Pathogenic Aeromonas hydrophila Serogroup O:14 and O:81 Strains with an S Layer

Abstract

Five autoagglutinating Aeromonas hydrophila isolates recovered from eels and humans were assigned to serogroups O:14 and O:81 of the Sakazaki and Shimada (National Institutes of Health) scheme. They had the following properties in common: positive precipitation after boiling, moderate surface hydrophobicity (salt-aggregation-test value around 1.2), pathogenicity for fish and mice (50% lethal dose, 10^4.61 to 10^7.11), lipopolysaccharides that contained O-polysaccharide chains of homogeneous chain length, and an external S layer peripheral to the cell wall observed by electron microscopy. A strong cross-reactivity was detected by immunoblotting between the homogeneous O-polysaccharide fraction of O:14 and O:81 strains but not between them and the lipopolysaccharide of A. hydrophila TF7 (O:11 reference strain). Outer membrane fractions of these strains contained a predominant 53- to 54-kDa protein which was glycine extractable under low-pH (pH 2.8) conditions and was identified as the surface array protein. The S-layer proteins of the O:14 and O:81 A. hydrophila strains seemed to be primarily different from those previously purified from strains A. hydrophila TF7 and Aeromonas salmonicida A450 on the basis of colony hybridizations with both the structural genes vapA and ahsA. This is the first report of the presence of an S layer in mesophilic Aeromonas strains not belonging to serogroup O:11.

Aeromonas hydrophila, a gram-negative motile rod, is considered one of the most important bacterial pathogens of aquatic animals in temperate areas, as it has been isolated from diseased fish, eels, frogs, and turtles (10, 25, 38, 47). A. hydrophila has been increasingly reported as one of the most common Aeromonas species associated with human intestinal disease (1, 24, 36) and systemic illnesses in immunocompromised patients (14, 15, 45). One group of A. hydrophila strains with high virulence for trout was reported (8, 32). At the same time, Janda and collaborators described an identical group of A. hydrophila, Aeromonas sobria, and Aeromonas veronii biovar sobria strains associated with systemic infections in humans (18, 19, 39). All these Aeromonas strains had a common phenotypic feature, autoagglutination in liquid medium by self-pelleting or by precipitation after boiling (18, 32).

At present, autoagglutinating (AA+) motile Aeromonas cells have been segregated into two subgroups: the first includes strains that belong to a single lipopolysaccharide (LPS) serogroup (O:11) (7, 32, 39), and the second includes all non-O:11-autoagglutinating strains which belong to diverse O-antigen LPS serogroups (21). These previous studies stated that only the O:11 autoagglutinating Aeromonas strains shared several additional features, including enhanced virulence for animals (50% lethal dose in the range of 10^4.50 to 10^7.43), LPS containing O-polysaccharide chains of homogeneous chain length (7), and the presence of a crystalline surface array protein in the form of an S layer which lies peripheral to the cell wall (8, 39). The S layers of motile O:11 Aeromonas strains are composed of subunits of a single surface array protein of around 52 to 55 kDa molecular mass (8, 21, 22). Moreover, they are very similar morphologically to the A. salmonicida surface array but appear to be unrelated genetically (35). These protein sacs are strategically positioned to interact with the tissues and body fluids of the host and to influence the outcome of a host-parasite interactions (4). Thus, in the aeromonads generally but mainly in the species A. salmonicida, the S layer facilitates association with macrophages (48), binds porphyrins and immunoglobulins (20, 40), and contributes to the organism's resistance to the bactericidal activity of both nonimmune and immune serum (30, 33).

The non-O:11 autoagglutinating motile Aeromonas isolates studied, which belonged to diverse serogroups (O:3, O:22, O:34, and O:36), displayed low virulence for animals (50% lethal dose in the range of 10^7.68 to 10^8.50), showed an LPS composed of O-polysaccharide side chains of heterogeneous lengths, and lacked the surface array protein, i.e., the S layer (21, 39). The present report describes for the first time the presence of an S layer in pathogenic non-O:11 autoagglutinating Aeromonas isolates which belong to serogroups O:14 and O:81.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains used in this study are listed in Table 1. The source, phenotypic identification, and autoagglutinating phenotype of each isolate have been described previously (9, 11, 16, 28, 50). Cultures of all strains were grown on tryptone soy agar (TSA) (Oxoid) for 18 h at 28°C unless otherwise stated.

