Detection of Flavobacterium psychrophilum from fish tissue and water samples by PCR amplification

Authors


Dr T. Wiklund, Fish Disease Laboratory, Danish Institute for Fisheries Research, Stigbøjlen 4, DK-1870 Frederiksberg C, Denmark (e-mail: tow@kvl.dk).

Abstract

Rainbow trout fry syndrome and cold-water disease, caused by Flavobacterium psychrophilum, are important diseases in farmed salmonids. Some of the presently available techniques for the detection of Fl. psychrophilum are either time consuming or lack sufficient sensitivity. In the present investigation, the possible detection of Fl. psychrophilum from fish tissue and water samples was examined using nested PCR with DNA probes against a sequence of the 16S rRNA genes. The DNA was extracted using Chelex® 100 chelating resin. The primers, which were tested against strains isolated from diseased fish, healthy fish, fish farm environments and reference strains, proved to be specific for Fl. psychrophilum. The obtained detection limit of Fl. psychrophilum seeded into rainbow trout brain tissue was 0·4 cfu in the PCR tube, corresponding to 17 cfu mg−1 brain tissue. The PCR-assay proved to be more sensitive than agar cultivation of tissue samples from the brain of rainbow trout injected with Fl. psychrophilum. In non-sterile fresh water seeded with Fl. psychrophilum the detection limit of the PCR-assay was 1·7 cfu in the PCR tube, corresponding to 110 cfu ml−1 water. The PCR-assay detected Fl. psychrophilum in water samples taken from a rainbow trout farm, but Fl. psychrophilum could not be isolated using inoculation on selective agar. The method presented here has the potential to detect low levels of Fl. psychrophilum in fish tissue and in water samples, and the technique can be a useful tool for understanding the epidemiology of Fl. psychrophilum.

Rainbow trout fry syndrome (RTFS) and cold-water disease, which both are caused by the bacterium Flavobacterium psychrophilum (former Cytophaga psychrophila and Flexibacter psychrophilus), are important infectious diseases in farmed fish ( Dalsgaard 1993; Holt et al. 1993 ; Lorenzen et al. 1997 ). The pathogen affects mainly salmonids (primarily rainbow trout (Oncorhynchus mykiss) and coho salmon (Oncorhynchus kisutch)), but it has also occasionally been isolated from non-salmonids, e.g. from an eel (Anguilla anguilla), cyprinids ( Lehmann et al. 1991 ) and pale chub (Zacco platypus) ( Iida & Mizokami 1996).

Flavobacterium psychrophilum has traditionally been detected from fish by cultivation of infected tissue on agar media and subsequent identification of isolates using biochemical and serological techniques. However, in recent studies done in this laboratory (unpublished results), yellow-pigmented colonies very similar to Fl. psychrophilum were isolated from fish. Based on phenotypic characteristics, correct identifications of these isolates sometimes proved to be difficult. Moreover, the identification was time-consuming and required a high degree of skill. In addition to agar isolation, methods for detection of bacterial cells in fish tissue using fluorescent polyclonal antibodies against Fl. psychrophilum and enzyme linked immunosorbent assay (ELISA) have been described ( Lorenzen & Karas 1992; Rangdale & Way 1995).

The presence and persistence of Fl. psychrophilum in natural aquatic environments has so far not been reported. A reason for this might be the lack of proper detection methods. Isolation of Fl. psychrophilum using agar cultivation is in many cases unsuitable due to the slow growth of the cells and overgrowth or inhibition by rapidly multiplying water bacteria. A recent study on the survival of Fl. psychrophilum in sterile laboratory microcosms, however, indicated that the cells were able to survive in significant numbers for several months in fresh and brackish water ( Wiklund et al. 1997 ).

However, some of the presented techniques are either time-consuming or lack sufficient sensitivity to allow the detection of low levels of Fl. psychrophilum from the environment, as well as from fish tissue. There is a need for sensitive and rapid techniques to detect Fl. psychrophilum in low numbers from fish and environmental samples in order to examine the epidemiology of this pathogen. A PCR-assay has been suggested as a suitable method for the discovery of fish pathogenic bacteria from environmental and tissue samples ( Gustafson et al. 1992 ; Brown et al. 1994 ; Arias et al. 1995 ; Coleman et al. 1996 ). Recently, DNA primers were developed and used in PCR-assays for the detection of Fl. psychrophilum in fish tissue samples ( Izumi & Wakabayashi 1997; Urdaci et al. 1998 ). However, the sensitivity of the method was not properly examined.

