Genetic variation among Flavobacterium psychrophilum isolates from wild and farmed salmonids in Norway and Chile



Patricia Apablaza, Fish Diseases Research Group, Department of Biology, University of Bergen, Norway, Post-Box 7803, 5020 Bergen, Norway. E-mail:



To aim of the study was to describe the genetic relationship between isolates of Flavobacterium psychrophilum with a main emphasis of samples from Chile and Norway. The isolates have been obtained from farmed salmonids in Norway and Chile, and from wild salmonids in Norway, but isolates from North America and European countries are also included in the analysis.

Methods and Results

The study is based on phylogenetic analysis of 16S rRNA and seven housekeeping genes (HG), gyrB, atpA, dnaK, trpB, fumC, murG and tuf, and the use of a multilocus sequence typing (MLST) system, based on nucleotide polymorphism in the HG, as an alternative to the phylogenies. The variation within the selected genes was limited, and the phylogenetic analysis gave little resolution between the isolates. The MLST gave a much better resolution resulting in 53 sequence types where the same sequences types could be found in Chile, North America and European countries, and in different host species.


Multilocus sequence typing give a relatively good separation of different isolates of Fl. psychrophilum and show that there are no distinct geographical or host-specific isolates in the studied material from Chile, North America and Europe. Nor was it possible to separate between isolates from ulcers and systemic infections vs isolates from the surface of healthy salmonids.

Significance and Impact of the Study

This study shows a wide geographical distribution of Fl. psychrophilum, indicating that the bacterium has a large potential for transmission over long distances, and between different salmonid hosts species. This knowledge will be important for future management of salmonids diseases connected to Fl. psychrophilum.


Flavobacterium psychrophilum has been detected worldwide since the first isolation in Washington, USA (Borg 1960). The bacterium has been reported from Norway (Bornø et al. 2009; Duesund et al. 2010; Nilsen et al. 2011a), Chile (Bustos et al. 1995; Valdebenito and Avendaño-Herrera 2009) and in the aquaculture of several other countries (Nematollahi et al. 2003). The disease caused by Fl. psychrophilum was first described by Davis (1946) and the pathological findings included open lesions in the peduncle area, which led to the former name of ‘peduncle disease’. Currently, the clinical syndromes caused by Fl. psychrophilum are named rainbow trout fry syndrome (RFTS) (Rangdale et al. 1997) and bacterial cold water disease (BCWD) (Bernardet and Kerouault (1989). Flavobacterium psychrophilum has been isolated from a variety of different hosts although Coho salmon Oncorhynchus kisutch (Walbaum) and rainbow trout Oncorhynchus mykiss (Walbaum) are suggested as the most susceptible species (Nilsen et al. 2011a). Other salmonids fish include Atlantic salmon (Salmo salar, L), and several Oncorhynchus, and Salvelinus spp (Nematollahi et al. 2003). During the later years, Fl. psychrophilum has emerged as a major salmonids pathogen in Norway and Chile causing economical losses as consequence of high mortalities and the increased costs derived from the use of antibiotics (Valdebenito and Avendaño-Herrera 2009; Olsen 2011).

Flavobacterium psychrophilum can be both horizontally (Madsen and Dalsgaard 1999; Madetoja et al. 2000) and vertically transmitted (Holt et al. 1993; Brown et al. 1997; Ekman et al. 1999; Amita et al. 2000; Cipriano 2005; Madsen et al. 2005). The bacterium has been isolated from salmonid eggs (Brown et al. 1997) even after routine standard iodophor disinfection (Cipriano 2005), from ovarian fluids and on the surface of eggs from Atlantic salmon (Salmo salar, L), and in ovarian fluids and milt from rainbow trout broodfish (Ekman et al. 1999; Madsen et al. 2005). Hence, it is reasonable to assume that movement of embryos from infected broodfish may play an important role in long distance transmission, by the means of international trade of fish eggs (Borg 1960; Yoshimizu 1996; Brown et al. 1997; Kumagai and Takahashi 1997; Ekman et al. 1999; Cipriano 2005; Madsen et al. 2005). Strain typing is an important aspect in any strategy for the prevention of transmission and assessing the best possible treatment and husbandry practice (Olive and Bean 1999; van Belkum et al. 2007).

In 2007, the complete genome of Fl. psychrophilum was published (Duchaud et al. 2007), making it possible to compare genes from individual isolates for developing genotyping systems with respect to virulence, host specificity and geographical distribution. Currently, difficulties remain in establishing a robust system for fast and accurate strain typing of Fl. psychrophilum (Izumi et al. 2003; Soule et al. 2005a,b; Hesami et al. 2008; Nicolas et al. 2008).

Maiden et al. (1998) proposed multilocus sequence typing (MLST) as a good tool for strain typing and mapping of strain diversity of Neisseria meningitides. This tool made it possible to perform comparative analyses based on data from different laboratories by the use of nucleotide sequence data. MLST has since been successfully used to differentiate individual clones within species assessing the variation in the DNA sequences of 5–10 housekeeping genes (HGs) (van Belkum et al. 2007). By comparing unique sequence types (STs) based on these HGs from different strains, it is possible to generate clusters in phylogenetic trees that separate for instance virulent clonal complexes from nonvirulent ones. This method has been used with good results on many species of bacteria including N. meningitides, which was the first species it was designed for (Maiden et al. 1998; Maiden 2006), Neisseria gonorrhoeae (Viscidi and Demma 2003) and Yersinia pseudotuberculosis (Ch'ng et al. 2011). Nicolas et al. (2008) applied a MLST system to 50 isolates of Fl. psychrophilum representing a worldwide distribution including several different host species and found several clonal complexes with marked association to host species. These clonal complexes were geographically distributed on different continents, and the author speculated that this association could reflect adaptive niche specialization and/or preferential routes of transmission. Part of the tools available in the MLST scheme developed by Nicolas et al. (2008) was used in this study to the different strains of Fl. psychrophilum.

