Genetic characterization at the species and symbiovar level of indigenous rhizobial isolates nodulating Phaseolus vulgaris in Greece

Phaseolus vulgaris (L.), commonly known as bean or common bean, is considered a promiscuous legume host since it forms nodules with diverse rhizobial species and symbiovars. Most of the common bean nodulating rhizobia are mainly affiliated to the genus Rhizobium, though strains belonging to Ensifer, Pararhizobium, Mesorhizobium, Bradyrhizobium, and Burkholderia have also been reported. This is the first report on the characterization of bean-nodulating rhizobia at the species and symbiovar level in Greece. The goals of this research were to isolate and characterize rhizobia nodulating local common bean genotypes grown in five different edaphoclimatic regions of Greece with no rhizobial inoculation history. The genetic diversity of the rhizobial isolates was assessed by BOX-PCR and the phylogenetic affiliation was assessed by multilocus sequence analysis (MLSA) of housekeeping and symbiosis-related genes. A total of fifty fast-growing rhizobial strains were isolated and representative isolates with distinct BOX-PCR fingerpriniting patterns were subjected to phylogenetic analysis. The strains were closely related to R. anhuiense, R. azibense, R. hidalgonense, R. sophoriradicis, and to a putative new genospecies which is provisionally named as Rhizobium sp. I. Most strains belonged to symbiovar phaseoli carrying the α-, γ-a and γ-b alleles of nodC gene, while some of them belonged to symbiovar gallicum. To the best of our knowledge, it is the first time that strains assigned to R. sophoriradicis and harbored the γ-b allele were found in European soils. All strains were able to re-nodulate their original host, indicating that they are true microsymbionts of common bean.

16S rRNA gene analysis. According to the BOX grouping results, seven isolates (PVIM1, PVIM10, PVKA6, PVMT25, PVMT26, PVPR1, and PVTN21) representing six different BOX patterns and originating from different geographic regions were chosen for subsequent analyses. Nearly full-length rrs gene sequences (> 1350 bp) were determined for all representative isolates and a region of 1308 bp was considered for the alignment. The 16S rRNA gene phylogenetic tree showed that all isolates were closely related to the defined species within the genus Rhizobium (Fig. 1).
Despite that the 16S rRNA gene is widely used as a molecular marker in the taxonomy of prokaryotes, it is not sufficient to differentiate closely related species within the genus Rhizobium since different type strains share identical rrs sequences 6,60 . In agreement with previous studies, our results showed that R. anhuiense, R. laguerreae, R. hidalgonense FH14 T , R. ruizarguesonis UMP1133 T and R. sophorae, as well as R. ecuadorense and R. pisi shared identical rrs sequences 8,13,61,62 . Multilocus sequence analysis of housekeeping genes. To clarify the 16S rRNA results, multilocus sequence analysis (MLSA) was performed using the housekeeping genes recA, atpD, gyrB, and glnII that have widely been used for delineation of Rhizobium species as well as for the identification of common bean nodulating rhizobia 14,15,46,47,63,65 . Ribeiro et al. (2009) described a useful MLST scheme for the identification and classification of rhizobial microsymbionts of common bean (Phaseolus vulgaris L.) by using housekeeping and symbiotic genes. Tong et al. (2018) demonstrated that a 97.36% threshold in MLSA of three housekeeping genes (~ 1055 bp), was concordant with the 95% ANI threshold for rhizobial species definition. Interestingly, recent genomic and phylogenomic studies have shown that several Rhizobium species are organized in welldefined genome clusters with ANI values > 96%, whereas others displayed a continuum of diversity with ANI values > 88% 67,68 . These findings indicated that a default ANI cut-off cannot be applied across all Rhizobium species and even more a general threshold for rhizobial species delineation in MLSA cannot be specified as we also www.nature.com/scientificreports/ pointed out previously 69 . Although phylogeny based on three core genes is not as accurate as the entire genome, ML analysis of few genes can still offer a demonstration for the taxonomic status of rhizobial strains. In the present study, partial fragments of recA, atpD, gyrB, and glnII were amplified from all representative isolates. The number of parsimony-informative sites for every selected gene was estimated within the test Rhizobium taxa to find those who were the most phylogenetically informative. In our analysis, gyrB had the best percentage of parsimony-informative characters (29.12%), as previously reported 64 , followed by recA (25.54%), atpD (23.13%), and glnII (21.5%).
