Vibrio anguillarum serogroup O3 and V. anguillarum-like serogroup O3 cross-reactive species—comparison and characterization

Journal of Applied Microbiology 1997,82, 21 1-21 8

Vibrio anguillarumserogroup 0 3 and V. anguillarum-like serogroup 0 3 cross-reactive species-comparison and characterization T. Tiainen, K. Pedersen and J.L. Larsen Laboratory of Fish Diseases, Department of Veterinary Microbiology, Royal Veterinary and Agricultural University, Frederiksberg, Denmark 5609/01/96: received 5 January 1996,revised 8 July 1996 and accepted 9 July 1996 T. TIAINEN, K. PEDERSEN AND J.L. LARSEN. 1997.Forty-five

Vibrio anguillarum-like isolates

reacting with V. anguillarum serogroup 0 3 antiserum were examined in 30 characters to clarify their phenotypical properties, while their genotype was examined by ribotyping. T h e strains were isolated from diseased and dead fish or from environmental sources such as water, sediment, plankton, and faeces and gills of healthy fish. Phenotypically, the similarity of all the strains was more than 90%. However, significant differences between the fish-associated and environmental strains were detected. Biochemically, deviations were found in the Voges-Proskauer test and lysine decarboxylase reaction. Clustering analysis of the ribotypes showed two distinct clusters with a similarity of only 32%. Two strains representing each of these groups were used in a LD,, study, which showed some difference also in the pathogenicity between environmental and fish strains. It is suggested that the environmental strains belong to another species than V. anguillarum, but serologically cross-reacting with the I/. anguillarum serogroup 0 3 . The ribotyping as u7ell as biochemical results indicated that the environmental strains possibly belong to Vibrio aestuarianus. T h e bona3de l? anguillarum serogroup 0 3 strains proved to be very homogeneous both phenotypically and genotypically, and the similarity of ribotypes was more than 96%. The V. anguillarum-like, serogroup 03-reactive strains from the environment were more heterogeneous in their biochemical behaviour, and showed an approximately 70% similarity in ribotypes.

fNTRODUCTlON

Vibriosis in fish caused by Vibrio anguillarum has mainly been attributed to serogroups 0 1 and 0 2 (Sm-ensen and Larsen 1986 ;Toranzo and Barja 1990 ; Larsen et al. 1994). For this reason, vaccines against vibriosis are generally composed of V. anguillarum serogroup 0 1 and 0 2 bacterin. Some of them contain the serogroup 0 2 cross-reacting Vzbrio ordalii instead of V. anguillarum 0 2 . However, there seems to be some host variation in the susceptibility to different 0-serogroups (Larsen et al. 1994). In eel (Anguilla anguilla), outbreaks of vibriosis seem to be frequently caused by V. anguillarum serogroup 0 2 and 0 3 strains (Larsen et al. 1994). In addition, outbreaks of vibriosis in sea bass (Dicentrarchus labrax) caused Correspondence to :Dr Karl Pedersen, Laboratory of Fish Diseases, Department of Veterinary Microbiology, Roya I Veterinarj and Agricultural University, Bulowsvej 13, I870 Frederiksberg C, Denmark

0 1997 The Society for Applied Bacteriology

by V. anguillarum serogroup 0 3 was recently reported from France and Italy, and vaccines containing this serogroup protected against disease (Vigneulle et al. 1993). Vibrio anguillarum serogroup 0 3 (equivalent to the Japanese serotype 5-0-2) has also been isolated in Japan from diseased ayu (Plecoglossus altivelis), and eel (Anguilla sp.) (Ezura et a f . 1980). Vibrio anguillarum serogroup 0 3 should therefore be considered as a potential pathogenic organism in some fish species. Studies on V. anguillarum serogroups 01 and 0 2 have been presented (Pedersen and Larsen 1995 ; Tiainen et al. 1995), and recently a characterization of the antigens of serogroup 0 3 was described (Santos et al. 1995). However, no studies including genotypical characterization of V. anguillarum serogroup 0 3 have to our knowledge at present been published. Recent data based on ribotyping (Pedersen and Larsen 1995) showed that V. anguillarum serogroup 0 1 strains iso-

212 T . T l A l N E N E T A L

lated from dead and diseased fish belonged to a clonal lineage different from that of environmental isolates. Preliminary results in this laborator! indicated that similar clonal differences between fish-associated strains and environmental isolates were recognized among T ; cinpSrri/l~irunrand C: ririguilluruni-like organisms reacting with the serogroup 0 3 antiserum (Rasmussen 1987). The purpose of the present in\-estigation was to extend our knowledge of the phenotypic and genot!-pic properties of I anpih?l-u7?zand I ~ i ~ ? ~ ~ ~ ~ i / ~ r / vorganisms / ~ ~ I / - l i kreacting e with serogroup 0 3 antiserum and to compare fish-associated and environmental isolates.