The aim for this study is to describe the genetic variations in a selection of seven HGs, analysed by phylogeny and MLST, from Fl. psychrophilum isolates collected in Norway and Chile. This should give insight into the variation among Fl. psychrophilum isolates connected to farming of salmonids in both countries and the possible presence of identical isolates in these two geographical separated areas.

Materials and methods

Isolation of Flavobacterium psychrophilum

This study includes 44 isolates of Fl. psychrophilum obtained from 2006 to 2011, 18 from Norway, 25 from Chile and the type strain NCIMB 1947T. The isolates are from different locations, salmonid species, wild and farmed, and tissues (Table 1). The bacteria were isolated either from salmonids in the field or from salmonids that were delivered to our laboratory. Some of the fishes appeared healthy and some showed signs of BCWD.

Table 1. Isolates of Flavobacterium psychrophilum from Norway and Chile collected between 2006 and 2011, from different water sources, fish species, farmed or wild and tissues
CodeCountryCountyDate isolationWater sourceHostWild/farmedTissue
  1. Code indicates the country of isolation, Chile (Ch), Norway (No), year (2006–2011), number of isolate (1–45), fish species, A. salmon (As), rainbow trout (Rt), trout (T) and tissue, gills (G), wound (W), kidney (K), eggs (Eg), milt (M), spleen (Sp), skin (SK), internal (int) and internal–external (int-ext) surface of fish. Nd, no information available.

Ch06-1-Rt-EChileCurarrehue (X)2006RiverRainbow troutFarmedExt
Ch07-2-As-EChileHornopiren (X)2007EstuaryAtlantic salmonFarmedExt
Ch07-3-As-EChileLlanquihue (X)2007LakeAtlantic salmonFarmedExt
Ch07-4-Rt-E-IChileRupanco (X)2007LakeRainbow troutFarmedExt-int
Ch09-5-Rt-EChileValdivia (X)2009RiverRainbow troutFarmedExt
Ch09-6-As-E-IChileValdivia (X)2009RiverAtlantic salmonFarmedExt-int
Ch07-7-As-EChileLlanquihue (X)2007LakeAtlantic salmonFarmedExt
Ch08-8-As-EChileLlanquihue (X)2008LakeAtlantic salmonFarmedExt
Ch07-9-Rt-E-IChileAntuco (VIII)2007Spring waterRainbow troutFarmedExt-int
Ch08-10-Rt-E-IChileMelipeuco (IX)2008Spring waterRainbow troutFarmedExt-int
Ch08-11-As-EChilePuerto Cisnes (XI)2009EstuaryAtlantic salmonFarmedExt
Ch09-12-Rt-WChilePuerto Rosales (X)2009LakeRainbow troutFarmedWound
Ch09-13-As-WChileHornohuinco (X)2009LakeAtlantic salmonFarmedWound
Ch10-14-Rt-GChilePanguipulli (X)2010Fresh waterRainbow troutFarmedGill
Ch10-15-As-FChileChinquihue (X)2010Fresh waterAtlantic salmonFarmedFin
Ch10-16-Rt-GChileQuillaipe (X)2010Fresh waterRainbow troutFarmedGill
Ch08-17-Rt-EChileTalca (VII)2008RiverRainbow troutFarmedExt
Ch10-18-Rt-GChileQuillaipe (X)2010Fresh waterRainbow troutFarmedGill
Ch08-19-As-EChilePuerto Cisnes (XI)2008EstuaryAtlantic salmonFarmedExt
Ch07-20-Rt-E-IChileQuellon (X)2007LakeRainbow troutFarmedExt-int
Ch10-21-Rt-WChileLlanquihue (X)2010LakeRainbow troutFarmedWound
Ch10-22-As-ndChileLlanquihue (X)2010LakeAtlantic salmonFarmedNd
Ch07-23-As-E-IChileRupanco (X)2007LakeAtlantic salmonFarmedExt-int
Ch07-24-Rt-E-IChileAntuco (VIII)2007Spring waterRainbow troutFarmedExt-int
Ch10-25-SpChileHornopiren (X)2010NdNdNdSpleen
No09-26-As-KNorwayMøre og Romsdal2009Fresh waterAtlantic salmonWildKidney
No09-27-As-SKNorwayMøre og Romsdal2009Fresh waterAtlantic salmonWildSkin
No09-28-Rt-KNorwayHordaland2009BrackishRainbow troutFarmedKidney
No09-29-AsNorwayHordaland2009Fresh waterAtlantic salmonWildNd
No09-30-T-SNorwayRogaland2009Fresh waterTroutWildSpleen
No09-31-As-ENorwayMøre og Romsdal2009Fresh waterAtlantic salmonWildEggs
No09-32-As-GNorwayMøre og Romsdal2009Fresh waterAtlantic salmonWildGills
No09-33-As-GNorwayMøre og Romsdal2009Fresh waterAtlantic salmonWildGills
No10-34-T-GNorwayMøre og Romsdal2010Fresh waterTroutWildGills
No10-35-T-WNorwayMøre og Romsdal2010Fresh waterTroutWildWound
No10-36-T-WNorwayMøre og Romsdal2010Fresh waterTroutWildWound
No10-37-T-WNorwayMøre og Romsdal2010Fresh waterTroutWildWound
No10-38-As-WNorwayNord-Trondelag2010Fresh waterAtlantic salmonFarmedWound
No10-39-As-WNorwayMøre og Romsdal2010Fresh waterAtlantic salmonFarmedWound
No10-42-As-KNorwayHordaland2010Fresh waterAtlantic salmonWildKidney
No10-43-As-KNorwayMøre og Romsdal2010Fresh waterAtlantic salmonWildKidney
No10-44-As-MNorwayHordaland2010Fresh waterAtlantic salmonWildMilt
No11-45-As-WNorwayHordaland2011Sea waterRainbow troutFarmedWound
NCIMB 1947T Nd1948Fresh waterCoho salmonNdKidney

