with antibodies and incubated at 37°C for 1 h before being placed on the erythrocytes and CHO cells. The ability of these antibodies to neutralize the biological activity of Act then was evaluated. Preimmune serum was used as a negative control in neutralization experiments.

(iii) Enterotoxic assay. Outbred Swiss-Webster mice (25 to 30 g; Taconic Farms, Inc., Germantown, N.Y.) were anesthetized with halothane (River Edge, N.J.). An abdominal incision was made, and a single 5-cm loop was constructed, as previously described with 00 silk suture (35). A 100-µl test sample was injected into the loop. After 6 h of observation, the animals were euthanized by cervical dislocation, and the intestinal loops were removed. The amount of luminal fluid was measured and expressed as microliters per centimeter.

LD₅₀ determination. The wild-type A. hydrophila, its transposon and isogenic mutants, and the revertant were grown in LB medium and centrifuged, and the cells were washed twice with PBS. The bacterial cells were resuspended in PBS and injected intraperitoneally (i.p.) into Swiss-Webster mice (6 to 10 mice per group) at various doses (10⁴ to 10¹⁰ CFU), and the mice were observed for death. The 50% lethal dose (LD₅₀) was determined by the method of Reed and Muench (37).

Electron microscopy. The mouse intestinal loops injected with culture filtrates from wild-type A. hydrophila and its isogenic mutants were removed after 6 h for electron microscopic studies. Small pieces (1 mm) of small intestine were fixed (17), and ultrathin sections were cut with a Sorvall MT-6000 ultramicrotome (RMC, Tucson, Ariz.) and stained with uranyl acetate and lead citrate. The sections were examined and photographed in a model 201 electron microscope (Phillips Electron Optics, Eindhoven, The Netherlands) at 60 kV.

SDS-polyacrylamide gel electrophoresis and Western blot analysis. Culture filtrates from A. hydrophila and its mutants were analyzed by SDS-10 to 12% polyacrylamide gel electrophoresis (30) and Western blot analysis (Bio-Rad) with toxin-specific antibodies developed in mice or rabbits. Goat anti-mouse or goat anti-rabbit immunoglobulin conjugated with alkaline phosphatase or horseradish peroxidase was used as the secondary antibody (diluted 1:3,000). The blots were developed with either an alkaline phosphatase substrate kit (Bio-Rad) or an enhanced chemiluminescence substrate (Pierce, Rockford, Ill.).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Transposon mutagenesis of chromosomal DNA of A. hydrophila SSU. Approximately 4,000 transposon mutants of A. hydrophila were screened for biological activity. Culture filtrates from five mutants displayed drastically reduced biological activities, as measured by hemolytic and cytotoxic assays (Table 1). All of these mutants were confirmed to be Aeromonas. We also performed dot blot hybridization of the total DNAs from wild-type A. hydrophila and its transposon mutants by using a 439-bp act-specific gene probe. The DNAs from these cultures exhibited a positive signal, while total DNA from E. coli did not hybridize with this probe. It was crucial to differentiate Aeromonas from E. coli used in the conjugation before proceeding further, because Aeromonas and E. coli exhibited very similar biochemical profiles. We indeed obtained streptomycin-, kanamycin-, and trimethoprim-resistant colonies with no hemolytic activity, but these were identified as E. coli. The frequency at which these colonies appeared was relatively high (25 to 30%). We also confirmed that the streptomycin-resistant spontaneous mutants of Aeromonas used in the conjugation experiment exhibited biological activity similar to that of wild-type Aeromonas.

View larger version (39K): [in this window] [in a new window]   FIG. 2.   Southern blot analysis of the genomic DNAs from wild-type and transposon mutants of A. hydrophila. The genomic DNA (15 µg) was digested with the SalI and BamHI restriction enzymes and subjected to Southern blot analysis. The probe used was a 1.4-kb XbaI/SalI DNA fragment (32P labeled), which depicts the coding region of the act gene. The blot was prehybridized, hybridized (5 × 106 cpm/ml), and washed as described in Materials and Methods. The lanes contained digested DNAs from the transposon mutants 353 (lane 1), 325 (lane 2), 312 (lane 3), 225 (lane 4), and 42 (lane 5) and from wild-type A. hydrophila SSU (lane 6) as a positive control.

