Skip to main content
NCBI home page
FEMS Microbiology Letters logoLink to FEMS Microbiology Letters
. 2024 Oct 22;371:fnae089. doi: 10.1093/femsle/fnae089

Whole-genome-based taxonomy as the most accurate approach to identify Flavobacterium species

Vincent Gélinas 1,2, Valérie E Paquet 3,4, Maude F Paquet 5,6, Antony T Vincent 7,8, Steve J Charette 9,10,✉
PMCID: PMC11562821  PMID: 39439205

Abstract

The genus Flavobacterium comprises a diversity of species, including fish pathogens. Multiple techniques have been used to identify isolates of this genus, such as phenotyping, polymerase chain reaction genotyping, and in silico whole-genome taxonomy. In this study, we demonstrate that whole-genome-based taxonomy, using average nucleotide identity and molecular phylogeny, is the most accurate approach for Flavobacterium species. We obtained various isolated strains from official collections; these strains had been previously characterized by a third party using various identification methodologies. We analyzed isolates by PCR genotyping using previously published primers targeting gyrB and gyrA genes, which are supposedly specific to the genus Flavobacterium and Flavobacterium psychrophilum, respectively. After genomic analysis, nearly half of the isolates had their identities re-evaluated: around a quarter of them were re-assigned to other genera and two isolates are new species of flavobacteria. In retrospect, the phenotyping method was the least accurate. While gyrB genotyping was accurate with the isolates included in this study, bioinformatics analysis suggests that only 70% of the Flavobacterium species could be appropriately identified using this approach. We propose that whole-genome taxonomy should be used for accurate Flavobacterium identification, and we encourage bacterial collections to review the identification of isolates identified by phenotyping.

Keywords: Flavobacterium, whole-genome, ANI, taxonomy, gyrB, phylogenetic tree

We demonstrate that whole-genome-based taxonomy is the most accurate method for identifying Flavobacterium species and recommend that bacterial collections review identifications made by phenotyping and PCR genotyping.

Introduction

As of January 2024, the genus Flavobacterium included 237 species with valid published names in the National Center for Biotechnology Information (NCBI) database. Bacteria of these species are yellow-pigmented, aerobic, gram-negative, and rod-shaped (Loch and Faisal 2015). Although flavobacteria are found in various environments, the main interest comes from their interactions with fish as commensal organisms or opportunistic pathogens, such as Flavobacterium psychrophilum (Sugahara et al. 2010, Kämpfer et al. 2020, Saticioglu et al. 2021).

Before bioinformatics and PCR amplification, phenotyping was the standard for the identification of flavobacteria. Phenotyping is based on colony pigmentation, the presence of certain chemical components such as fatty acids and enzymes catabolizing specific proteins, and certain metabolism processes like glucose degradation (Van Trappen et al. 2004, Yi et al. 2005, Bernardet and Nakagawa 2006). More recently, PCR genotyping using 16S rRNA gene sequence-based primers has been used to refine the taxonomy of the genus Flavobacterium (Bernardet et al. 1996, Bernardet and Bowman 2006). However, this method alone is not precise enough to reveal the genetic relationships of flavobacteria isolates and thus must be complemented by another identification tool (Bernardet and Bowman 2006). Some researchers have used 16S rRNA gene sequences paired with phenotypes (Sun et al. 2016, Xu et al. 2020, Seo et al. 2022, Sun et al. 2022), and others have used 16S rRNA gene sequences paired with an in vitro or whole-genome approach using in silico DNA–DNA hybridization (DDH) (Bernardet and Bowman 2006, Kayansamruaj et al. 2017, Khan et al. 2020).

The use of primers that detect the gyrB gene (encoding the subunit B protein of DNA gyrase) in combination with primers targeting the 16S rRNA gene by PCR genotyping has been proposed for more accurate identification (Izumi and Wakabayashi 2000, Izumi et al. 2005). Primers allowing gyrA gene PCR amplification can be used complementarily to specifically find isolates from the F. psychrophilum species (Fujiwara-Nagata et al. 2019). This gene encodes the subunit A protein of DNA gyrase. Furthermore, some recent methods using whole-genome sequences, such as Average Nucleotide Identity (ANI), can identify new species more precisely than 16S rRNA gene sequences (Ranjan et al. 2016). Recent papers on the genus Flavobacterium have opted for whole genome taxonomy (Cai et al. 2018, Jung et al. 2021), but they are still overshadowed in number by those that use taxonomy based on 16S rRNA gene sequences and phenotyping.

The genomic approach can also reveal identification errors, as it has been the case for other bacterial genera (Rekadwad and Gonzalez 2017, Mateo-Estrada et al. 2019). Considering that some flavobacteria isolates were characterized decades ago with potentially less precise methods, it is possible that their identities need to be updated. The objective of our study is to re-evaluate the taxonomy of previously identified flavobacteria isolates, while seeking the most accurate method of identification.

Isolates from the genus Flavobacterium, previously identified with various taxonomic methods, were gathered and reanalyzed using PCR genotyping and whole genome analysis. Some of the analyzed isolates were found to not be flavobacteria, while others were found to be a different species than initially predicted. The taxonomy of these newly uncovered species was confirmed using phylogenetic trees from whole-genome sequence.

Materials and methods

Bacteria growth and conditions

Isolates from Table 1 were obtained from different official collections: The German Collection of Microorganisms and Cell Cultures; The Félix D'Herelle Reference Center for Bacterial Viruses, and The National Agricultural Library-Agricultural Research Service Culture Collection (NRRL) or were obtained from academic research laboratories.

Table 1.

Isolates analyzed from official collections and research laboratories.

