Introduction

Burkholderia is a genus comprised by nearly 70 species with worldwide distribution in diverse environments including soil, water, plant, fungi and insects. Although many species are saprophytes, others are human, animal, or plant pathogens and are or have the potential to be opportunistic pathogens of humans. Recently, a number of plant-associated species were described and more are in the pipeline awaiting description (Gyaneshwar et al. 2011; Howieson et al. 2013). Thus in the last few years, Burkholderia is steadily growing as a genus with regard to species number. However, as the number of species described increases, it is becoming more obvious that the genus contains two or more lineages (Gyaneshwar et al. 2011; Suarez-Moreno et al. 2012l; Estrada-de los Santos et al. 2013). One distinct lineage contains: (1) members of the B. cepacia complex (BCC), which are opportunistic pathogens of humans; (2) the human and animal pathogens of the B. pseudomallei group; (3) the phytopathogenic species; and (4) some saprophytic species. The second lineage, which is divided into two sublineages, contains saprophytic and plant-associated species, including several of which nodulate legumes (Vandamme et al. 2002; Chen et al. 2003, 2006, 2007, 2008). B. andropogonis and B. rhizoxinica, which do not fit into either of the large lineages, may represent new genera.

The genus also includes species with the ability to carry out PGP (Plant Growth Promoting) activities (Balandreau and Mavingui 2007). These species have the potential for agrobiotechnological applications due to such traits as siderophores synthesis, phosphate solubilization, or nitrogen fixation ability (Caballero-Mellado et al. 2007; Angus et al. 2013). Nitrogen fixation was first discovered in Burkholderia when B. vietnamiensis was described (Gillis et al. 1995). Afterwards, this ability was detected in other newly described species, including B. unamae, B. tropica, B. xenovorans and B. silvatlantica (Estrada-de los Santos et al. 2001; Caballero-Mellado et al. 2004; Reis et al. 2004; Goris et al. 2004; Perin et al. 2006b). In addition, Burkholderia contains 11 species that nodulated legumes effectively; B. tuberum, B. phymatum, B. caribensis, B. mimosarum, B. nodosa, B. sabiae, B. diazotrophica, B. symbiotica, B. phenoliruptrix, B. sprentiae and B. rhynchosiae (Vandamme et al. 2002; Chen et al. 2003, 2006, 2007, 2008; Sheu et al. 2012, 2013; de Oliveira-Cunha et al. 2012; De Meyer et al. 2013a, b). B. fungorum is also able to nodulate legume plants although ineffectively (Ferreira et al. 2012).

Many of the saprophytic Burkholderia species degrade xenobiotic compounds (Caballero-Mellado et al. 2007), making them likely bioremediation agents. The different Burkholderia species and the substrates they utilize have been extensibely reviewed by Denef (2007). Burkholderia species with PGP activities and biodegradation abilities could play a beneficial role in agriculture, but concerns have been expressed about the use of Burkholderia species for agriculture because some species are opportunistic pathogens, especially species in the BCC (Hauser et al. 2011). Yet, this may not be a serious problem in the future because evidence is accruing that indicates that the plant-associated Burkholderia are not closely related to the BCC, making the plant-associated species useful for both agriculture and bioremediation.

During a survey of Burkholderia species with potential for agrobiotechnological applications, several strains were isolated from the tomato rhizosphere and rhizoplane (Caballero-Mellado et al. 2007). Five strains were closely related to the trichloroethylene (TCE) degrading and N-fixing bacteria B. kururiensis (Zhang et al. 2000; Estrada-de los Santos et al. 2001). This group was named the Bkr group. A taxonomic analysis of the Bkr strains together with a second group of seven strains isolated from the same area showed that all 12 strains actually belong to a new Burkholderia species, for which the name Burkholderia caballeronis sp. nov. is proposed.

Materials and methods

Bacterial isolation

Bacteria were isolated from Saladet variety tomato (Lycopersicon esculentum) plants, collected in Nepantla, State of Mexico, Mexico (Table 1). All strains were isolated at the same time, but only five were previously described (Caballero-Mellado et al. 2007). The remaining strains were kept in our in-house collection until this description.

