Clonality of Vibrio anguillarum strains isolated from fish from the Scandinavian countries, Sweden, Finland and Denmark

Authors


Abstract

In order to investigate whether outbreaks of vibriosis in the Baltic region were caused by the spread of certain pathogenic clones, 291 Vibrio anguillarum isolates from Finland (n = 156), Sweden (n = 88) and Denmark (n = 47) were studied with respect to serogroup, ribotype, plasmid content, and biochemical phenotypes as expressed with the PhenePlate (PhP) typing system. For comparison, 54 V. anguillarum serogroup O1 from other countries worldwide were included. Most isolates from Finland, Sweden and Denmark belonged to serogroup O1 (255), followed by O2 (30). Four Finnish isolates cross-reacted strongly with antisera against two new serogroups VaNT2 and VaNT4, whereas two strains were non-typeable. The serogroup O1 isolates displayed ten different ribotype patterns, whereas the other strains were considerably more diverse with respect to ribotypes. Most of the O1 isolates carried the 67 kb virulence plasmid and a group of Finnish isolates, in addition, carried an 86 kb plasmid. Additional plasmids with molecular weights of 63, 76, 135 or 260–290 kb were found in single O1 isolates. With few exceptions, strains of serogroup O2 either had no plasmids or carried one or two small plasmids. PhenePlate typing revealed considerable diversity within the species, serogroup O1 being the most homogeneous. A few PhP types were dominant, whereas other types were observed only in one to four isolates. The prevalence of the different types changed significantly from one year to another but in Finland, one clonal lineage became increasingly important from 1992 (20% of isolates) to 1996 (80%). Remaining clones were mostly restricted to specific geographic areas. By cluster analysis, it was demonstrated that most of the isolates from Finland, Sweden and Denmark belonged to two clusters, and most of the strains from Southern Europe fell into two other, distinct clusters. Most isolates from the UK, North America, Chile and Tasmania grouped together in a distinct cluster. For the typing of V. anguillarum, O-serotyping should be the primary method. For isolates belonging to serogroups other than O1, plasmid profiling in combination with ribotyping gives a very good discrimination between strains, whereas for serogroup O1, another method is required. It is concluded that PhP typing is a tool that provides a good discrimination between O1 isolates.

Correspondence to: Karl Pedersen, Laboratory of Fish Diseases, Department of Veterinary Microbiology, The Royal Veterinary and Agricultural University, Bülowsvej 13, DK-1870, Frederiksberg C, Denmark (e-mail: KPE@SVS.DK).

*Present address: Danish Veterinary Laboratory, Hangøvej 2, DK-8200 Aarhus N, Denmark.

Introduction

Vibrio anguillarum is an important fish pathogen in many countries where fish are grown in aquaculture ( Austin & Austin 1993), including the coastline of the Baltic Sea in Finland, Sweden and Denmark. Vibrio anguillarum is still causing outbreaks of disease in Finnish and Swedish farmed salmonids in spite of vaccination programmes. In Finland, vibriosis is enzootic in all brackish waters except for the northern-most part of the Gulf of Bothnia, where the salinity is close to 0%. Vibrio anguillarum is isolated every year from diseased farmed fish in the Gulf of Bothnia, the Baltic Sea and the Gulf of Finland. In Sweden and Denmark, vibriosis has previously been a problem on fish farms, but the frequency of outbreaks has decreased during the 1990s, probably due to the implementation of proper prophylactic measures, including vaccination. Vibrio anguillarum is also a naturally occurring bacterium in the marine environment and can be isolated in large numbers during the summer season from water, sediment and plankton ( Larsen 1985, 1990) as well as from the intestine, gills and mucus of healthy fish ( Larsen 1990 ; Pedersen & Garcia 1996). However, these environmental isolates are usually not virulent for fish ( Larsen 1990 ; Pedersen & Larsen 1995 ; Pedersen & Garcia 1996 ; Pedersen et al. 1997 ). An environmental habitat for the virulent forms, and the way in which they are spread from one fish farm to another, is not known in any detail. The purpose of the present study was to evaluate diversity among V. anguillarum strains isolated from a certain geographical region over time, and to follow the clonality of V. anguillarum in this area. For this purpose, 291 V. anguillarum isolates from fish along the coast of the Baltic Sea in Sweden, Finland and Denmark were studied. For comparison, 54 strains from other regions were included.