Difco) were also examined. All the media were standardized to contain l0/o NaCl. Serological tests

Serogrouping, sero-subgrouping and production of antisera 1% ere performed according to Sm-ensen and Larsen (1986) and Larsen et 01. (1994). Slide agglutination with an antiserum against !I anguillrzrum 0 3 type strain (ATCC 43307) was used for the primary identification. An additional antiserum against V. mpilkorurn-like strain from water (1.1.2.13/2) was produced, and these two antisera were cross absorbed (Larsen ct ( I / . 1994).

MATERIALS AND METHODS

Lipopolysaccharide (LPS) profiling and Western blotting

Bacterial strains

Strains showing the t!-pical colony morphology of C; mguilIrrvun2, i.e lox, comes, semitranslucent and haemolytic colonies, additionall!- being semiti\-e to the \-ibriostatic agent O/ 129 (2,4-diamino-6,7-diisopropylpteridine phosphate; Rosco Diagnostics. Taastrup, Il)enniark), motile, non-pigmented and non-swarming, positive for catalase and oxidase, degrading glucose b!- the fermentative and the oxidative pathways, and reacting with antiserum against I ringuilluruni serogroup 0 3 , were selected for further studies. -4total of 45 strains, isolated from dead or moribund fish ( n = 33), sediment and particles ( a = 2 ) , water ( n = I, fish faeces ( 1 1 = 2 ) and gills ( n = 1) were examined (Table 1). Strains \$-ereisolated in the authors' own laborator!. or received from other laboratories, and stored a t - 80°C in tryptic soy broth supplemented with O.!io;o NaCl and including 2.i-50°/o glycerol. Strains were cultiT-ated in marine broth (Difco, Detroit, LS.4) or on marine agar (Difco) supplemented ~ i t hj O o calf blood if not othern-ise stated.

LPS \$as extracted using a Proteinase K method modified from Hitchcock and Brown (1983) by Grskov and Grskov (1984), and subjected to SDS-PAGE in 12% gels (Laemmli 1970). Gels were stained with siher stain kit (BioRad, Hercules, C.4, USA). For Western blotting, LPS was transferred to nitrocellulose by semidry blotting (Kem-En-Tec, Copenhagen, Denmark) (Kyhse-Anderson 1984 ; Bjerrum and Schafer-Nielsen 1986). Blocking and subsequent incubations were performed at room temperature. All the strains were tested with both unabsorbed sera, while only the strains chosen for antiserum production were examined with the cross-absorbed antisera. Alkaline phosphatase-labelled swine anti-rabbit IgG (Dako, Glostrup, Denmark) was used as secondar! antibod!, and the labelled bands were visualized with 5-bromo-4-chloro-3-indolyl phosphate (BCIP ; Roehringer, llannheim, Germany) and nitroblue tetrazolium (NBT ; Boehringer). Plasmid analysis

Phenotypic characters

Strains were examined in 30 tests for their phenotypic characters. These \%ere: decarboxylation of arginine, I!-sine and ornithine (hbller) (Barrow and Feltham 1993), nitrate reduction (Lautrop t t al. 1978), acetoin from glucose in the 1-oges-Proskauer test, citrate (Simmons), pellicle formation, indole, h!-drol!-sis of casein, gluconate, meth!-l red, Tween 80, ONPG (o-nitrophen\-l-B-D-galactop!ranoside) and serum digestion (Barrow and Feltham 1993).Acid from arabinose, cellobiose, cellulose, fructose, galactose, mannitol, mannose, sucrose, sorbitol, starch and trehalose were examined in sugar-free broth (1% beef extract, l ? peptone, ~ 04% Na,HPO,, 0.0024°~~ bromoth!-mol blue), supplemented with O..j0,o of the sugars, at pII 7.4, Production of urease, gelatinase and phosphatase (Barrow and Feltham 1993), production of lecithinase (Colwell 1984) and DNasc (4.2% DNase test agar,

Plasmid DNA was isolated according to Kado and Liu (1981). T h e visualization of the plasmids and the estimation of the plasmid sizes was performed as previously described (Tiainen et af. 1995). Ribotyping

Chromosomal DNA was isolated by the method of Saunders r i rtl. (1988), slightly modified as described by Pedersen and L,arsen (1993). DNA mas digested with HzndIII (Promega, lladison, WI, USA) according to the instructions of the manufacturer. Digested DNA was electrophoresed in 0.8% agarose (SeaKem G T G , Rockland, ME, USA). DNA fragments were transferred to nylon hybridization membranes (Hybond-N, Amersham, UK) and fixed by baking for 1 h at 80°C. Hybridization with a digoxigenin-labelled probe

0 1997 The Society for Applied Bacteriology, Journal of Applied Microbiology82, 21 1-218