Table 1 Burkholderia caballeronis strains isolated from Saladet variety tomato plants cultivated in Nepantla, State of Mexico
Full size table

Genotypic characterization

The 12 isolates were grouped using Amplified rRNA Restriction Analysis (ARDRA) (Caballero-Mellado et al. 2007). The nearly full 16S rRNA gene sequences (~1,533 nucleotide positions) from the strains were amplified with the universal primers fD1/rD1 (Weisburg et al. 1991), as previously described (Perin et al. 2006a). The 16S rRNA genes were cloned into the pCR2.1 vector according to the manufacturer’s instructions (Invitrogen). The sequences of the 16S rRNA genes were obtained from Macrogen (www.macrogen.com) using universal primers. The sequences accession numbers are include in Fig. 1. A multiple alignment was performed with Muscle 3.57 (Edgar 2004) and a phylogenetic analysis was carried out with maximum likelihood (ML) using PhyML program (Guindon and Gascuel 2003). Among-site, rate variation was modeled by a gamma distribution with four rate categories (Yang 1996), with each category being represented by its mean, under the GTR+G model. Tree searches were initiated from a BioNJ seed tree, retaining the best tree among those found with NNI (Nearest Neighbor Interchange). The robustness of the ML topologies was evaluated using a Shimodaira-Hasegawa (SH)-like test (Anisimova and Gascuel 2006). ML trees were visualized with program MEGA version 5 (Tamura et al. 2011). Genomic DNA from strains TNe-841T, TNe-878, TNe-834, the most closely related type species, B. cepacia and other randomly selected Burkholderia species was extracted from liquid cultures grown in BSE for the DNA-DNA hybridization (DDH) experiments. These experiments were performed in duplicate as described previously (Estrada-de los Santos et al. 2001). The DNA G+C content determination of strain TNe-841T was carried out at BCCM/LMG. The strain was cultivated on LMG medium 203. Genomic DNA was extracted according to a modification of the procedure of Wilson (Wilson 1987). The DNA G + C content was determined using the HPLC technique (Mesbah et al. 1989). The given value is the mean of three independent analyses. The novel strains were analyzed using fingerprinting techniques, BOX-PCR (BOX dispersed-repeated motif) (Versalovic et al. 1994). The PCR conditions for the BOX element (BOXA1) were carried out according to a previous description (Estrada-de los Santos et al. 2011). An internal fragment of 320 pb from nodC gene was amplified using the primers nodC-F (5′-GAYATGGARTAYTGGCTNGC-3′) and nodC-R (5′-ANGTRCTBCGBGCCCAVC-3′). PCR conditions were carried out with AccuPrime Taq DNA Polymerase HF (Invitrogen), with an initial denaturing step at 95 °C for 2 min, followed by 35 cycles of 95 °C for 30 s, 56 °C for 30 s and 68 °C for 30 s; finishing with a final elongation step at 68 °C for 5 min.

Fig. 1
figure 1

Maximum likelihood tree based on the 16S rRNA gene sequences of selected Burkholderia species. The accession numbers of the sequences used in this analysis are shown in parenthesis. The robustness of the ML topologies was evaluated using a Shimodaira-Hasegawa (SH)-like test (Anisimova and Gascuel 2006), which is shown at branches nodes. The bar represents the number of expected substitutions per site under the GTR + G model

Full size image

The PCR reaction mix was made in 40 μl, with the following proportions : genomic DNA, 1 μl (50 ng); 4 μl Buffer 10X, 2.5 μl each primer (25 μM); water, 30 μl. The amplified fragments were sequenced at Instituto de Biotecnología-UNAM (http://www.ibt.unam.mx/) and deposited to the GenBank, NCBI (National Center for Biotechnology Information).

Phenotypic and biochemical characterization

Phenotypic features for strains TNe-841T, TNe-878, TNe-834, B. kururiensis KP23T, B. mimosarum PAS44T, B. sacchari IPT101T, B. ferrariae FeGI01T and B. cepacia ATCC 25416T were determined with the API 20NE systems according to the manufacturer’s instructions (bioMérieux). For the analysis, bacterial cells were pre-cultured on BSE medium. Additionally, the novel isolates were grown on (1) BSE agar medium at 15, 29 and 37 °C and BSE liquid medium at 42, 46 and 50 °C; (2) MacConkey agar medium at 29, 37 and 42 °C; and (3) LB in the presence of 0.5, 1.0, 1.5, 3.0. 5.0, 7.5, 10, 15 and 20 % NaCl at 29 °C. The similarity among the 12 novel strains was tested by comparison of the whole-cell protein patterns. The bacterial cultures were grown in BSE medium with reciprocal shaking (200 rpm) for 15 h at 29 °C, and 1.0 ml samples were harvested by centrifugation at 12,300×g for 10 min. The pellets were resuspended in 70 μl of 0.125 M Tris–HCl, 4 % SDS, 20 % glycerol, and 10 % mercaptoethanol at pH 6.8. Aliquots of 10 μl were used for SDS-PAGE performed as described by Laemmli (1970).