Collection/reference Original taxonomy Taxonomic method Genbank accession number Taxonomy based on whole genome ANI results (%) PCR genotyping
Isolate ID Environment Origin gyrA gyrB
UW10123 Cow feces Wisconsin HER1317; a Flavobacterium johnsoniae Phenotypes JBEWQE000000000 New (F. tistrianum) 89.75 − +
UW101 NA Wisconsin HER1371; a F. johnsoniae Phenotypes JBEWQF000000000 F. johnsoniae 99.98 − +
LBUM151 NA Moncton Moncton University* Flavobacterium spp. 16S rRNA JBGGGM000000000 New (F. pectinovorum) 88.61 − +
NRRL B-14 732 Gills of diseased salmon Michigan USDA; b F. hydatis DNA–rRNA
16S rRNA		 JBEWQG000000000 F. hydatis 99.97 − +
FPG 6.1.0 Ulcerative dermatitis, Oncorhynchus mykiss Ontario Guelph University*; c F. oncorhynchi 16S rRNA JBGGGN000000000 F. oncorhynchi 96.27 − +
DSM 18 644 Kidney of diseased salmon Washington DSMZ; b F. columnare DNA–rRNA
16S rRNA		 JBEWQH000000000 F. columnare 99.51 − +
DSM 25 092 Ictalurus punctatus Alabama DSMZ; b F. columnare DNA–rRNA
16S rRNA		 JBEWQI000000000 F. covae 98.79 − +
FPG100 Kidney, systemic disease, O. mykiss Ontario Guelph University*; c F. psychrophilum 16S rRNA JBEWQK000000000 F. psychrophilum 99.81 + +
FPG48 Kidney, systemic disease, O. mykiss Ontario Guelph University*; c F. psychrophilum 16S rRNA JBEWQL000000000 F. psychrophilum 99.82 + +
FPG101 Kidney, systemic disease, O. mykiss Ontario Guelph University*; c F. psychrophilum 16S rRNA JBEWQM000000000 F. psychrophilum 99.79 + +
DSM 21 280 Kidney of a diseased O. mykiss fry Seine-Maritime DSMZ; b F. psychrophilum DNA–rRNA
16S rRNA		 JBEWQN000000000 F. psychrophilum 99.87 + +
NRRL B-14 842 Soil NA USDA; d F. aquatile Phenotypes JBGGGO000000000 Microbacterium paraoxydans 92.84 − −
NRRL B-4264 Fish NA USDA; e F. fuscum Phenotypes JBEWQO000000000 Stenotrophomnas koreensis 97.10 − −
NRRL B-2648 Soil NA USDA; f F. resinovorum Phenotypes JBGGGP000000000 Novosphingobium resinovorum 93.53 − −
NRRL B-14 859 Soil Illinois USDA; g Flavobacterium spp. Phenotypes JBEWQP000000000 Chryseobacterium gallinarum 86.03 − −
Open in a new tab

HER: Félix d'Hérelle Reference Center for Bacterial Viruses; USDA: The National Agricultural Library; DSMZ: The German Collection of Microorganisms and Cell Cultures; NRRL: USDA-ARS Culture Collection (NRRL); a: Chang et al. 1984; b: Bernardet et al. 1996; c: Hesami et al. 2008; Jarau et al. 2018; d: Weeks, 1954; e: Soda and Misono, 1968; f: Delaporte and Daste, 1956; g: Hou, 1994; NA: not available. *Details about contributing laboratories are indicated in acknowledgements section.

Isolates from Table 1 were thawed from 15% glycerol stock, streaked on FLPA (Flavobacterium psychrophilum agar) medium (4 g/l tryptone, 0.4 g/l yeast extract, 0.5 g/l CaCl2, 0.5 g/l MgSO4, 0.5 g/l glucose, and 1.2% bactoagar) (Daskalov et al. 1999), and incubated at 15°C for 3 days.

DNA lysate and PCR genotyping

The protocol for preparing DNA lysate was adapted from another study (Charette and Cosson 2004). A quarter of each bacterial lawn from the Petri dish (Table 1) was resuspended in 1 ml single worm lysis buffer containing 1 µl proteinase K at 20 mg/ml. The bacteria were homogenized by passing the suspension several times through a 23-gauge needle with a syringe. The resuspensions were incubated at 95°C for ≥15 min, with a pause halfway through to vortex, and then were centrifuged at 10 700 g for 5 min. The supernatant was transferred to a new 1.5 ml tube and the DNA concentration in the lysate was quantified by Nanodrop (Thermo Scientific, Ca, USA), and adjusted to 100 ng DNA/µl with 10 mM Tris-HCl pH 8.

Each bacterial DNA lysate (1 µl containing 100 ng DNA) was analyzed by PCR with Promega GoTaqTM DNA polymerase (Fisher Scientific, Saint-Laurent, QC, Canada) according to the manufacturer's protocol with a few modifications. In brief, the concentration of each reagent in the final master mix was: 1 × completely thawed Green Go Taq reaction buffer, 0.2 mM each dNTP from Promega polymerase (Fisher Scientific, Saint-Laurent, QC, Canada), 1.25 U GoTaqTM DNA polymerase, 0.9 µM each forward and reverse primer, and nuclease-free water to make a final volume of 20 µl. The program used for the PCR was 95°C for 5 min; followed by 30 cycles of 95°C for 30 s, 55°C for 30 s, and 68°C for 30 s; with a final extension at 68°C for 10 min. The primers targeting the gyrA gene (forward: 5′-GAAACCGGTGCACAGAAGG-3′; reverse: 5′-CCTGTGGCTCCGTTTATTAA-3′) and gyrB (forward: 5′-GTTGTAATGACTAAAATTGGTG-3′, reverse: 5′-CAATATCGGCATCACACAT-3′) were previously developed (Nicolas et al. 2008, Shah et al. 2012). Primers were synthesized by IDT (Coralville, Iowa) and the expected size of each amplicon was 395 bp (gyrA) and 1100 bp (gyrB). The PCR products (5 µl) were loaded on a 1% agarose gel and separated by electrophoresis for 30 min at 90 V, and then visualized with 0.5 µg/ml ethidium bromide under ultraviolet light.

DNA extraction, sequencing, and read assembly

A portion of the bacterial biomass grown on FLPA medium was resuspended in 180 µl ATL buffer the DNeasy blood and tissue kit (Qiagen, Valencia, CA, USA) and lysed by passing the suspension several times through a 23 gauge needle with a syringe. DNA extraction was performed according to the manufacturer's protocol for gram-negative bacteria, except that 5 µl of RNase A (Ambion, Burlington; Canada, 1 mg/ml) was added to each sample to eliminate residual RNA before proteinase K was added.