V . ANGUILLARUM SEROGROUP 0 3 213

Table 1 Strains of Vibrio anguillarum serogroup 0 3

No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Strain RVAU 820806-1/1 RVAU 890209-1/2 RVAU 890209-2/ 1 RVAU 2337 RVAU 820806-1/3 RVAU 820806-1/4 RVAU 860129-1/2 RVAU 860129-1/4 RVAU 890209-1/1 RVAU 890209-1/3 RVAU 900214-1/1 RVAU 900214-1/3 RVAU 900214-1/8 RVAU 920327-1/2 RVAU 920327-1/4 RVAU 920327-1/6 RVAU 920327-1/7 RVAU 920327-1/9 ATCC 43307 RVAU 890209-2/2 RVAU 920327-1/1 RVAU 920327-1/5 CNEVA NB11008 UB 910/90 UB 960/90 UB 1032/90 UB 1075/90 RVAU 9 1-9-2 15 RVAU 91-9-222 LMG 13584 RVAU 91-7-143 RVAU V 2 4/3 RVAU V 2 1/4 RVAU 890315-1/2 RVAU P T 493 RVAU B 1.1.2/4 RVAU V 2 3/6 RVAU 1-il-G/ 1 RVAU F 1 1/6 RVAU F 1 3/6 RVAU 2.1.3.17/2 RVAU 2.2.2.21 P/2 RVAU 1.1.2.13/2 RVAU 1.1.2.20/1 RVAU 5.3.1.94/2

Source Trout, DK Eel, DK Eel, DK Trout, I Trout, DK Trout, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Eel, DK Trout, DK Eel, DK Eel, DK Eel, DK Sea bass, F Sea bream, I Sea bass, I Sea bass, I Sea bass, I Trout, FIN Trout, FIN Fish, J Trout, DK Water, DK Water, DK Eel, DK Ayu, J Water, DK Water, DK Eel, DK Faeces, DK Faeces, DK Sediment, DK Particle, DK Water, DK Water, DK Water, DK

Ribotype

Biotype

1 1 1 1 1 1

1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 3 4 4 4 5 6 6 6 6 7 8 8 6 6 9 4 7 10 11 12 13 14 15 16 17 18 19

1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 3 3 4

4 5 6 7 7 7 7 8 8 8 9

Plasmid

4.2

50

Laboratory source : RVAU, Royal Veterinary and Agricultural University, Copenhagen, Denmark ; ATCC, American Type Culture Collection, Rockville, MD ; UB, University of Barcelona, Barcelona, Spain ; LMG, Laboratorium voor Microbiologie, Riijksuniversiteit, Gent, Belgium ;CNEVA, Centre National d’Etudes Veterinaire et Alimentaire, Brest, France. Isolation country : DK, Denmark ; FIN, Finland ;I, Italy ;J, Japan ; F, France.

0 1997 The Society for Applied Bacteriology,Journal of Applied Microbiology82, 21 1-21 8

214 T. TlAlNEN E T A L

complementar) to 16s and 23s rRU.4 of EsclrcriLlrra colt (Boehringer) and visualization of h! bridized fragments was performed according to Pcder5en ct a / . (1994).

Colony hybridization

All environmental strains and one of the fish-associated strains were tested in colon!- h!-bridization. Digoxigeninlabelled (Roehringer) 24 basepair DN,4 oligonucleotide probe for I/. unguilluruiii, V a l 3 (DK.l-technology, :\rhus, Denmark ; Martinez-Picado et a / . 1994) was used for hybridization. Examined strains were grown on nylon membrane (Iiybond-N) placed on marine agar (Difco) for -18 h a t 20°C. Llembrane was placed on moistened LVhatman no. 3 filters, colonies upwards, and treated as follow : 5 min in 1000 SDS, 20 min in 0..Po NaOH, and 3 x 1 min in 1 mol 1-’ Tris, pH 7. Colonies uere fixed to the membrane b!- baking 30 min at 80°C. Membrane u-as hybridized (10 p1 probe (400 ng ml I ) in 3 ml of hybridization buffer), and visualized as preF-iously described for ribot!-ping (Saunders rt rrl. 1988 ; Pcdersen et of. 1993) except for the following modifications: extra prehybridization step (30 min a t 56°C) followed by removal of surplus colon!- material with a soft paper towel was included prior to the actual prchybridization, and in the 1-isualization,the membrane \\-as washed a t 56°C for 1 min, instead of 13 min. A T,’. anguillrruni serogroup 01 strain M-as used as a positive control, and an E. ioli strain as a negative control.