The fatty acid profiles from strains TNe-841T and TNe-878 was carried out at BCCM/LMG. In short, the cells were grown for 24 h at 28 °C on LMG medium 203 (10 g mannitol, 0.5 g KH2PO4, 5 g sodium glutamate, 50 mg NaCl, 10 ml solution A, 1 ml solution B, 1 ml solution C, 1 g yeast extract, and 20 g agar, per 1 L distillated water, pH 6.8. Solution A consists of 1 % MgSO4·7H2O, solution B is composed of 5.28 % CaCl2·2H2O, and solution C is 0.666 % FeCl3.6H2O). Inoculation and harvesting of the cells, and extraction and analysis of the fatty acids were performed according to the recommendations of the commercial identification system MIDI (Microbial Identification Systm, Inc., DE, USA). The whole-cell fatty acid composition was determined by gas chromatography. For nodulation tests, Phaseolus vulgaris cultivar Negro Jamapa seeds were surface sterilized with 25 % bleach for 25 min and rinsed with sterile water. The seeds were placed on LB agar plates during 3 days in the dark at 30 °C for germination and sterility before planting. The germinated seeds were transferred to 250 ml aluminumfoil-wrapped Erlenmeyer flasks containing 200 mL of Fahraeus solution (Fahraeus 1957) solidified with 0.8 % agar. The germinated seeds were inoculated with 1 mL of a bacterial suspension adjusted to 0.2 OD600 and kept in a plant growth chamber at 26 °C with a photoperiod of 12 h light/12 h dark. After 3 weeks, nodules were tested for acetylene reduction (Mascarua-Esparza et al. 1988). Nodules were surface sterilized with 25 % bleach for 2 min and rinsed with sterile water. The nodules were macerated and streaked on BAc media (Estrada-de los Santos et al. 2001). The isolated bacteria were compared to the original strain by ARDRA. The nodulation experiments were performed in triplicate and repeated in two independent experiments.

Results

Novel species characterization and identification

The percentage of 16S rRNA similarity among the novel strains was 99.8 % and the Burkholderia species most closely related to the TNe-841T strain were B. kururiensis KP23T (97.4 %, AB024310) and B. mimosarum PAS44T (97.1 %, AY752958). Based on an analysis of the 16S rRNA gene sequence in the EzBioCloud server (http://www.ezbiocloud.net/, Kim et al. 2012). Other closest Burkholderia species, such as B. sacchari, B. ferrariae, B. acidipaludis or B. cepacia, were less than 97.0 % similar to strain TNe-841T.

The relationship of the novel strains was next determined in a ML tree based on a set of Burkholderia species, which not only included the closest species, but also B. cepacia (Fig. 1). An additional ML tree was constructed with all the so-far known Burkholderia species (Fig. S1). This analysis revealed that the novel strains were integrated into the plant-associated and saprophytic Burkholderia lineage we described earlier (Estrada-de los Santos et al. 2013). DDH values between strain TNe-841T and the other novel strains were higher than the cut-off value used to define a bacterial species (70 %); for example, 90.6 % to TNe-878 and 88.8 % to TNe-834; indicating that these three strains are related at the species level (Tindall et al. 2011). Low DDH values were found relative to other Burkholderia species; for example, 12.7 % to B. kururiensis KP23T, 40.5 % to B. mimosarum PAS44T, 13.2 % to B. sacchari IPT101T, 10.1 % to B. ferrariae FeGI01T, 13.8 % to B. unamae MTl-641T and 15.1 % to B. cepacia ATCC 25416T. These results confirm that the novel group of Burkholderia species isolated from the tomato rhizosphere and rhizoplane belong to a distinct species.

The DNA G+C content for strain TNe-841T was 66.0 %. Previously, we showed that the DNA G+C content of certain saprophytic and plant-associated Burkholderia species is lower than that of the the BCC and other pathogenic species (Gyaneshwar et al. 2011; Estrada-de los Santos et al. 2013). However, despite the fact that these novel strains are included within the clade of plant-associated bacteria, according to the 16S rRNA gene analysis, the DNA G+C content is higher and more similar to that found in the members of the BCC and the plant and human pathogens Burkholderia species. These results strongly suggest that the genus Burkholderia deserves a more extensive as well as deeper taxonomical analysis to understand the position of the different lineages in the genus.