The aims of the present study were to apply a nested PCR-assay for the detection of Fl. psychrophilum using a simple DNA extraction, and to evaluate the ability of the method to detect low levels of the pathogen from experimentally infected fish tissues and water samples. The primers for the detection of Fl. psychrophilum were initially examined for specificity against a number of strains of Fl. psychrophilum isolated from different sources, related and non-related fish pathogenic bacteria, and yellow-pigmented bacteria isolated from fish and fish farm environments.

MATERIALS and METHODS

Bacterial strains

The bacterial strains examined are listed in Table 1. Some of the strains belong to the culture collection at the Fish Disease Laboratory, Frederiksberg, Denmark and Institute of Parasitology, Åbo, Finland. Of the strains examined, 81 had been previously biochemically and serologically identified as Fl. psychrophilum, two were previously identified as Fl. johnsoniae-like organisms ( Rintamäki-Kinnunen et al. 1997 ), and 33 were identified as yellow-pigmented bacteria (YPB) biochemically and serologically different from Fl. psychrophilum. In addition, reference strains from type culture collections were included in the study. The strains were originally isolated from farmed fish, both from diseased fish and from fish without any disease signs, and from water or sediment samples from fresh water fish farms ( Table 1). A virulent strain 950106–1/1 ( Madsen & Dalsgaard 1998), which was previously isolated from a disease outbreak of RTFS in a Danish freshwater fish farm in 1995, was used in the seeding experiments. The bacterial strains in the present study were grown in tryptone yeast extract salts (TYES) broth ( Holt et al. 1993 ), with shaking, at 15 °C for 2 days. When Fl. psychrophilum was cultivated on solid media, 1·1% agar was added to the TYES broth. For the DNA extraction, the cells were harvested from the broth culture by centrifugation at 10 000 rev min−1 (Biofuge, Fresco, Heraeus Instruments, Hanau, Germany) for 10 min at 5 °C, then washed once with filtered (0·2 μm) and autoclaved (121 °C, 15 min) distilled water. The cells were re-suspended in distilled water ( Toyama et al. 1994 ) to a cell concentration of 10 mg wet weight ml−1. A sub-sample (10 μl) of the cell suspension was either used immediately for DNA extraction, followed by PCR amplification using the primers PSY1 and PSY2 (see below), or the cell suspension was stored at −20 °C until required.

Table 1.  Bacterial strains included in the present study and the results from the PCR amplification
OrganismSource
 YearCountryDisease outbreakNumber testedPCR positive
  • Strains kindly provided by:

  • *

    Varpu Hirvelä-Koski, EELA, Oulu, Finland (3474/94, 5563/95, 255/96).

  • Anders Hellström, SVA, Uppsala, Sweden (SVA1013, SVA1045, SVA1081, SVA1090, SVA1283, SVA1289, SVA1298, SVA1324).

  • Elizabeth Crump, University of Victoria, Victoria, B.C., Canada (N9701, N9702, N9703, N9704, N9705, N9706).

  • §

    Päivi Rintamäki-Kinnunen, University of Oulu, Department of Biology, Oulu, Finland (CR6/95, CR26/95).

Strains isolated from fish
Flavobacterium psychrophilumRainbow trout1994–98Denmark+1414
Flavobacterium psychrophilumRainbow trout1995–98Denmark1616
Flavobacterium psychrophilumRainbow trout, lake trout1993–96Finland *+1313
Flavobacterium psychrophilumRainbow trout, Atlantic salmon, sea trout1996Sweden +88
Flavobacterium psychrophilumChinook-, coho salmon1997Canada +66
Flavobacterium johnsoniae-like Atlantic salmon1995Finland §+20
Yellow-pigmented coloniesRainbow trout1994–98Denmark200
Strains isolated from environmental samples
Flavobacterium psychrophilumWater1997–98Denmark±2424
Yellow-pigmented coloniesWater, sediment1997–98Denmark±130
Reference strains
Flavobacterium psychrophilum NCIMB 1947TCoho salmon USA  +
Flavobacterium columnare NCIMB 2248TChinook salmon USA  
Flexibacter maritimus NCIMB 2153Black sea bream Japan  
Flavobacterium aquatile NCIMB 8694TWell water UK  
Flavobacterium branchiophilum ATCC 35035TYamame Japan  
Aeromonas salmonicida subsp. salmonicida NCIMB 1102TAtlantic salmon UK  
Aeromonas salmonicida subsp. achromogenes NCIMB 1110TSea trout UK  
Vibrio anguillarum NCIMB 6TCod Norway  
Yersinia ruckeri ATCC 29473TRainbow trout USA  