Bacterial isolates were obtained from several tissues including kidney, spleen, gonads and wound margins. All isolates were cultured on tryptone–yeast extract–salts agar and broth supplemented with glucose (Cepeda et al. 2004). Agar plates were incubated at 15°C in a KB 5410 (Thermaks, Bergen, Norway) incubator for a period up to 3 weeks. After the colony screening by morphology and phenotypic characteristics (colonies features, direct microscopy of smears, Gram staining), the pure culture was verified as Fl. psychrophilum by the amplification of the 16S rRNA gene. All the Fl. psychrophilum isolates were stored in liquid nitrogen or −80°C by adding of glycerol as cryopreservant.

The type strains of Flavobacterium frigidarium (NCIMB13737T) and Fl. psychrophilum (NCIMB 1947T) were obtained from NCIMB, United States.

Genotypic characterization

For the DNA extraction, a loop from a pure culture of bacteria from agar plates was transferred to 180 μl buffer ATL and DNA extraction continued using the DNeasy kit (Qiagen, Hilden, Dusseldorf, Germany) as described by the manufacturer's protocol. Extracted DNA was stored at −20°C for later use.

Specific primers for Fl. psychrophilum were chosen based on the study by Nicolas et al. (2008). Two sets of primers were used to amplify trpB, atpA, gyrB, fumC, murG, dnaK and tuf, while 16S rRNA gene sequence was obtained by primers published by Giovannoni (1991). The primers were used in both the PCR and sequencing process (Table 2). Amplification was based on standard reaction of a mixture containing 10× ThermoPol buffer (BioLabs, Ipwich, MA, USA), 2·5 mmol l−1 dNTP (TaKaRa, St Louis, MO, USA), 10 μmol l−1 of forward and reverse primers, 2 μl DNA, 1·5 units (0·3 μl) Taq DNA polymerase (BioLabs) and distilled water was added to a final solution of 50 μl. Amplification was performed as described in Nicolas et al. (2008) with slight modifications in Tm as described in Table 2. All PCR products were visualized by gel electrophoresis using 1% agarose. Afterwards, the PCR product was purified using EZNA Cycle-Pure Kit (Omega Bio-Tek, Norcross, GA, USA) and ExoSAP-IT (usb, Cleveland, OH, USA) as described by the manufacturer.

Table 2. Overview of specific primers, target genes and annealing temperatures for the housekeeping genes of Flavobacterium psychrophilum
Target genePrimerDirectionAnnealing °C/TPrimer sequenceSource
16S rRNAEug B27FForward58°C/30 s5′-AGAGTTTGATCmTGGCTCAG-3′Giovannoni (1991)
Eug A1518RReverse58°C/30 s5′-AAGGAGGTGATCCAnCCRCA-3′Giovannoni (1991)
Eug 20FForward50°C/40 s5′-AGAGTTTGATCATGGCTC AG-′3Metselaar et al. (2010)
Eug U1510RReverse50°C/40 s5′-GGTTACCTTGTTACGACTT-′3Metselaar et al. (2010)
atpAatpA-FForward52°C/30 s5′-CTTGAAGAAGATAATGTGGG-3′Nicolas et al. (2008)
atpA-RReverse52°C/30 s5′-TGTTCCAGCTACTTTTTTCAT-3′Nicolas et al. (2008)
Fsp-ATPA-F1Forward54°C/30 s5′-TGGAATCAAAGAAGGATCwAC-3′Present study
Fsp-ATPA-R1Reverse54°C/30 s5′-TTrATyGCwGGACGAACyCC-3′Present study
dnaKdnaK-FForward52°C/30 s5′-AAGGTGGAGAAATTAAAGTAGG-3′Nicolas et al. (2008)
dnaK-RReverse52°C/30 s5′-CCACCCATAGTTTCGATACC-3′Nicolas et al. (2008)
fumCfumC-FForward52°C/30 s5′-CCAGCAAACAAATACTGGGG-3′Nicolas et al. (2008)
fumC-RReverse52°C/30 s5′-GGTTTACTTTTCCTGGCATGAT-3′Nicolas et al. (2008)
Fsp-FUMC-F1Forward48·5°C/30 s5′-AAGAAATCRTHKAWGGWTTTGC-3′Present study
Fsp-FUMC-R2Reverse48·5°C/30 s5′-AGATGATCCAGGYTCGTTTTC-3′Present study
gyrBgyrB-FForward52°C/30 s5′-GTTGTAATGACTAAAATTGGTG-3′Nicolas et al. (2008)
gyrB-RReverse52°C/30 s5′-CAATATCGGCATCACACAT-3′Nicolas et al. (2008)
murGmurG-FForward52°C/30 s5′-TGGCGGTACAGGAGGACATAT-3′Nicolas et al. (2008)
murG-RReverse52°C/30 s5′-GCATTCTTGGTTTGATGGTCTTC-3′Nicolas et al. (2008)
Fsp-MURG-F1Forward51·2°C/30 s5′-MAAGACAAAATGGAAATGCAAA-3′Present study
Fsp-MURG-R1Reverse51·2°C/30 s5′-GGCGACGGAATRWAAATNACTGG-3′Present study
trpBtrpB-FForward52°C/30 s5′-AAGATTATGTAGGCCGCCC-3′Nicolas et al. (2008)
trpB-RReverse52°C/30 s5′-TGATAGATTGATGACTACAATATC-3′Nicolas et al. (2008)
Fsp-TRPB-F1Forward53°C/30 s5′-TTGTATTTTGCAGAGCGTTTRTC-3′Present study
Fsp-TRPB-R1Reverse53°C/30 s5′-ATTGCTGGAATHAANCCTTC-3′Present study
tuftuf-FForward52°C/30 s5′-GAAGAAAAAGAAAGAGGTATTAC-3′Nicolas et al. (2008)
tuf-RReverse52°C/30 s5′-CACCTTCACGGATAGCGAA-3′Nicolas et al. (2008)