The intragenic diversity from the novel species was analyzed with BOX fingerprintings, and compared to related Burkholderia species. Although some have indicated that DNA fingerprinting methods have limited value for species description (Tindall et al. 2011), we found in this study that the novel species were very homogeneous in their fingerprinting pattern and that it is significantly different from the patterns of related species (Fig. S2).

The novel strains were also phenotypically characterized. The protein patterns among the novel strains were found to be very similar, but were different from B. kururiensis and B. mimosarum, the closest related species (Fig. S3). Identical or very similar protein patterns in a group of strains usually correlate with high levels of genome similarity, and when the protein patterns differ, the similarity diminishes (Vandamme et al. 1996). Hence, this result provides additional evidence for the placement of these novel strains into a new Burkholderia species.

The biochemical features from the novel strains were also analyzed. The data from strains B. unamae MTl-641T and B. mimosarum was collected from the original description (Caballero-Mellado et al. 2004; Chen et al. 2006), which was performed with the same methodology herein. Additionally, the ability of strains TNe-841T, TNe-878 and TNe-834 to assimilate different carbon sources was tested with the API 50CH system. The following activities were negative for the novel strains: nitrate and nitrite reduction, β-galactosidase, indole production, esculine and gelatin hydrolisis, d-glucose fermentation and assimilation of d-maltose, citrate, erythritol, methyl-βd-xylopyranoside, dulcitol, methyl-αd-glucopyranoside, amydgaline, arbutine, salicine, d-lactose, inuline, d-melazitose, d-raffinose, amidon, glycogene and xylitol. The following activities were present: arginine dihydrolase, urease, catalase, nitrogen fixation and assimilation of d-glucose, dl-arabinose, d-mannose, d-mannitol, N-acetyl glucosamine, gluconate, capric acid, malate, acetate, d-ribose, d-xylose, d-adonitol, d-galactose, d-fructose, l-ramnose, inositol, d-sorbitol, d-cellobiose, d-turanose, d-xylose, d-fucose, d-arabitol, potassium 2-ketogluconate, and postassium 5-ketogluconate. Oxidase activity was weak for strain TNe-841T and TNe-878, and negative for strain TNe-834. The assimilation of different carbon sources was strain-dependant (adipate, glycerol, l-xylose, methyl-α-d-mannopyranoside, esculine, d-melibiose, d-threalose, gentibiose, d-tagatose and l-arabitol). The twelve novel strains were able to grow on MacConkey agar plates at 29 and 37 °C, but at 42 °C the growth was strain-dependant. All strains grew on LB and BSE agar plates at 15, 29, 37 and 42 °C, and on LB plates at 29 °C with up to 5.0 % NaCl. The distinctive phenotypic features between the novel strains and the closest Burkholderia species are displayed in Table 2.

Table 2 Phenotypic differences between B. caballeronis and related Burkholderia type species
Full size table

The following fatty acids were detected in strain TNe-841T and TNe-878, respectively: C14:0 (4.46, 4.66), C16:0 (21.77, 20.14), C16:0 2OH (2.3, 2.5), C16:0 3OH (6.2, 8.0), C16:1 2OH (3.81, 4.5), C17:0 cyclo (12.43, 10.88), C18:1 2OH (1.5, 1.4), C18:1 w7c (16.62, 16.79), C19:0 cyclo w8c (14.89, 11.18), summed feature 2 (5.9, 8.8) and summed feature 3 (8.3, 10.5). Summed feature two corresponds to C14:0 3OH and/or 16:1 ISO I, an unidentified fatty acid with equivalent chain length value of 10.928, 12:0 ALDE or any combination of these fatty acids. Summed feature three corresponds to C16:1 w7c and/or C15:0 ISO 2OH. The large amount of C16:0, C17:0 cyclo and C18:1 w7c is typical for Burkholderia species (data not shown).

Because the group of twelve novel Burkholderia strains was related to B. kururiensis, which can degrade TCE, the novel strains were tested for their ability to grow in the presence of different xenobiotic compounds, including TCE. However, none of the strains grew in liquid mineral media (0.04 % K2HPO4, 0.04 % KH2PO4, 0.02 % MgSO4-7H2O, 0.5 % NH4Cl, pH 6.5) supplemented with 4.23 mM of nitrobenzene, nitrotoluene, TCE, cumene, ethylbenzene, o-xylene, toluene, m-xylene or phenol after 15 days of incubation at 29 °C. Approximately 50 % of the strains grew, albeit poorly, with benzene as a carbon source. By contrast, B. kururiensis KP23T grew in the presence of phenol, m-xylene, o-xylene, toluene and ethylbencene.