DNA extraction

The DNA was extracted from the bacterial cells according to a modified method described by Izumi & Wakabayashi (1997). In short, a sub-sample of the bacterial cell suspension or fish tissue suspension was mixed with 300 μl of a 5% Chelex® 100 suspension (chelating resin, iminodiacetic acid, Sigma-Aldrich Chemie, Steinheim, Germany) ( Walsh et al. 1991 ), vortexed at high speed for 10–15 s, incubated for 30 min at 56 °C, vortexed, boiled for 20 min, cooled on ice, vortexed, and centrifuged at 10 000 rev min−1 for 10 min at 5 °C. A 5 μl aliquot of the supernatant fluid was used for PCR amplification of the extracted DNA.

Primers

Species-specific primers complementary to sequences in two variable regions of the 16S rRNA genes of Fl. psychrophilum, previously developed and described by Toyama et al. (1994 ), were used. The primers are referred to as PSY1 and PSY2. For the nested PCR, two universal primers, 20F and 1500R ( Toyama et al. 1994 ), complementary to conserved regions of most eubacterial 16S rRNA ( Weisburg et al. 1991 ), were included in the amplification of the extracted DNA ( Table 2). The universal primers were used in the first PCR step (PCR1), and the specific primers were used in the second step (PCR2) of the nested PCR and in a standard PCR in the analysis of the bacterial strains. All oligonucleotides were synthesized by T-A-G-Copenhagen (Copenhagen, Denmark).

Table 2.  The sequences and positions of primers for the amplification of Flavobacterium psychrophilum using nested PCR
Primer *Position Sequence (5′−3′)Product length (bp)
  • *

    Primers 20F and 1500R were used in the first PCR step and primers PSY1 and PSY2 were used in the second step of the nested PCR amplification.

  • Position of primers corresponding to the numbers of the Escherichia coli 16S rRNA.

20F8–27AGA GTT TGA TCA TGG CTC AG 
1500R1510–1492GGT TAC CTT GTT ACG ACT T 
PSY1190–206GTT GGC ATC AAC ACA CT 
PSY21278–1262CGA TCC TAC TTG CGT AG1089

PCR amplification

The PCR amplifications were done using a commercial kit, Ready–To–Go® PCR beads (Amersham Pharmacia Biotech, Millwaukee, USA), which included all necessary reagents. All amplifications were done in a 25 μl reaction mixture, which contained 1·5 units of Taq DNA polymerase, 10 mmol l−1 Tris-HCl (pH 9·0 at room temperature), 50 mmol l−1 KCl, 1·5 mmol l−1 MgCl2, 200 μmol l−1 of each dNTP and stabilizers. In PCR1, 5 pmol of each of the universal primers 20F and 1500R and 5 μl of the DNA extractions from the Chelex® 100-treated samples were used. The product from PCR1 was diluted to a 5% solution in sterile TE buffer (10 mmol l−1 Tris–HCl, 1 mmol l−1 EDTA, pH 8·0). In PCR2, 8 pmol of the primer PSY1, 3 pmol of the primer PSY2 and as a template, 5 μl of the diluted product from PCR1, were used.

The reaction mixtures were amplified in a Perkin–Elmer GeneAmp PCR System 2400 thermal cycler (PE Applied Biosystems, Foster City, CA, USA). Identical amplification conditions were used in both PCR reactions of the nested PCR. The samples were initially denatured at 95 °C for 5 min, followed by amplification cycles including denaturation of the DNA for 30 s at 95 °C, annealing of primers for 30 s at 57 °C and primer extension for 60 s at 72 °C. After the last cycle, the PCR mixture was incubated for 5 min at 72 °C. In the amplification of DNA from the bacterial strains, only the PCR2 step was used and subjected to 28 amplification cycles. In the amplification of DNA extracted from water or fish tissue, nested PCR was used with 30 or 35 cycles of amplification of both PCR1 and PCR2.

