|Home | About | Journals | Submit | Contact Us | Français|
The genus Burkholderia includes pathogens of plants and animals and some human opportunistic pathogens, such as the Burkholderia cepacia complex (Bcc), but most species are nonpathogenic, plant associated, and rhizospheric or endophytic. Since rhizobacteria expressing ACC (1-aminocyclopropane-1-carboxylate) deaminase may enhance plant growth by lowering plant ethylene levels, in this work we investigated the presence of ACC deaminase activity and the acdS gene in 45 strains, most of which are plant associated, representing 20 well-known Burkholderia species. The results demonstrated that ACC deaminase activity is a widespread feature in the genus Burkholderia, since 18 species exhibited ACC deaminase activities in the range from 2 to 15 μmol of α-ketobutyrate/h/mg protein, which suggests that these species may be able to modulate ethylene levels and enhance plant growth. In these 18 Burkholderia species the acdS gene sequences were highly conserved (76 to 99% identity). Phylogenetic analysis of acdS gene sequences in Burkholderia showed tight clustering of the Bcc species, which were clearly distinct from diazotrophic plant-associated Burkholderia species. In addition, an acdS knockout mutant of the N2-fixing bacterium Burkholderia unamae MTl-641T and a transcriptional acdSp-gusA fusion constructed in this strain showed that ACC deaminase could play an important role in promotion of the growth of tomato plants. The widespread ACC deaminase activity in Burkholderia species and the common association of these species with plants suggest that this genus could be a major contributor to plant growth under natural conditions.
Burkholderia is a common genus in the bacterial communities present in agricultural and polluted soils (24, 56, 63) and includes over 40 properly described species (18). While some of these species are opportunistic pathogens of humans (for example, species in the Burkholderia cepacia complex [Bcc]) and others are phytopathogenic (17), most species have neutral or beneficial interactions with plants (18). For a long time, the ability of bacteria belonging to the genus Burkholderia to fix N2 was found only in the species B. vietnamiensis (36), a member of the Bcc (17). At present, several diazotrophic plant-associated Burkholderia species have been validly described, including B. unamae (6), B. xenovorans (41), B. tropica (62), and B. silvatlantica (60), all of which are able to colonize the rhizosphere and/or the endophytic environment of a wide range of host plants (6, 7, 36, 59, 60, 62). In addition, legume-nodulating N2-fixing strains have formally been classified as novel species, including B. phymatum, B. tuberum (68), B. mimosarum, and B. nodosa (12, 13). It is worth noting that most of the plant-associated Burkholderia species are phylogenetically distant from the Bcc species and exhibit potential activities of interest in agrobiotechnology (7). Among the plant-associated Burkholderia, B. unamae has relevant features, such as colonization of the rhizosphere and internal tissues of taxonomically unrelated host plants, including maize, coffee, sugarcane, and tomato, has a wide distribution in different geographical regions, and exhibits several potential activities involved in plant growth promotion, bioremediation, or biological control (6, 7, 59).
The bacterial enzyme ACC (1-aminocyclopropane-1-carboxylate) deaminase promotes plant growth by lowering plant ethylene levels (37, 39). This enzyme catalyzes the conversion of ACC, the immediate precursor of ethylene synthesis in plants, to ammonia and α-ketobutyrate (α-KB). ACC is exuded from seeds or plant roots and then metabolized by bacteria expressing ACC deaminase activity, which stimulates plant ACC efflux, decreasing the root ACC concentration and root ethylene evolution and increasing root growth (39). Moreover, it has been reported that some ACC deaminase-producing bacteria promote plant growth under a variety of stressful conditions, such as flooding (42), saline conditions (54), and drought (55). In addition, this enzyme has been implicated in enhancing nodulation in pea (51) and alfalfa (50) plants.
Growth on a minimal medium with ACC as the sole N source is indicative of ACC deaminase-containing bacteria (38). Based on this criterion, it has been postulated that soil bacteria capable of degrading ACC are relatively common (2, 38).
The acdS gene, encoding ACC deaminase, has been isolated from different species and strains of genera belonging to the Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria, as well as the Firmicutes and Actinobacteria (3, 25, 28, 44, 64). However, the identification of most acdS genes has been based only on partial sequences, and some of the strains analyzed have been poorly characterized taxonomically. Moreover, it is noteworthy that the beneficial effect of ACC deaminase on plant growth has usually been tested using wild-type isolates with ACC deaminase activity; only a few studies have used ACC deaminase-negative mutants (49, 51, 67).