View larger version (57K): [in this window] [in a new window]   FIG. 3.   Northern blot analysis of the total RNAs isolated from the wild-type and transposon mutants of A. hydrophila. The total RNA was isolated by using a total RNA isolation kit (Qiagen). A 439-bp 32P-labeled DNA fragment, which represented the 5' end of the toxin structural gene, was used as a probe in these blots. The blot was prehybridized, hybridized, and washed as described in Materials and Methods. The blots were exposed to X-ray film at 70°C for 12 h. The lanes contained RNAs from mutant cultures 353 (lane 1), 325 (lane 2), 312 (lane 3), 225 (lane 4), and 42 (lane 5) and from wild-type A. hydrophila SSU (lane 6).

Lethality studies with the Act-deficient Aeromonas strain A52 and transposon mutants of A. hydrophila SSU. Isolate A52 of Aeromonas is naturally deficient in the act gene, as confirmed by Southern analysis and biological activity measurements. Both A. hydrophila SSU (Act positive) and Aeromonas strain A52 (Act negative) were injected i.p. into mice to demonstrate the role of Act in Aeromonas-mediated infections. The LD₅₀ of Aeromonas strain A52 was calculated to be 3.9 × 10⁹. However, the LD₅₀ of wild-type A. hydrophila SSU was 2.5 × 10⁷, indicating Act's role in the organism's virulence. The difference in the LD₅₀s between these two cultures was statistically significant (P = 0.01) by the Fisher exact test. The sera from animals that survived the challenge with A. hydrophila SSU contained toxin-specific antibodies; however, the sera from animals challenged with Aeromonas strain A52 were devoid of toxin-specific antibodies as determined by Western blot analysis (data not shown).

Characterization of double-crossover mutants of Aeromonas with an altered act gene. The strategy used to develop an isogenic mutant of Aeromonas is depicted in Fig. 1. The single-crossover transconjugants obtained after conjugation of wild-type, rifampin-resistant A. hydrophila SSU with E. coli S17-1 harboring plasmid pXHC97.2 did not grow in the presence of 5% sucrose, as it induces the sacB gene, coding for levan sucrase, which is lethal to cells when produced in large amounts. The hemolytic activity of these single-crossover mutants was similar to that of wild-type Aeromonas.

View larger version (168K): [in this window] [in a new window]   FIG. 4.   Electron microscopy of intestinal tissues of mice injected with culture filtrates from wild-type Aeromonas and its isogenic mutant. Mouse ligated loops were placed in fixative and cut into 1-mm pieces. Ultrathin sections were stained and photographed in a Philips 201 electron microscope. (A) After administration of Act (contained in culture filtrate) from wild-type Aeromonas, enterocytes were completely destroyed. (B) Mouse loops challenged with culture filtrate from a double-crossover mutant. Normal enterocytes with intact brush borders surround the lumen. Mucus is being emptied from a goblet cell into the lumen in the upper layer of cells. Bars, 1 µm.

Southern blot analysis of the chromosomal DNAs of A. hydrophila SSU and its isogenic mutants. Southern blot analysis of the genomic DNAs of wild-type Aeromonas and its mutants was performed to confirm the identity of isogenic mutants. It is evident from Fig. 5A, lane 6, that a 2.8-kb chromosomal DNA fragment reacted with the act gene probe in wild-type Aeromonas. Total DNA from a single-crossover transconjugant showed bands at 9.3 and at 2.8 kb (Fig. 5A, lane 5). The band at 2.8 kb represented the native act gene of Aeromonas, whereas the band at 9.3 kb contained a truncated act gene and the suicide vector. A fragment of 4.0 kb reacted with the act gene probe in the digested DNAs of the double-crossover mutants (Fig. 5A, lanes 3 and 4), instead of a 2.8-kb fragment observed with the wild-type A. hydrophila (Fig. 5A, lane 6). The size shift was due to insertion of a kanamycin resistance gene cassette in the act gene. When similar DNA samples were probed with a kanamycin gene cassette (Fig. 5B), only the double-crossover and single-crossover mutants showed signals at 4.0 and 9.3 kb, respectively (Fig. 5B, lanes 1 to 4). Genomic DNA from wild-type Aeromonas did not react with this probe (Fig. 5B, lanes 5 and 6). When plasmid pJQ200KS was used as a gene probe, only the digested DNAs from single-crossover mutants reacted (Fig. 5C, lanes 3 and 4). As predicted, no band was detected in the digested DNAs from double-crossover mutants, indicating loss of the suicide vector sequences (Fig. 5C, lanes 1 and 2). Likewise, genomic DNA from wild-type A. hydrophila did not react with this probe (Fig. 5C, lanes 5 and 6).