Detection of PCR products

The amplified PCR products were detected by electrophoresis using 1% agarose gels. A sample (10 μl) of the PCR products was mixed with 3 μl of tracking dye and subjected to electrophoresis. After electrophoresis, the gel was stained with ethidium bromide and photographed at u.v. transillumination. The primers PSY1 and PSY2 were designed to give a product of 1089 bp.

Seeding of fish tissue

For the seeding experiment, rainbow trout (weight: approximately 10 g) with no known history of infection with Fl. psychrophilum were used. The brain of the fish was chosen as the target tissue for the detection of Fl. psychrophilum. The dorsal skull of the rainbow trout was opened aseptically and the brain was transferred into pre-weighed sterile Eppendorf tubes. The brain tissue was weighed and stored at −20 °C until used for the seeding experiment. The tissue was homogenized with sterile phosphate-buffered saline (PBS) to a 50% (w/v) suspension by repeated pipetting. Aliquots (50 μl) of the brain suspension were transferred into sterile Eppendorf tubes and subsequently seeded with an equal volume of a bacterial suspension described below. The final dilution of the brain was 25% (w/v). For the seeding, Fl. psychrophilum (strain 950106–1/1) was grown for 2 days in TYES broth at 15 °C. The bacterial culture was washed once in sterile PBS and serially 10-fold diluted in sterile PBS. At this point, the number of bacterial cells was estimated using plate counts. Aliquots of the bacterial dilutions (50 μl) were seeded to the brain suspensions described previously, and to an equal volume of sterile PBS to serve as positive controls for the brain–bacteria suspensions. Washed bacteria, non-seeded brain suspension and PBS buffer were also included as one positive and two negative control samples for the PCR reaction. All suspensions were washed twice in sterile PBS. A 5 μl sample of the brain–bacteria, PBS–bacteria and control suspensions, and 1 μl of the washed bacterial suspension, were transferred into Eppendorf tubes, containing 300 μl of a Chelex® 100 suspension, for DNA extraction and treated as previously described. In order to examine the influence of the number of amplification cycles on the results, 30 and 35 cycles for both the first and the second step of the nested PCR in all combinations were studied.

Challenge of rainbow trout

For the detection of Fl. psychrophilum in brain tissue from challenged fish using PCR-assay, rainbow trout (average weight 1·0 g) were infected by intraperitoneal (i.p.) injection, co-habitation or bath challenge. Two strains of Fl. psychrophilum were used: in the i.p. injection experiments, strain 950106–1/1 and 900406–1/3, and in the co-habitation and bath challenge experiments, strain 950106–1/1. The i.p. injection challenge method described by Madsen & Dalsgaard (1998) was used. In short, three groups of fish were infected with 1·2 × 104 cfu (strain 950106–1/1), and 2·0 × 105 and 2·0 × 106 cfu (strain 900406–1/3), respectively. In the co-habitation experiment, non-injected rainbow trout (co-habitants) were introduced into the same aquarium as the fish infected i.p. with strain 950106–1/1. In the bath challenge experiment, 100 ml TYES broth with a 2 day culture of bacteria was diluted 1 : 10 with tap water. The fish were exposed to an aerated bacterial solution, containing 2 × 107 cfu ml−1, for 0·5 h. Fish in the control tank were bathed for 0·5 h in sterile TYES broth that was diluted 1 : 10 with tap water.

The brain, spleen and kidney of the dead and surviving fish were examined for the presence of Fl. psychrophilum using inoculation of tissue samples on TYES agar. For PCR amplification, the brain was transferred into a pre-weighed, sterile Eppendorf tube and stored at −20 °C until examined. The brain was homogenized to a 25% (w/v) suspension as previously described. A sample containing 5 μl of the brain suspension, 1 μl of a washed, diluted, bacterial (strain 950106–1/1) suspension (positive control) and 5 μl sterile PBS (negative control) was subjected to DNA extraction using the Chelex® 100 method as previously described. Samples (5 μl) of the supernatant fluid from the extractions were used for nested PCR amplification (35 cycles of both PCR1 and PCR2) of DNA from Fl. psychrophilum.