The sample DNA was sequenced using the Illumina MiSeq (Illumina; San Diego, CA, USA) system at the Plateforme d'Analyse Génomique of the Institut de Biologie Intégrative et des Systèmes (Université Laval, Québec City, QC, Canada). De novo assemblies were performed with Shovill version 1.1.0 (https://github.com/tseemann/shovill) using default parameters. Adapters were removed, and reads were filtered using fastp version 0.20.1 (Chen et al. 2018, Chen 2023). Whole-genome assembly evaluation was done with QUAST version 5.2.0 (Mikheenko et al. 2018).

Taxonomy with the ANI

The samples were submitted to the Type Isolate Genome Server (TYGS), and the genus and species of each isolate were established via the MApping SHort reads (MASH) algorithm (Meier-Kolthoff et al. 2022). ANI was used to validate the species against a curated database. To do so, whole-genome sequences of the 237 Flavobacterium species available on NCBI (January 2024) were extracted to form the database (Table S1). Then, ANIb (ANI calculated with BLAST+) values were computed using pyani version 0.2.9 (https://github.com/widdowquinn/pyani) with a database comprising of the 11 Flavobacterium isolates analyzed in this study and the entire genus Flavobacterium. Species were defined by a minimum threshold of 95%-96% sequence identity (Yoon et al. 2017). In parallel, for each of the 4 isolates identified to be outside the Flavobacterium genus (within the genera Microbacterium, Stenotrophomonas, Novosphingobium, and Chryseobacterium), ANIb was also computed, but by using the whole-genome sequences of every species of the corresponding genus for each isolate (115 Microbacterium spp., 24 Stenotrophomonas spp., 53 Novosphingobium spp., and 113 Chryseobacterium spp.) (January 2024) (Table S2).

Phylogenetic trees

Two molecular phylogenetic analyses were performed to determine the phylogenetic position of the different isolates identified as flavobacteria, or as not flavobacteria, within their respective previously identified genera. The first tree was made using whole-genome sequences of 45 Flavobacterium representative type strains of the genus's diversity (January 2024) (Table S3). Whole-genome sequences of the 11 isolates gathered from bacterial collections and previously identified as Flavobacterium by pyani were also added to the phylogenetic analysis. Whole-genome sequences from the genus Myroides (M. odoratus NCT11036, GCA_900453865.1; and M. profundi D25, GCA_000833025.1) were used as outgroups. The 59 whole-genome sequences were annotated using Prokka version 1.14.5 (Seemann 2014), and orthologs were found using the amino acid sequences submitted to the Clusters of Orthologous Groups (COG) algorithm through GET_HOMOLOGUES version 20 220 516 (Contreras-Moreira and Vinuesa 2013). A total of 944 softcore gene sequences (sequences found in ≥95% of the database) were identified, which excluded 188 paralogous sequences. These softcore sequences were aligned using MAFFT version 7.505 (Katoh and Standley 2013) and were then translated back into nucleotides using TranslatorX version 1.1 (Abascal et al. 2010). The alignments were filtered for repetitive or uninformative sites using BMGE version 1.12 (Criscuolo and Gribaldo 2010), and filtered alignments were concatenated into a single partition file with AMAS (Borowiec 2016). Finally, an evaluation for the best-fit model of each partition and an analysis of maximum likelihood phylogeny were performed using IQ-TREE version 2.2.0 with 10 000 bootstraps (Nguyen et al. 2015) (Table S4). A consensus tree was visualized using FigTree version 1.4.4 with an outgroup method (https://github.com/rambaut/figtree).

Using BLASTn, both primers used for gyrB PCR amplification were compared against all genomes used in the molecular phylogenetic analysis to assess if genotyping could theoretically work with diverse flavobacteria representatives (Nicolas et al. 2008). The GC content and melting temperatures of the primer sequences were calculated using the IDT oligo analyzer (https://www.idtdna.com/calc/analyzer).

Another tree was made with the sequences of the four isolates classified as outside the genus Flavobacterium genus based on the TYGS database analysis, using a different taxon sampling than the one used for the first phylogenetic tree. Chosen randomly, the whole-genome sequences of 10 species from each genus (Chryseobacterium spp., Microbacterium spp., Novosphingobium spp., and Stenotrophomonas spp.) were collected. The same was done for 22 Flavobacterium species, chosen according to the previous phylogeny (Table S5). The same homology and phylogenetic analyses were performed as described previously with certain adjustments. Specifically, 164 softcore genes were used, excluding 532 paralogous sequences. Because this phylogenetic tree was constructed with amino acid sequences instead of nucleotide sequences, translation back to nucleotides and filtration of the alignments was not performed. The evaluation of the best-fit model for each partition was also performed with 10 000 bootstraps (Table S6). The consensus tree was visualized using a midpoint method.

Results

Fifteen isolates were retrieved from various official collections and academic research laboratories (Table 1). Genotyping using gyrB and gyrA PCR with primers designed by Nicolas et al. (2008) and Shah et al. (2012) revealed that four isolates (FPG100, FPG48, FPG101, and DSM 21280), initially identified as F. psychrophilum, tested positive for both PCR amplifications. This indicates that these isolates are indeed from the genus Flavobacterium, specifically belonging to the F. psychrophilum species. The ANI results from the whole-genome sequencing compared with all genomes of the genus Flavobacterium indicated that these four isolates clustered with F. psychrophilum strain 160401–1/5 N, with respective sequence identities of 99.81%, 99.82%, 99.79%, and 99.87%.