Gene sequences for Rhizobium type/reference strains were retrieved from the GenBank and correctly trimmed. The lengths of the alignments used were 462 bp, 441 bp 594 bp, and 465 bp for recA, atpD, gyrB, and glnII, respectively. Phylogenetic trees based on four individual housekeeping genes were constructed and the percentage identity of each gene was also calculated (Supplementary Figs. S3-S6, Supplementary Tables S3-S4). The lengths of the alignments used were 462 bp, 441 bp 594 bp, and 465 bp for recA, atpD, gyrB, and glnII, respectively.
The analysis of the concatenated sequences of housekeeping genes recA, atpD, gyrB, and glnII provided more robust phylogenies of the test strains and congruent with those of the individual gene trees (Fig. 2, Supplementary Figs. S3-S6). The test strains were grouped into five well-supported clades containing defined Rhizobium species, except for clade 1 that included the strains PVMΤ25, PVKA6, and PVΙΜ10 originated from three different geographical regions of Greece (Fig. 2, Table 1). Phylogenetic analysis showed that the above strains belong to a   Table S5). This identity value was lower than those found among Rhizobium type strains analysed in the dataset of the present study (Supplementary Table S5). In our pairwise analysis, four pairs of Rhizobium type strains showed identity values in the recA-atpD-glnII-gyrB concatenated sequences higher than 95.2%, which were presented between the pairs of R. azibense 23C2 T and R. mongolense USDA 1844 T (97.25%), R. gallicum R602sp T and R. azibense 23C2 T (96.69%), R. pisi DSM 30132 T and R. fabae CCBAU 33202 T (97.6%), R. aethiopicum HBR26 T and R. aegyptiacum 950 T (99.24%). These results, together with the position of PVMΤ25, PVKA6, and PVΙΜ10 in the phylogenetic tree suggested that they might constitute a putative novel genospecies within Rhizobium. Previously, MLSA and whole-genome analyses defined 25 species or genospecies among the bean-nodulating rhizobia, while species affiliations for some previously named strains were reassigned 6 . Comparison of our strains with the defined genospecies and those isolated previously from bean root-nodules in various countries was also performed to determine their relationships. Since not all gene sequences were available for all strains, a concatenated phylogenetic tree based on the recA and atpD sequences was constructed ( Supplementary Fig. S7). To avoid confusing the reader, in our analysis the grouping of strains taken from the literature did not correspond to the given species names at the time of their deposition, since many bean-nodulating strains were inaccurately assigned at the species level and therefore misnamed due to weak characterization. Interestingly, our isolates were closely related (> 99.9%) to those of Rhizobium sp. M1 and M10 isolated from nodules of P. vulgaris in China 70 . Recently, the latter two strains were assigned to an unidentified genospecies named as Rhizobium sp. I, based on genomic data 6 71 . All strains were grouped in a well-supported cluster (Clade 2) containing R. sophoriradicis CCBAU 03470 T as well as the strains JJW1, L101, Kim5, and IE4803, which were recently assigned to R. sophoriradicis based on genomic data 6 . Therefore, all strains of clade 2 should be assigned to R. sophoriradicis. To the best of our knowledge, it is the first time that strains belonging to R. sophoriradicis were found in European soils. The wide distribution of R. sophoriradicis in P. vulgaris nodules all over the world suggests that this species is likely well adapted to different environmental conditions and various bean varieties.