Given the similarity of the novel species to B. mimosarum, strains TNe-841T and TNe-834 were tested in nodulation experiments on Phaseolus vulgaris. B. mimosarum PAS44T, B. nodosa Br3437T and Rhizobium etli CFN42T were included as positive controls and an uninoculated plant served as a negative control. Surprisingly, the two strains and the positive controls nodulated bean roots (Fig. S4), and, the nodules were effectively fixing nitrogen. The inoculated strains corresponded to the original strain (analyzed by ARDRA, data not shown). This result fulfilled the Koch’s postulates. Up to now, all nodulating Burkholderia species have been isolated from legume nodules, but this novel species was isolated from tomato rhizosphere and rhizoplane. This finding opens up the possibility that other Burkholderia species, not recognized as nodulating strains, might harbor the machinery needed to nodulate legume plants. Accordingly, the twelve novel strains were analyzed for the amplification of a 320 pb nodC gene fragment using the primers nodC-F/nodC-R. All novel strains were positive for the amplification of nodC gene fragment. To corroborate the nodC identity, the DNA fragments were sequenced directly from PCR at the Instituto de Biotecnología-UNAM. The analysis of the sequences from strains TNe-841 and TNe-834 revelead the nodC gene, 99 % identical to nodC of B. mimosarum. The sequences were deposited at the GenBank with the following accession numbers, KF484909 for TNe-841 and KF484910 for TNe-834.

Taking together, the phenotypic and genotypic analyses of the group of 12 Burkholderia strains presented in this study, it is proposed that they should be considered a new species, for which we have proposed the name Burkholderia caballeronis sp. nov.

Description of Burkholderia caballeronis sp. nov

Burkholderia caballeronis: ca.bal.le’ro.nis. N.L. gen. masc. n. caballeronis, of Caballero, named after the Mexican microbiologist Jesús Caballero Mellado who contributed significantly to the knowledge on plant-associated Burkholderia species.

Cells are Gram negative and aerobic. Growth occurs in the range of 15–42 °C and in the presence of up to 5 % NaCl. Oxidase activity and assimilation of adipate is strain dependant. The strains are unable to reduce nitrate and nitrite, produce indol, hydrolyze esculine and gelatine, ferment d-glucose and assimilate d-maltose, citrate, erythritol, methyl-βd-xylopyranoside, dulcitol, methyl-αd-glucopyranoside, amydgaline, arbutine, salicine, d-lactose, inuline, d-melazitose, d-raffinose, amidon, glycogene and xylitol. β-galactosidase activity is negative. The following activities are present: arginine dihydrolase, urease, catalase, nitrogen fixation and assimilation of d-glucose, dl-arabinose, d-mannose, d-mannitol, N-acetyl glucosamine, gluconate, capric acid, malate, acetate, d-ribose, d-xylose, d-adonitol, d-galactose, d-fructose, l-ramnose, inositol, d-sorbitol, d-celobiose, d-turanose, d-lyxose, d-fucose, d-arabitol, potassium 2-ketogluconate and postassium 5-ketogluconate. The assimilation of adipate, glycerol, L-xylose, methyl-αd-mannopyranoside, esculine, d-melibiose, d-threalose, gentibiose, d-tagatose and l-arabitol is strain dependant. The major fatty acids are C16:0, C18:1 w7c, C19:0 cyclo w8c and C17:0 cyclo. The novel species is able to effectively nodulate Phaseolus vulgaris plants.

The strains TNe-841T and TNe-878 have been deposited at BCCM/LMG and CIP with the following identifiers: LMG 26416T = CIP 110324T and LMG 26417 = CIP 110325, respectively.

The type strain is TNe-841 (=LMG 26416T = CIP 110324T) and was isolated from the rhizosphere of tomato plants growing in Nepantla, State of Mexico, Mexico. Phenotypic features are the same as described above for the species. Additionally, the oxidase activity is weak and the assimilation of glycerol, methyl-αd-mannopyranoside, esculine, d-melibiose, d-threalose, gentibiose, l-arabitol and dipate is negative. The assimilation of l-xylose and d-tagatose are positive. Cells are coccoïd rods (0.9 × 1.3–2.0 μm), single or in pairs, non motile. The DNA G+C content of strain TNe-841T is 66.0 %.