TABLE 1.

Major characteristics of the Aeromonas strains used^b

Strain (origin)	Biochemical identificationa	16S ribosomal DNA RFLP identification	Phenotype			O serogroup		50% lethal dose (log10 CFU/g)
			SP	PAB	SAT	NIPHEH system	NIH system	Eel	Trout	Mice
TW1 (freshwater, Spain)	A. hydrophila	A. hydrophila	−	+	1.4	O:19	O:14	6.67	ND	ND
A19 (European eel, Spain)	A. hydrophila	A. hydrophila	−	+	1.2	O:19	O:14	6.47	4.61	6.15
E37 (European eel, Spain)	A. hydrophila	A. hydrophila	−	+	1.2	O:19	O:14	7.11	ND	6.38
E40 (European eel, Spain)	A. hydrophila	A. hydrophila	−	+	1.0	O:19	O:14	6.30	ND	ND
AH290c (human, The Netherlands)	A. hydrophila	ND	−	+	ND	O:19	O:81	ND	ND	ND
TF7d (trout, Canada)	A. hydrophila	ND	+	+	ND	ND	O:11	ND	4.50e	5.85
A450 (brown trout, France)	A. salmonicida	ND	+f	+	0.02f	ND	ND	ND	ND	ND

^aData from Esteve (9) and Valera and Esteve (50).

^bSP, self-pelleting; PAB, precipitation after boiling; NIPHEH, National Institute of Public Health and Environmental Hygiene, Bilthoven, The Netherlands; NIH, National Institute of Health, Tokyo, Japan; ND, not determined; SAT, salt aggregation test (values represent the lowest molarity of ammonium sulfate yielding a strong aggregation of bacteria).

^cReference strain for serogroup O:19 (National Institute of Public Health and Environmental Hygiene, Bilthoven, The Netherlands) (16).

^dThis strain belongs to serogroup O:11 (National Institute of Health, Tokyo, Japan) (23).

^eData from Mittal et al. (32).

^fData from Olivier (37).

Identification by 16S rDNA restriction fragment length polymorphisms.

All strains which had been previously identified by biochemical methods (9, 50) were now identified on the basis of the restriction fragment length polymorphism patterns of PCR-amplified 16S rRNA genes following a previously described method (5, 13).

Virulence for fish and mice.

The virulence of selected strains was measured by their mean lethal dose (50% lethal dose), evaluated according to the method of Reed and Muënch (41). The virulence trials were performed as previously described (10, 12).

Serological testing.

All motile Aeromonas strains had been previously tested by the O-serogrouping system of the National Institute of Public Health and Environmental Hygiene (NIPHEH), Bilthoven, The Netherlands (11, 16). Strains were grown on TSA slants overnight at 30°C, harvested with phosphate-buffered saline (>10⁹ cells/ml), and heated for 1 h at 100°C. After being heated, 20 μl of the boiled cell suspensions (thermostable O antigen of the strains) was mixed with 20 μl of each specific rabbit antiserum (O:1 to O:30; NIPHEH system) in ceramic rings on agglutination glass sides. The mixtures were rotated for 2 min, and the degree of agglutination (0 to 2+) was recorded. Two negative controls were used, boiled cell suspensions mixed with phosphate-buffered saline and boiled cell suspensions mixed with rabbit serum obtained from nonimmunized animals.

In addition, Aeromonas isolates were serotyped by the tube agglutination method (44) with polyvalent antisera at the National Institute of Health (Tokyo, Japan) by E. Arakawa. This typing scheme included antisera specific for O:1 to O:97, as the O-serogrouping system of Sakazaki and Shimada (42) has recently been extended (34).

Electron microscopy.

To determine whether individual Aeromonas isolates possessed a surface layer, negatively stained sections of each bacterium were prepared as previously described (30). In addition, negative staining (2% phosphotungstic acid) of glycine extracts from S-layer-containing strains was also observed by transmission electron microscopy.

Hydrophobicity test.