LD,, trial

Rainbow trout (0nr.orh)m-husn/,ykiss),average weight of 15 g, were kept in aerated Lvater tanks (20 1) a t room temperature. Two-thirds of the water were changed dail!-. T h e two strains (-4TCC43307 and 1.1.2.13/2) used for antisera production, representing the t\ro groups of I : atiguillurrcm serogroup 0 3 , were examined for pathogenicit!-. Strains svere passaged in fish by injecting three to four fish intraperitoneally with 0.1 ml of washed bacterial culture containing approximately 10‘ cells ml (.%TCC -13307) or lo8 cells ml-’ (1.1.2.13/2). Bacteria recmered from the kidneys of the dead or moribund fish were further passaged twice, and after passages directl!. used for the LD,,, experiments. Six dilutions of bacteria were injected intraperitoneally into groups of six fish, 0.1 ml per fish. For strain . \ T K 43307, inocula ranged from 2.29 x 1 0 to 2.34 x colon!-forming units (cfu) and for strain I.1.2.13/2 from 3.28 x 10’ to 1.05 x 10’. -4control group was injected with 0.1 ml of NaCl (0.9”/0). Dead and moribund fish were remored from the tanks, and after 7 d all surviving fish were sacrificed. -111 the fish were subjected to bacteriological and gross pathological examination. LDj,) values tyere estimated according to Reed and Muench (1938).

’

Statistical analysis

The similarity between the strains was estimated both from the 30 phenotypical characters and from the size of bands in the ribotype patterns by simple matching using a computerized taxonomy programme (TAXAN-4, version 4.0, University of Maryland, MD, USA). Clustering of the strains from the similarity matrix was carried out (TAXAN-4) by the unweighted pair group method with arithmetic averages (UPGMA) (Priest and Austin 1993). Fischer’s exact test (SAS, version 6.04, Cary, USA) was used to examine the correlation between the isolation source and the phenotypic, as well as genotypic characters. RESULTS

A11 the strains produced acid from arabinose, cellobiose, cellulose, fructose, galactose, mannitol, mannose, sucrose, sorbitol, starch and trehalose. All strains degraded casein, produced gelatinasc, DNase, lipase (Tween SO), were ONPG-positive and reduced nitrate to nitrite. All the strains were negative in ornithine decarboxylase. None of the strains converted gluconate to 2-ketogluconate or produced urease. T h e 10 variable phenotypic characters are presented in Table 2. All in all, 19 different biochemical profiles were detected, which were ghen the biotype numbers of 1-19 (Table 1). Strains isolated from the fish showed eight different biotypes (I-8), most of the strains belonging to biotype 2, while each of the en>ironmental strains had their own biotype (9-19). All strains shared an identical LPS profile, although some Table 2 Variable phenotypic characters of Vthrio anguzNarurn serogroup 03-reactive strains from different isolation sources

Isolation source

Pellicie Decarboxylation of: -1rginine

Lysine Indole Serum digestion Xleth! 1 red Voges-Proskauer Citrate (Simmons) Phosphatase Lecithinase

Fish

Other

25/33*

10/12

32/33 0/33t 33/33 33/33 6/33 32/331 33/33 16/33 32/33

11/12 8/12 11/12 10/12 2/ 12 2/ 12 11/12 7 / 12 11/12

+Positile strains/total. ?Significant difference in the reaction betmeen the strains from different isolation sources (P< 0.0002). f Significant difference in the reaction between the strains from different isolation sources (P < 0.0001).

0 1997 The Society for Applied Bacteriology, Journal of Applied Microbiology82, 21 1-21 8

V. A N G U l L L A R U M S E R O G R O U P 0 3 215

gel to gel variation was detected. LPS of the two strains used for antiserum production were also indistinguishable. All strains reacted with V. anguillarum serogroup 0 3 antiserum (ATCC 43307) ; however, there was clear difference in the intensity of the reactions in the western blotting : the environmental strains gave a weaker and slower reaction with this antiserum (Fig. lb). There was no difference in the reaction intensity when the strains were tested with the antiserum against the strain 1.1.2.13/2 (Fig. lc). When antiserum against ATCC 43307 was cross-absorbed with 1.1.2.13/2 antigen, only reaction with ATCC 43307 antigen was detected (Fig. Id). T h e antiserum against 1.1.2.13/2 absorbed with ATCC 43307 antigen had very weak and slow reaction with 1.1.2.13/2 antigen (Fig. le). Only two environmental strains had plasmids, a small one of 4.2 kbp, and a larger one of approximately 50 kbp, respectively (Table 1). By ribotyping, the strains showed nine different profiles which could be grouped by clustering analysis into two clusters (Fig. 2, Table 1). The strains clustered according to their isolation source, except for two environmental strains, V.2.3/6 and 5.3.1.94/2, that clustered together with the fish strains. Furthermore, a comparison with 50 ribotypes of the type strains of various Vibrio species showed a remarkable similarity between the environmental strains, and the type strain of Vibrio aestuarianus ATCC 35048, which clustered together with the environmental strains. All the tested strains, except for the negative control, gave a strong positive reaction in the colony hybridization. The LDio experiment showed some difference in pathogenicity of the two strains tested. ATCC 43307 had an LD,, value of 1.5 x lo’, while 1.1.2.13/2 showed a somewhat higher LDjo,i.e. 3.0 x lo8.