Although the presence and activity of ACC deaminase in few nonpathogenic Burkholderia strains (65, 7) and in some Bcc strains (3) have been analyzed, nothing is known about the expression of this enzyme in association with plants. Similarly, the effect of ACC deaminase-negative mutants of plant-associated Burkholderia strains on plant growth is unknown.
In this work, the ACC deaminase activities and acdS gene sequences of most of the novel rhizospheric, endophytic diazotrophic, and legume-nodulating Burkholderia species, as well as of non-N2-fixing Burkholderia strains (mainly plant associated), were analyzed. In addition, we analyzed the effect of an acdS knockout mutant of B. unamae and the expression of the ACC deaminase gene by use of a gusA (uidA) reporter gene fusion on tomato plants grown under different conditions. The colonization patterns of a constitutive gusA-marked strain in tomato roots were examined as well.
Bacterial strains and plasmids used in this work are listed in Table Table11 and Table Table2,2, respectively. Since most of the Burkholderia species listed in Table Table11 were proposed on the basis of one to three isolates analyzed (59), in this work only the type strain of the majority of the species was examined. In addition to type strains, several previously characterized strains of B. unamae (6), B. xenovorans (41), B. tropica (62), B. silvatlantica (60), B. vietnamiensis (32, 36), and B. kururiensis (7) were included in the analysis (Table (Table1).1). For phenotypic identification of the acdS gene, Burkholderia isolates were grown on salts medium (SM) plates (7) supplemented with 3.0 mM ACC; the plates were incubated for 5 to 6 days at 29°C. Burkholderia strains able to grow on ACC as a sole nitrogen source were analyzed for the presence of the acdS gene. For triparental mating, B. unamae strains, Escherichia coli DH5α (donor), and E. coli M607 (with helper plasmid pRK600) were cultured in Luria-Bertani medium, and antibiotics were added as required at the following concentrations: kanamycin, 50 μg/ml; tetracycline, 15 μg/ml; gentamicin, 30 μg/ml; and chloramphenicol, 20 μg/ml. These experiments were performed as described previously (11). Growth kinetics for wild-type and derivative strains were determined at least three times. Strains were grown in liquid BSE medium or SM as described previously (7). Cells were collected and washed, and the initial optical density at 600 nm (OD600) was adjusted to 0.02. The cultures were incubated at 29°C with reciprocal shaking (250 rpm), and the growth over time was determined by quantifying the number of cells.
Cultures of Burkholderia strains used for ACC deaminase assays were grown as described previously (7). ACC deaminase activity was measured by measuring the production of α-KB as described by Honma and Shimomura (43). The protein concentration in cell extracts was determined by the Bradford method (4).
For PCR amplification of partial acdS genes, primers 5′ACC and 3′ACC were used, as described previously (7). To obtain the complete acdS gene sequences, the following two sets of degenerate primers were designed: primers F-acdS (5′ATGAAYCTSCARCGHTTY3′) and R-acdS (5′TYARCCGTYSCGRAARRT3′); and primers NF-acdS (5′ATGAAYCTSCARMRHTTYC3′) and NR-acdS (5′TYARCCGTYGCGRAARATV3′). PCR assays were performed using 50-μl reaction mixtures with PFX polymerase (Invitrogen) under the following conditions: initial denaturation for 5 min at 95°C, followed by 30 cycles of 30 s at 95°C, 30 s at 50°C, and 1 min at 72°C and then a final 5-min elongation at 72°C. The amplified products were cloned into the vector pCR 2.1 (Invitrogen), and the acdS gene sequences were determined at the Biotechnology Institute, UNAM (Mexico). Multiple-sequence alignments of acdS sequences were constructed with CLUSTALW2 software (48). The tree topology was inferred by the neighbor-joining method, and a distance matrix analysis was performed as described by Jukes and Cantor using the program MEGA, version 3.1 (47).