Materials and methods

Bacterial strains and culture conditions

A total of 156 Finnish, 88 Swedish and 47 Danish strains of V. anguillarum, isolated in the Baltic area during the period 1982–1995, was investigated. All strains, except for five, were from diseased fish and most of them from outbreaks of vibriosis in fish farms. Only one isolate from each outbreak was included. For comparison, 54 strains of V. anguillarum serogroup O1 from other countries, all isolated from diseased fish, were included : Norway (n = 5), Italy (n = 28), Greece (n = 3), Spain (n = 3), France (n = 3), Germany (n = 1), UK (n = 2), Canada (n = 4), USA (n = 2), Chile (n = 1), Tasmania (n = 1) and Taiwan (n = 1). Therefore, the total number of isolates included in the study was 345.

Serotyping

O-serotyping on the basis of heat stable O-antigens was carried out as described by Larsen et al. (1994) using rabbit antisera against the serogroups O1–O10 ( Sørensen & Larsen 1986) and the four additional serogroups, VaNT2, VaNT4, VaNT5 and VaNT7, introduced by Austin et al. (1995) .

Ribotyping

rRNA gene restriction pattern analysis (ribotyping) was performed as described by Pedersen & Larsen (1993) and Pedersen et al. (1996) . Isolated total DNA was digested with HindIII and subjected to electrophoresis in 0·8% agarose gels in TAE (Tris 40 mmol l−1, sodium acetate 5 mmol l−1, EDTA 1 mmol l−1) buffer, pH 7·8. Separated DNA fragments were blotted onto nylon membranes (Hybond-N, Amersham, UK, or Magna Nylon, Micron Separations Inc., Westborough, MA, USA) and hybridized with a digoxigenin-labelled probe complementary to 16S and 23S RNA from Escherichia coli. Labelled fragments were visualized by incubation with alkaline phosphatase-labelled anti-digoxigenin immunoglobulin (Boehringer Mannheim) followed by the addition of nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl-phosphate (Boehringer Mannheim). The ribotype patterns of serogroup O1 strains were given numbers referring to the patterns previously described by Pedersen & Larsen (1993), whereas the serogroup O2 strains were given arbitrary pattern designation letters A–R. The remaining strains, not belonging to serogroup O1 or O2, were given arbitrary pattern designations, λ, μ, ν, π, σ and ω.

Plasmid profiling

Plasmids were isolated by the method of Kado & Liu (1981) and separated by electrophoresis in 0·8% agarose gels in TAE buffer. DNA was visualized by staining with ethidium bromide and photographed at 254 nm u.v. transillumination ; plasmid molecular weights were then calculated as described by Tiainen et al. (1995) . Distinction between covalently closed circular forms and open circular forms was done by the u.v. irradiation method described by Hintermann et al. (1981) .

PhenePlate (PhP) typing

Biochemical fingerprinting using the PhenePlate system (BioSys inova, Stockholm, Sweden) was carried out as described by Kühn et al. (1990, 1996, 1997) and Möllby et al. (1993) . Briefly, the quantitative kinetics of 11 biochemical reactions, selected to give high discriminatory power for V. anguillarum strains, were measured spectrophotometrically at 620 nm in microtitre plates. The 11 biochemical reactions were : degradation of l-arabinose, galactose, mannose, cellobiose, trehalose, sucrose, gentobiose, sorbitol, mannitol, gluconate and starch. Plates were incubated at 30 °C and the reactions read after 16, 40 and 64 h. The mean absorbance in each well was calculated and the similarity between the isolates was then calculated as the correlation coefficient between the 11 reactions. Strains were assigned to the same PhP type if they had a similarity higher than 0·975 to each other. Clustering of strains was performed by the unweighted pair group method with arithmetic averages ( upgma) ( Sneath & Sokal 1973). Phenotypic diversity was calculated as the Simpson’s index of diversity (Di) ( Atlas 1984). PhenePlate types were given specific designations if more than one strain had this type ; strains with unique PhP types were designated as single (S).