Statistical analysis showed significant difference between the strains from the fish and from the environment in VogesProskauer (P< 0.001) and lysine decarboxylase (P < 0.01), as well as in ribotype (P < 0.001). DISCUSSION

In the present investigation of a number of V. anguillarum serogroup 0 3 and serogroup 0 3 reactive V. anguillarum-like isolates, the existence of two distinct bacterial populations was evident. The strains in one of the populations were commonly isolated from disease outbreaks from various fish species. These strains were very homogeneous in their biochemical reactions showing typical biochemical characters of V. anguillarum. Strains were serologically reactive with I? anguillarum serogroup 0 3 antiserum, and had ribotype patterns that resembled previously shown ribotypes of V. anguillarum (Olsen and Larsen 1993). Furthermore, the ribotypes were nearly identical, having a similarity of more than 96%. Hence, this group of strains can be identified as bonajde V. anguillarum serogroup 0 3 . The other group, although being more heterogeneous, fulfilled most of the biochemical criteria for V. anguillurum. However, most of the strains were lysine decarboxylase-positive and Voges-Proskauer-negative. Some of the strains were previously examined by Pazos et al. (1993) and phenotypically clustered together with some ?i anguillarum-related strains. They had a ribotype pattern distinct from that of the bonajdt! V. anguillarum strains, and possessed a remarkable homology with the ribotype of Vibrio aestuariunus ATCC 35048. Nevertheless, these strains reacted with V. anguillarum serogroup 0 3 antiserum, and also showed a positive reaction with the V. anguillarum-probe, VaV3.

Fig. 1 The SDS-PAGE of the LPS profiles and Western blotting of the strains ATCC 43307 (A) and 1.1.2.13/2 (B). (a) Silver-stained LPS ; (b) Western blotting with anti-ATCC 43307 ; (c) Western blotting with anti-43307 absorbed with 1.1.2.13/2; (d) Western blotting with anti-1.1.2.13/2; (e) Western blotting with anti-1.1.2.13/2 absorbed with ATCC 43307. Arrowheads indicate the reactive structures in the filter 0 1997 The Society for Applied Bacteriology, Journal of Applied Microbiology 82, 21 1-21 8

216 T. TlAlNEN E T A L .

Fig. 2 Hind111 ribotj-pes and their clustering analysis of bunri./i(ie C 7brrb angrrillnrunz serogroup 0 3 and V. nnguzllarum-like serogroup 0 3 cross-reactiw environmental strains. The figures in parentheses indicate the number of strains in each ribotype. The fragment sizes of the digosigenin-labelled i phage marker digested nith Hind111 are shown in kilobases at the bottom. 35048 = Vibrio urstuarirrnus t!pe strain .-ZTCC35048

In a previous stud!- in this laboratory (Larsen et a/. 1995), it was discovered that the VaV3 probe reacted with some of the tvpe strains of Tibrio species other than 1’. ringirillurum. ‘The type strain of I/: uestuarianus ATCC 35048 was one of those strains. Vibrio uestuariunus was first described b!- Tison and Seidler (19831, and has by nov been documented from water and shellfish in LSA, but has not previousl!- been described from Denmark. While comparing the phenotypic characters of our environmental strains with the known characters of V. (lestuirri(inus(Alsina and Blanch 1994), 15 out of 16 reactions supported the identification of the emironmental strains of the present study as C’. nestuarianus. Hence, we belie\ e that the t w sero-subgroups of 0 3 represent two different I’ihrio species, the honu j d e I”. anguilliirirni 0 3 and I ‘. irestunrimus. However, the type strain of C: aestunrianus did not react \vith an!- of the two antisera, neither in slide agglutination nor in Western blotting, although its LPS profile to some extent resembled that of the used type strains of fish-associated and environmental strains (data not shown). Also Santo et a/. (1995), like Rasmussen (1987), discovered two serological subgroups, 03.4 and 03B, within the P’. rmguillarurn serogroup 0 3 . These two groups seem to coincide with the two groups presented in the present studj~,having fish-associated strains in one group (03.4) and environmental strains in another ( 0 3 U ) . However, unlike these authors we \\-ere unable to \-isuall!- detect differences in the LPS profiles of the mentioned subgroups, but like Santos et a / . (1995) who found differences in the dot-blot analysis and quantitative agglutination between these two groups, we found differences in the intensit!. of the reaction in Western blotting. Furthermore, the reactions with absorbed antisera in Western

