Flavobacterium is a genus of gram-negative bacteria belonging to the family Flavobacteriaceae (order: Flavobacteriales, class: Flavobacteria, phylum: Bacteroidetes) [1]. The term Flavobacterium was first introduced by Bergey et al. in 1923 [1]. Since then, a number of Flavobacterium species have been identified, and the classification of the genus has been continuously amended. At present, it is widely accepted that the genus Flavobacterium contains bacteria that are yellow-pigmented, aerobic, and generally rod-shaped [2], and some Flavobacterium species mobilize through a gliding motility mechanism [3]. Flavobacterium strains could be isolated from various habitats, including freshwater, soil, plants, and food [3, 4].

The 16S rRNA gene has been widely used in many studies concerned with the identification or classification of microorganisms. However, using only one housekeeping gene as a single marker is not enough to distinguish some bacterial species, especially when they are closely related. In addition, gene-specific changes caused by horizontal gene transfer or recombination could be an obstacle to the phylogenetic studies that are based on a single gene marker [5, 6]. To overcome the limitation, multilocus sequence analysis (MLSA) has been proposed as an alternative method in many studies. In reality, MLSA is a microbial genotyping method used to characterize a variety of microorganisms, including the genera of Pantoea [7], Bifidobacterium [8], Borrelia [9], Bradyrhizobium [10], and Bacteroides [11]. In MLSA, approximately seven housekeeping genes are required, but a standard for the given number and size of genes has not been established. This concise approach is based on concatenated sequences of multiple housekeeping genes, uses the concatenated sequences to identify the phylogenetic aspects of desired species, and investigates the patterns of genotypes between each species cluster [12].

This study was designed to classify the genetic diversity of collected Flavobacterium species. In addition, we also investigated their relationship using phylogenetic analysis and compared the results for phylogenetic markers.

In this study, to determine the phylogenetic relationship of the five closely related Flavobacterium species that were previously isolated from a freshwater reservoir in Chonan, Korea [13, 14], we constructed a maximum-liklihood phylogenetic tree, but the tree showed a very low bootstrap value (<25%) between them. Previously, Nicolas et al. [15] recommended several housekeeping genes to analyze the phylogenetic relationship among fish-pathogenic F. psychrophilum strains. This approach is called MLSA, which is very useful for distinguishing closely related species and showing population structure. We have chosen seven housekeeping genes as phylogenetic markers-atpA, dnaK, glyA, gyrB, murG, tuf (elongation factor Tu), and trpB (tryptophan synthase, subunit B)-because these loci contain the largest amounts of polymorphisms among the 50 F. psychrophilum strains [15].

This study is the first to apply Flavobacterium species isolated from a freshwater reservoir. First, we tried to amplify seven housekeeping genes from five Flavobacterium isolates [13, 14], which are closely related to each other based on the 16S rRNA gene. However, we could not amplify the trpB gene from our samples from a freshwater reservoir. Therefore, we collected the sequences of six housekeeping genes and the 16S rRNA gene from five Flavobacterium isolates [13, 14], isolated from a freshwater reservoir in Cheonan, Korea, and two Flavobacterium reference strains, F. aquatile ATCC 11947 and F. frigidimaris F3-JCM122, as interspecies out-groups. In addition, we searched for more Flavobacterium species that were available for the six housekeeping genes from the GenBank database. We found that 52 Flavobacterium species were available for all six housekeeping gene sequences in the database. Among them, 50 Flavobacterium species were F. psychrophilum that were identified and studied by Nicolas et al. [15]. The 50 F. psychrophila were assigned to 33 sequence types. Consequently, to evaluate the resolution and efficiency of MLSA, we aligned and trimmed each housekeeping gene sequence (555 bp of gyrB, 680 bp of dnaK, 833 bp of tuf, 725 bp of murG, 796 bp of atpA, and 823 bp of glyA genes) from the five Flavobacterium isolates, 33 distinct F. psychrophila, two other Flavobacterium species (F. branchiophilum FL-15 and F. indicum GPTSA100-9), and two reference strains. As shown in Fig. 1, we constructed two different phylogenetic trees of the Flavobacterium species, based on only 16S rRNA gene sequences (Fig. 1A) and based on the concatenated sequences of the six housekeeping genes (Fig. 1B). The bootstrap values of the tree branches clearly improved in Fig. 1B compared with those in Fig. 1A. Therefore, we could state that the phylogenetic tree of the genus Flavobacterium, produced by the MLSA approach, is more reliable than the tree based on a single locus (e.g., 16S rRNA gene).

To investigate the nucleotide diversity of each gene among the 10 Flavobacterium species, we analyzed the pairwise distances determined by MEGA5.05 (Table 2). Among the six housekeeping genes and the 16S rRNA gene, the highest average pairwise distance was 0.303 in murG, while the lowest average pairwise distance was 0.043 in the 16S rRNA gene. We found that the average pairwise distance within the gyrB, dnaK, tuf, atpA, and glyA genes was 0.218, 0.141, 0.115, 0.150, and 0.215, respectively, among Flavobacterium species. In addition, we estimated the pairwise distance of the six concatenated housekeeping genes, and thus, the pairwise distance was 0.164 on average. The pairwise distance of the six concatenated housekeeping genes was significantly greater than the pairwise distance of the 16S rRNA gene (0.043). Next, we analyzed G + C contents and variable nucleotide sites of six housekeeping genes in our MLSA data and 16S rRNA gene (Table 2). The G + C contents of the genes ranged from 35.4% (murG) to 48.9% (16S rRNA) (Table 2). The G + C content among housekeeping genes was found to be variable.

As a result, this study has demonstrated that MLSA, based on multiloci, produces more reliable phylogenetic trees among Flavobacterium species compared with the tree based on a single locus (e.g., 16S rRNA). When a high number of sequential similarities exist in genetic markers, it might be difficult to identify the phylogenetic relationship between microorganisms of the same genus. Therefore, MLSA will be a good approach to classify the genetic relationships among microorganisms.