View larger version (41K): [in this window] [in a new window]   FIG. 5.   Southern blot analysis of the chromosomal DNAs from A. hydrophila SSU and its isogenic mutants. Total DNAs (15 µg) from A. hydrophila and its mutants were digested with the BamHI restriction enzyme and subjected to Southern blot analysis. The probes used were a 439-bp XbaI/SalI DNA fragment, which depicts part of the coding region of the act gene (A), a 1.2-kb kanamycin resistance gene cassette (B), and a 4.9-kb pJQ200 suicide vector (C). The blots were probed and washed as described in Materials and Methods. (A) Digested DNAs from E. coli with suicide vector and truncated act gene (lane 1), double-crossover mutants of A. hydrophila (lanes 3 and 4), a single-crossover mutant of A. hydrophila (lane 5), and wild-type A. hydrophila (lane 6). (B and C) Digested DNAs from double-crossover mutants of A. hydrophila (lanes 1 and 2), single-crossover mutants of A. hydrophila (lanes 3 and 4), and wild-type A. hydrophila (lanes 5 and 6).

Reintegration of the native act gene in the double-crossover mutants with an inactive act gene. The truncated act gene in the double-crossover mutant was replaced with the functionally active act gene by homologous recombination. To perform this experiment, the 2.8-kb BamHI DNA fragment from plasmid pXHC95 was made blunt ended and ligated to the blunt-ended suicide vector pMW1823 at the EcoRI restriction site (Fig. 1). The recombinant plasmid pXHC97.3 was transformed into E. coli S17-1. After the identity of chloramphenicol-resistant recombinant clones was confirmed, the plasmid pXHC97.3 was transferred from E. coli into an isogenic mutant of Aeromonas (rifampin resistant) which contained the truncated act gene. Colonies resistant to rifampin were inoculated on a blood agar plate and observed for a surrounding zone of hemolysis after overnight incubation at 37°C. Aeromonas revertants exhibiting hemolytic activity were obtained at a frequency of 0.05%. The hemolytic activities in the culture filtrates of wild-type and revertant strains of Aeromonas were identical, indicating that the revertant had regained the biological activity of Act (Table 2). All of the Aeromonas mutants exhibited similar growth rates in synthetic M-9 medium (data not shown).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Aeromonas species, like many other bacterial pathogens, secrete a number of extracellular proteins which play important roles in the pathogenesis of disease (3). Hemolysins have been shown to be produced by many gram-negative bacteria, and Welch and Falkow (41) were able to establish a correlation between the hemolytic titer in E. coli and lethality in rats. In contrast, Wright and Morris (42) noted that the cytolysin (with hemolytic and cytotoxic activities) produced by Vibrio vulnificus had a minimal effect on the pathogenesis of V. vulnificus infections.

Asao et al. (4) noted that the hemolysin produced by Aeromonas had multiple biological activities, including hemolytic, cytotoxic, and enterotoxic activities and lethality in mice, similar to the case for Act (14). Chakraborty et al. (9) reported that their aerolysin from A. trota reacted with antibodies to hemolysin isolated by Asao et al. (4). Hirono and Aoki (21) reported another hemolysin from Aeromonas which exhibited minimal homology with aerolysin. By marker exchange mutagenesis, Chakraborty et al. (10) showed that aerolysin-deficient mutants were less virulent in mice than wild-type Aeromonas. Molecular cloning and DNA sequence analysis of our act gene from A. hydrophila (14) revealed that it differed significantly from the aerolysin gene of A. trota and from the hemolysin purified by Asao et al. (4). The differences included the inability of one of the neutralizing Act monoclonal antibodies to react with the two other proteins in Western blot analysis and its failure to neutralize the hemolytic activity of these toxins. Site-directed mutagenesis within the act gene revealed many other differences between Act and aerolysins (16). We also have demonstrated that Act stimulated the chemotactic activity of human leukocytes and inhibited the phagocytic function of mouse phagocytes (27), clearly indicating a role for Act in Aeromonas-mediated infections. Further, in a clinical study, A. hydrophila was isolated as the sole enteropathogen from patients' diarrheal stools and from the ready-to-eat shrimp cocktail that those patients had ingested (2), indicating a definitive epidemiological link between diarrhea and direct exposure to Aeromonas.

The LD₅₀ of Aeromonas strain A52 (Act negative) was almost two logarithmic doses greater than the LD₅₀ of A. hydrophila SSU when injected i.p. However, since we were unaware of the various virulence factors produced by A52 compared to SSU, we opted to generate transposon and isogenic mutants of wild-type A. hydrophila SSU. At present, no oral-challenge models are available for Aeromonas, and therefore, the current model has limitations in mimicking the true disease process in humans. Regardless, this is the first report of a study in which an act gene-deficient mutant was prepared from an authentic strain of A. hydrophila to unequivocally establish the role of Act in Aeromonas-mediated infections in mice after i.p. challenge.