blotting indicated differences in the epitopes of the two type strains. Western blotting reactions of the two absorbed antisera indicate that almost all the epitopes of the 1.1.2.13/2 strain are common for both type strains, while ATCC 43307 has also some additional epitopes, specific for itself (Fig. 1). When suggesting that the two serogroup 0 3 subgroups actuallv belong to two different Vzbrio species, we also suggest the existence of common antigens between two species. Common antigens between two Vibrio species have also been detected previously. Thus all Vzbrio ordulii studied so far react \I-ith V. anguillurum serogroup 0 2 (Smensen and Larsen 1986 ; Larsen et a/. 1994). Additionally, C< furnissii and i7 J7uziulis (Shimada et al. 1991), as well as V. cholerae and V. niirniciis (Sakazaki and Donovan 1984) have been shown to share some common antigens, and this has led to the use of common serogrouping schemes for these species. Toranzo et al. (1987) evaluated the application of slide agglutination in diagnosis of bacterial fish pathogens. Using two antisera prepared against the type strains of V. anguillurum serogroups 0 1 and 0 2 they could recognize most of the strains causing vibriosis. With respect to V. anguillurum 0 3 , the present study shows the possibility of receiving false positives due to cross-reactions with non-pathogenic environmental strains. Thus, the results emphasize some limitations of serotyping as a diagnostic tool, and suggest that this technique is merely used for further typing and verification of diagnosis. Two isolates originating from water had ribotypes similar to the fish-associated strains. This may be explained by the excretion of bacteria from the fish during an outbreak of \-ibriosis, or alternatively that the fish-associated strains are

0 1997 The Society for Applied Bacteriology, Journal of Applied Microbiology 82, 21 1-21 8

V . A N G U l L L A R U M S E R O G R O U P 0 3 217

able to survive outside the fish and persist in the sea water (Muroga et al. 1986), or have an environmental habitat (Larsen 1985). Previous studies have shown plasmids in the environmental V. anguillarum serogroup 0 3 strains (Pazos et al. 1993 ;Olsen and Larsen 1993), but no plasmids have been detected in the fish-associated V. anguillarum serogroup 0 3 strains. These results were further supported in the present study. In a previous study in the authors’ laboratory (Pedersen et al. 1995), the antibiotic resistance of 520 V. anguillarum strains was investigated. Significant difference in susceptibility to colistin and ampicillin was demonstrated between serogroup 0 1 and 0 2 isolates. When comparing the previous susceptibility results considering the two populations of V. anguillarum 0 3 , similar significant difference in susceptibility to colistin ( P < 0,001) and ampicillin ( P < 0.005) was discovered. T h e bona jide V. anguillarum strains showed resistance to colistin, and susceptibility to ampicillin, and the opposite behaviour of the environmental strains. T h e susceptibility of the bona jide V. anguillarum serogroup 0 3 strains to ampicillin should be taken into account, if the recently developed medium for identifying V. anguillarum (Alsina et al. 1994) containing this antibiotic is used for primary isolation. Disease outbreaks caused by the serogroup 0 3 have been reported from Japan, France and Denmark (Ezura et al. 1980 ; Vigneulle et al. 1993; Larsen et al. 1994). Nevertheless, in the present study the V. anguillarum serogroup 0 3 type strain showed only low pathogenicity to rainbow trout-100-fold lower than reported by Pazos et al. (1993). This difference may be due to the differences in fish size, housing conditions or experimental design. Strains can also have lost some of their virulence during the preservation in the laboratory. However, both studies showed that the pathogenicity of serogroup 0 3 strains is remarkably lower than what has been reported for serogroups 0 1 and 0 2 (Pazos et al. 1993). This could be explained by a host specificity, as 0 3 strains are most commonly isolated from the disease outbreaks in eels, and more rarely in rainbow trout (Larsen et al. 1994). Previous studies on V. anguillarum have shown that the environmental strains are generally non-pathogenic (Pazos et al. 1993 ; Pedersen et al. 1997). T h i s was also supported by the present study, although only one of the environmental strains was tested for pathogenicity. T h e LD,,, of strain 1.1.2.13/2 was approximately 20 times higher than that of A T C C 43307. I n the clustering analysis of the biochemical results it was found that the order of the data input affected the clustering of some of the strains (strains 24, 25, 26, 27, 30 and 43). All these strains, or only strains 30 and 43, clustered together, separately from the two other clusters. This variation is likely to reflect the fact that there were only 10 variable characters in the data matrix, and hence, a single test result could have significant impact on the final clustering. However, the main

tendency of the biochemical behaviour was that the strains clustered according to their isolation source. No such problems were detected in clustering analysis of the ribotypes. This is due to the clear difference in the ribotype profiles between the two populations, and further supports the use of ribotyping in characterization of bacterial species.