After visualization and cleaning of PCR products, sequencing was performed as described in Brevik et al. (2011). All sequences were compared using Vector NTI® Software (Invitrogen, Norway) and the Multiple Sequence Alignment Editor & Shading Utility (GeneDoc) (Copyright by Karl Nicholas ©2006). The accession numbers for all the sequences are presented in the Table 3.

Table 3. Overview of all Flavobacterium psychrophilum strains included in the genotypic analysis. Each housekeeping gene is presented with accession number from the GenBank. Host and origin codes were explained above
IsolateatpAdnaKfumCgyrBmurGtrpBtuf16S rRNAHostOrigin
Ch06-1-Rt-E JN164128 JN164166 JN164202 JN164040 JN164054 JN164092 JN164240 JN173012 RtCh
Ch07-2-As-E JN164121 JN164159 JN164195 JN164009 JN164047 JN164085 JN164233 JN173003 AsCh
Ch07-3-As-E JN164122 JN164160 JN164196 JN164038 JN164048 JN164086 JN164235 JN173004 AsCh
Ch07-4-Rt-E-I JN164125 JN164163 JN164199 JN164024 JN164051 JN164089 JN164237 JN173007 RtCh
Ch09-5-Rt-E JN164129 JN164167 JN164203 JN164026 JN164055 JN164093 JN164241 JN173013 RtCh
Ch09-6-As-E-I JN164130 JN164168 JN164204 JN164013 JN164056 JN164094 JN164242 JN173014 AsCh
Ch07-7-As-E JN164124 JN164162 JN164198 JN164039 JN164050 JN164088 JN164236 JN173006 AsCh
Ch08-8-As-E JN164127 JN164165 JN164201 JN164012 JN164053 JN164091 JN164239 JN173011 AsCh
Ch07-9-Rt-E-I JN164123 JN164161 JN164197 JN164023 JN164049 JN164087 JN164235 JN173005 RtCh
Ch08-10-Rt-E-I JN164131 JN164169 JN164205 JN164014 JN164057 JN164095 JN164238 JN173015 RtCh
Ch08-11-As-E JN164126 JN164164 JN164200 JN164011 JN164052 JN164090 JN164243 JN173010 AsCh
Ch09-12-Rt-W JN164140 JN164178 JN164214 JN164044 JN164066 JN164083 JN164252 JN173002 RtCh
Ch09-13-As-W JN164139 JN164177 JN164213 JN164033 JN164065 JN164103 JN164251 JN173001 AsCh
Ch10-14-Rt-G JN164134 JN164172 JN164208 JN164031 JN164060 JN164098 JN164246 JN173018 RtCh
Ch10-15-As-F JN164135 JN164173 JN164209 JN164017 JN164061 JN164099 JN164247 JN173019 AsCh
Ch10-16-Rt-G JN164133 JN164171 JN164207 JN164041 JN164059 JN164097 JN164245 JN173017 RtCh
Ch08-17-Rt-E JN164132 JN164170 JN164206 JN164016 JN164058 JN164096 JN164244 JN173016 RtCh
Ch10-18-Rt-G JN164136 JN164174 JN164210 JN164032 JN164062 JN164100 JN164248 JN173020 RtCh
Ch08-19-As-E JN164119 JN164156 JN164193 JN164010 JN164081 JN164082 JN164230 JN173008 AsCh
Ch07-20-Rt-E-I JN164155 JN164157 JN164229 JN164008 JN164045 JN164083 JN164231 JN173009 RtCh
Ch10-21-Rt-W JN164137 JN164175 JN164211 JN164042 JN164063 JN164101 JN164249 JN173021 RtCh
Ch10-22-As-nd JN164138 JN164176 JN164212 JN164043 JN164064 JN164102 JN164250 JN173022 AsCh
Ch07-23-As-E-IHM443882 HM443792 HM443805 HM443819 HM443831 HM443844 HM443856 HM443869 ASCh
Ch07-24-Rt-E-I HM443881 HM443791 HM443804 HM443818 HM443830 HM443843 HM443855 HM443868 RtCh
Ch10-25-nd-Sp JN164120 JN164158 JN164194 JN164030 JN164046 JN164084 JN164232 JN173000 ndCh
No09-26-As-K HM443887 HM443797 HM443810 HM443823 HM443834 HM443849 HM443861 HM443887 AsNo
No09-27-As-SK HM443888 HM443798 HM443811 HM443824 HM443835 HM443850 HM443862 HM443888 AsNo
No09-28-Rt-K HM443885 HM443795 HM443808 HM443821 HM443832 HM443847 HM443859 HM443885 RtNo
No09-29-As-nd HM443886 HM443796 HM443809 HM443822 HM443833 HM443848 HM443860 HM443886 AsNo
No09-30-T-Sp HM443889 HM443799 HM443812 HM443825 HM443836 HM443851 HM443863 HM443889 TNo
No09-31-As-E JN164146 JN164183 JN164220 JN164027 JN164072 JN164110 JN164257 JN173028 AsNo
No09-32-As-G JN164147 JN164184 JN164221 JN164028 JN164073 JN164111 JN164258 JN173029 AsNo
No09-33-As-G JN164148 JN164185 JN164222 JN164037 JN164074 JN164112 JN164259 JN173030 AsNo
No10-34-T-G JN164142 JN164180 JN164216 JN164025 JN164068 JN164106 JN164254 JN173024 TNo
No10-35-T-W JN164143 JN164181 JN164217 JN164034 JN164069 JN164107 JN164266 JN173025 TNo
No10-36-T-W JN164144 JN164192 JN164218 JN164036 JN164070 JN164108 JN164255 JN173026 TNo
No10-37-T-W JN164145 JN164182 JN164219 JN164035 JN164071 JN164109 JN164256 JN173027 TNo
No10-38-As-W JN164150 JN164187 JN164224 JN164019 JN164076 JN164114 JN164261 JN173032 AsNo
No10-39-As-W JN164153 JN164189 JN164226 JN164020 JN164078 JN164116 JN164263 JN173034 AsNo
No10-40-As-K JN164152 JN164190 JN164227 JN164022 JN164079 JN164117 JN164264 JN173035 AsNo
No10-42-As-K JN164141 JN164179 JN164215 JN164018 JN164067 JN164105 JN164253 JN173023 AsNo
No10-43-As-K JN164149 JN164186 JN164223 JN164029 JN164075 JN164113 JN164260 JN173031 AsNo
No10-44-As-M JN164154 JN164191 JN164228 JN164021 JN164080 JN164118 JN164265 JN173036 AsNo
No11-45-As-W JN164151 JN164188 JN164225 JN164015 JN164077 JN164115 JN164262 JN173033y AsNo