A 785-bp segment containing the central region of the acdS gene was amplified by PCR (using primers 5′ACC and 3′ACC) and cloned into the Kmr suicide vector pUX19 (71). The resulting plasmid (pUX19acdS) was transferred by triparental mating into B. unamae MTl-641T, and recombinant strains were selected in the presence of kanamycin. The total DNA of a single-recombinant strain was digested with restriction enzyme BclI (Invitrogen); the BclI restriction site was absent in the pUXacdS vector, which allowed incorporation of sequences upstream and downstream of the acdS gene into the vector. The digested DNA was religated and transformed into E. coli DH5α. Ten Kmr colonies were purified, and the clones were digested with BclI. One of these clones (pUX-UD) was sequenced by primer walking to verify that sequences surrounding the acdS gene were cloned. In silico analysis of the resulting sequence was performed with the programs FGENESB, BPROM (http://linux1.softberry.com/berry.phtml), and virtual footprint promoter analysis version 3.0 (http://www.prodoric.de/vfp/vfp_promoter.php).
A DNA fragment containing the structural and regulatory region of the acdS gene was amplified by PCR using primers Xba-404 (5′GCTCTAGACCAGGCCACACCATCATC-3′) and Xba-2631 (5′GCTCTAGAAGTTACCAGTTGCCAGTT3′); the 2,261-bp PCR product was cloned in pACYC184, and the acdS gene was then disrupted with the Kmr gene of pHP45Ω-Km (34). The interrupted gene was cloned in the suicide vector pJQ200mp18 (61). The suicide plasmid harboring the interrupted gene was introduced into strain MTl-641T, and double crossover, Kmr sucrose-resistant clones were selected. The gene replacement was confirmed by Southern blot hybridization. To complement the acdS mutant, an 1,864-bp fragment harboring the complete acdS gene and its 5′ regulatory sequence was cloned in the broad-host-range vector pFAJ1700 (27), and the resulting plasmid (pFacdS+) was used for complementation experiments.
To determine bacterial acdS gene expression in plants, a transcriptional fusion was constructed. A 653-bp fragment from MTl-641T containing the upstream acdS regulatory region, as well as 150 bp of the acdS coding sequence, was fused to gusA as follows. The gusA gene from pVO155 (57) was inserted 102 nucleotides downstream of the start codon of the acdS gene to obtain a transcriptional fusion, which was subsequently cloned into pFAJ1700 (pFacdSp-gusA) and introduced into strain MTl-641T (acdSp-gusA). To obtain a constitutive gusA-marked strain, the promoter region (839 bp) of the s7 ribosomal gene from B. xenovorans LB400T was cloned into the promoterless pFgusA vector, resulting in pFs7p-gusA. To generate a negative control for gusA expression, the gusA gene was cloned into the promoterless vector pFAJ1700, resulting in the pFgusA− plasmid. Fusions were introduced by conjugation into B. unamae MTl-641T, and the strains were designated the s7p-gusA, acdSp-gusA, and gusA strains (Table (Table2).2). Transcriptional gene fusions were sequenced to verify the correct DNA sequences.
Stationary-phase cultures of B. unamae were diluted to obtain an OD600 of 0.02 in 6 ml of fresh BSE liquid medium without antibiotics and cultivated for 8 h, and 100-μl aliquots of these cultures were inoculated into fresh BSE liquid medium and incubated for 24 h; this procedure was repeated once, but the culture was incubated for 48 h, and then samples were diluted and plated on BAc agar (32) without antibiotics. Two hundred colonies were picked onto plates with tetracycline or without an antibiotic, and the pFAJ1700 stability frequency in B. unamae derivatives was based on the total number of recovered colonies on medium without an antibiotic compared to the number of colonies resistant to tetracycline.
The Burkholderia acdSp-gusA and gusA strains were grown for 18 h at 29°C in BSE liquid medium with reciprocal shaking (200 rpm); the initial OD600 of the cultures were adjusted to 0.02 (approximately 6.30 log CFU/ml) in SM broth supplemented with ACC (100 nM to 3 mM) or NH4Cl (3 mM) as a single nitrogen source, and the cultures were grown for 6, 18, and 24 h at 29°C. The cultures were centrifuged and resuspended in a salt wash solution supplemented with 100 μg/ml chloramphenicol. Quantitative β-glucuronidase assays were performed with the p-nitrophenyl glucuronide substrate as described previously (73). The protein concentration was determined by the Bradford method (4). Three independent experiments were performed in duplicate.
B. unamae MTl-641T and derivative strains were grown in BSE liquid medium (supplemented with tetracycline or kanamycin when necessary) and incubated at 29°C for 18 h. The cells were collected and resuspended in sterile 10 mM MgSO4 · 7H2O, and the concentration was adjusted to 7 log CFU/ml. Germinated seeds were sown after immersion in the bacterial suspension for 1 h.