Selected strains were evaluated for their relative cell surface hydrophobicity by the salt aggregation test (27). Briefly, strains were grown overnight on TSA, harvested, and washed twice in 0.002 M phosphate buffer (pH 6.8) prior to adjustment to an A₆₆₀ of 1.0 yielding, 1 × 10⁹ to 3 × 10⁹ CFU/ml. A 10-μl portion of this suspension was mixed with a 25-μl volume of various molarities of (NH₄)₂SO₄ (2.0 to 0.2 M; a total of 10 different molarities) on glass agglutination slides containing 14-mm rubber rings. After the mixing, the slides were rotated at 80 rpm for 2 min and read against a black background. The salt aggregation test value was defined as the lowest molarity of ammonium sulfate which caused a strong aggregation of bacteria.

Preparation of LPSs and OMPs.

LPS extractions were prepared by a modification of the procedure of Hitchcock and Brown (17). Whole-cell suspensions obtained from 18-h-old cultures in tryptone broth (1% [wt/vol] tryptone, 1% [wt/vol] NaCl, pH 7.2) were centrifuged, and the dried pellet was suspended in 50 μl of phosphate-buffered saline, mixed with double concentrated electrophoresis sample buffer (26) at a ratio of 1:1, and boiled for 10 min. Finally samples were digested with 30 μl of proteinase K (0.25% [wt/vol] protease type XI; Sigma-Aldrich Corporation, St Louis, Mo.) at 60°C for 75 min. Outer membrane proteins (OMPs) were obtained by incubating membrane suspensions with 3% Sarkosyl in 20 mM Tris-HCl buffer (pH 8.0) for 20 min at room temperature, as previously described (11).

Isolation of S-layer sheets.

The S-layer sheet material was obtained by a modification of the procedure of Dooley and Trust (8). Cells were grown overnight in 1,000 ml of Luria broth (LB) with agitation (200 rpm), harvested by centrifugation (12,000 × g, 20 min), and washed twice in 20 mM Tris-HCl (pH 8.0). They were suspended in 100 ml of 0.2 M glycine HCl (pH 2.8) and stirred at 4°C for 30 min. The cells were removed by a single centrifugation at 12,000 × g for 20 min. The S-layer sheet material was collected by centrifugation at 40,000 × g for 60 min, suspended in 500 μl of 20 mM Tris-HCl (pH 8.0), and frozen at −20°C.

Electrophoresis.

Sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) was performed by the procedure of Laemmli (26), as previously described (11). The separated LPSs were visualized by silver staining, as previously described (49). The separated protein bands were visualized by Coomassie brilliant blue staining (11).

Antisera.

Anti-O11 LPS serum was obtained and assayed as previously described for other LPSs (29). Anti-purified S-layer antiserum was obtained and assayed as previously described (31). Polyclonal antisera against A. hydrophila cells of strains A19 and AH290 were obtained as previously described (43).

Western immunoblotting.

After SDS-PAGE, immunoblotting was carried out by transfer to polyvinylidene difluoride membranes (Millipore Corp., Bedford, Mass.) at 1.3 A for 1 h in the buffer of Towbin et al. as previously described (30, 43). The membranes were then incubated sequentially with 1% bovine serum albumin (Sigma-Aldrich Corporation), specific antiserum (1:500), alkaline phosphatase-labeled goat anti-rabbit immunoglobulin G (Bio-Rad Laboratories Inc.), and 5-bromo-4-chloro-indolylphosphate disodium-nitroblue tetrazolium (Sigma-Aldrich Corporation). Incubations were carried out for 1 h, and washing steps with 0.05% Tween 20 in phosphate-buffered saline were included after each incubation step.

Colony blot hybridizations.

Strains were inoculated onto LB agar plates. After 5 h of incubation at 37°C, each plate was maintained at 4°C for 30 min and then a nylon membrane (Boehringer Mannheim Biomedical Products) was placed over the surface of the plate. The membrane was removed after 1 min and air dried. The membrane was then sequentially placed with the colony side up on a pad of several absorbent filter papers soaked with 10% SDS (3 min), denaturing solution (0.5 M NaOH, 1.5 M NaCl) (15 min), neutralizing solution (1.5 M NaCl, 1 M Tris-HCl, pH 7.4) (15 min), and 2× SSC buffer (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) (pH 7.0) (10 min). DNA material was fixed to the membrane by UV cross-linking by placing the side of the membrane containing the colony blot down on a transilluminator for 7 min. The membrane was then placed in a 0.2% proteinase K solution in 2× SSC buffer at 37°C for 60 min to eliminate cellular debris. For hybridization, structural genes vapA (3) and ahsA (46) were used as probes. Probe labeling, hybridization, and detection were carried out with the enhanced chemiluminescence labeling and detection system (Amersham) according to the manufacturer's instructions.