The strain PVMT26, representing seven isolates from one region (Table 1), showed high sequence relatedness to R. hidalgonense FH14 T in all individual gene phylogenies, with identity values ranging from 99.3 to 100%, and in combined sequences of the four genes (99.6%) (Fig. 2 Tables S3-S4). Although this type strain was isolated from nodules of Phaseolus vulgaris grown in Mexico 77 , it did not form nodules on its original host P. vulgaris and other tested legumes evidenced the loss of its nodulation ability 13 . Despite that nodC gene was not amplified from the strain FH14 T , it is present in the genome sequence of FH14 T (NZ_LODW01000075). Strains closely related to R. hidalgonense have also been isolated from nodules of Phaseolus vulgaris grown in Spain (LBM1212, LBM1123, LCS0303, LCS0401, LCS0411, LEV0613 and RPVR24) 46,47 , Mexico (NH05) 77 , China (CCBAU 65761) 65 , Iran (Hm1) 5 , Kenya (NAK 327, 321, 334) 74 , and Croatia (25 T and 26 T) 73 . Noteworthy, strains closely related to R. hidalgonense have also been isolated from other legumes including Acacia gummifera 78 , Indigofera arrecta in Ethiopia 3 , Trifolium spp. in Ethiopia 79 , T. semipilosum in Kenya 80 , Vicia faba in Ethiopia and China 78,81 . The concatenated analysis of recA-atpD showed that all these strains formed a highly bootstrapped cluster with R. hidalgonense FH14 T and displayed high nucleotide identities of recA-atpD (> 99.4%). Therefore, several strains previously named as R. leguminosarum, such as LBM1212, LBM1123, LEV0613, WSM2012, NH05, and CCBAU 65761, or Rhizobium sp., such as NAK 327, 321, 334, LCS0401, LCS0411, and RPVR24 might be reclassified in the future as R. hidalgonense taking into account phenotypic and chemotaxonomic data.
Phylogenetic analysis based either on the individual or concatenated gene trees showed that PVIM1, representing seven isolates (Table 1), was clustered together with Rhizobium azibense 23C2 T , isolated from common bean nodules in Tunisia 7,36 . Based on the pair-wise comparisons of concatenated sequences of four genes, PVIM1 displayed 99.75% identity to R. azibense 23C2 T and consequently was assigned to this species (Fig. 2, Supplementary Table S5). Strains belonging to R. azibense have also been isolated from nodules of P. vulgaris ( Supplementary Fig. S7), such as IE4868 from Mexico 43 , 8C-3, and GR42 from Spain 7,36,45 . The strain 8C-3 was originally classified as R. gallicum 45 but it was recently reassigned to R. azibense based on genomic data 6 . Interestingly, the strains IE4868, 8C-3 and GR42 formed a separate well-supported sub-clade closely related to R. www.nature.com/scientificreports/ azibense 23C2 T with identity values of recA-atpD concatenated sequences ranged from 96.1% to 96.4%, while the isolate PVIM1 displayed 99.88% identity. Therefore, the Spanish isolates appeared to be more similar to the Mexican ones, while the Greek isolates were phylogenetically closer to the Tunisian strain suggesting that the two sub-clades may represent distinct lineages within R. azibense species with a different origin.
Concerning the distribution of our isolates in different regions of Greece, Clade 1 isolates, possibly belonging to genospecies Rhizobium sp. I, were found in three regions with different soil textures (SCL, CL and SL) and pH ranging from 6.9 to 7.9 ( Supplementary Fig. S1 and Supplementary Table S6). Interestingly, isolates of Clade 2 belonging to R. sophoriradicis were predominant in Tinos (soil SCL, pH 8.1), although one isolate was isolated from another region (Metsovo) with different soil textures (SL) and pH 6.9. Despite that Clade 3, 4, and 5 isolates were identified solely in Preveza, Metsovo, and Imathia, respectively, these findings could not rule out the existence of similar isolates in other regions if more isolates were examined or genomic approaches were used. Therefore, the present study cannot provide conclusive evidence for the association of the rhizobial diversity with the edaphic parameters or host genotypes at our sampling sites. To define the factors influencing the distribution of different species or genospecies in Greek soil, further studies are required.