Seeds of tomato variety saladet were surface disinfected by soaking them in 1.5% sodium hypochlorite for 15 min, thoroughly rinsed in sterile distilled water, and germinated on agar plates. Germinated seeds were soaked for 1 h in a 10 mM MgSO4 · 7H2O solution (noninoculated) or in a bacterial suspension (wild-type strain or the acdS::ΩKm mutant). After inoculation, three germinated seeds were aseptically transferred to 500-ml plastic pots containing sterilized river sand (approximately 1 kg) moistened with 140 ml Fahraeus solution (33) supplemented with NH4Cl (7.5 mg N/pot), and 1 ml of a bacterial suspension (7 log CFU/ml) was added to each seed. The plants were maintained under greenhouse conditions, and 80 ml of sterile water was applied every 3 days. After 25 days of growth, the plants were treated either with 80 ml of a sterile saline solution (100 mM NaCl) every 3 days or with 80 ml of sterile water (water saturation) daily; 80 ml of sterile water applied every 3 days was used for the control treatment. The plants were harvested 15 days after each treatment, and the root length, chlorophyll content, and dry weight were analyzed. The chlorophyll content was determined as described by Wellburn (72).
Tomato seeds were surface disinfected as described above and transferred to petri dishes containing filter paper moistened with 5 ml of a Farheus-bacterium suspension containing wild-type, acdS::ΩKm, or acdS::ΩKm(pFacdS+) bacteria or with Farheus solution without bacteria. After 5 days of incubation in the dark at 29°C, root lengths were measured.
The expression of the gusA reporter gene was monitored in plants inoculated with gusA-marked strains (the acdSp-gusA, s7p-gusA, and gusA strains). One assay was carried out as described above for the root elongation test, but β-glucuronidase activity was analyzed instead of root length. In addition, inoculated germinated tomato seeds were sown aseptically in assay tubes containing sand and Fahraeus solution with NH4Cl (1.0 mg N/tube). The seedlings were grown in a greenhouse for 3 weeks. Localization of bacterial β-glucuronidase activity was determined using X-Gluc (5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid) as the substrate, as described by Jefferson et al. (46). After the samples were washed in phosphate buffer, the plant tissues were examined and photographed using a light microscope (Zeiss).
Enumeration of root-associated bacteria was performed in all of the plant-inoculation assays. Roots of seedlings were weighed and homogenized in a sterile 10 mM MgSO4 solution with a mortar and pestle. Homogenates were serially diluted and plated in selective BAc medium. After incubation for 3 days at 29°C, the number of colonies of each strain was determined. In addition, amplified 16S rRNA gene restriction profiles for five colonies grown on the selective medium inoculated with the highest dilution were determined as previously described (32) in order to confirm the identity of the inoculated strain.
Unless otherwise indicated, plant and bacterial population data were analyzed by one-way analysis of variance, followed by Tukey's analysis, using the Minitab 15 statistical software. Bacterial population data were log transformed before statistical analysis. All analyses were performed using a P value of ≤0.05. The numbers of replicates and seedlings or plants sampled in each experiment for statistical data analysis are indicated in Tables Tables33 and and44.
Twenty-one acdS gene sequences (15 complete sequences and 6 partial sequences) have been deposited in the EMBL/GenBank database under accession numbers EU886299 through EU886320. Specific accession numbers of each strain are shown in Fig. Fig.11.
Of 45 Burkholderia strains belonging to 20 species (Table (Table1),1), 38 were able to grow on ACC as a sole nitrogen source. These 38 strains, belonging to 18 species, expressed ACC deaminase activity at levels ranging from 2 to 15 μmol α-KB/h/mg protein (Table (Table1).1). All of the B. tropica strains and B. ambifaria 6991 were not able to grow on ACC, and neither of these species expressed ACC deaminase activity.
PCR amplification of acdS genes encoding ACC deaminase from 25 representative strains analyzed belonging to 18 Burkholderia species was performed using three sets of primers. Although ACC deaminase activity was found in B. tuberum STM678T, the acdS gene sequence was not obtained with any of the primer sets tested. The sequences of Burkholderia species were highly conserved, with levels of identity between 76 and 99% at the nucleotide level. Phylogenetic analysis of the acdS gene sequences showed that there were robust clusters containing strains having different origins and from different sources, including plant-associated diazotrophic, non-N2-fixing species and opportunistic pathogens (Fig. (Fig.1).1). Phylogenetic trees based on acdS and 16S rRNA gene sequences of Burkholderia species showed similar topologies (Fig. (Fig.11).