ACKNOWLEDGEMENTS

T h e help of D r Stephania Koblavi, Institut Pasteur, Paris, France, and D r Linda Verdonck, University of Gent, Gent, Belgium, in characterization of the environmental strains is greatly appreciated. T h e authors would also like to thank colleagues, who have provided strains for the study. They are grateful to M r s Maj-Britt I-Zlrjgird, Ms Annette Rasmussen and M r Svend-Erik Jensen for their skilful technical assistance. T h e study was supported by the Danish Agricultural and Veterinary Research Council, grant no. 13-4710-1, and the European Union grant no. AIR-CT92-0341.

REFERENCES

Alsina, M. and Blanch, A.R. (1994) A set of keys for biochemical identification of environmental Vzbrio species. Journal of-4pplied Bacteriology 76, 79-85. A h a , M., Martinez-Picado, J., Jofre, J. and Blanch, A.R. (1994) A medium for presumptive identification of Vibrio ~ ~ n g z d l ~ i r u m . Applied and Environmental Microbiology 60, 1681-1683. Barrow, G.I. and Feltham, R.K.A. (ed.) (1993) Cowan crnd Steel’s Manual f i r the Ident$cation ofMedica1 Bacteria, 3rd edn. pp. 33 1. Cambridge : Cambridge University Press. Bjerrum, O.J. and Schafer-Nielsen, C. (1986) Buffer systems and transfer parameters for semi-dry electroblotting with a horizontal apparatus. In Electrophoresis ’86 ed. Dunn S.D. pp. 315-327. Weinheim : VCH Verlagsgesellschaft. Cowell, R.R. (1984) Vibrios in the Environment ed. Colwell, R.R. pp. 634. New York : John Wiley. Ezura, Y., Tajima, K., Yoshimizu, M. and Kimura, T. (1980) Studies on the taxonomy and serology of causative organisms of fish vibriosis. Fish Pathology 14, 167-179. Hitchcock, P.J. and Brown, T.M. (1983) Morphological heterogeneity among Salmonella lipopolysaccharide chemotypes in silver stained polyacrylamide gels. Journal of Bocteriolo~y154, 269277. Kado, C.1 and Liu, S.-T. (1981) Rapid procedure for detection and isolation of large and small plasmids. Journal of Bacteriology 145, 1365- 1373. Kyhse-Andersen, J. (1984)Electroblotting of multiple gels : a simple apparatus without buffer tank for rapid transfer of proteins from polyacrylamide to nitrocellulose. Journal of Biochemical and Bioph.ysical Methods 10, 203-209. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteria phage T4. Nature 227, 68C685. Larsen, A.R., Dehlsen, L.V., Kjaersgaard, H.D., Dalsgaard, A. and

0 1997 The Society for Applied Bacteriology, Journal of Applied Microbiology 82, 21 1-21 8

218 T. TlAlNEN E T A L .

Pedersen K. (1995) Testing of Vibrio atigziillariim specific DN.4 probe. Dunsk ~~eteri7iaertidskrl~i 78, 163-166. Larsen, J.L. (1985) I ’ibrio iinguilhriim : predence of typical and atypical strains in marine recipients with special reference to carbohydrate pollution. .-lc.ta I keriniiria Scundinii rinr 26, H9460. Larsen, J.L., Pedersen, K. and hlsgaard, I. (1994) Vibrio anguillicrum serogroups associated with vibriosis in fish. Journal of’ Fish Disecisrs 17, 259-267. Lautrop, H., Hsibl-, N.,Bremmelgaard, .%. and Korsager, B. (1979) Bn k~crioliyiskt~iirrtlt~r~ugr/sesnre~oder. pp. 416. Copenhagen : FA%I)L’sForlag. 1Iartinez-Picado, J., Blanch, .-\.R. and Jofre, J. (1994) Rapid detection and identification of Lrbrin q p l l i i r u t n b!- using a specific oligonucleotide probe complementary to 16s rRN.4. .Applied and Fn-.rroninental . Itfic.rohrr//og)* 60, 732-737. ~

’