Results

The distribution of the isolates from Sweden, Finland and Denmark on serotypes, ribotypes, plasmid content and PhP types is shown in Table 1 (serogroup O1 only) and Table 2 (other serogroups), and of the strains from other countries, in Table 3.

Table 1. Distribution of ribotypes, plasmids and PhP types on serogroup O1 Vibrio anguillarum isolates from Sweden, Finland and Denmark. The number of strains is shown in brackets
CountryRibotypePlasmidsPhP type
Finland (151)1 (111)pJM1 (107)1 (79)
2b* (1)
3 (6)
4 (20)
Single† (1)
  73 kb (1)1 (1)
  – (3)1 (3)
 4 (40)rjM1 (25)1 (3)
2 (1)
4 (13)
6 (7)
S4/S71 (1)
  pJM1, 86 kb (13)2 (10)
2b (2)
4 (1)
  45 kb, pJM1 (1)4 (1)
  – (1)4 (1)
Sweden (n = 73) 1 (29)pJM1 (28)1 (20)
1A (1)
2 (1)
4 (5)
4B (1)
  pJM1, 135 kb (1)1A (1)
 4 (33)pJM1 (32)1 (1)
4 (13)
4B (7)
4C (1)
D4 (2)
D28 (2)
D82 (2)
D99 (2)
Single (2)
63, 76 kb (1)4 (1)
 6 (11)pJM1 (10)1 (9)
NF (1)
  260–290 kb (1)1 (1)
Table 2. Distribution of serotypes, ribotypes, plasmids and PhP types on Vibrio anguillarum isolates from Sweden, Finland and Denmark not belonging to serogroup O1. The number of strains is shown in brackets
CountrySerotypeRibotypePlasmidsPhP type
FinlandO2BD65
(n = 5) VaNT2/VaNT4ν6·5 kb13
VaNT2/VaNT4πS
VaNT2/VaNT4σ2·613
VaNT2/VaNT4ωS
SwedenO2AD55
(n = 15) O2B7·8 kbD65
O2BD65
O2C7
O2CS
O2C5·5 kb7
O2D22 kbS
O2DS
O2E4·2 kbS
O2E4·8 kb7
O2FD55
O2GS
O2H4·2 kb7
NTλD55
NTμS
DenmarkO2ID32
(n = 16) O2IS
O2ID32
O2IS
O2J4·2S
O2J4·2, 5·1D40
O2J4·2D40
O2K4·2D36
O2K4·2D36
O2L5·11
O2MS
O2N>200S
O2OD43
O2PD43
O2Q13
O2RS
Table 3. Distribution of ribotypes, plasmids and PhP types on Vibrio anguillarum serogroup O1 isolates from outside the Baltic region. The number of strains is shown in brackets
CountryRibotypePlasmidsPhP type
Norway (5)1 (4)pJM1 (4)1 (2)
6 (1)
Single (1)
6 (1)33 kb, pJM1 (1)1 (1)
Italy (28)1 (24)pJM1 (19)IT3 (19)
89, 90 kb (1)IT4 (1)
80, 90 kb (1)IT4 (1)
pJM1, 86 kb (3)IT4 (2)
Single (1)
 3 (4)pJM1 (4)IT5 (4)
Greece (3)1 (3)pJM1 (3)IT3 (2)
Single (1)
Spain (3)1 (3)pJM1 (3)IT5 (3)
France (3)1 (2)pJM1 (1)Single (1)
36 kb, pJM1 (1)IT3 (1)
5 (1)pJM1 (1)IT5 (1)
Germany (1)1 (1)pJM1 (1)1 (1)
UK (2)1 (2)pJM1 (1)Single (1)
5·3, 50 kb (1)Single (1)
Canada (4)4 (2)pJM1 (2)Cn7 (2)
13 (2)pJM1 (2)Cn6 (2)
USA (2)1 (2)pJM1 (1)1 (1)
50, pJM1 (1)Single (1)
Chile (1)1 (1)pJM1 (1)NF (1)
Tasmania (1)1 (1)pJM1 (1)Single (1)
Taiwan (1)1 (1)pJM1 (1)IT4 (1)

