Members the genus Chryseobacterium are widely distributed microorganisms that can be recovered from a wide variety of environments, such as fresh water, sewage and wastewater, soil or food sources, such as milk, poultry and meat and dairy products 1. Some species of Chryseobacterium have been involved in human infections, acting as sporadic but severe opportunistic nosocomial pathogens 23. In veterinary medicine, chryseobacteria are not relevant pathogens for domestic animals, but they are widely distributed in aquatic environments and fish farms 14. Until recently members of the genus Chryseobacterium were not commonly associated with fish infections. However, there has been an increase in the frequency of clinical cases in which Chryseobacterium sp. strains have been isolated from different fish species. Thus, Chryseobacterium balustinumChryseobacterium scophtalmum and Chryseobacterium joostei have been isolated from diseased fish 456. More recently, Chryseobacterium piscicola has been reported to produce mortalities in farmed Atlantic salmon (Salmo salar) and rainbow trout (Oncorhynchus mykiss) in Chile and Finland 789, Chryseobacterium arothri was isolated from the kidneys of the pufferfish Arothron hispidus in Hawaii 10 and Chryseobacterium chaponense from diseased farmed Atlantic salmon in Chile 11. In fact, some Chryseobacterium species are considered potentially emerging pathogens in fish 4. However, many chryseobacteria isolated from diseased fish are usually identified only at the genus level due to the difficulty of their correct identification by phenotypically based laboratory methods alone 45, which limits the knowledge of the diversity of species associated with fish disease.

All isolates gave shiny, round, yellow-pigmented colonies on Anacker and Ordal agar, a characteristic that lead to the presumptive diagnosis of infection by F. psychrophilum, but none of the isolates gave the expected amplicon product of 1,089 bp, specific of F. psychrophilum. Moreover, cells of trout isolates were straight short Gram negative rods after Gram staining. Comparative analysis of the 16 S rRNA gene sequences revealed that the isolates shared 99.8-100 % sequence similarity between each other, thus demonstrating their high phylogenetic relatedness, 99.2-99.8 %, with the type strain of C. shigense (GUM-Kaji^T; Figure 1) and only 81.8-81.9 % with F. psychrophilum NCIMB 1947^T (GenBank accession n D12670). The 16 S rRNA gene sequences of the isolates included in this study have been deposited in GenBank under the accession numbers indicated in Figure 1.

Phenotypically all trout isolates were catalase and oxidase positive, non-motile, grew on nutrient agar with yellow and shiny colonies but not on MacConkey agar, produced flexirubin-type pigment, were able to grow at 5–30 °C but not at 37 °C, and hydrolysed starch, casein and gelatin. With the API 20NE system they exhibited homogeneous biochemical characteristics displaying the numerical profiles 3452205, while C. shigense CCUG 61059^T gave the numerical profile 2456204. With the APY ZYM strips, the trout isolates, as well as the type strain of C. shigense CCUG 61059^T, expressed activity for alkaline phosphatase, leucine arylamidase, trypsin, acid phosphatase and naphthol-AS-BI-phophohydrolase but not for esterase C4, lipase C14, cystine arylamidase, α-chymotrypsin, α-galactosidase, β-galactosidase, β-glucuronidase, α-glucosidase, N-acetyl-β-glucosaminidase, α-mannosidase and α-fucosidase. Clinical isolates of C. shigense expressed activity for valine arilamidase and not for ester lipase C8 and β-glucusidase, while C. shigense CCUG 61059^T gave opposite results for these tests.

After genetic characterization by random amplified polymorphic DNA, both oligonucleotides generated reproducible patterns, but an appropriate number on bands were produced with oligonucleotide P2 (Figure 2). The seven C. shigense trout isolates showed undistinguishable RAPD fingerprints with amplifications bands ranging from 600 to 2500 bp, indicating genetic homogeneity among them. On the other hand, the strain CCUG 61059^T yielded a completely different fingerprint (Figure 2).