Seven of the 15 isolates were positive for gyrB by PCR genotyping but negative for gyrA, showing that they are likely Flavobacterium species but are not F. psychrophilum. These isolates are UW10123, UW101, LBUM151, NRRL B-14732, FPG 6.1.0, DSM 18644, and DSM 25092. According to the whole-genome taxonomy, these isolates clustered with already described Flavobacterium species (Table 1). Two of these isolates (UW10123 and LBUM151) are new species because they have < 95% genetic identity with their closest relative. UW10123 was previously identified as F. johnsoniae, which is not correct according to the ANI results. Furthermore, results suggest that DSM 25092, previously identified as F. columnare, is actually F. covae.

The 11 isolates identified as Flavobacterium in the present analysis were included in a phylogenetic tree with multiple Flavobacterium genomes (Fig. 1). They are all clustered with their corresponding species indicated by the ANI results. Moreover, they all clustered within the Flavobacterium clade. BLASTn results of all genomes included in this phylogenetic analysis against gyrB primers show that 31 out of 45 flavobacteria isolates should be compatible with gyrB PCR amplification. However, four isolates, and potentially an additional 10 isolates, should not be compatible with gyrB PCR amplification because of mismatches in the 3′ part of the forward primer (Fig. 1) (Table S7).

Figure 1.

Figure 1.

Open in a new tab

Phylogeny of the isolates that are flavobacteria. Phylogenetic tree of the Flavobacterium genus using the softcore genomes of 45 species and the 11 isolates identified as Flavobacterium in this study. Two species of the genus Myroides served as outgroups in this analysis. Branches in red represent the isolates from this study that were included in the phylogeny. The numbers at the nodes indicate the level of bootstrap values (only values <100% are indicated) based on 10  000 replications. The scale represents the number of substitutions per nucleotide. BLASTn results for the gyrB primers are indicated with colored circles (Table S7). Green circles mean that the gyrB primers should hybridize because no mutations were found on the 3′ end. Brown circles mean that the gyrB primers should not work because there is an impactful mutation found on the 3′ end: a GGTG changing into a GGAG. Gray circles mean that the PCR amplification should not work because of significant discrepancies between the target sequence and the primers.

The four remaining isolates (NRRL B-14842, NRRL B-4264, NRRL B-2648, and NRRL B-14859) were identified as members of the genera Microbacterium, Stenotrophomonas, Novosphingobium, and Chryseobacterium, respectively, by the TYGS database. Additionally, these isolates were negative for both gyrB and gyrA PCR genotyping (Table 1). A phylogenetic tree was constructed with the whole-genome sequences of these four isolates and representatives of the Flavobacterium, Chryseobacterium, Microbacterium, Novosphingobium, and Stenotrophomonas genera. The four isolates did not cluster with Flavobacterium, but did cluster with their respective genera identified in the TYGS database (Fig. 2). Moreover, the ANI results showed that NRRL B-14842 seems to be a new species from the genus Microbacterium, with a sequence identity of 92.84%. NRRL B-14859 and NRRL B-2648 are new species from the Novosphingobium and Chryseobacterium genera, respectively, with genetic identities of 93.53% and 86.03%. NRRL B-4264 is a new isolate clustering with Stenotrophomonas koreensis with 97.10% identity. These four isolates were previously suggested to be part of the genus Flavobacterium (Table 1).

Figure 2.

Figure 2.

Open in a new tab

Phylogeny of the isolates that are not flavobacteria. Phylogenetic tree of the five genera (22 whole genomes of Flavobacterium, blue; 10 of Chryseobacterium, brown; 10 of Stenotrophomonas, green; 10 of Novosphingobium, pink; and 10 of Microbacterium, yellow) and isolates NRRL B-14842, NRRL B-4264, NRRL B-2648, and NRRL B-14859. The tree was made using amino acid sequences, and a midpoint method was applied. The numbers at the nodes indicate the level of bootstrap values (only values <100% are indicated) based on 10 000 replications. The scale represents the number of substitutions per amino acid.

Discussion

In this research, 15 isolates previously identified to be Flavobacterium were gathered from various bacterial collections or research laboratories (Table 1). Their taxonomy was previously established by other methodologies, such as 16S rRNA genotyping, in vitro DNA–rRNA hybridization, and phenotyping.

Genotyping using primers targeting gyrB and gyrA genes was performed on all isolates. Results showed that of the 15 tested isolates, every PCR amplification resulted in the same identification as the whole-genome approach; primers targeting the gyrB gene correctly identified Flavobacterium species while primers previously designed for the gyrA gene specifically identified F. psychrophilum (Nicolas et al. 2008, Shah et al. 2012), and primers were restrictive enough with no false positives.

Genotyping primers for gyrB and gyrA genes were designed by Shah et al. (2012) and Nicolas et al. (2008), respectively. Their studies were performed using 50 Flavobacterium isolates (for gyrB) and 12 isolates (for gyrA) from diverse fish species and organ sources of different countries (Nicolas et al. 2008, Shah et al. 2012). Considering a possible bias for flavobacteria strains linked to fish environments in these studies, we performed a BLASTn analysis to determine the capacity of gyrB primers to amplify all the species included in the phylogenetic tree in Fig. 1, which included flavobacteria from various origins. The BLASTn results show that 4 species out of 45 (8.8%) should not be compatible with the gyrB primers (F. sediminis, F. fontis, F. lotistagni, and F. succinicans; gray circles in Fig. 1) (Table S7), indicating that, although gyrB PCR amplification is an effective alternative for first screening, it does have limitations when it comes to certain species. Additionally, 10 species out of 45 species (22.2%) should not be compatible with these primers because a mismatch at the second-closest nucleotide to the 3′-terminal nucleotide has a strong negative impact on the PCR results (Bru et al. 2008) (brown circles in Fig. 1). Furthermore, instead of an A–T amplification, these species should have an A–A amplification (a purine–purine amplification), which has been shown to have the most detrimental effect: this mutation should cause a negative result for the PCR amplification on the gyrB gene (Stadhouders et al. 2010). Because these results are based in silico and PCR mismatches can have subtle exceptions (Klungthong et al. 2010), experimental confirmation of the reliability of the gyrB primers in the laboratory is recommended. In particular, research teams having strains of these species that were not amplified with the gyrB primers used in this study should carry out this validation exercise.