Seeding of non-sterile water with Fl. psychrophilum

For the seeding experiments of non-sterile water, fresh water was collected from a small lake without any connection to fish farms. The number of resident bacteria in the lake water was estimated by plate count on TYES agar. For the experiments, strain 950106–1/1 was treated as previously described for the seeding experiments in brain tissue. Aliquots (100 μl) of the bacterial dilutions were added to 900 μl portions of non-sterile lake water in Eppendorf tubes. Sterile PBS (100 μl) added to the lake water (900 μl) served as a negative control. The samples were centrifuged (13 000 rev min−1, 15 min, 5 °C) and washed twice with sterile PBS. After the last washing, the supernatant fluid was discarded (20 μl PBS was retained in the tube) and 300 μl of a 5% Chelex® 100 suspension were added to the tubes, which were treated as previously described. Samples (5 μl) of the extracted DNA solution were used for the amplification using nested PCR and 35 cycles of PCR1 and PCR2.

Water samples

In order to determine the presence of Fl. psychrophilum in fish farm environments, water samples from one freshwater fish farm cultivating rainbow trout were collected on two occasions (in July 1998 and September 1998). Samples were taken from the inlet water, from a fishpond (pond water), and from a channel collecting water from the whole farm (outlet water). The presence of Fl. psychrophilum in the collected water samples was examined using nested PCR as well as agar plate inoculation. For the PCR-assay, 2 ml of the water were transferred into sterile tubes and centrifuged (13 000 rev min−1, 15 min, 5 °C), the pellet was washed twice with sterile, filtered PBS and treated as previously described. Aliquots (5 μl) of the Chelex® 100-treated samples were used in the PCR-assay. For the plate count, aliquots of the water samples were inoculated on TYES agar containing 50 μg sulphadiazine ml−1 and 10 μg trimethoprim ml−1. Yellow-pigmented colonies showing characteristics similar to those of Fl. psychrophilum were chosen for further biochemical and serological characterization.

Results

Specificity of the primers

The primers PSY1 and PSY2 generated a fragment of identical size (1089 bp) from all tested strains of Fl. psychrophilum including the type strain NCIMB 1947T ( Table 1, Fig. 1). PCR amplification products were not detected when DNA from the non-identified YPB, Fl. johnsoniae-like strains, and the other reference strains, were used.

Figure 1.

Agarose gel electrophoresis patterns of amplification products from DNA extracted from pure cultures of lysed bacteria. Lane A: molecular weight marker (123 bp); lane B: Flavobacterium psychrophilum NCIMB 1947T; lane C: Fl. columnare NCIMB 2248T; lane D: Flexibacter maritimus NCIMB 2153; lane E: Flavobacterium aquatile NCIMB 8694T; lane F: Fl. branchiophilum ATCC 35035T; lanes G–I: Fl. psychrophilum, Danish strains 950106–1/1, 62/4 and 68/4 A; lanes J–L: yellow-pigmented bacteria, strains 24/1 A, 51/1 A and 87/3; lanes M–N: Fl. psychrophilum isolated from water, strains 1–145 and 1–150; lanes O–Q: Fl. psychrophilum Finnish strains T1–1, V9 and P3–3/96; lane R: negative control; lane S: molecular weight marker (123 bp)

Seeding experiments with brain tissue

In the seeding experiments, a fragment of the expected size (1089 bp) was amplified, using nested PCR, from the brain tissue, and sterile PBS seeded with Fl. psychrophilum in different concentrations and from a pure culture of Fl. psychrophilum in the positive control tube ( Fig. 2). Amplification products were not obtained from the negative controls. The experiments showed that as few as 0·4 cfu could be detected in the PCR tube containing DNA from both brain suspensions and PBS inoculated with bacteria ( Fig. 2). When using 35 amplification cycles for both PCR1 and PCR2, a strong signal was obtained for 0·4 cfu seeded into both brain tissue and PBS. When using 35 cycles for PCR1 and 30 cycles for PCR2, a weaker band was visualized in the gel electrophoresis from the tube containing 0·4 cfu of bacteria in brain suspension, compared with the product from the tube containing identical number of bacteria in PBS ( Fig. 2). When the amplification was run with 30 cycles for PCR1 and 30 or 35 cycles for PCR2, no product from the tube containing 0·4 cfu was visible in the gel electrophoresis. The lowest number of bacteria (0·4 cfu) detected in the PCR tube corresponded to about 17 cfu of Fl. psychrophilum mg−1 brain tissue.