RESULTS

Major characteristics of the Aeromonas strains used.

A. hydrophila strains TW1, A19, E37, and E40, which had been previously identified by biochemical methods (9, 50), were identified by 16S ribosomal DNA restriction fragment length polymorphisms and displayed an autoagglutinating phenotype (AA⁺) (Table 1). The 18-h-old static cultures of these strains in brain heart infusion broth (Oxoid) settled after boiling for 1 h (PAB⁺), but none of them displayed self-pelleting (SP) properties (Table 1). Their cell surface was relatively hydrophobic, as measured by the salt aggregation test. These strains presented salt aggregation test values ranging from 1.0 to 1.4 M ammonium sulfate (Table 1). The identical pattern (PAB⁺ and SP⁻) was observed for the reference strain A. hydrophila AH290, but strains A. hydrophila TF7 (serogroup O:11) and A. salmonicida A450 were PAB⁺ and SP⁺ (Table 1).

Table 1 shows the O-serogrouping of these motile aeromonads obtained with the NIH and NIPHEH systems (16, 34, 42). A. hydrophila strains TW1, A19, E37, and E40 reacted with the O:19 (NIPHEH system) and the O:14 (NIH system) polyclonal antisera, respectively. However, the reference strain of serogroup O:19 (NIPHEH system), which is AH290 (16), was assigned to serogroup O:81 by the NIH system (Table 1).

A. hydrophila strains TW1, A19, E37, and E40 were virulent for fish and mice. The mean lethal dose observed for fish (eel and trout) was in the range of 10^4.61 to 10^7.11 whereas that observed for mice was around 10⁶ (Table 1).

Structural properties of Aeromonas strains used.

LPSs (proteinase K-digested whole-cell lysates) from aeromonads were electrophoresed on SDS-containing gels and analyzed by silver staining (Fig. 1A). The LPS profiles of all A. hydrophila strains tested displayed a slow-migrating band, a small number of well-resolved faster-migrating bands, and the fast-migrating lipid A core oligosaccharide fraction (Fig. 1A). This pattern has previously been referred to an LPS composed of homogeneous-length O-polysaccharide side chains (type A) (21).

FIG. 1.

(A) Silver stain of LPS on SDS-12% PAGE from autoagglutinating Aeromonas strains belonging to different serogroups: O:11 (lane 1), O:14 (lanes 2, 3, 5, and 6), and O:81 (lane 4). Each lane contains about 10 μg of LPS. Shown are strains TF7 (lane 1), TW1 (lane 2), E40 (lane 3), AH290 (lane 4), A19 (lane 5), and E37 (lane 6). (B) SDS-12% PAGE of OMP profiles of serogroup O:14 and O:81 strains. Each lane contains about 30 μg of protein. Shown are strains TW1 (lane 1), A19 (lane 2), E37 (lane 3), E40 (lane 4), and AH290 (lane 5).

The SDS-PAGE OMP profiles of our strains as well as strain AH290 are shown in Fig. 1B. All these strains showed a major protein of 50 kDa, which was poorly resolved in 12% acrylamide gels, appearing as a large smear. We proceeded to remove this protein from the cells by washing with a low-pH buffer, as was previously described for mesophilic Aeromonas strains from serogroup O:11. Initially, cells were treated with 0.2 M glycine-HCl (pH 4.0) for 15 min (8), but S-layer material was collected from strain TF7 (serogroup O:11) (Fig. 2, lane 1) but not from the other A. hydrophila strains. Nevertheless, a surface array protein was removed from TW1, A19, E37, E40, and AH290 when the experimental conditions were modified by decreasing the pH of the buffer (0.2 M glycine-HCl) to pH 2.8 and increasing the time of treatment (30 min). Low-pH extracts of these A. hydrophila cells showed no signs of cellular lysis and were composed predominantly of a single protein of 53 to 54 kDa (Fig. 2, lanes 2 to 6).

FIG. 2.