Ideally, strains of all Flavobacterium species should be in the same collection to provide the opportunity to test currently available primers, and also to test new primers that could allow effective genotyping of the entire genus. Such a PCR approach could revolutionize work on flavobacteria. Unfortunately, it is unlikely that such a tool could be implemented for logistical and genomic reasons because of the diversity of flavobacteria strains.

Considering all the species for which PCR amplification is possible (green circles in Fig. 1), ∼70% of the flavobacteria species analyzed could have a positive signal with this PCR approach. Although our strategy is far from perfect, it represents an interesting option for initial screening of strains, not only for bacteria originating from environments inhabited by fish, and further indicates the importance of whole-genome sequencing for identification. Additional research should identify an alternative set of primers encompassing all flavobacteria species, which could be used in parallel with the gyrB primers analyzed here.

BLASTn analyses show that the two representatives of the genus Myroides should also be amplified with gyrB primers (Fig. 1). Although it is a small sample size, the gyrB primers do not seem to be specific to the genus Flavobacterium but could theoretically hybridize with other species from close genera (García-López et al. 2019). Because of the experimental gaps shown, we propose a combination of this PCR approach with genome sequencing of isolates that have a positive signal during PCR analysis. Sequencing costs are now relatively low (∼$100 per genome and less), and having the genomic sequence opens the door to many other analyses that could further justify the sequencing effort.

Table 1 shows that 6 of 15 isolates initially identified as Flavobacterium had their taxonomy modified with our whole genome analysis. Five of these isolates receiving an updated taxonomy were initially identified using the phenotyping identification approach. Phenotyping was clearly the least precise methodology with only one isolate of six correctly identified (Table 1). The identification strategy based on the 16S rRNA gene, with or without complementary analyses, was more precise, with 8 of 9 isolates appropriately identified when compared with the whole-genome analysis. Only DSM 25092 was revealed to be misclassified; it was previously suggested to be F. columnare instead of F. covae. The phylogenetic tree in Fig. 1 shows that this isolate is phylogenetically closer to F. covae than to F. columnare, and whole-genome taxonomy supports this genetic resemblance, meaning that isolate DSM 25092 is likely to be F. covae. The previous misidentification could be because these two species, along with F. davisii and F. oreochromis, were all considered as F. columnare before 2022 (LaFrentz et al. 2022). All F. columnare isolates collected prior to 2022 should be examined to determine if they are indeed F. columnare or one of its sister species.

Isolate UW10123, identified as a close relative of F. tistrianum by the proposed strategy in this study, is likely a new species in the genus Flavobacterium because it has <95% sequence identity with all species of Flavobacterium. As shown in Fig. 1, the cluster of isolates containing UW10123 is within the genus Flavobacterium instead of outside of it, meaning that the isolate is indeed part of the genus Flavobacterium.

Another isolate, LBUM151, was determined to be a new species with <95% genetic identity when compared with all Flavobacterium spp. Its position in the phylogenetic tree shows that it is indeed a flavobacteria, and is closely related to F. pectinovorum. LBUM151 was retrieved from a research laboratory and was previously labeled as a Flavobacterium spp. by 16S rRNA gene sequences analysis. The lack of precision of the previous identification was likely because the database was much less rich than it is now. To demonstrate this hypothesis, the 16S rRNA gene sequence of LBUM151, retrieved from the annotation by Prokka, was compared using BLASTn with the 16S rRNA gene database on NCBI. The resulting identity indicates that this isolate is a new species of the genus Flavobacterium, with a corresponding identity of 98.14% to its closest relative, F. pectinovorum. This new identification agrees with that found using our new approach, as both show that LBUM151 is a new species and is closer to F. pectinovorum.

The whole-genome analytical strategy revealed that four isolates (NRRL B-14842, NRRL B-4264, NRRL B-2648, and NRRL B-14859) previously classified as Flavobacterium are, in fact, from other genera (Table 1). This statement was further affirmed by Fig. 2, which shows that these four isolates do not cluster with any of the Flavobacterium species, but instead cluster with their respective genera. Interestingly, NRRL B-2648 was identified as Novosphingobium resinovorum by whole-genome taxonomy and phylogenetic analysis. This result is aligned with a study where NRRL B-2648 was also shown to N. resinovorum by in vitro DDH (Lim et al. 2007). Note that N. resinovorum was previously known as F. resinovorum until it was reclassified by Lim et al. in 2007. It is possible that the NRRL collection has not updated their records, assuming the strain originated from the type strain.

Our results indicate that genotyping using the 16S rRNA gene alone does not seem to lead to the misclassification of Flavobacterium isolates, as mentioned earlier. However, this strategy lacks precision when identifying isolates at the species level (Johnson et al. 2019); therefore, it is not recommended to use this method alone. Our results also show that 16S rRNA gene analysis coupled with variations of DDH, such as DNA–rRNA hybridization, does seem to work for Flavobacterium isolates and could be used for taxonomic analysis.

Rather than using the TYGS database for genus identification, the Genome Taxonomy Database Toolkit offers another viable option (Chaumeil et al. 2022, 2020). In our analysis, this tool performed equally well as TYGS for genus identification, though it was less effective for species-level resolution or the identification of novel species (data not shown).

In conclusion, this study demonstrates, in the specific context of the genus Flavobacterium, a classification strategy based on whole-genome analysis is the most accurate method for species identification. This outcome was expected, as whole-genome taxonomy has long been recognized for its potential as a powerful taxonomic tool (Konstantinidis and Tiedje 2005, Kim et al. 2014). Our results provide a concrete example within the genus Flavobacterium, emphasizing the need for collections and research laboratories to update the taxonomy of their Flavobacterium isolates, particularly those previously classified using phenotypic methods.

Supplementary Material

fnae089_Supplemental_Files
fnae089_supplemental_files.zip (160.4KB, zip)

Acknowledgments

The authors thank the various groups that provided or helped to provide the bacterial strains for this project: German Collection of Microorganisms and Cell Cultures (DSMZ GmbH), USDA-ARS Culture Collection (NRRL), Félix d'Hérelle Reference Center for Bacterial Viruses, Martin Filion (University of Moncton), John Lumsden (University of Guelph), and Brian Dixon (University of Waterloo).