Muroga, K.. Iida, M.,Xlatsumoto, H. and Nakai, T. (1986) Detection of L ibrro ringcrillariini from waters. Bulletin of’ the Japanese Socieiy o/Scientific Fisheries 52, 641447. Olsen, J.E. and Larsen, J.L. (1993) Ribotl-pes and plasmid contents of fi‘brio ongui/lariinz strains in relation to serovar. ripplied and Enrironmental Mir.ri~bio/i~gy 59, 3863-3870. Orskov, F.and !3rsko\, I. (1984) Serotyping of Escherirhia coh. In .&/hods in . ~ ~ i r r o h i i d u ~ed. q ) ~Bergan, T. pp. 43-1 12. London : lcademic Press. Pazos, F.,Santos, Y..llagariiios, B., Bandin, I., Suiiez, S. and Toranzo, -4.E. (1993) Phenot!-pic characteristics and \-irulence of I‘ibrio nnguillurum-related organisms. *-lpplied rind Enrironniental i2licrobiolo.gl 59, 2969-2976. Pedersen. K. and Larsen, J.I.. (1993) rRNri gene restriction patterns of 1FIbrio uagrii//~rirniserogroup 01. Diseases yf.4qnatic Organisnu 16, 121L126. Pedersen, Ii. and Larscn, J.L. (1995) Evidence for the existence of distinct populations of C ibrio unguillaruni serogroup 0 1 based on plasmid contents and ribotypes. .4pphed and Enrironwienral .If/c.r//b/I/I/~g)l 61, 229-2296. Pcdersen. K., Ceschia, G. and Larsen, J.L. (1994) Ribotl-pcs of F’ibtio i~irgrii//urrim01 from Ital! and Grecce. Current .tlicrobio/ug)~ 28, 97 99. Pedersen, K., Tiainen, T. and Larsen, J.L. (1995) Antibiotic resistancc of I ibrio iing‘iii/lnr.uni, related to serovar and plasmid contents. . 4 c 1 ~I~’c/rrin~iriri .Srutir/itia~irii36, -5544. Pedersen, K., Gram, Id., Austin, D.:-\. and tiustin, B. (1997) Pathogenicit! of I ibrro iirzgiiillorutn serogroup 01 strains as compared to plasmid- and outer membrane protein profiles and siderophore production. ]uiit-tzu/ o/.-fpp/zef/.bfrc-robiohg)~ (in press).

Priest, F. and Austin, B. (1993) Modern Bacterial Taxonomy,2nd edn. pp. 227. London : Chapman & Hall. Rasmussen, H.B. (1987) Surface structures in gram-negative bacteria. Immunoelectrophoretic analyses of Vibrio anguillarum surface polysaccharides. PhD Thesis, Copenhagen, Royal Veterinar!- and Agricultural University. 45 pp. Reed, L.J. and Muench, H. (1938) A simple method for estimating fifty percent endpoints. American Journal ufHygiene 27,493497. Sakazaki, R. and Donovan T.J. (1984) Serology and epidemiology of Vibrio cholerrie and Vibrio mimicus. In Methods in Microbiology ed. Bergan Y. Vol. 16, pp. 271-289. London: Academic Press. Santos, Y., Pazos, F., Bandin, I. and Toranzo, A.E. (1995) Analysis of antigens present in the extracellular products and cell surface of C’ibrio nnguillarum serotypes 0 1 , 0 2 , and 0 3 . Applied and Enrironmental MicrobioloKy 61, 2493-2498. Saunders, N.A., Harrison, T.G., Kachwalla, N. and Taylor, A.G. (1988) Identification of species of the genus Legionella using a cloned rRN.% gene from Legionella pneumophila. Journal oJ’ General itlicrobiology 134, 2363-2374. Shimada, T., Kosako, Y., Inoue, K., Ohtorno, M., Matsushita, S., Yamada, S . and Kudoh, Y. (1991) Vibrio fluvialis and Vzbriu furnissii serotyping scheme for international use. Current Microblolog), 22, 335-337. Song, Y.-L., Chen, S.-N. and Kou, G.-H. (1988) Serotyping of Vibriomguillurum strains isolated from fish in Taiwan. Fish PathO / O ~ ) 23, I 185-189. Ssrensen, U.B.S. and Larsen, J.L. (1986) Serotyping of Vibrio a n g t i i l h m . Applied and Environmental Microbiologj 51, 593-597. Tiainen, T., Pedersen, K. and Larsen, J.L. (1995) Ribotyping and plasmid profiling of Vibrio anguillarum serovar 0 2 and Vibrio ordalii. Journal of Applied Bacteraology 79, 384-392. Tison, D.L. and Seidler, R.J. (1983) Vibrio aestuarianus: a new species from estuarine waters and shellfish. InternationalJnurnal uf Systematic Bacteriology 33, 699-702. Toranzo, A.E. and Barja, J.L. (1990) A review of the taxonomy and seroepizootiology of Vibrio anguillarum with special reference to aquaculture in northwest of Spain. Diseases of Aquatic Organisms 9, 73-82. Toranzo, ,%.E.,Baya, A.M., Roberson, B.S., Barja, J.L., Grimes, J.D. and Hetrick, F.M. (1987) Specificity of slidc agglutination test for detecting bacterial fish pathogens. Aquaculture 61, 81-97. \$,meulle, M., Breuil, G., Ceschia, G. and Blanch, A. (1993) Vaccination trials of sea bass (Dicentrarchus Iabrax) against two serotypes of Vibrio anguillarum. p. 66. Abstract from European .%ssociation of Fish Pathology, Sixth International Conference ‘Diseases of Fish and Shelljish’, Brest, France, 5-10 September 1993.

0 1997 The Society for Applied Bacteriology,Journal of Applied Microbiology82,21 1-218