Phylogenetic analysis of symbiosis genes nodC and nifH. Currently, the nodC gene is commonly used to define symbiovars within rhizobial species. P. vulgaris is considered to be a promiscuous host since it can be nodulated by different rhizobial species and symbiovars 1,2 . At least thirty rhizobial species and eight symbiovars have been reported to nodulate common bean so far 2,31,32 . However, most bean-nodulating rhizobia, regardless of their species affiliation, belong to sv. phaseoli, which also exclusively nodulates P. vulgaris 12,82 . Previously, the sv. phaseoli was divided into three sub-clades, representing different alleles of nodC designated α, γ-a, and γ-b 5,39,74 . The γ nodC allele is considered to be the most widely distributed worldwide, implying a distribution of this allele together with bean seeds from their American distribution centers 39,46,82,83 .
Partial nucleotide sequences of nodC and nifH were amplified and sequenced for all representative strains and their phylogenetic trees are shown in Figs. 3 and 4, respectively. In the nodC and nifH trees, most Greek isolates were placed into two well-supported clades that corresponded to symbiovars phaseoli and gallicum. The inclusion of representative strains carrying different nodC alleles from previous works in our phylogenetic analysis allowed us to define the nodC alleles of the studied strains (Fig. 5). Interestingly, isolates belonging to sv. phaseoli were clustered into three subgroups coincident with the previously described alleles α, γ-a, and γ-b 74 .
The α allele was found in the Greek strains closely related to R. hidalgonense and Rhizobium sp. I. The α allele is considered to have originated in America and was distributed to Europe and other continents with bean seeds 39,48,64,83 . The strain PVMT26, assigned as R. hidalgonense, carried the α nodC allele, which was identical to that of the type strains R. hidalgonense (Mexico), R. etli (Mexico), and R. phaseoli (USA), and displayed 99.8% identity to the putative new lineages PVIM10, PVMT25, and PVKA6 (Fig. 3). The α allele has also been found in strains of the undescribed species Rhizobium sp. I (M1, M10, H4, 1648, 1652, NAK 299, 26 T), Rhizobium sp. II (N541), Rhizobium sp. IX (FA23), R. esperanzae (TAL182), R. phaseoli (NAK 299, Ch24-10) and Rhizobium sp. RPVR04 and HBR42 (Fig. 5). For simplification, not all strains carrying the α allele were included in the nodC phylogenetic tree. The identities of α nodC alleles found in various strains isolated from various countries ranged between 99.2 and 100%. In European soils, the α allele has been found in strains affiliated to R. hidalgonense in Croatia 73 , R. etli in Spain 46 , and R. leguminosarum in Poland 84 .
The strain PVPR1 assigned to R. anhuiense harbored the γ-a nodC allele, which was identical to those of the type strains R. vallis, and R. ecuadorense isolated from bean nodules in China and Ecuador, respectively 8,23 . The γ-a allele was also harbored by the type strains of R. acidisoli (Mexico), R. esperanze (Mexico), and R. sophorae (China) sharing 99-99.5% identity with that of PVPR1. The γ-a allele is also present in strains belonging to other species, such as R. etli, R. leguminosarum, R. lusitanum, R. phaseoli, and R. sophoriradicis with identity values among strains ranging from 97.2 to 100% (Fig. 5). Therefore, this allele was not only found in strains isolated from P. vulgaris nodules in various countries from all continents but also was the most prevalent within the rhizobial species nodulating common bean. In European soils, the γ-a nodC allele is the most frequent among bean-nodulating rhizobia regardless of the species to which they belong 12,18,39,[46][47][48]73,82,85 . Considering that the sv. phaseoli evolved with common beans in America 39,86 and probably disseminated worldwide along with bean seeds 2,87 , it is possible that native rhizobia in various countries have acquired symbiotic genes typical of sv. phaseoli through horizontal gene transfer in the rhizosphere or within nodules 88,89 .