SDS-12% PAGE of glycine hydrochloride extraction of surface array protein from autoagglutinating Aeromonas strains belonging to different serogroups: O:11 (lane 1), O:14 (lanes 2, 3, 4 and 6), and O:81 (lane 5). Shown are strains TF7, 40 μg of protein (lane 1); TW1, 40 μg of protein (lane 2); E40, 20 μg of protein (lane 3); A19, 40 μg of protein (lane 4); AH290, 8 μg of protein (lane 5); E37, 30 μg of protein (lane 6); and protein standards (119, 98, 52, 37, 30, 22, and 8 kDa) (lane 7).

The antigenic cross-reactivity between the LPS from the autoagglutinating A. hydrophila strains was examined by immunoblotting (Fig. 3). SDS-PAGE-separated LPS from strains TF7, A19, and E37 displayed relative cross-reactivity with antiserum against purified LPS from strain TF7 (anti-O.11) (Fig. 3A, lanes 1, 3, and 4), whereas no reaction was detected for the LPS of strains AH290, TW1, and E40 with the same antiserum (Fig. 3A, lanes 2, 5, and 6). The results in Fig. 3B show a strong immunoblot reaction obtained when SDS-PAGE-separated LPS from strains A19, E37, TW1, E40, and AH290 (reference strain) reacted with antiserum prepared in rabbits to A. hydrophila AH290 heat-killed cells. Identical results were obtained when the immunoblots reacted with anti-A19 serum (Fig. 3C). In addition, a negative reaction was observed when LPS from strain TF7 reacted with anti-AH290 or anti-A19 sera (Fig. 3B, lane 6; Fig. 3C, lane 4).

FIG. 3.

Immunochemical analysis of the LPSs from autoagglutinating Aeromonas strains. (A) Western blot analysis of LPSs reacted with antiserum prepared against purified LPS from strain TF7 (serogroup O:11). The dilution of antiserum used was 1:500. Shown are TF7 (lane 1), AH290 (lane 2), A19 (lane 3), E37 (lane 4), TW1 (lane 5), and E40 (lane 6). (B) Western blot analysis of LPSs reacted with antiserum prepared against heat-killed cells of strain AH290. The dilution of antiserum used was 1:500. Shown are A19 (lane 1), E37 (lane 2), TW1 (lane 3), E40 (lane 4), AH290 (lane 5), and TF7 (lane 6). (C) Western blot analysis of LPSs reacted with antiserum prepared against heat-killed cells of strain A19. The dilution of antiserum used was 1:1,000. Shown are A19 (lane 1), E37 (lane 2), AH290 (lane 3), TF7 (lane 4), TW1 (lane 5), and E40 (lane 6).

On the other hand, strains TW1, A19, E37, E40, and AH290, which belong to serogroups O:14 and O:81 (NIH system) (Table 1), clearly possessed an extra layer peripheral to the cell wall (Fig. 4A and C). Moreover, the surface array protein present in low-pH extracts of these A. hydrophila strains disposed in sheets following a tetragonal pattern (Fig. 4B and D). As expected, we also found an S layer on the reference strain TF7 (serogroup O:11) (Fig. 4E and F).

FIG. 4.

Presence of S layer in autoagglutinating Aeromonas strains belonging to different serogroups: O:14 (A, B), O:81 (C, D) and O:11 (E, F). (A) External layer on strain A19. (B) Negative stain of glycine extracts obtained from strain A19. (C) External layer on strain AH290. (D) Negative stain of glycine extracts obtained from strain AH290. (E) External layer on strain TF7. (F) Negative stain of glycine extracts obtained from strain TF7. Bar: 50 nm (A and C), 100 nm (E), and 200 nm (B, D, F).

Relatedness among surface array proteins of Aeromonas strains used.

The distribution of the surface array protein genes of A. hydrophila TF7 and A. salmonicida A450 among our autoagglutinating and S-layer-positive strains was assessed by colony hybridization. The most striking finding was the lack of hybridization with the structural genes for the Aeromonas S layers, vapA (3) and ahsA (45), in our O:14 and O:81 A. hydrophila strains, although these isolates possessed an S layer (Table 2). As expected, colony hybridization assays detected the presence of the ahsA and vapA genes in A. hydrophila TF7 and A. salmonicida A450 (Table 2).

TABLE 2.

Relatedness among surface array proteins (SAPs) of Aeromonas strains used

Species or serogroup (strain[s])	Size of SAP (kDa)	Presence of cross- reactive epitopes in the SAP proteins		Hybridization for the gene encoding:
		Anti-TF7 SAP	Anti-A450 SAP	AhsA	VapA
A. hydrophila O:14 (TW1, A19, E37, E40)	54	−	+	−	−
A. hydrophila O:81 (AH290)	53	NDa	−	−	−
A. hydrophila O:11 (TF7)	52	+	−	+	−
A. salmonicida (A450)	48	−	+	−	+

^aND, not determined.