Figure 2.

Agarose gel electrophoresis patterns of amplification products from DNA extracted from rainbow trout brain tissue (25% w/v) and sterile PBS seeded with different numbers of Flavobacterium psychrophilum and from brain tissue of rainbow trout artificially infected with Fl. psychrophilum. The DNA from Fl. psychrophilum seeded into brain tissue or PBS was amplified using nested PCR and 35 and 30 cycles for the first, respectively, and second amplification. Lane A: molecular weight marker (123 bp); lane B: fish brain seeded with 35·5 cfu; lane C: fish brain seeded with 3·6 cfu; lane D: fish brain seeded with 0·4 cfu; lane E: fish brain seeded with 0·04 cfu; lane F: fish brain without bacteria; lane G: PBS seeded with 35·5 cfu; lane H: PBS seeded with 3·6 cfu; lane I: PBS seeded with 0·4 cfu; lane J: PBS seeded with 0·04 cfu; lane K: PBS without bacteria; lane L: Fl. psychrophilum, strain 950106–1/1; lane M: fish brain from a control fish; lane N: fish brain from a fish challenged (i.p.) with Fl. psychrophilum strain 950106–1/1; lane O: molecular weight marker (123 bp)

Challenge experiments in rainbow trout

The results from the detection of Fl. psychrophilum using PCR amplification and agar isolation from brain tissues of challenged rainbow trout are shown in Table 3. of the fish challenged by i.p. injection, DNA from Fl. psychrophilum was detected, using nested PCR, in all brain tissue samples that were positive in agar cultivation (n = 15). In addition, Fl. psychrophilum was also detected using PCR in seven out of 14 examined brain tissues that were negative in agar cultivation. The presence of the bacterium in some of these fish was confirmed by the isolation of Fl. psychrophilum from internal organs. However, the isolation on agar of the injected pathogen from the brain was, in some cases, hampered by overgrowth on the plate with different colonies not showing yellow colour, or with a colony appearance different from Fl. psychrophilum. No differences in the detection using PCR were observed between the two strains used for the i.p. injection. DNA or bacteria could not be amplified from the examined co-habitants using either PCR or agar inoculation. Detection of Fl. psychrophilum in fish that were bath challenged gave a slightly improved result using PCR (six positive out of 15 tested) compared with agar cultivation (five positive out of 15 tested).

Table 3.  Detection of Flavobacterium psychrophilum from rainbow trout, challenged by either i.p. injection, co-habitation or bath, on TYES agar inoculated with tissue from the brain and internal organs (spleen and/or kidney) and PCR amplification of brain tissue
 Isolation of Flavobacterium psychrophilum on TYES agar from 
Mode of infectionNumber of fish examinedbrainspleen and/or kidneyPCR positive (n)
i.p. Injection10101010
i.p. Injection5505
i.p. Injection3032
i.p. Injection11005
Co-habitation5000
Bath5555
Bath2021
Bath8000
Control10000
Total59202028

Seeding experiments with Fl. psychrophilum in non-sterile water

In the experiments with Fl. psychrophilum seeded into non-sterile lake water, the obtained detection limit was 1·7 cfu in the PCR tube using nested PCR amplification of the extracted DNA. This corresponded to 110 cfu ml−1 of Fl. psychrophilum in the water sample ( Table 4). The negative and positive controls gave corresponding negative and positive amplification results. The number of resident bacteria in the lake water used for the seeding experiments was estimated to be 4·5 × 104 cfu ml−1.

Table 4.  The results of PCR amplification of water samples seeded with 10-fold dilutions of Flavobacterium psychrophilum. The concentration of bacteria in the DNA extraction tube was calculated on a final sample volume of 320 μl (20 μl after the centrifugation of the water sample and 300 μl Chelex 100 added to the tube)
 Bacterial dilutions
Number of bacteria in water sample (1 ml)1100110111cfu
Concentration of bacteria in DNA extraction tube3·40·340·0340·0034cfu μl−1
Estimated number of bacteria in PCR tube17·21·70·170·017cfu
PCR amplification result++ 

Detection of Fl. psychrophilum in water from fish farm

In the analysis of water sampled from a freshwater fish farm, PCR amplification products were detected from the inlet, pond and outlet samples on both sampling occasions. Corresponding isolation attempts of Fl. psychrophilum on selective agar plates seeded with water from the farm were unsuccessful.