The Greek strains identified as R. sophoriradicis and represented by PVTN21 harbored the γ-b allele, which is present in the type strains of R. aethiopicum and R. sophoriradicis (Fig. 5). Noteworthy, all γ-b nodC alleles found in various strains were identical (100%) and were found in Asia (China, Iran), Africa (Ethiopia, Kenya, Morocco), and America (USA, Mexico) 5,63,65,70,74,90,91 . Most strains carrying this allele were closely related to R. sophoriradicis (Kim5, IE4803, RHM67, RHM19, NAK368, NAK378, NAK387, L1, S1, G1, B1, 1706, 1587, 1617, and 1532), except for strain L101 that carried the γ-a allele and the strain IE4771 harbored a nodC gene similar to the sv. gallicum. Moreover, this allele is also present in R. anhuiense strains, such as JX3 Y27, S10, C15, J3 from China 6,70 , in Rhizobium sp. I (e.g. Rhizobium sp. G2) from Iran 5 and in Rhizobium sp. strains Mar-10 and HBR22 from Nepal and Ethiopia, respectively 90,92 . Therefore, this allele seems to be restricted to a few rhizobial species with prevalence in R. sophoriradicis. To the best of our knowledge, this is the first time that the γ-b allele was found in European soils and within isolates assigned to R. sophoriradicis.
Finally, strains identified as Rhizobium azibense and represented by PVIM1 harbored nodC genes identical (100%) to sv. gallicum, which is present in R. azibense 23C2 T , and R. gallicum R602sp T isolated from bean nodules in Tunisia and France, respectively 7,12,36 . However, the R. azibense strains 8C-3, and GR42, isolated from bean nodules in Spain belong to sv. phaseoli harboring the γ-a allele 7,44,45,93 as shown in Fig. 5. Strains belonging to sv. gallicum have also been isolated from common bean in Austria 51 96 , it remains to be investigated whether the European and African strains nodulating common bean possess a better symbiotic efficiency since they carry more divergent nodC genes. Noteworthy, the sv. gallicum has also been reported to effectively nodulate legumes belonging to the genera Leucaena, Macroptilium, Onobrychis, Sesbania, Caliandra, Gliricidia, Leucaena, and Piptadenia 12,26,44,45,52,78,[97][98][99] . The nodC gene sequences of our isolates were also identical to those found in sv. gallicum strains isolated from nodules of other legumes, such as the strains Rhizobium sp. AC91a from Calliandra calothyrsus in Ethiopia 78 , R. tarimense AS1-101a and SPT1 from Ammopiptanthus in China, and Rhizobium sp. UPRM 8060 from Piptadenia flava in Puerto Rico 100 . For simplification, not all strains from other legumes were included in the nodC phylogenetic tree. The wide distribution of sv. gallicum in different continents in combination with its broad host range and its presence in different rhizobial species makes it a promising multi-host inoculant.
Phylogenetic analysis based on partial nifH sequences (726 bp) grouped the isolates into two clades that corresponded to symbiovars phaseoli and gallicum (Fig. 4). The phaseoli clade consisted of two sub-clades with an identity 99.3%. One sub-clade included the isolates PVIM10, PVKA6, PVMT25, PVMT26, and PVPR1, which shared identical nifH sequences to those of R. hidalgonense FH14 T , R. phaseoli ATCC 14482 T , R. etli CFN42 T , R.  Supplementary Table S2. Type strains are indicated by superscript "T" and GenBank accession numbers of their sequences are indicated within parentheses. Bootstrap values (greater than 50%) were calculated for 500 replications and are shown at the nodes. The scale bar shows the number of nucleotide substitutions per site. Phylogenetic analysis was conducted in MEGA 6 104 (https:// www. megas oftwa re. net/) using the maximum likelihood algorithm with the Tamura 3-parameter model plus invariant site (T92 + I). The genus names are abbreviated as follows: R., Rhizobium. www.nature.com/scientificreports/ ecuadorense CNPSO 671 T , and R. vallis CCBAU 65647 T . Strain PVTN21 was separately clustered along with R. sophoriradicis CCBAU 03470 T displaying identical nifH sequences. Strain PVIM1 had an identical nifH sequence to that of R. azibense 23C2 T and formed a clade that corresponded to symbiovar gallicum. Overall, the phylogenetic analysis of nifH was congruent with that of nodC phylogeny.