The antigenic diversity of the S-layer proteins produced by these Aeromonas strains was examined. SDS-PAGE-separated surface array proteins of A. hydrophila and A. salmonicida strains were analyzed with antiserum prepared in rabbits to several SDS-denatured surface array proteins (Table 2). Western blot analysis with antiserum to the S-layer protein of reference strain TF7 (serogroup O:11) showed that the S-layer proteins of A. hydrophila strains of serogroup O:14 (NIH system) did not share cross-reactive epitopes with the similar protein of strain TF7 (Table 2). In contrast, the S-layer proteins of these O:14 strains displayed reactivity by Western blot analysis with the antiserum against the A-layer protein of A. salmonicida A450 (Table 2 and Fig. 5, lanes 2, 3, 4, and 5). Strains TF7 and AH290 showed surface array proteins that were unable to react with this antiserum against the A-layer protein of A. salmonicida A450 (Table 2 and Fig. 5, lanes 7 and 8).

FIG. 5.

Antigenic relatedness among the S-layer proteins of A. hydrophila strains (lanes 2-5, 7, 8) and the A. salmonicida strain (lane 6). (A) SDS-PAGE analysis of S-layer proteins stained by Coomassie blue. Each lane contains 20 μg of protein. (B) Western blot analysis of S-layer proteins reacted with antiserum prepared against SDS-denatured A-layer protein of strain A. salmonicida A450. The dilution of serum used was 1:500. Shown are protein standards (97, 67, 43, 30, 20, and 14 kDa) (lane 1), A19 (lane 2), E37 (lane 3), E40 (lane 4), TW1 (lane 5), A450 (lane 6), AH290 (lane 7), and TF7 (lane 8).

DISCUSSION

Until now, the possession of a peripheral proteinaceous surface array had been related to A. salmonicida (A layer) and to several motile Aeromonas species (S layer) whose strains had in common an LPS type belonging to serogroup O:11 of the NIH system (21, 22, 23, 39). Moreover, all previous studies had stated that non-O:11 autoagglutinating motile Aeromonas strains lack the surface array protein, but these reports included mesophilic Aeromonas strains from serogroups O:3, O:22, O:34, and O:36 of the NIH system (21, 39). The present report indicates the presence of an S layer in autoagglutinating (AA⁺) Aeromonas hydrophila strains which belong to serogroups O:14 and O:81 of the NIH system (34, 42). These pathogenic isolates have been accurately identified to the species level by numerical taxonomy (9, 50) and restriction fragment length polymorphism patterns of PCR-amplified 16S rRNA genes (genotypic study), including as a reference most of the recently proposed Aeromonas species (5, 13).

The distribution of an S layer in the genus Aeromonas is strain and not species specific, although many A. salmonicida isolates possess this extra layer (8, 18, 21, 22, 37, 39). However, all S-layer-positive Aeromonas strains have in common the presence of an LPS containing homogeneous-chain-length O-polysaccharides in their cell wall (6, 7, 21, 22, 37). Regarding this, our O:14 and O:81 S-layer-positive A. hydrophila strains also contained an LPS with homogeneous-length O-polysaccharide side chains. This is the first report describing the presence of this kind of LPS in non-O:11 autoagglutinating Aeromonas isolates, as this morphological feature has so far been associated with motile Aeromonas isolates of serogroup O:11 (7, 21, 22, 39).

The O-antigen polysaccharides on LPS from O:14 and O:81 A. hydrophila strains displayed strong antigenic cross-reactivity both by slide agglutination and immunoblotting among them. In fact, the A. hydrophila AH290 strain, which belonged to O:81 serogroup (NIH system) (34, 42), and is the reference strain of serogroup O:19 (NIPHEH system) (16) prompted us to previously assigned all our O:14 A. hydrophila strains to serogroup O:19 by with the NIPHEH O-serotyping scheme (11). Interestingly, the cross-reactions observed between these LPSs (somatic antigens) have been described previously by Shimada and Kosako (44), who stated that somatic antigens O:14 (NIH system) (34, 42) and O:19 (NIPHEH system) (16) presented an a,b-a,c type relationship. In contrast, a null or weak antigenic cross-reactivity was essentially obtained among LPSs of O:14 and O:81 strains and that of the strain TF7 which belongs to serogroup O:11. These findings indicate that our A. hydrophila strains with an homogeneous LPS isolated from eels and humans possess somatic antigens which are different to those of the serogroup O:11.