Discussion

The PCR-assay with the primers PSY1 and PSY2 was found to be specific for Fl. psychrophilum, yielding a 1089 bp DNA product. Only DNA from the Fl. psychrophilum strains was amplified, and amplification products were not recovered from any other examined reference strain or yellow-pigmented strain isolated from fish or fish farm environments. These results are in agreement with those reported by Toyama et al. (1994 ) and Bader & Shotts (1998). Although these primers have been tested against a collection of different strains, care is recommended when interpreting the results from field studies. Only the bacteria that can be isolated on agar can be tested with the primers, and a large majority of the bacteria present in the water environment cannot be isolated and thus, cannot be tested ( Hiney & Smith 1998). Consequently, a positive PCR product from environmental samples cannot exclusively be considered as amplification products from the supposed target cells, but from non-cultivable bacterial cells, that cannot be tested against the primers used. Hiney & Smith (1998) discussed this problem in detail. Urdaci et al. (1998 ) recently reported that the primers developed by Toyama et al. (1994 ) revealed some non-specific amplification using their PCR conditions. Consequently, Urdaci et al. (1998 ) designed new primers slightly different from those used by Toyama et al. (1994 ) for the amplification of DNA from Fl. psychrophilum. Bader & Shotts (1998) achieved better amplification using different thermal cycle conditions but identical primers to those designed by Toyama et al. (1994 ). In our initial pilot experiments, non-specific binding of the primers PSY1 and PSY2 to DNA occurred from other bacterial species and from the YPB (results not shown). However, the amplified products were always of different sizes from the expected 1089 bp, and optimized amplification conditions eliminated these non-specific bindings. Identification of Fl. psychrophilum by biochemical characterization is a rather time-consuming and slow procedure, partly due to the slow growth of the pathogen. The use of PCR significantly enhances the identification of Fl. psychrophilum, which can be completed during one day.

One of the main problems concerning the PCR-assay is the possible inhibition of the enzymatic amplification reaction by different compounds present in the samples containing the target DNA. Blood containing haemoglobin, serum proteins and other components may act as strong inhibitors of the PCR reaction ( Wilson 1997). In order to minimize the possible inhibiting action of the blood, the brain was chosen as target tissue for the detection of Fl. psychrophilum in the PCR amplification. The amount of blood in the brain is low compared with other internal organs such as the kidney and spleen. Additionally, Fl. psychrophilum has been reported to be readily isolated from the brain tissue or cranial space of infected fish ( Kent et al. 1989 ; Holt et al. 1993 ).

The results from the seeding of brain tissue with Fl. psychrophilum indicate that the method used is extremely sensitive, detecting as few as 0·4 cfu in the PCR tube, which corresponds to 17 cfu mg−1 brain tissue. In other studies, detection of Fl. psychrophilum in kidney, skin and spleen tissue from fish using PCR amplification was reported ( Izumi & Wakabayashi 1997; Urdaci et al. 1998 ). Izumi & Wakabayashi (1997) indicated that the limit of detection of Fl. psychrophilum in kidney tissue was 1·5 cfu in the PCR reaction tube. Unfortunately, the description of the methods used was rather limited. In a study of kidney tissue seeded with Aeromonas salmonicida, approximately 10 cfu mg−1 tissue could be detected using standard PCR ( Gustafson et al. 1992 ). In investigations of other bacteria, detection limits of one bacterial cell have been reported ( Arias et al. 1995 ; Lee et al. 1995 ). Our results on the specificity of the primers indicated that the PCR-based assay is probably more specific compared with immunological methods in which antigenic cross-reactivity could be a problem ( Lorenzen & Karas 1992).