Conclusions
In summary, the present study provides the first analysis on the phylogenetic diversity of indigenous rhizobia nodulating P. vulgaris in Greece by identifying them at the species and symbiovar level. Strains were affiliated to R. anhuiense, R. azibense, R. hidalgonense, R. sophoriradicis, and to a putative new genospecies consisting of various strains all over the world and provisionally named as Rhizobium sp. I 6 . Most strains belonged to symbiovar phaseoli carrying the α-, γ-a and γ-b alleles of nodC gene, while few of them belonged to symbiovar gallicum. To the best of our knowledge, it is the first time that strains assigned to R. sophoriradicis and harbored the γ-b allele were found in European soils. All strains formed effective symbioses with bean plants, suggesting that they are true symbionts of common bean. The analysis of the symbiovar phaseoli nodC alleles is congruent with previous findings in other European countries suggesting the American origin of sv. phaseoli. Moreover, the presence of nodC alleles in diverse rhizobial strains regardless of the species to which they belong raises the possibility that local rhizobia have acquired symbiosis genes via lateral gene transfer in the rhizosphere or within nodules. However, the Rhizobium azibense isolates were closely related and grouped together with African strains in both MLSA and nodC phylogenies suggesting their common evolutionary histories. Consequently, the current study increases the knowledge of the diversity, geographic distribution, and evolution of common bean-nodulating rhizobia in European soils and further provides a natural resource for the selection of highly efficient rhizobia that are more competitive and adapted to the local conditions.

Μethods
Nodule and soil sampling. Nodules were collected from local common bean varieties grown in five different geographical regions of Greece, namely as Imathia, Metsovo, Preveza, Tinos, and Karpathos ( Supplementary  Fig. S1). The sampling sites were located in fields with no history of rhizobial inoculation. The soil samples were slightly acidic to alkaline, with pH range 6.9 to 8.1.

Isolation and purification of nodules and rhizobial strains.
Four nodules per plant were randomly selected from four plants of each region and at least three isolates were retained from each nodule. A great number of isolates were non-nodulating bacterial strains which were probably nodule endophytes or contaminants and they were not analyzed further. Finally, a total of 50 rhizobial strains were isolated in pure culture. Standard routine laboratory techniques were applied for the isolation of strains from the nodules 101 . Briefly, the nodules were surface disinfected by immersion in 70% ethanol for 60 s and then in 3-5% (v/v) solution of sodium hypochlorite for 2-4 min and were washed six times with sterile ddH 2 O. To check the absence of surface contamination, sterilized nodules were rolled over yeast-mannitol agar (YMA) plates 101 and aliquots of water from the last washing step were also spread on YMA plates and incubated at 28 °C for 2-5 days. Sterilized nodules were crushed in a drop of sterile distilled water and the nodule juice was streaked onto YMA plates and incubated under the same conditions as the control plates. Only nodules without any contaminants were considered for the isolation of rhizobial strains. Single colonies were subsequently purified by repeated streaking on YMA medium supplemented with Congo red until pure cultures of the isolates were obtained. Cultures of pure isolates were maintained in 20% glycerol-YMA broth at − 80 °C.