Contributor Information

Vincent Gélinas, Institut de Biologie Intégrative et des Systèmes (IBIS), Université Laval, Québec City, QC G1V 0A6, Canada; Département de Biochimie, de Microbiologie et de Bio-Informatique, Faculté des Sciences et de Génie, Université Laval, Québec City, QC G1V 0A6, Canada.

Valérie E Paquet, Institut de Biologie Intégrative et des Systèmes (IBIS), Université Laval, Québec City, QC G1V 0A6, Canada; Département de Biochimie, de Microbiologie et de Bio-Informatique, Faculté des Sciences et de Génie, Université Laval, Québec City, QC G1V 0A6, Canada.

Maude F Paquet, Institut de Biologie Intégrative et des Systèmes (IBIS), Université Laval, Québec City, QC G1V 0A6, Canada; Département de Biochimie, de Microbiologie et de Bio-Informatique, Faculté des Sciences et de Génie, Université Laval, Québec City, QC G1V 0A6, Canada.

Antony T Vincent, Institut de Biologie Intégrative et des Systèmes (IBIS), Université Laval, Québec City, QC G1V 0A6, Canada; Département des Sciences Animales, Faculté des Sciences de l'Agriculture et de l'Alimentation, Université Laval, Québec City, QC G1V 0A6, Canada.

Steve J Charette, Institut de Biologie Intégrative et des Systèmes (IBIS), Université Laval, Québec City, QC G1V 0A6, Canada; Département de Biochimie, de Microbiologie et de Bio-Informatique, Faculté des Sciences et de Génie, Université Laval, Québec City, QC G1V 0A6, Canada.

Conflict of interest

The authors have no conflict of interest to declare.

Funding

This project is funded by a grant number FPQ-058 from the Quebec Fisheries Fund.

Data Availability

The sequences used in this study are publicly available at the accession numbers shown in Table 1.