Alignments were constructed for the 16S rRNA gene (1254 nt) and the seven HGs (atpA-978 nt, dnaK-897 nt, fumC-849 nt, gyrB-1149 nt, murG-792 nt, trpB-890 nt and tuf-939 nt) using the AlignX program in the Vector NTI® Software package (Invitrogen). Available 16S rRNA gene sequences of Fl. psychrophilum that had a sufficient length (>1250) were obtained from the GenBank and included in the alignment. In addition, a 16S rRNA gene sequence from a Flavobacterium columnare (accession number AB078047) isolate was included as out-group. The alignments of the seven HGs were also supplemented with homologous GenBank sequences from 50 Fl. psychrophilum isolates that originated from the study by Nicolas et al. (2008). Sequences of Fl. frigidarium (NCIMB13737T) were included in the HG alignments to provide an out-group in the phylogenetic analysis. Aligned sequences were manually adjusted to equal size, and large gaps were removed or modified using GeneDoc. A search for recombination was performed on all alignments using the Recombination Detection Program (RDP3) (Martin et al. 2010) at default settings.

The evolutionary model and substitution rates were calculated for all alignments using jModelTest: phylogenetic model averaging with the Bayesian information criterion option was chosen (Posada 2008). Phylogenetic trees for the 16S rRNA gene and the seven HGs were constructed using the Bayesian method, using the Beast package v1.6.1 (Drummond and Rambaut 2007). Housekeeping gene alignments were analysed using the calculated nucleotide evolutionary model (tuf, trpB, murG, dnaK and atpA = HKY + I + G, fumC = GTR + G, gyrB = GTR + I + G and 16SrRNA = HKY + G) with a relaxed lognormal molecular clock (Drummond et al. 2006). The Markov chain Monte Carlo were run for 100 000 000 generations, and when necessary, a second run of 500 000 000 generations was performed to attain effective sample size (ESS) values higher than recommended (>200). ESS values were inspected using Tracer v1.4 (Rambaut and Drummond 2007). A maximum clade credibility tree was obtained using a 10% burn-inn in TreeAnnotator and viewed using FigTree. Nodes with <60 posterior probability values were collapsed.

Multilocus sequence typing

The different allele types (AT) within each of the seven HGs were coded as a discrete character (i.e. 1–9, A–H). Fl. psychrophilum MLST Database ( was used for coding homologous sequences in data set of the current study. The AT was used to create the ST for each isolate. These STs were used to construct a data matrix with 94 Fl. psychrophilum isolates, 44 isolates from the current study together with the 50 isolates published by Nicolas et al. (2008). The data matrix was exported as a nexus file into PAUP 4.0 (Swofford, D. L. 2003, PAUP* (Phylogenetic Analysis Using Parsimony* and Other Methods), version 4; Sinauer Associates, Sunderland, Massachusetts) where a dendrogram was constructed using the neighbour-joining (NJ) distance method.