Nodulation tests. The nodulation capability of each isolate was tested by inoculating seedlings of its original host grown in a greenhouse. Seeds were surface sterilised in 3% sodium hypochlorite for 10 min and rinsed six times. Surface-sterilized seeds were germinated on moist sterile filter paper in the dark at 22 °C for 3-4 days and then transferred to 250 ml pots containing vermiculite and watered with 0.5Χ Hoagland nutrient solution without nitrogen 102 . Each seedling was inoculated with 1 ml of rhizobial suspension (∼10 9 cells ml −1 ). Three replicates were performed per isolate and plants were grown in greenhouse. Unfertilized and uninoculated seedlings were included as negative controls and uninoculated, nitrogen fertilized (5 mM KNO 3 ) seedlings were used as positive controls. Six weeks after inoculation, one nodule per plant was excised and rhizobia were reisolated as described above and their identity was confirmed by BOX-PCR fingerprinting. Nodulation capacity was recorded as positive (Nod+) when nodules were present and negative (Nod−) if were absent. Nitrogen fixation was considered effective when nodules were pink (Fix+) and ineffective if nodules were white (Fix−). PCR amplification and sequencing. The DNA fragments of 16S rRNA, recA (DNA recombination protein), atpD (ATP synthase subunit beta), gyrB (DNA gyrase B) and glnII (glutamine synthetase II) were amplified by PCR, using the primer pairs described in Supplementary Table S1. PCR amplification and sequencing were carried out as previously described 69 . Primers taken from the literature or designed in the present study were slightly modified in such a way to include at their 5′ ends either T7 or SP6 primer sequence to facilitate direct sequencing of the amplicons. Each PCR mixture contained the following: approximately 50 ng genomic DNA, 20 pmol each primer, 200 µM dNTPs (Invitrogen), Phusion High Fidelity DNA polymerase (Thermo Fisher Scientific), and the respective 10X polymerase buffer in a final reaction volume of 50 µl. The PCR conditions for the amplification of each gene fragment are described in Supplementary Table S1. PCR products from the aforementioned genes were purified using the PureLink™ Quick Gel Extraction kit (Thermo Fisher Scientific). Purified DNA fragments were directly sequenced on both strands using the standard primers attached in the corresponding primer sequences. All PCR products were commercially sequenced by CEMIA (cemia.eu), Greece.

DNA isolation and BOX-PCR fingerprinting.
Phylogenetic analyses. The sequences of rrs genes were compared with those of bacterial type strains using the EzTaxon-e server (http:// eztax on-e. ezbio cloud. net). BLAST searches were done at the National Center for Biotechnology Information (NCBI) server using BLASTN (http:// www. ncbi. nlm. nih. gov/ blast). Sequences from closely related type strains, as listed on the List of Prokaryotic Names with Standing in Nomenclature (LPSN) (www. bacte rio. net), and reference strains were retrieved for phylogenetic analyses from the GenBank database (http:// www. ebi. ac. uk/ Tools/ sss/ fasta/ nucle otide. html). For pairwise distance matrixes, the multiple sequence alignments were performed using the algorithm CLUSTAL Omega (https:// www. ebi. ac. uk/ Tools/ msa/ clust alo/) provided by the European Bioinformatics Institute (EMBL-EBI). For phylogenetic analyses, the partial gene sequences obtained in this study, together with sequences retrieved from GenBank were aligned using the www.nature.com/scientificreports/ CLUSTALW software in the MEGA 6.0 software package 104 . Phylogenetic trees were constructed using either the neighbor-joining (NJ) or Maximum likelihood (ML) methods in MEGA 6.0 software package. The gene sequences were appropriately trimmed and were concatenated. The best-fit models of nucleotide substitution were determined in MEGA 6 and the most appropriate were selected for the construction of ML trees as referred in the figure legends.
Nucleotide sequence accession numbers. All sequences from common bean isolates were deposited in the GenBank database and the accession numbers are listed in Supplementary Table S2.
Ethics approval. This article does not contain any studies with human participants and/or animals performed by any of the authors. The formal consent is not required in this study.

Data availability
Sequence data that support the findings of this study have been deposited in GenBank (https:// www. ncbi. nlm. nih. gov/ genba nk/) with the accession codes: MT476928-MT476934 and MT503467-MT503508. Sequence data MT503467-MT503508 will be publicly available upon article publication but are available from the corresponding author on reasonable request.