Genomic identification of rhizobia-related strains and threshold of ANI and core-genome for family, genus and species

It is well known that the symbiotic bacteria (rhizobia) and the tumor-inducing phytopathogenic bacteria (agrobacteria) in Rhizobiaceae family are phylogenetically intermingled in some genera, even in the same species. Originally, the symbiotic bacteria were all grouped within the genus Rhizobium, which was established in 1890 with Rhizobium leguminosarum as the type species [1, 2]; and the tumor-inducing phytopathogenic bacteria were designed as the genus Agrobacterium which was first proposed by Conn including Agrobacterium tumefaciens (tumor-inducing), Agrobacterium radiobacter (no tumor) and Agrobacterium rhizogenes (hairy root) based on their phytopathogenic symptoms [3]. Later, Agrobacterium rubi (from Rubiaceae plants), Agrobacterium vitis (from Vitis plants) and Agrobacterium larrymoorei (from Ficus plants) were established [4-6], which were divided into Biovars I, II and III [7]. Based upon the phylogeny of 16S rRNA gene, the genus Agrobacterium and a later described genus Allorhizobium [8] were officially immerged into Rhizobium [9]. However, this combination caused frequently argument because their different affection on plants, and their divergent phylogenetic relationships of 16S rRNA, 23S rRNA and recA genes [10-14], as well as the fatty acid profiles [15]. With description of more and more symbiotic and non-symbiotic species in the combined genus Rhizobium, its polyphylic feature was further apparent.

 Meanwhile, some novel molecular techniques have been developed for estimating the phylogenetic relationships, such as the multilocus sequence analysis (MLSA) and whole genome sequencing. Recently, the taxonomy of Agrobacterium/Rhizobium group was dramatically revised again based upon the MLSA data of four or six protein-coding housekeeping genes [16-17], which led the split of Agrobacterium/Rhizobium group into five sister genera, Agrobacterium, Allorhizobium, Neorhizobium, Pararhizobium and Rhizobium. In the recently emended Agrobacterium genus, Ag. radiobacter and Ag. rubi are phytopathogenic species, while Ag. nepotum, Ag. pusense, and Ag. skierniewicense were new combinations transferred from the former Rhizobium species. The emended Allorhizobium covered the phytopathogenic species Al. vitis (formerly Agrobacterium vitis), and the symbiotic or endophytic species Al. taibaishanense, Al. paknamense, Al. oryzae, Al. pseudoryzae and Al. borbori. The genus Neorhizobium included the species N. galegae, N. vignae, N. huautlense and N. alkalisoli transferred from the former Rhizobium species [16]. Pararhizobium included P. giardinii, P. capsulatum, P. herbae and P. sphaerophysae [17], which were all transferred from the former Rhizobium species. After the reversion, the species represented by Rhizobium leguminosarum are maintained in the genus Rhizobium, and the phytopathogenic species R. rhizogenes (former Agrobacterium rhizogenes) was also included in this genus.

Despite the nomenclature change or taxonomic reversion, the pathogenic (for plants and human being), symbiotic, endophytic and saprophytic bacterial species are intermingled in the five Agrobacterium/Rhizobium sister genera [16-18]. Furthermore, these four living states or characters even can be found in the single species Ag. radiobacter [19] or in the same strains of R. rhizogenes [20]. Although the recent reversions have resolved the nomenclature argument about the symbiotic Rhizobium species and the phytopathogenic Agrobacterium species, the phylogenetic relationships between the symbiotic species and phytopathogenetic species were still not sufficiently revealed because only several housekeeping genes have been considered [16-17]. To obtain an insight view in the phylogenetic relationships among the members of Agrobacterium/Rhizobium, the whole genome comparison would be valuable. 

Previously, we isolated some arsenite-oxidizing or antimonite tolerant strains and they were primitively identified as unnamed species within Agrobacterium and Sinorhizobium based on the 16S rRNA gene sequence analyses [21-23]. Aiming at further identifying them, as well as developing a rapid, confident/stable, high-throughput identification method, we performed this study by using the genome data. In particular, the average nucleotide identity (ANI) and core-genome [24] were estimated to ascertain the phylogenetic relationships among the 45 strains in the family Rhizobiaceae. The results offered accurate identification of our test strains and generated some valuable taxonomic clues.