The phylogenetic analysis of the 16S rRNA gene was performed using an alignment of 56 partial sequences (1254 bp) showing variation in 34 positions (2·7%) (Table 4). The 16S sequences from a collection of isolates from Japan (Accession no: AB297671, AB297672, AB297674, AB297675, AB297676, AB297483, AB297484) contain a number of insertions and deletions that are not found in any of the other Fl. psychrophilum isolates, and when these deletions are removed, there is a variation in eighth positions only. The resolution between the sequences was relatively poor resulting in only two supported clades (clades: 16S rRNA-I and 16S rRNA-II) (Fig. 1). The clade, 16S rRNA-I, consists of seven sequences including two separate sequences of the type species (NCIMB1947) from Coho salmon in North America. Four of the remaining five isolates are from wild Atlantic salmon and trout (Salmo trutta) in Norway, while the fifth isolate is from Atlantic salmon in Chile. These five isolates were all collected in the period 2009–2010. The 16S rRNA gene sequences of the four isolates from wild salmonids in Norway are identical to the type species (NCIMB1947), while the Chilean isolates differ in one nucleotide position only. Clade 16S rRNA-II consists of the remaining isolates of Fl. psychrophilum, and if the isolates from Japan are excluded, there is variation in two nucleotide positions only.

Table 4. Overview of the number of variable nucleotide and putative amino acid positions in the alignments of the 16S rRNA gene and the housekeeping gene sequences
Gene N No. nucleotidesVariable positions%No. amino acidsVariable positions%
  1. N = the number of Flavobacterium psychrophilum sequences.

Multilocus sequence typing936 4941251·92164211·0
Figure 1.

The phylogenetic positioning of the 44 Flavobacterium psychrophilum isolates from the present study and selected GenBank sequences based on an analysis of 1254 nt within the 16S rRNA gene. The phylogenetic tree was obtained by Bayesian method using the Beast package v1.6.1. Posterior probability values below 50% are not shown. The codes for the isolates are explained in the Table 1.

Phylogenetic analyses of tuf, murG, fumC, trpB, dnaK and atpA resulted in only 2–3 clades, while the analyses of gyrB resulted in five clades (Fig. 2). The clades did not represent specific geographical groups of isolates, but contained isolates from different continents in most cases. The gyrB-I and gyrB-II groups include only Norwegian isolates from wild and farmed Atlantic salmon.

Figure 2.

The phylogenetic positioning of the 44 isolates Flavobacterium psychrophilum isolates from the present study and sequences from 50 isolates from Nicolas et al. (2008) obtained from GenBank. The phylogeny was constructed from an analysis on 1149 nt within the gyrB gene. The phylogenetic tree was obtained by Bayesian method using the Beast package v1.6.1. Posterior probability values below 50% are not shown. The abbreviations are explained above. The codes are described in the Table 1; the new codes for this figure are the country of isolation, Japan (Ja), Tasmania (Ta), United States (Us), British Columbia (BC), France (Fr), Switzerland (Sw), Denmark (Dn), Germany (Gn), Israel (Is), Scotland (Sc), Spain (Sp); year of isolation (1947–2011); number of the isolate; fish species, ayu (Ay), coho salmon (Cs), brown trout (Bt), Chinook salmon (CHs), carp (Ca), tench (Te), eel (Ee), rainbow trout (Rt); tissue, eroded fin (EF), spinal cord (Sc), liver (L), peduncle lesion (PL), dorsal lesion (DL), mouth lesion (ML). Nd, no information available.

Multilocus sequence typing

The MLST study includes 94 Fl. psychrophilum isolates from around the world, that is, 45 from Europe, 10 from North America, 28 from Chile and 10 from Japan, Tasmania and Israel. The bacteria were isolated from 10 different hosts: Atlantic salmon (N = 27); trout (N = 7); rainbow trout (N = 39); Coho salmon (N = 8); Chinook salmon, Oncorhynchus tshawytscha (Walbaum) (N = 1); cutthroat trout, O. clarki (Richardson) (N = 1); carp, Cyprinus carpio (L.) (N = 2); tench, Tinca tinca (L.) (N = 4); eel, Anguilla anguilla (L.) (N = 2); ayu, Plecoglossus altivelis (T. and S.) (N = 2); and not determined (N = 1). The isolates were collected in a period from 1947 until 2011 with a large number collected in Chile and Norway in the period from 2007 to 2010. Comparison of the seven HG sequences from the Fl. psychrophilum isolates included in this study revealed 125 nucleotide position showing variation across the 6494 nucleotides used in this study (Table 4). The highest number of variable positions (2·9%) was seen in the atpA locus and the lowest in fumC (0·8%). Of the 125 variable positions in the HGs, only 21 resulted in amino acid substitutions suggesting that most of the variations observed are selectively neutral. The number of distinct alleles ranged from 12, at locus fumC, to 24 at locus gyrB. The combination of the different ATs at the seven loci made it possible to distinguish 53 STs among the 94 isolates (Fig. 3).

Figure 3.

Dendrogram showing sequence type relations of the 44 isolates of Flavobacterium psychrophilum from the present study and the isolates from Nicolas et al. (2008). The multilocus sequence typing analysis was constructed using 6494 nt from the genes: gyrB, atpA, dnaK, trpB, fumC, murG and tuf. The analysis was performed using the distance method in PAUP 4.0. The codes for all the isolates are explained in the Table 1 and Fig. 2.

Of the 53 STs, 38 were only sampled once. ST6\had the highest number of isolates and contained isolates from Europe (France, Sweden, Denmark and Norway) only, collected in the period from 1988 to 2011. All but one isolates (No11-45-As-W) were sampled from rainbow trout. One ST, ST9, contained seven isolates from Chile, where five isolates were collected from Atlantic salmon and two isolates from rainbow trout and Coho salmon, respectively. Another ST, ST20, contains five isolates from Atlantic salmon and rainbow trout in Chile and one isolate from rainbow trout in Germany. Two STs, ST42 and ST49, contained five and four isolates, respectively, all from Norway. ST 42 contained three isolates from trout and two isolates from Atlantic salmon, while the isolates in ST 49 were all from Atlantic salmon.