Analysis of the sequence of 2.2-kb pUX-UD revealed two putative open reading frames, one of which is 1,017 bp long and encodes an ACC deaminase with high levels of identity (69 to 87%) with other complete ACC deaminases (8, 66, 67). The second open reading frame (501 bp), transcribed in the opposite direction from acdS, encodes a putative polypeptide that showed 74% identity to Lrp (leucine-responsive regulatory protein) of Pseudomonas putida UW4. In silico analysis of the acdS-lrpL intergenic region showed that these genes contain putative σ70-dependent promoters and that the transcriptional start sites of acdS and lrpL were located 75 and 42 bp upstream of the corresponding ATG initiation codon. Two putative Lrp-binding sites were located upstream of acdS. L-B1 overlapped the −10 box, and L-B2 was localized 84 bp upstream of the ATG initiation codon (Fig. (Fig.2);2); these sites had 11 and 10 identical nucleotides compared to the consensus sequence (YAGHAWATTWTDCTR) in E. coli (22).
ACC deaminase activity was not detected in the acdS mutant (acdS::ΩKm). However, in a liquid rich medium, the growth of the acdS mutant and the growth of wild-type strain MTl-641T were not significantly different (P ≤ 0.05, Student's t test) after 12 h of incubation, reaching concentrations of 8.61 ± 0.02 and 8.74 ± 0.09 log CFU/ml, respectively. In the complemented acdS::ΩKm(pFacdS+) strain the ACC deaminase activity was restored (4.32 ± 0.73 μmol α-KB/h/mg protein) to a level similar to that in the wild-type strain (4.87 ± 0.34).
Plasmid frequency analysis showed that there was 100% stability of pFAJ1700 in B. unamae without selective pressure; the results were virtually identical for all of the cultures analyzed. The growth patterns of the acdSp-gusA strain in liquid SM medium with 3 mM ACC after 48 h of incubation were similar to those of wild-type strain MTl-641T. While the acdSp-gusA strain reached concentrations of 8.25 ± 0.03 log CFU/ml, the wild-type strain concentrations were on the order of 8.18 ± 0.08 log CFU/ml; the concentrations were not statistically different (P ≤ 0.05, Student's t test).
Transcriptional fusions were evaluated in the presence of ACC and NH4Cl. In the presence of ACC, transcriptional acdS activity was detected at 6, 18, and 24 h; in contrast, no activity was detected with NH4Cl (Fig. (Fig.3A).3A). These results were in agreement with the ACC deaminase activity of wild-type strain MTl-641T, since the ACC deaminase did not produce α-KB when the bacterium was grown with 3 mM NH4Cl as the sole nitrogen source (Fig. (Fig.3B).3B). Transcriptional assays with different ACC concentrations (100 nM to 1,000 μM) showed that the maximal acdS activity (1,030 ± 107.61 nmol p-nitrophenyl/min/mg protein) was obtained with 1,000 μM ACC; 1,000 μM NH4Cl and ACC concentrations lower than 0.5 μM did not induce acdS transcription (Fig. (Fig.3C).3C). At the highest concentration (1,000 μM) and at 1 μM ACC, no transcriptional activity was detected in the promoterless gusA strain (Fig. (Fig.3C3C).
The B. unamae MTl-641T (wild type) and acdS::ΩKm mutant populations on tomato plants grown under different conditions were not significantly different (Table (Table3).3). The shoot and root dry weights of tomato plants inoculated with strain MTl-641T were significantly higher (P ≤ 0.05) than those of plants inoculated with the acdS::ΩKm mutant and of noninoculated plants for all the treatments (Table (Table3).3). In addition, the chlorophyll contents of plants inoculated with both strains were statistically increased compared to those of noninoculated plants (Table (Table3).3). Moreover, statistical differences (P ≤ 0.05) were not found among control plants without stress and plants grown either in the presence of NaCl or with water saturation treatments (Table (Table33).
Root growth was enhanced by all bacterial treatments compared to the growth of noninoculated seeds (Table (Table4).4). However, root length was significantly greater with the wild-type and acdS::ΩKm(pFacdS+) treatments than with the acdS::ΩKm mutant treatment, although the bacterial populations associated with the roots from inoculated seedlings were not significantly different (Table (Table44).