An external S layer peripheral to the cell wall was observed in all O:14 and O:81 A. hydrophila strains by thin-section electron microscopy. Regarding OMP analysis, these A. hydrophila strains show a major protein of 52 to 53 kDa, which was poorly resolved in 12% acrylamide gels, appearing as a large smear. This effect was possibly due to the presence of the O-antigen LPS, which has been shown to migrate at around the same area of the gel, promoting the smear in the protein resolution (7). Removal of the 52- to 53-kDa protein from the cell surface was only possible by pH 2.8 glycine extraction, indicating a stronger LPS-S layer interaction than that described in A. hydrophila TF7.

It is worth noting that LPS with homogeneous O-side chains is mostly important for anchoring the surface array in A. salmonicida cells but not in A. hydrophila TF7 (2, 8). In fact, we observed cross-reactive epitopes between the S-layer protein of our O:14 A. hydrophila strains and that of the A. salmonicida A450 S layer, but not between them and the S layer of A. hydrophila TF7 (serogroup O:11). On the other hand, tetragonal S-layer material in the form of intact sheets was readily seen during low-pH extraction of whole cells from our strains, as reported by other authors in several S-layer-positive motile aeromonads (8, 21, 39). These S-layer sheets were easily isolated by centrifugation, and SDS-PAGE analysis showed that the predominant protein in the S-layer sheet material was the 52- to 53-kDa protein. This size is consistent with those reported for the surface array proteins of aeromonads, including the 48- to 53-kDa S-layer proteins of A. salmonicida (37) and the 52- to 55-kDa S-layer proteins of motile Aeromonas species always belonging to serogroup O:11 (8, 21, 22). Furthermore, the S-layer proteins of the O:14 and O:81 A. hydrophila strains seemed to be primarily different from those previously purified from strains A. hydrophila TF7 (46) and A. salmonicida A450 (3) on the basis of colony hybridizations with both structural genes vapA (3) and ahsA (46).

The S-layer-positive A. hydrophila strains of serogroups O:14 and O:81 displayed common phenotypic features so far assigned to O:11 strains (18, 32). They were etiological agents of hemorrhagic disease in fish (10), presented enhanced virulence for fish and mice (50% lethal dose of 10^4.61 to 10^7.11), possessed both LPS composed of homogeneous-length O-polysaccharide side chains and surface array protein (S layer), and showed an autoagglutinating (AA⁺) phenotype by precipitation after boiling (PAB⁺). At present, the autoagglutinating phenotype refers to two different features presented by static cultures of motile Aeromonas strains in BHI broth; one is the spontaneous self-pelleting of bacterial cells during growth, and the other is the pelleting of bacterial cells after boiling 18-h-old cultures for 1 h (precipitation after boiling) (18, 32).

It is worth noting that the first reports on the cell surface characteristics of virulent motile Aeromonas strains whose LPS contained O-polysaccharide chains of homogeneous chain lengths (7, 32) used the term “autoagglutination in broth culture” to name the phenotype PAB⁺, as they compared this phenotype with the characteristic self-pelleting observed in A. salmonicida cultures (37). Later, some reports pointed out the existence of self-pelleting (SP⁺) motile Aeromonas strains (18, 39), which were assigned to the autoagglutinating phenotype even though most SP⁺ strains did not present homogeneous LPS or enhanced virulence for animals (18, 39). At this point, the use of the autoagglutinating phenotype for screening for S-layer-containing motile Aeromonas strains is controversial because most SP⁺ strains do not have this extra layer (21). Nevertheless, all S-layer-positive motile Aeromonas isolates of serogroups O:11, O:14 and O:81 (NIH system) studied were autoagglutinating strains by precipitation after boiling, and so they mostly had the common phenotype SP⁻ and PAB⁺ (Table 1) (18, 28, 32, 39). Thus, precipitation after boiling is, in our opinion, the proper marker to search for virulent and S-layer-positive motile Aeromonas strains from epizootic and clinical samples.