The largest number of isolates included in this study are from Norway (N = 19) and Chile (N = 28), and these isolates group into 10 and 16 STs, respectively. Isolates from both Norway and Chile are present in one ST (ST36) only, where both isolates are from rainbow trout. The Norwegian isolates were collected in 1996, while the Chilean isolates were collected in 2009. Another ST, ST45, contains isolates from Chile, USA and Canada (British Columbia), all collected from Coho salmon. Two of the 10 STs with Norwegian isolates contain isolates also from other countries. The six isolates from Japan group into four different STs (ST1, ST18, ST41 and ST 43) where the first two contain two isolates from ayu and rainbow trout, respectively, and ST41 and ST43 isolates from Coho salmon. All three isolates from Atlantic salmon in Tasmania grouped together in ST 53. Three of the four isolates from tench grouped in ST 46, while the fourth was the only isolate in ST 50. Both isolates, from eel, collected in France and Greece, grouped in ST 37.

A large number, 48, of the isolates have been obtained from internal organs, kidney, spleen, liver, spinal cord, and gonads products, while 38 isolates have been associated with surface tissues like gills, fins and skin ulcers. It is not known from which tissues the other isolates have been obtained. The MLST analysis of the isolates gives no clear separation between systemic isolates and isolates from the host surfaces.


The need and efforts to perform genetic characterization of Fl. psychrophilum to find virulence markers that can be used to differentiate isolates with respect to mechanism of pathogenesis, to establish a bases for vaccine development, treatment and control of flavobacteriosis, and to find genetic markers that can be used to trace spreading of isolates during disease outbreak to establish transmission patterns have been extensively reported in the literature: (i) DNA restriction patterns and PFGE used on Chilean isolates (Avendaño-Herrera et al. 2009; Castillo et al. 2012), (ii) PFGE on isolates from rainbow trout (Chen et al. 2008), (iii) genotyping using PCR-RFLP (Izumi et al. 2003), (iv) ribotyping (Chakroun et al. 1997), (v) plasmid profile in ayu (Kim et al. 2010) and, recently, (vi) a study of polymorphism of the gyrA gene to distinguish between virulent and nonvirulent Fl. psychrophilum (Fujiwara-Nagata et al. 2012). Currently, there is not a standard method for the genetic characterization of Fl. psychrophilum available. However, the most promising approach to establish an understanding of the population structure of Fl. psychrophilum was given by Nicolas et al. (2008). They examined the nucleotide polymorphisms at 11 protein-coding loci in the genome from 50 isolates. In this study, we have used the system developed by Nicolas et al. (2008) to identify variation in Fl. psychrophilum isolates from Norway and Chile in an attempt, understand the distribution pattern and possibly identify virulence markers.

The PCR amplification and sequencing of the 16S rRNA gene is commonly used for detection and identification of different pathogenic fish bacteria including members of the genus Flavobacterium (Tiirola et al. 2002; Bader et al. 2003; Figuereido et al. 2012). However, as this and other studies show, there is not enough variation in the nucleotides sequences of the 16S rRNA gene of Fl. psychrophilum to allow the separation between different strains (Soule et al. 2005a,b; Ramsrud et al. 2007). Phylogenetic analyses based on the 16S rRNA gene sequences yielded only two supported clades, 16S-I and 16S-II (Fig. 1), with the type species from North American, a Chilean isolate, and a collection of isolates from wild salmonids in Norway in the former. These isolates group together due to a shared inversion of six nucleotides (Soule et al. 2005a), while most of the remaining variation is due to insertion, deletions and substitutions in a collection of isolates from Japan published as sequences in the GenBank only.

In the study of nucleotide polymorphisms, Nicolas et al. (2008) used 11 protein-coding loci, and in this study, we selected seven of these HGs (fumC, murG, tuf, atpA, trpB, dnaK and gyrB) for differentiation of Fl. psychrophilum isolates from Norway and Chile. Each of these seven genes provided limited information on the relationships between the isolates when analysed separately as was also found by Nicolas et al. (2008). The highest resolution was obtained in the analysis of gyrB gene, which resulted in five supported clades (Fig. 2). The analysis of gyrB sequencing has been previously reported in literature as an identification method for Fl. psychrophilum (Izumi et al. 2005; Nakagawa et al. 2009). Recently, Shah et al. (2012), in a phylogenetic study of Fl. psychrophilum quinolone resistance in Norway, suggested a clonal relationship between rainbow trout isolates. In the current study, the grouping of some isolates from Norwegian Atlantic salmon in the gyrB-I and gyrB-II clades shows that the phylogeny of gyrB was able to establish a close relationship among some isolates from one species and from one country.

Nicolas et al. (2008) tested 11 protein-coding loci and recommended that a ML sequence type (MLST) scheme should rely on the seven loci used in our study. Based on the combination of AT at these seven loci, the isolates were divided into 23 STs (Nicolas et al. 2008). The resulting number of ST increased to 53 when we added the 44 isolates from Norway and Chile. The majority of the ST of Fl. psychrophilum has been obtained from rainbow trout (N = 39) and Atlantic salmon (N = 27), in addition to 20 isolates from other salmonids, and eight isolates from other fish species (eel, tench and carp). As the natural host range for Fl. psychrophilum is unclear, that is, the importance of nonsalmonid species as natural reservoirs for transmission to salmonids species, the discussion about host specificity is to a large extent limited by the isolates available. Recently, in a similar study, Siekoula-Nguedia et al. (2012) present 15 STs among 66 isolates from rainbow trout in France pointing out the predominance of a clonal complex in this species, which make sense because the study was carried out in only one species and in a restricted geographical area compared with the current study.