The detection of 0·4 cfu of Fl. psychrophilum using PCR is close to, or below, the theoretical detection limit of one single-stranded copy of target DNA. This discrepancy can be explained in different ways. (i) Possible dead and/or injured cells in the suspension will not be detected when estimating the number of cfu in the bacterial dilutions using agar inoculation; these dead and/or injured cells will be detected in the PCR-assay. (ii) Broth cultures of Fl. psychrophilum often form aggregations or short chains of cells, which will subsequently produce only one colony on agar. Due to the processing of the bacterial–brain suspension, it is possible that the aggregates disintegrate, theoretically producing more than one cell from one cfu. (iii) In the seeding experiments, actively growing cells were used that were inoculated into the brain suspension. Further multiplication of the bacterial cells could have occurred during processing; alternatively, more than one genome was present in the growing cells. However, due to the slow multiplication rate of Fl. psychrophilum this is rather unlikely. In a previous study of Renibacterium salmoninarum,McIntosh et al. (1996 ) concluded that dead and/or damaged cells, which failed to produce colonies on agar, most probably contributed to the detection of DNA in tissue samples using PCR amplification. Kapperud et al. (1993 ) suggested that further multiplication of the examined cells of Yersinia enterocolitica may have occurred, slightly lowering the obtained sensitivity of a PCR detection in naturally contaminated samples containing weakened cells.

In the i.p. injection experiments, the injected bacteria were found in brain tissue more frequently (76%, 22 out of 29 brains examined) using PCR amplification of DNA, compared with isolation on agar (52%, 15 out of 29 brains examined) ( Table 3). These negative results from the agar cultivation were possibly partly due to fast growing contaminating bacterial colonies inhibiting the multiplication of cells of Fl. psychrophilum on the inoculated agar plates. The contaminating bacteria were probably non-pathogens that invaded the dead fish, but these bacteria did not affect the PCR reaction and an amplification product was detected.

The results indicate that the PCR method used in this study is more sensitive than agar cultivation for the detection of Fl. psychrophilum from infected rainbow trout. In addition, the PCR-assay is faster, giving results in 24 h, and labour-saving compared with the traditional method with isolation of the pathogen on agar and subsequent identification using biochemical characterization.

In the co-habitation experiments, Fl. psychrophilum could not be detected in the brain or any other tissue of the examined co-habitants using agar or PCR amplification, suggesting that the bacterium was most probably not transferred from infected fish to the uninfected co-habitants. There was no mortality of the co-habitants and they did not show any disease signs.

So far, there are no reports of the isolation of Fl. psychrophilum from water and environmental samples using agar inoculation or serological methods. The present study indicates that Fl. psychrophilum can be detected from the water in low levels using PCR amplification. The measured detection level in the PCR tube of the bacteria in non-sterile fresh water was 1·7 cfu, which is slightly higher than that obtained for the brain–bacteria suspension (0·4 cfu). Since the method includes a centrifugation step for concentrating the bacteria in the sample, Fl. psychrophilum in low concentrations can be detected by centrifuging larger water volumes (5–50 ml) than those used in the present study. The use of agar inoculation of water samples for the isolation of Fl. psychrophilum is rather unsuitable due to the slow multiplication rate of Fl. psychrophilum and the presence of a large number of different bacteria in fish farm environments, especially during the summer. In this study, agar plates with incorporated antibiotics were used in order to inhibit the growth of ‘contaminating’ bacteria, but Fl. psychrophilum could not be isolated from water samples from the fish farm. In the PCR assay, a fragment of the expected size (1089 bp) was amplified from the water samples, suggesting that Fl. psychrophilum was present. During previous sampling occasions in spring and winter from the same farm, Fl. psychrophilum was isolated on agar plates from inlet, pond and outlet water (results not shown). The present results indicate that the PCR-assay is a suitable tool for the detection of Fl. psychrophilum in water samples.

The nested PCR assay described in the present study allows the direct and specific detection of Fl. psychrophilum in fish tissue and water samples, within 24 h, compared with several days and up to 1 week for isolation, and 1–2 further weeks for identification, of Fl. psychrophilum using plating and biochemical and serological methods. The Chelex® extraction procedure is a very simple and fast procedure, reducing the use of lengthy organic extractions and alcohol precipitation for the DNA extraction, and it has the potential to detect low levels of the pathogen in fish tissue and particularly from fish farm environments. So far, there is a lack of other specific and sensitive methods for the detection of Fl. psychrophilum from samples highly contaminated with other bacteria, and the present PCR method can be a useful tool for future understanding of the epidemiology of Fl. psychrophilum.

Acknowledgements

Kirsten Kaas, Farah S. Bahrani and Ann Charlot Ejsing are acknowledged for excellent technical work. The authors are indebted to their colleagues for kindly providing bacterial strains. This work was supported by the Ministry of Food, Agriculture and Fisheries, Denmark.

Ancillary