The transposon mutants of A. hydrophila SSU with dramatically reduced biological activity were not lethal to mice at a dose of 5 × 10⁷ compared to the wild type. Southern blot data suggested that transposition might not have occurred within the structural gene for Act in these mutants. Our Northern blot data demonstrated that transcription of the act gene in the transposon mutants was affected (Fig. 3). The exact location of the transposition in these mutants has not been determined and is under investigation. It is plausible that the transposition might have occurred in some regulatory element whose product was essential for the transcription of the act gene. Earlier, Chakraborty et al. (9) used transposon insertions to demonstrate that the DNA sequences flanking the aerolysin structural gene (aerA) in both the 5' (referred to as aerC) and 3' (referred to as aerB) regions in A. trota were important for the expression of the aerolysin gene. However, Howard et al. (24) noted that the expression of their aerolysin gene from A. bestiarum was not affected when Tn5 insertions were introduced immediately downstream of the stop codon for the aerolysin structural gene. Although regulation of the act gene is a subject of intense investigation in our laboratory, at present nothing is known about the act operon in A. hydrophila. It is therefore plausible that transposition in these mutants, although not within the structural gene, may be in the act operon.

Transposition may lead to polar mutations, and the possibility that some other virulence genes might have been affected in the transposon mutants due to a polar effect cannot be ruled out. Therefore, we generated an isogenic mutant of Aeromonas. The frequency of double-crossover events was very low (0.01%). We obtained at a high frequency colonies which acquired sucrose resistance but still were gentamicin resistant. This could have occurred as a result of various types of mutations within the sacB gene (36). Further, the DNA sequences flanking the act gene had to be increased significantly (1.9 kb at the 5' end and 0.9 kb at the 3' end) to obtain double-crossover mutants.

Originally, we removed 63% of the coding region of the toxin by using the BstXI restriction enzyme, which resulted in a flanking 422 bp of the DNA sequence at the 5' end and 179 bp at the 3' end of the act gene (14). Donnenberg and Kaper (15) similarly removed 66% of the eae gene of E. coli and had 519 and 120 bp of flanking DNA sequences in order for the double-crossover event to occur. They reported successful isolation of double-crossover mutants. However, this strategy did not provide us any genuine double-crossover mutants, although single-crossover mutants were obtained. These single-crossover transconjugants were grown without antibiotic selection to the late logarithmic phase, allowing second recombination events to accumulate. Although we obtained the desired phenotype (gentamicin sensitivity and sucrose resistance), we noticed that the suicide vector was indeed not lost and that the biological activity of the toxin remained intact. Thousands of colonies were screened on the blood agar plates for the loss of hemolytic activity, without any success. These data indicated mutations in the gentamicin resistance and sacB genes.

A dramatic difference between the LD₅₀s of the isogenic mutant and wild-type Aeromonas when injected into animals was noted. Even with the construction of an isogenic mutant, it is possible that unlinked mutations might influence the biological effects of Act. We therefore reintroduced the native act gene in the isogenic mutant to restore the biological activity of Act. Indeed, full hemolytic, cytotoxic, and enterotoxic activities were regained by this Aeromonas revertant. Further, the revertant was as virulent in mice as wild-type A. hydrophila SSU, indicating that there was no polar effect in the isogenic mutants. Finally, we have demonstrated that Act was produced during the infection process, since antisera obtained from mice surviving infection with wild-type A. hydrophila and the revertant had Act-specific antibodies. However, the sera from animals injected with the isogenic mutant did not show an Act-specific band in Western blots.

In conclusion, we have demonstrated that elimination of the biological effects of Act by either transposon or marker exchange mutagenesis significantly affected the pathogenicity of Aeromonas in mice. These data were substantiated by using an isolate of Aeromonas that naturally did not produce Act. These observations are very provocative, since Aeromonas increasingly has been isolated from patients with peritonitis and urinary tract infections, and there have been reports in which A. hydrophila has been shown to cause multilobular lung abscesses and a fatal bacteremia with myonecrosis and gas gangrene in a hemodialysis patient treated with deferoxamine (25, 29, 31, 33). The reported isolation of Aeromonas worldwide from 5 to 7% of individuals suffering from gastroenteritis, the presence of this organism in a variety of foods, and the prevalence of Act-related molecules in most Aeromonas isolates examined have resulted in an increased awareness of their association with human infections (8, 11, 12, 18, 22). Overall, our data indicate that Act has an impact on virulence in mice, but further studies will be necessary to clearly correlate this observation with human illness.