The 18 Norwegian isolates were collected from Atlantic salmon (12), trout (5) and rainbow trout (1) and two STs (ST 6, ST 42) contained isolates from more than one host species. The two isolates in ST 6 came from farmed rainbow trout and A. salmon in the same fjord system where the two localities belonged to the same company. These two isolates probably arrived at the two localities when the fish were put to sea, that is, most likely from the smolt production site owned by the company. However, the fjord may, at times with heavy rainfalls, become very brackish and, hence, possibly provide an opportunity for horizontal transmission between sites in the fjord, but the Fl. psychrophilum isolates (No09-29As, No10-42As and No10-44As) collected, in the same time period, from wild salmonids migrating through the fjord on the way to the river all belongs to different STs. The five isolates in ST 42 all come from wild trout and Atlantic salmon in the same river in north western Norway, which means that horizontal transmission of isolates between the two species in the river could be a frequent event as they are the only two host species in this river system. The other ST from Norway contains a few isolates only. Nicolas et al. (2008) found a marked association between the type of isolate and host species, but the isolates from Norway, collected from Atlantic salmon, trout and rainbow trout, cannot be used to support such a hypothesis suggesting host-specific isolates for these salmonids. Nor can the 28 isolates from salmonids in Chile be used to support such a hypothesis as the STs (ST 9, ST 16 and ST 20) containing several isolates include Fl. psychrophilum from both A. salmon and rainbow trout. It is to be expected that the pathogen populations of salmonids must be influenced by the extensive farming activity, which may have led to an artificial mixing of previous host-specific isolates. Looking at a larger number of Fl. psychrophilum isolates and isolates from other nonsalmonid species may give a clear picture of a possible pattern of host-specific isolates as suggested by other authors (Chakroun et al. 1997; Soule et al. 2005a,b; Arai et al. 2007; Avendaño-Herrera et al. 2009; Kim et al. 2010).

In Chile, there are no native species of salmonids, all of them have been introduced from North America and Europe. The first introduction to Chile occurred in 1905 with the import of 400 000 fertilized eggs of Atlantic salmon, rainbow trout and trout from Germany (Avila et al. 1994). After that the main suppliers of fertilized eggs from salmonids have been USA, Ireland, Denmark, Scotland, Sweden and Norway (Avila et al. 1994). Connected to the movement of embryos is always a risk to introducing unwanted organisms as parasites, virus and bacteria (Vike et al. 2009; Mutoloki and Evensen 2011). Accordingly, the high prevalence of Fl. psychrophilum in Chilean aquaculture could be a result of introduction of this bacterium in connection with the import of embryos (Bustos et al. 1995; Taylor 2004; Cipriano 2005; Kumagai and Nawata 2011). No occurrence of Fl. psychrophilum in native fish species from Chile has been established to our knowledge, and there are no reports of flavobacteriosis in native fish species in Chile. However, absence of reports does not necessarily mean absence of Fl. psychrophilum in native species from Chile, but if the known isolates from Chile at some point were introduced, then they should have close relatives in Europe and North America. As can be seen from the MLST dendrogram (Fig. 3), all Chilean isolates are intermingled with isolates from Europe and North America, that is, there are no indications that the farmed salmonids in Chile have been infected with distinct local strains of Fl. psychrophilum. The isolates of Fl. psychrophilum from Norway and Chile do not support a hypothesis suggesting the existence of distinct geographical isolates. A similar result has also been obtained using random amplified polymorphic DNA analysis of Fl. psychrophilum isolates (Chakroun et al. 1997).

Flavobacterium psychrophilum has been isolated from gills, skin, fins and ulcers in addition to a long range of different tissues and organs of the host fish (Lumsden et al. 1996; Miwa and Nakasayu 2005; Nilsen et al. 2011b). The fish may show no signs of disease or the signs may range from surface ulcers and fin erosion to moribund fish with serious gill damages or heavy muscle bleedings (Ekman and Norrgren 2003), suggesting that there could be a large variation in virulence of the different isolates. In a study of 232 isolates of Fl. psychrophilum from four families of fish, the isolates were genotyped using a single nucleotide polymorphisms (SNPs) analysis based on the gyrA gene, and it was suggested that strains virulent for ayu could be identified using this system (Fujiwara-Nagata et al. 2012). In the present study, Fl. psychrophilum was isolated from what seemed to be perfectly healthy to moribund salmon with gill pathology, ulcers (saddleback disease, fin erosions and bleedings in the head region and along the lateral line) and large muscle bleedings. The bacterium was also isolated from the gonads of spawning Atlantic salmon. Of these isolates, 48 were obtained from internal organs/tissues, systemic infection, and 38 from fins, gills and skin lesions. However, no clear pattern can be seen when these isolates are arranged according to ST based on the HGs. The present MLST scheme cannot be used to identify potential pathogenic/virulent isolates of Fl. psychrophilum. A similar result was obtained after phenotypic and genotypic analysis of Fl. psychrophilum isolates from North America (Hesami et al. 2008). Another approach, possibly the system used by Fujiwara-Nagata et al. (2012), has to be used in the search for virulence markers in isolates from salmonids fish in Norwegian and Chilean aquaculture.


To Veterquímica, Chile, that kindly provided all the Chilean Fl. psychrophilum isolates with the exception of Ch09-13-As-W and Ch09-12-Rt-W.