The Finnish strains were all isolated from diseased salmonid fish, mostly rainbow trout, during the 5 year period 1992–1996. The majority of these isolates belonged to serogroup O1 (151 of 156), whereas only one strain was O2. Four isolates cross-reacted strongly with antisera against the new serogroups VaNT2 and VaNT4. The serogroup O1 isolates belonged to the two ribotype patterns 1 (111 of 151) and 4 (40 of 151). Irrespective of ribotype, all but five of these isolates carried the 67 kb pJM1 plasmid. One isolate carried a 73 kb plasmid, although this was shown by restriction analysis to be a derivative of pJM1, and four serogroup O1 strains had no plasmids. Thirteen isolates carried an 86 kb plasmid in addition to the 67 kb plasmid. These 13 isolates all belonged to ribotype pattern 4. A total of eight PhP types was recorded among the serogroup O1 isolates, of which PhP type 1 (n = 86), 2 (n = 11), 4 (n = 36) and 6 (n = 7) were the most frequent. Among the strains carrying the 86 kb plasmid, 12 out of 13 belonged to PhP type 2 or 2b. Among ribotype 1, one PhP type (PhP type 1) was dominant (83 of 111) and a second type (PhP type 4) was found in 20 of 111 isolates. Among ribotype 4, PhP type 4 was the most common (16 of 40) but also, PhP type 2 was found in 13 isolates. It therefore seems that although some correlation existed between ribotype and PhP type, these two properties did not always coincide.

Only five of the Finnish strains were of single PhP types, including the single serogroup O2 isolate, but this strain shared PhP type as well as ribotype with two Swedish strains.

The four VaNT2/VaNT4 cross-reacting isolates all had different, although very similar, ribotype patterns. Two of them shared PhP type whereas the other two had unique PhP types ( Table 2). These four strains were isolated on four different fish farms.

The Swedish strains were mostly isolated from diseased rainbow trout but 16 were from other fish species. Like the Finnish isolates, the majority belonged to serogroup O1 (73 of 88). Thirteen isolates belonged to serogroup O2, whereas two isolates were non-typeable. The serogroup O1 strains belonged to three different ribotypes, 1 and 4 as in Finland, and ribotype 6, and they all carried plasmids. Out of 73 serogroup O1 isolates, 71 had the pJM1 plasmid. One of these isolates had an additional large plasmid of 135 kb. Furthermore, one isolate had a very large plasmid of 260–290 kb. The last O1 isolate had two plasmids, 63 and 76 kb, one of which was a derivative of the 67 kb pJM1 plasmid. The Swedish O1 isolates displayed more diversity with respect to PhP types than the Finnish isolates, but the most common types were the same as the most common Finnish PhP types. The major PhP types were 1 and 4 containing 31 and 19 isolates, respectively. Only eight isolates were single types and five of those were O2 strains. The 13 serogroup O2 isolates were divided into eight different ribotypes, which sometimes corresponded to the PhP types and sometimes not ( Table 2). One of these ribotypes, B, was identical to the ribotype of the single Finnish O2 isolate, and the three Swedish and Finnish isolates belonging to this ribotype also had identical PhP type.

The Danish strains belonged to serogroup O1 (31 of 47) and O2 (16 of 47). Most of the strains (39 of 47) were isolated from rainbow trout, but one strain was isolated from eel and two from turbot. Five strains were environmental isolates. Among the O1 strains, six different ribotypes were found ( Table 1). However, the strains with the ribotypes 7, 10 and 11 were the environmental strains, and the strain with ribotype pattern 2 has later been shown to be non-pathogenic to fish, although it was originally isolated from the kidney of a rainbow trout ( Pedersen et al. 1997 ). Thus, all fish pathogenic strains belonged to the two ribotypes 1 and 6. Most of the serogroup O1 isolates carried the pJM1 plasmid, and most of these isolates shared PhP types with isolates from Finland and Sweden. The 16 O2 strains belonged to 10 different ribotypes, none of which were found in Swedish or Finnish isolates ( Table 2).

Twelve of the total number of 30 O2 strains carried one or two small plasmids of 4·2–7·8 kb, one carried a medium sized plasmid of 22 kb, and one strain a large plasmid >200 kb.