References

  1. Abascal F, Zardoya R, Telford MJ. TranslatorX: multiple alignment of nucleotide sequences guided by amino acid translations. Nucleic Acids Res. 2010;38:7–13. 10.1093/nar/gkq291. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bernardet J-F, Bowman J. The prokaryotes: a handbook on the biology of bacteria. Proteobacteria: delta and epsilon subclasses. Deeply Rooted Bacteria. 2006;7:481–532. [Google Scholar]
  3. Bernardet J-F, Nakagawa Y. An introduction to the family Flavobacteriaceae. In: Dworkin M, Falkow S, Rosenberg E, et al. (eds.), The Prokaryotes, Vol. 7. New York, NY: Springer, 2006,455–80. 10.1007/0-387-30747-8_16. [DOI] [Google Scholar]
  4. Bernardet J-F, Segers P, Vanva Met al. Cutting a gordian knot: emended classification and description of the genus Flavobacterium, emended description of the family Flavobacteriaceae, and proposal of Flavobacterium hydatis nom. nov. (Basonym, Cytophaga aquatilis Strohl and Tait 1978). Int J Syst Evol Microbiol. 1996;46:128–48. 10.1099/00207713-46-1-128. [DOI] [Google Scholar]
  5. Borowiec ML. AMAS: a fast tool for alignment manipulation and computing of summary statistics. PeerJ. 2016;4:e1660. 10.7717/peerj.1660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bru D, Martin-Laurent F, Philippot L. Quantification of the detrimental effect of a single primer-template mismatch by real-time PCR using the 16S rRNA gene as an example. Appl Environ Microb. 2008;74:1660–3. 10.1128/AEM.02403-07. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cai H, Zeng Y, Wang Yet al. Flavobacterium cyanobacteriorum sp. nov., isolated from cyanobacterial aggregates in a eutrophic lake. Int J Syst Evol Microbiol. 2018;68:1279–84. 10.1099/ijsem.0.002664. [DOI] [PubMed] [Google Scholar]
  8. Chang LE, Pate JL, Betzig RJ. Isolation and characterization of nonspreading mutants of the gliding bacterium Cytophaga johnsonae. J Bacteriol. 1984;159:26–35. 10.1128/jb.159.1.26-35.1984. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Charette SJ, Cosson P.. Preparation of genomic DNA from Dictyostelium discoideum for PCR analysis. BioTechniques. 2004;36:574–5. 10.2144/04364BM01. [DOI] [PubMed] [Google Scholar]
  10. Chaumeil P-A, Mussig AJ, Hugenholtz Pet al. GTDB-Tk: a toolkit to classify genomes with the genome taxonomy database. Bioinformatics. 2020;36:1925–7. 10.1093/bioinformatics/btz848. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chaumeil P-A, Mussig AJ, Hugenholtz Pet al. GTDB-Tk v2: memory friendly classification with the genome taxonomy database. Bioinformatics. 2022;38:5315–6. 10.1093/bioinformatics/btac672. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chen S, Zhou Y, Chen Yet al. fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics. 2018;34:884–90. 10.1093/bioinformatics/bty560. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Chen S. Ultrafast one-pass FASTQ data preprocessing, quality control, and deduplication using fastp. iMeta. 2023;2:e107. 10.1002/imt2.107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Contreras-Moreira B, Vinuesa P. GET_HOMOLOGUES, a versatile software package for scalable and robust microbial pangenome analysis. Appl Environ Microb. 2013;79:7696–701. 10.1128/AEM.02411-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Criscuolo A, Gribaldo S. BMGE (Block Mapping and Gathering with Entropy): a new software for selection of phylogenetic informative regions from multiple sequence alignments. BMC Evol Biol. 2010;10:210–31. 10.1186/1471-2148-10-210. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Daskalov H, Austin DA, Austin B.. An improved growth medium for Flavobacterium psychrophilum. Lett Appl Microbiol. 1999;28:297–9. 10.1046/j.1365-2672.1999.00522.x. [DOI] [PubMed] [Google Scholar]
  17. Delaporte B, Daste P., Une bactérie du sol capable de décomposer la fraction fixe de certaines oléorésines Flavobacterium resinovorum n. sp. C R Acad Sci. 1956;242:831–4. [Google Scholar]
  18. Fujiwara-Nagata E, Shindoh Y, Yamamoto Met al. Distribution of Flavobacterium psychrophilum and its gyrA genotypes in a river. Fish Sci. 2019;85:913–23. 10.1007/s12562-019-01355-7. [DOI] [Google Scholar]
  19. García-López M, Meier-Kolthoff JP, Tindall BJet al. Analysis of 1,000 type-strain genomes improves taxonomic classification of bacteroidetes. Front Microbiol. 2019;10:2083–157. 10.3389/fmicb.2019.02083. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hesami S, Allen KJ, Metcalf Det al. Phenotypic and genotypic analysis of Flavobacterium psychrophilum isolates from Ontario salmonids with bacterial coldwater disease. Can J Microbiol. 2008;54:619–29. 10.1139/w08-058. [DOI] [PubMed] [Google Scholar]
  21. Hou CT. Production of 10-ketostearic acid from oleic acid by Flavobacterium sp. isolate DS5 (NRRL B-14859). Appl Environ Microb. 1994;60:3760–3. 10.1128/aem.60.10.3760-3763. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Izumi S, Fujii H, Aranishi F. Detection and identification of Flavobacterium psychrophilum from gill washings and benthic diatoms by PCR-based sequencing analysis. J Fish Dis. 2005;28:559–64. 10.1111/j.1365-2761.2005.00663.x. [DOI] [PubMed] [Google Scholar]
  23. Izumi S, Wakabayashi H. Sequencing of gyrB and their application in the identification of Flavobacterium psychrophilum by PCR. Fish Pathol. 2000;35:93–94. 10.3147/jsfp.35.93. [DOI] [Google Scholar]
  24. Jarau M, Di Natale A, Huber PEet al. Virulence of Flavobacterium psychrophilum isolates in rainbow trout Oncorhynchus mykiss (Walbaum). J Fish Dis. 2018;41:1505–14. 10.1111/jfd.12861. [DOI] [PubMed] [Google Scholar]
  25. Johnson JS, Spakowicz DJ, Hong BYet al. Evaluation of 16S rRNA gene sequencing for species and isolate-level microbiome analysis. Nat Commun. 2019;10:5029–40. 10.1038/s41467-019-13036-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Jung EJ, Choi SY, Lee S-Met al. Flavobacterium dauae sp. nov., isolated from rhizosphere soil of a tomato plant. Int J Syst Evol Microbiol. 2021;71:1–7. 10.1099/ijsem.0.005033. [DOI] [PubMed] [Google Scholar]
  27. Kämpfer P, Irgang R, Glaeser SPet al. Flavobacterium salmonis sp. nov. Isolated from Atlantic salmon (Salmo salar) and formal proposal to reclassify Flavobacterium spartansii as a later heterotypic synonym of Flavobacterium tructae. Int J Syst Evol Microbiol. 2020;70:6147–54. 10.1099/ijsem.0.004510. [DOI] [PubMed] [Google Scholar]
  28. Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol. 2013;30:772–80. 10.1093/molbev/mst010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kayansamruaj P, Dong HT, Hirono Iet al. Comparative genome analysis of fish pathogen Flavobacterium columnare reveals extensive sequence diversity within the species. Infect Genet Evol. 2017;54:7–17. 10.1016/j.meegid.2017.06.012. [DOI] [PubMed] [Google Scholar]
  30. Khan I, Debnath SC, Yan Cet al. Flavobacterium ajazii sp. nov., isolated from seaweed of Gouqi Island, China. Curr Microbiol. 2020;77:2925–32. 10.1007/s00284-020-02112-1. [DOI] [PubMed] [Google Scholar]