For the seedlings grown on filter paper and inoculated with the s7p-gusA strain (constitutive gusA strain) there was an intense blue color on the whole surface of germinated tomato seeds and at the base of the radicle (Fig. (Fig.4A,4A, seedling 1), while for the seedlings inoculated with the acdSp-gusA strain there was a very light blue color on the seed surface and a blue color at the base of radicle that was more intense than the color on the seed surface (Fig. (Fig.4A,4A, seedling 2) but less intense than the color observed with the s7p-gusA strain. As expected, no blue color was detected on seedlings inoculated with a promoterless gusA-marked strain (Fig. (Fig.4A,4A, seedling 3). Similar numbers of the three gusA-marked strains colonized the seedlings (7.37 ± 0.08, 7.32 ± 0.24, and 7.41 ± 0.25 log CFU/g root for the acdSp-gusA, s7p-gusA, and gusA strains, respectively). The root systems of seedlings grown on sand and inoculated with the s7p-gusA strain were well colonized (6.44 ± 0.12 log CFU/g roots), as revealed by intense blue staining of the tissue. The blue staining was most intense at the stem base and where lateral roots emerged (Fig. 4B and C). In contrast, the root systems of seedlings treated with the acdSp-gusA strain appeared to have less intense blue staining on fewer root sections (Fig. (Fig.4D).4D). However, the root colonization by this strain (6.59 ± 0.14 log CFU/g root) was not statistically different from the root colonization of the plants inoculated with the s7p-gusA and gusA (6.59 ± 0.21 log CFU/g roots) strains used as positive and negative gusA expression controls, respectively. No blue color was detected on seedlings inoculated with the gusA strain (data not shown).
Considering the limited availability of Burkholderia strains, and since the genomic and phenotypic characteristics of strains of a particular bacterial species are the same or highly similar (1, 6, 12), only the type strains of most Burkholderia species were analyzed in this work. The results showed that ACC deaminase activity and the acdS gene are widely distributed in Burkholderia species. Since an activity of 20 nmol α-KB/h/mg protein has been reported (58) to be sufficient to cause plant growth promotion and activities of 2 to 15 μmol α-KB/h/mg protein were detected for the strains tested, it appears that most of the plant-associated Burkholderia strains analyzed in this study, either rhizospheric (e.g., B. xenovorans, B. phenoliruptrix, and B. graminis) or endophytic (B. unamae, B. silvatlantica, and B. kururiensis), could be considered potential plant growth-promoting bacteria. Similarly, the legume-nodulating species B. phymatum and B. tuberum could contribute to enhanced nodulation through ACC deaminase activity, as observed in pea plants inoculated with Rhizobium leguminosarum bv. viciae (51) and in alfalfa plants inoculated with a Sinorhizobium meliloti strain expressing an introduced acdS gene (50). Homologs of acdS genes have been found in whole-genome sequences of B. xenovorans LB400T (accession no. CP000271), B. phymatum STM815T (accession no. CP001046), and B. cenocepacia J2315T (accession no. AM747721), and their ACC deaminase activities were demonstrated in this work.
Interestingly, the topology of the acdS phylogenetic tree was fairly congruent with that of the bacterial 16S rRNA tree. Phylogenetic trees based on 16S rRNA and acdS sequences show separation of the genus Burkholderia into two major clusters, one of which is represented mainly by human-pathogenic species, including the opportunistic Bcc species (17). The second major cluster, phylogenetically distant from the Bcc, is formed solely by environmental nonpathogenic species, including mainly plant-associated, rhizospheric and/or endophytic bacteria, many diazotrophic organisms (53), and/or legume-nodulating species (13, 30, 68). Phylogenetic analysis indicated that acdS gene sequences of Burkholderia strains were highly conserved, even though the strains were isolated from different sources and from distant geographic regions. It is worth noting that there is tight clustering of the Bcc species analyzed in this work; their acdS gene sequences were clearly distinct from those of the plant-associated Burkholderia species, even B. unamae and B. silvatlantica. Although the cluster formed by Burkholderia species (Betaproteobacteria) was well separated from clusters containing a variety of Alphaproteobacteria (R. leguminosarum bv. viciae and Mesorhizobium loti) and Gammaproteobacteria (Pseudomonas spp. and Enterobacter cloacae), Pseudomonas sp. strain ACP (accession no. M73488) formed a cluster with B. xenovorans strains, and the acdS sequences of these strains were closely related to those of B. phytofirmans and B. terricola. Analysis based on nucleotide and protein sequences revealed high levels of identity (90 to 