Ribotype 6 was found only in strains from Sweden (n = 11), Norway (n = 1) and Denmark (n = 1). Strains of this ribotype displayed interesting PhP reactions by showing slightly varying results at repeated testings. However, their PhP types were most similar to PhP type 1 and they were therefore allocated to this type. Such variations were never recognized in their ribotype or plasmid profiles.

Although many of the PhP types among the O1 isolates were found in all three countries, some were found only in one country and there was considerable variation in the prevalence of these PhP types between countries and between years of isolation. The year of isolation was not known for all Danish isolates, but Figs 1 and 2 show the prevalence of the most important combinations of ribotypes and PhP types in Finland and Sweden, respectively, in different years. In Finland, the clonal lineage of strains with ribotype 1 and PhP type 1 has become considerably more prevalent during the period from 1992 to 1996, increasing from approximately 30% of the isolates to approximately 80%. During the same period, the proportions of the other clonal lineages have decreased accordingly. The combination of ribotype 4 and PhP type 2, which are the strains carrying pJM1 and the 86 kb plasmid, were found only in 1993, 1994 and 1996 ( Fig. 1). Considerable variations in prevalence were also noticed among the Swedish serogroup O1 isolates. The combination of ribotype 1 and PhP type 1 first decreased from almost 40% of the V. anguillarum strains isolated in 1985–1986 to 0 in 1991–1992, and then increased again to 40% in 1993–1995 ( Fig. 2).

Figure 1&.

emsp;Prevalence of combinations of the most important ribotypes and PhP types in Finland during the period 1992–1996. P and R indicate PhP type and ribotype, respectively : (▒), P1-R1 ; (), P4-R1 ; (▓), P4-R4 ; (▪), P2-R4p ; (□), other

Figure 2&.

emsp;Prevalence of combinations of the most important ribotypes and PhP types in Sweden during the period 1985–1995. P and R indicate PhP type and ribotype, respectively : (▒), P1-R1 ; (), P4-R1 ; (▓), P4-R4 ; (▪), P1-R6 ; (□), other O1

The serogroup O1 strains from outside Finland, Sweden and Denmark displayed marked differences from the O1 strains from these three countries with respect to PhP types and sometimes also ribotypes ( Table 3). Only four of the five Norwegian isolates, one isolate from the USA and one isolate from Germany shared combinations of ribotype and PhP type with strains from Finland, Sweden and Denmark. Remaining isolates had different PhP types. Most isolates belonged to ribotype 1, but ribotypes that were not detected among the strains from Finland, Sweden and Denmark were found in strains from Italy (ribotype 3), France (ribotype 5) and Canada (ribotype 13, which has not previously been described). Most of the strains from Italy, Greece, Spain and France (34 of 37) belonged to three PhP types, IT3, IT4 and IT5. Apart from a single isolate from Taiwan, which had PhP type IT4, these PhP types were not found outside the Mediterranean area in the present material. The Canadian isolates belonged to two different PhP types, not found outside Canada, and two of these isolates also had the unique ribotype 13, not previously described.

On the basis of PhP typing, a dendrogram of strains of different geographic origin was generated ( Fig. 3). It was demonstrated that most of the isolates from Finland, Sweden and Denmark belonged to two major clusters, and most of the strains from southern Europe fell into two other distinct clusters. Most isolates from UK, North America, Chile and Tasmania grouped together in a distinct cluster. A minor group of Finnish isolates clustered together with the Italian isolates belonging to the PhP type IT4. Three of the five Norwegian isolates together with one isolate from the USA were found in one of the major clusters of Finnish, Swedish and Danish isolates, whereas a fourth Norwegian isolate was located in the second major cluster of Finnish, Swedish and Danish isolates. The last Norwegian isolate was located in the cluster of isolates from UK, North America, Chile and Tasmania.