  31. Kim M, Oh HS, Park SCet al. Towards a taxonomic coherence between average nucleotide identity and 16S rRNA gene sequence similarity for species demarcation of prokaryotes. Int J Syst Evol Microbiol. 2014;64:346–51. 10.1099/ijs.0.059774-0. [DOI] [PubMed] [Google Scholar]
  32. Klungthong C, Chinnawirotpisan P, Hussem Ket al. The impact of primer and probe-template mismatches on the sensitivity of pandemic influenza A/H1N1/2009 virus detection by real-time RT-PCR. J Clin Virol. 2010;48:91–95. 10.1016/j.jcv.2010.03.012. [DOI] [PubMed] [Google Scholar]
  33. Konstantinidis KT, Tiedje JM. Towards a genome-based taxonomy for prokaryotes. J Bacteriol. 2005;187:6258–64. 10.1128/JB.187.18.6258-6264.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. LaFrentz BR, Králová S, Burbick CRet al. The fish pathogen Flavobacterium columnare represents four distinct species: F lavobacterium columnare, Flavobacterium covae sp. nov., Flavobacterium davisii sp. nov. and Flavobacterium oreochromis sp. nov., and emended description of Flavobacterium columnare. Syst Appl Microbiol. 2022;45:126293. 10.1016/j.syapm.2021.126293. [DOI] [PubMed] [Google Scholar]
  35. Lim YW, Moon EY, Chun J. Reclassification of Flavobacterium resinovorum delaporte and Daste 1956 as Novosphingobium resinovorum comb. nov., with Novosphingobium subarcticum (Nohynek et al. 1996) Takeuchi et al. 2001 as a later heterotypic synonym. Int J Syst Evol Microbiol. 2007;57:1906–8. 10.1099/ijs.0.64852-0. [DOI] [PubMed] [Google Scholar]
  36. Loch TP, Faisal M. Emerging flavobacterial infections in fish: a review. J Adv Res. 2015;6:283–300. 10.1016/j.jare.2014.10.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Mateo-Estrada V, Graña-Miraglia L, López-Leal Get al. Phylogenomics reveals clear cases of misclassification and genus-wide phylogenetic markers for Acinetobacter. Genome Biol Evol. 2019;11:2531–41. 10.1093/gbe/evz178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Meier-Kolthoff JP, Sardà Carbasse J, Peinado-Olarte RLet al. TYGS and LPSN: a database tandem for fast and reliable genome-based classification and nomenclature of prokaryotes. Nucleic Acid Res. 2022;50:801–7. 10.1093/nar/gkab902. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Mikheenko A, Prjibelski A, Saveliev Vet al. Versatile genome assembly evaluation with QUAST-LG. Bioinformatics. 2018;34:42–50. 10.1093/bioinformatics/bty266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Nguyen LT, Schmidt HA, von Haeseler Aet al. IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol Biol Evol. 2015;32:268–74. 10.1093/molbev/msu300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Nicolas P, Mondot S, Achaz Get al. Population structure of the fish-pathogenic bacterium Flavobacterium psychrophilum. Appl Environ Microb. 2008;74:3702–9. 10.1128/AEM.00244-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Ranjan R, Rani A, Metwally Aet al. Analysis of the microbiome: advantages of whole genome shotgun versus 16S amplicon sequencing. Biochem Biophys Res Commun. 2016;469:967–77. 10.1016/j.bbrc.2015.12.083. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Rekadwad BN, Gonzalez JM.. Correcting names of bacteria deposited in National Microbial Repositories: an analysed sequence data necessary for taxonomic re-categorization of misclassified bacteria-ONE example, genus Lysinibacillus. Data Brief. 2017;13:761–78. 10.1016/j.dib.2017.06.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Saticioglu IB, Ay H, Altun Set al. Flavobacterium turcicum sp. nov. And Flavobacterium kayseriense sp. nov. Isolated from farmed rainbow trout in Turkey. Syst Appl Microbiol. 2021;44:126186. 10.1016/j.syapm.2021.126186. [DOI] [PubMed] [Google Scholar]
  45. Seemann T. Prokka: rapid prokaryotic genome annotation. Bioinformatics. 2014;30:2068–9. 10.1093/bioinformatics/btu153. [DOI] [PubMed] [Google Scholar]
  46. Seo J, Peng Y, Jiang Let al. Flavobacterium endoglycinae sp. nov., an endophytic bacterium isolated from soybean (Glycine max L. cv. Gwangan) stems. Int J Syst Evol Microbiol. 2022;72:1–9. 10.1099/ijsem.0.005220. [DOI] [PubMed] [Google Scholar]
  47. Shah SQ, Nilsen H, Bottolfsen Ket al. DNA gyrase and topoisomerase IV mutations in quinolone-resistant Flavobacterium psychrophilum isolated from diseased salmonids in Norway. Microb Drug Resist. 2012;18:207–14. 10.1089/mdr.2011.0142. [DOI] [PubMed] [Google Scholar]
  48. Soda K, Misono H. L-Lysine:alpha-ketoglutarate aminotransferase. II. Purification, crystallization, and properties. Biochemistry. 1968;7:4110–9. 10.1021/bi00851a046. [DOI] [PubMed] [Google Scholar]
  49. Stadhouders R, Pas SD, Anber Jet al. The effect of primer-template mismatches on the detection and quantification of nucleic acids using the 5' nuclease assay. J Mol Diagn. 2010;12:109–17. 10.2353/jmoldx.2010.090035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Sugahara K, Fujiwara-Nagata E, Eguchi M. Dynamics of the bacterial cold-water disease pathogen, Flavobacterium psychrophilum, in infected fish organs and rearing water after warmed water treatment. Fish Pathol. 2010;45:58–65. 10.3147/jsfp.45.58. [DOI] [Google Scholar]
  51. Sun H, Zheng H, Wang Xet al. Flavobacterium coralii sp. nov., a marine bacterium isolated from coral culture seawater. Int J Syst Evol Microbiol. 2022;72:1–6. 10.1099/ijsem.0.005201. [DOI] [PubMed] [Google Scholar]
  52. Sun J-Q, Xu L, Liu Met al. Flavobacterium suaedae sp. nov., an endophyte isolated from the root of Suaeda corniculata. Int J Syst Evol Microbiol. 2016;66:1943–9. 10.1099/ijsem.0.000967. [DOI] [PubMed] [Google Scholar]
  53. Van Trappen S, Vandecandelaere I, Mergaert Jet al. Flavobacterium degerlachei sp. nov., Flavobacterium frigoris sp. nov. And Flavobacterium micromati sp. nov., novel psychrophilic bacteria isolated from microbial mats in Antarctic lakes. Int J Syst Evol Microbiol. 2004;54:85–92. 10.1099/ijs.0.02857-0. [DOI] [PubMed] [Google Scholar]
  54. Weeks OB. Flavobacterium aquatile (Frankland and Frankland) Bergey et al., type species of the genus Flavobacterium. J Bacteriol. 1954;69:649–58. 10.1128/jb.69.6.649-658.1955. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Xu L, Wang H-T, Zhang J-Xet al. Flavobacterium alkalisoli sp. nov., isolated from rhizosphere soil of Suaeda salsa. Int J Syst Evol Microbiol. 2020;70:3888–98. 10.1099/ijsem.0.004255. [DOI] [PubMed] [Google Scholar]
  56. Yi H, Oh H-M, Lee J-Het al. Flavobacterium antarcticum sp. nov., a novel psychrotolerant bacterium isolated from the Antarctic. Int J Syst Evol Microbiol. 2005;55:637–41. 10.1099/ijs.0.63423-0. [DOI] [PubMed] [Google Scholar]
  57. Yoon S-H, Ha S, Lim Jet al. A large-scale evaluation of algorithms to calculate average nucleotide identity. Antonie Van Leeuwenhoek. 2017;110:1281–6. 10.1007/s10482-017-0844-4. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

fnae089_Supplemental_Files
fnae089_supplemental_files.zip (160.4KB, zip)

Data Availability Statement

The sequences used in this study are publicly available at the accession numbers shown in Table 1.

Articles from FEMS Microbiology Letters are provided here courtesy of Oxford University Press

RESOURCES