Diagnosis of bacterial fish diseases is not possible purely on the basis of the clinical signs and symptoms observed in diseased fish, because many of them can be caused by more than one etiological agent. Presumptive diagnosis is usually based on previous epidemiological data and a rough bacteriological analysis of cultured organisms from affected animals. In the present study, isolates were recovered from diseased trout suspected of F. psychrophilum infection because the fish farm had previous records of infections with this fish pathogen and the macroscopic characteristics of the colonies obtained on Anacker and Ordal agar were also compatible with that presumptive diagnosis. However, cell morphology of the trout isolates following Gram staining was different of the filamentous rods exhibited by F. psychrophilum 18, they were non-motile and none gave a positive reaction with a F. psychrophilum species-specific PCR assay 13. This lack of amplification is consistent with the low similarity observed (81.8-81.9 %) between the 16 S rRNA gene sequences of trout isolates and F. psychrophilum NCIMB 1947^T. The trout isolates exhibited the highest 16 S rRNA sequences similarities with C. shigense GUM-Kaji^T (99.2-99.8 %), percentages higher than the 99 % sequence similarity used as the criterion for species identification 19. In addition, most of the phenotypic characteristics of the C. shigense trout isolates were consistent with the current description of this species based solely in the strain DSM 17126^T 20, which support the identification based on 16 S rRNA gene sequencing. Despite the overall phenotypic similitude, some differences were found between the C. shigense trout isolates and the type strain C. shigense CCUG 61059^T. Thus, trout isolates reduced nitrate, assimilated citrate but not mannitol and produce the enzyme valine arylamidase but not the enzymes ester lipase C8 and β-glucosidase, while the type strain of C. shigense CCUG 61059^T gave opposite reactions for these traits.

Widely spread microorganisms are usually genetically heterogeneous 21. Members of the genus Chryseobacterium are isolated from a wide range of environments 1 and therefore, it could be expected to be genetically diverse. However, trout isolates exhibited undistinguishable RAPD fingerprints indicating that they represent a single strain. This fact, together with their isolation in pure culture from internal organs might suggest a clinical significance; however the role of C. shigense as the causative agent of the disease episodes in trout cannot be undoubtedly established in absence of experimental infections trials.

Some members of the family Flavobacteriaceae, as Chryseobacterium, usually are opportunistic pathogens 4, because they require the existence of different predisposing factors such as coinfections with other bacteria or virus 7 or other environmental or husbandry sources of stress. No infestations or viral infections were detected previously to the disease episodes, but these usually occurred after transportation of rainbow trout fry from the hatchery to fish farm, or in tanks with elevated stock densities. These circumstances represent stressful conditions for fish 22 and might have increased the susceptibility of fry trout to infection.

Several species of Chryseobacterium, such as C. piscicola 8, C. chaponense 11, and C. shigense in this study, have been isolated from diseased fish in which F. psychrophilum infections were initially suspected. Although there are no clear evidences for considering these species as consistent pathogens for fish, they should be considered for a differential diagnosis in those cases with a suspicious of F. psychrophilum infection. Table 1 shows some phenotypic characteristics that can be useful for their differentiation.

^a Data for C. shigense are taken from this study, for C. piscicola from reference Ilardi et al. 8, for C. chaponense from reference Kämpfer et al. 11 and for F. psychrophilum from references Bernardet et al. 23, Cipriano and Holt 18 and Hesami et al. 24. ^bF. psychrophilum differs also from C. shigense by its inability to utilize glucose as sole carbon source, to hydrolyze aesculin and to produce a brown diffusible pigment on tyrosine agar. +, Positive reaction; -, negative reaction; V, variable reaction; w, weak reaction.

Since the initial description of C. shigense from a fresh lactic acid beverage in Japan 20, no further isolations of this species have been reported. Consequently, the isolation of C. shigense from trout shows that it can also occur in a very different habitat. To our knowledge, this is the first description of the isolation of C. shigense from clinical specimens.

In this work we describe by first time the recovery of C. shigense from clinical specimens in trout, showing that it can also occur in a very different habitat to fresh lactic acid beverage where it was initially isolated.

LZ carried out the phenotypic and genetic analyses and participated in the analysis of the data and drafting of the manuscript. AIV participated in the design of the study, in phylogenetic analysis and drafting of the manuscript. MAP participated in the recovery of clinical specimens. LD participated in the design of the study. JFJG conceived the study and participated in its design, drafting and coordination. All the authors read and approved the final manuscript.