Figure 3&.

emsp;Dendrogram of 72 Vibrio anguillarum serogroup O1 isolates of different geographic origin. NO = Norway, US = USA, SW = Sweden, FL = Finland, DK = Denmark, GE = Germany, IT = Italy, GR = Greece, FR = France, TW = Taiwan, SP = Spain, UK = United Kingdom, TS = Tasmania, CN = Canada and CH = Chile. (•), Strains from Finland, Sweden and Denmark ; (▪), strains from the Mediterranean ; (□), strains from other countries. Sal, Rb, Tu, Csa, Sba, Sbr, Mul, Mf, Sm and Ssa indicate that the strains were isolated from Atlantic salmon, rainbow trout, turbot, chinook salmon, sea bass, sea bream, mullet, milkfish, Baltic salmon and sockeye salmon, respectively

Between strains assigned to each PhP type, there are minor variations that will affect the diversity index. When comparing PhP type 1 strains from Finland with PhP type IT3 strains from Italy, all belonging to serogroup O1 and ribotype 1, the Finnish strains showed much higher diversity (Di = 0·51) than the Italian strains (Di = 0·019) ( Fig. 4).

Figure 4&.

emsp;Dendrogram of Vibrio anguillarum serogroup O1, ribotype pattern 1 isolates of PhP type 1 from Finland (VF-marked strains), PhP type IT3 from Italy (IT-marked strains), PhP type IT3 from Greece (GR36 and GR35), together with PhP single type from Greece (GR34), three PhP type IT5 strains from Spain (SP-marked strains) and one PhP single type from France (FR40). All strains carried the 67 kb pJM1 plasmid. Sba, Sbr, Mul and Tu indicate that the strains were isolated from sea bass, sea bream, mullet and turbot, respectively

Discussion

The V. anguillarum isolates included in the present study were regarded as representative of the clonal lineages of this species that cause vibriosis in mainly Swedish and Finnish fish farms. In Denmark, the problems with vibriosis are less pronounced because of effective vaccination of all sea-reared rainbow trout and turbot prior to transfer to the sea. The Danish strains of V. anguillarum were therefore collected over a longer period of time than the Finnish and Swedish strains, and also included some non-pathogenic strains isolated from water and from the gills, faeces and mucus of healthy rainbow trout.

The distribution of serogroups with the dominance of O1 followed by O2, and the presence of low numbers of non-typeable isolates and isolates belonging to other serogroups, was in accordance with previous reports ( Larsen et al. 1994  ; Tiainen et al. 1994 ). The serological reaction of four strains that cross-reacted very strongly with VaNT2 and VaNT4 serum was remarkable and has not been reported before, nor has it been observed in our laboratory before, even though we have serotyped thousands of V. anguillarum isolates. However, in a previous investigation of Finnish V. anguillarum strains isolated during the period 1989–1991, two strains reacting only with VaNT2 serum were described ( Tiainen et al. 1994 ). Also remarkable were the facts that each of these four strains had a unique ribotype and usually also a unique PhP type, and that they were isolated on four different farms. This indicates that they cannot be considered as the emergence of a new virulent clone with a serotype not recognized before, but may represent a hitherto unrecognized serogroup of V. anguillarum that, under certain circumstances, may be pathogenic for fish. The pathogenic properties of these strains are now under further investigation.

Several isolates from Finland carried an 86 kb plasmid. This plasmid was the same as the 90 kb plasmid isolated from one of the Danish strains and the 86 and 90 kb plasmids isolated from five Italian strains. This plasmid had previously been described by Larsen & Olsen (1991), Pedersen & Larsen (1995) and Pedersen et al. (1996) among strains from Denmark, France and Italy, and possibly also from other countries. However, all of the 13 Finnish isolates carrying this plasmid belonged to ribotype 4, and 12 of these were of the PhP type 2 or the related 2b, whereas the previously reported strains from Denmark, France and Italy, that carried this 86 kb plasmid, belonged to ribotypes 1, 5 or 6 ( Pedersen & Larsen 1995 ; Pedersen et al. 1996 ). The Finnish strains carrying the 86 kb plasmid were all isolated in 1993, 1994 and 1996 and were not reported in the investigation of strains from 1989–1991 ( Tiainen et al. 1994 ). The combination of ribotype 4, plasmid profile pJM1 plus the 86 kb plasmid, and PhP types 2 or 2b, was only found in Finnish strains, indicating that this is a true clonal lineage which was probably not introduced from outside Finland. Several other examples of unique combinations of ribotypes, plasmid profiles and PhP types were present, many of which were restricted to one region or even the same fish farm, indicating that the strains that caused infections were not introduced from outside the area but were local strains recruited from the environment or other unknown habitats or reservoirs. F. ex., eight serogroup O1 strains from Sweden belonged to the four unique PhP types, D4, D28, D82 and D99, with two strains in each group. In two of these groups, the two strains were isolated on the same fish farm but in two different years, indicating that the strains responsible for the outbreak were not introduced from outside the farm but may most likely have been resident in the area or on the farm. Alternatively, the same clone may have been introduced repeatedly from the same external source, e.g. the same hatchery.

The function of the 67 kb pJM1 plasmid is well known. It was first shown that this plasmid encodes an iron-sequestering system consisting of a siderophore, anguibactin, and a siderophore-binding outer membrane protein, OM2 ( Crosa 1980 ; Crosa & Hodges 1981 ; Actis et al. 1985 ) ; it was soon understood that this system was important for the virulence of V. anguillarum strains of serogroup O1 ( Crosa et al. 1980 ). Later, it has been demonstrated that the entire iron-sequestering system is somewhat more complicated, involving genes on the pJM1 plasmid as well as genes on the chromosome ( Köster et al. 1991  ; Chen & Crosa 1996 ; Chen et al. 1996 ). The function of the 86 kb plasmid is at present unknown, but as it is found in strains from several countries and seemingly, also in different clonal lineages, this deserves to be investigated. It seems, however, not to be involved in virulence ( Pedersen et al. 1997 ).

It is known that the pJM1 plasmid exists only in V. anguillarum strains belonging to serogroup O1 and that the majority of the virulent strains carry this plasmid ( Pedersen & Larsen 1995 ; Pedersen et al. 1996 ). Most of the serogroup O1 isolates included in the present investigation contained pJM1. Five O1 strains isolated from diseased fish had no plasmids but it is not known whether these strains had the plasmid at the time of isolation and subsequently lost it during laboratory procedures ( Pedersen 1997). The 86 kb plasmid has also been isolated exclusively from O1 strains, and only those strains as also carried the pJM1 plasmid.

One Finnish isolate carried a plasmid of 73 kb. The normal size of the pJM1 plasmid is 67 kb, but variations caused by insertions or deletions have previously been described ( Pedersen et al. 1996 ), and it has been demonstrated that virulence is not necessarily affected by such variations ( Pedersen et al. 1997 ). The 73 kb plasmid was shown to be such a derivative of the pJM1 plasmid.

Several isolates of serogroup O2 carried one or two small cryptic plasmids, varying in size from 4·2 to 7·8 kb, in accordance with the observation of Tiainen et al. (1995) who found these plasmids only in strains from Scandinavia and not among strains from southern Europe. The strains belonging to serogroup O2 were considerably more heterogeneous than serogroup O1 strains, displaying 18 different ribotypes and 15 different PhP types among 30 strains. Eight of the 13 Swedish O2 strains were from fish species other than rainbow trout.

Previous investigations of V. anguillarum serogroup O1 have so far demonstrated the existence of 12 different ribotypes. Six of these ribotype patterns (2, 7, 8, 9, 10 and 11 according to Pedersen & Larsen 1993, 1995) have been found in non-pathogenic and environmental strains whereas the remaining types (1, 3, 4, 5, 6 and 12) have been found among fish pathogenic isolates ( Pedersen & Larsen 1993, 1995 ; Pedersen et al. 1994  ; Skov et al. 1995 ). Pattern 1 was shown to be the most prevalent, an observation that is in accordance with the results of the present study. Pattern 3 has previously been demonstrated only in strains from Italy ( Pedersen & Larsen 1993, 1995 ; Pedersen et al. 1994  ; Skov et al. 1995 ), and that observation was confirmed in the present study. Pattern 4 has hitherto only been demonstrated in strains from salmonid fish from the USA and Canada ( Pedersen & Larsen 1993, 1995 ; Pedersen et al. 1994  ; Skov et al. 1995 ), but the present study showed that this ribotype was also prevalent in Sweden and Finland. However, the PhP types of these isolates were different from those of the two Canadian ribotype 4 strains included in this study. Previous studies have demonstrated the presence of ribotype 6 only in Denmark and Norway ( Pedersen & Larsen 1993, 1995 ; Pedersen et al. 1994  ; Skov et al. 1995 ), but the present study showed that this ribotype was also present in Sweden, and most of these Swedish isolates, as well as the Norwegian ribotype 6 isolate, belonged to the same PhP type. Two Canadian isolates had a ribotype pattern not previously seen. This pattern was designated pattern 13 and these strains also shared an identical PhP type not observed in other strains.

Recent investigations using pulsed-field gel electrophoresis, plasmid analyses and ribotyping have shown that strains from northern Europe and southern Europe seem to belong to different clonal lineages, and that Japanese and American strains are different from these ( Skov et al. 1995  ; Tiainen et al. 1995  ; Pedersen et al. 1996 ). Likewise, in a study using a combination of different typing methods ( Kühn et al. 1997 ), it was found that comparing strains from close geographic locations yielded a lower diversity than comparing strains from distant locations, and strains from certain fish species showed lower diversity than strains from other sources. The present results on the basis of PhP typing confirmed these observations. As indicated by the dendrogram ( Fig. 3), strains from different geographic regions, with few exceptions, clustered separately.

An interesting observation was that the prevalence of different clonal lineages, as judged by their ribotype patterns and PhP types, varied from one year to another and in Finland, it seemed that one clonal lineage, the combination of ribotype 1 and PhP type 1, became increasingly important, its prevalence increasing from 20% of the isolates in 1992 to approximately 80% in 1996. The cause of this development is now known but, as previously mentioned, the fact that most combinations of ribotype and PhP type were restricted to specific geographic areas, together with the fact that the prevalence of specific types varies from one year to another, suggests that either the clones change their properties quickly or they are present locally, in some hitherto unknown reservoir, from where they can be recruited and cause infections whenever the conditions are right.

The much higher diversity of the Finnish PhP type 1 strains compared with the Italian IT3 strains, which were all identical with respect to serotype, ribotype and plasmid profile, can be explained by a changing effect ; a specific clone will develop minor deviations that increase with time. Thus, a develops into b that develops into c, etc., where a and b are very closely related and b and c are very closely related, but a and c are less closely related. A high diversity index was also noticed among other clonal lineages from Finland, Sweden and Denmark, suggesting that these clonal lineages have been present there for long enough time to develop descendants that have subsequently developed separately and undergone small evolutionary changes. In contrast, the very low diversity index among the IT3 isolates shows that this clone must be a new one that has emerged and spread quickly and has not yet had time to develop any significant diversity.

The first study of ribotypes of V. anguillarum serogroup O1 was carried out by Pedersen & Larsen (1993). These authors concluded that ribotyping could be a valuable tool in epidemiological investigations of this serogroup. Later, a combination of ribotyping and pulsed-field gel electrophoresis was found to increase the discriminatory power significantly ( Skov et al. 1995 ). However, the present study showed that ribotyping alone was not sufficiently discriminative to distinguish between strains from different locations. PhenePlate typing proved to be more discriminative and allowed us to follow the evolution of clones and the development of minor deviations over time, and by using a combination of these two techniques, excellent discrimination between strains was obtained. On the basis of the results from this study, different typing methods for V. anguillarum are recommended, depending on the collection of strains as well as the purpose of the study. We would certainly recommend O-serotyping as the primary typing method as very important properties, especially virulence, are associated with the O-serogroup. Similarly, we found important information in the plasmid profiles of the strains, especially those belonging to serogroup O1. A more detailed plasmid analysis could have included restriction analysis of the plasmids but was not carried out in the present study. For the typing of serogroups other than O1, ribotyping yielded very good discrimination and for those isolates, plasmids and PhP typing yielded little further information. However, for serogroup O1, ribotyping yielded very limited information as the vast majority of isolates belonged to only three different ribotypes. For this serogroup, the PhP data were very discriminative and provided extremely useful information that allowed us to draw important conclusions as to the clonality, ecology and epizootiology for this serogroup.

Acknowledgements

The technical assistance of Ms Farah Bahrani and the financial support of The Danish Agricultural and Veterinary Research Council, grant no. 9503658, and the Swedish Council of Forestry and Agricultural Research, grant no. 722.0985/93, is gratefully appreciated.

Ancillary