91% and 94 to 96%, respectively) between B. xenovorans strains CAC-124 and CCUG 2844 obtained in this work and B. xenovorans LB400T (accession no. CP000271.1), TCo-382, and TCo-26 described previously (7). In contrast, alignment of the acdS sequence of Pseudomonas sp. strain ACP and reported sequences of P. putida UW4 (accession no. AF0477710), Pseudomonas fluorescens 17 (accession no. UFU37103), P. fluorescens F113 (accession no. DQ125256), and Pseudomonas sp. strain 6G5 (accession no. M80882) showed levels of identity of 73 to 74% and 81 to 82% for nucleotide and protein sequences, respectively. Based on these data, lateral transfer of the acdS gene between Betaproteobacteria (Burkholderia) and Gammaproteobacteria (Pseudomonas) has been suggested (3). Nevertheless, although the acdS gene sequence of strain ACP has been well characterized, we cannot exclude the possibility that this strain belongs to the genus Burkholderia, since strain ACP was tentatively classified as Pseudomonas sp. using a very limited set of phenotypic tests (43); to our knowledge, the exact taxonomic status of this strain has never been defined. Recently, a partial acdS gene sequence (797 nucleotides; accession no. DQ125247) of B. caledonica LMG 19076T was reported (3). However, our results show that B. caledonica forms a tight cluster with B. phenoliruptrix and B. graminis (Fig. (Fig.1),1), but not with several Pseudomonas species described by Blaha et al. (3). Moreover, in a previous study (7) and in this study, we were unable to detect ACC deaminase activity and presence of the acdS gene in our original strain B. tropica BM-273 (31), as previously reported (3). The absence of ACC deaminase activity and of the acdS gene was confirmed using other six B. tropica strains (62). Our analyses based on 15 complete and 6 partial acdS sequences from Burkholderia strains belonging to 15 species and on the genomes of four other species (B. xenovorans LB400T, B. phytofirmans PsJNT, B. phymatum STM815T, and B. cenocepacia J2315T) do not support the lateral transfer hypothesis for such a gene in this genus, as previously suggested (3), although lateral transfer appears to have happened between bacteria belonging to other genera (44).
Although the N2-fixing species B. unamae exhibited low ACC deaminase activity (however, the activity was high enough to promote plant growth), this species was chosen to assess the effect of the enzyme on tomato plant growth due to its phylogenetically distant relationship with the Bcc species and relevant features (6, 7, 59) compared to other Burkholderia species analyzed in this work. B. unamae wild-type strain MTl-641T and derivative strains had very similar growth patterns in culture media and very similar abilities to colonize tomato roots. On this basis, the higher shoot and root dry weights of tomato plants inoculated with wild-type strain MTl-641T could be attributed to the ACC deaminase activity, since the values for these parameters were significantly lower for plants inoculated with the ACC deaminase-negative mutant (acdS::ΩKm) and for noninoculated plants. A beneficial effect on tomato root elongation was consistently observed in plants inoculated with the wild-type strain and the complemented acdS mutant [acdS::ΩKm(pFacdS+)], as these two strains were similarly able to promote root elongation compared to the noninoculated plants and plants inoculated with the acdS mutant (acdS::ΩKm) strain. However, an additional plant growth-promoting mechanism seems to be expressed by B. unamae MTl-641T, since the presence of the acdS::ΩKm mutant also increased, although at a lower rate than the wild type and the complemented strain, the root length of inoculated plants compared to noninoculated plants. A similar effect on the chlorophyll content of the plants inoculated with the acdS::ΩKm mutant compared to noninoculated plants was observed as well. Such an additional beneficial mechanism could be related to the ability of B. unamae MTl-641T to synthesize auxins like indoleacetic acid (IAA), which is produced in culture media (unpublished results). The participation of bacterial IAA in plant growth promotion has been demonstrated in plant-rhizobacterium interactions in many studies (26). Thus, whether IAA synthesized by B. unamae MTl-641T partially masked the positive effect of ACC deaminase on tomato plant growth could be demonstrated by engineering an acdS and IAA-negative mutant in further studies. Although it has been reported that some ACC deaminase-producing bacteria promote plant growth under stressful conditions, such as flooding (42) and saline conditions (54), in this work these stress conditions caused no significant differences in shoot or root dry weights between controls and stress treatments. It is possible that the method used to apply stress treatments was not strong enough to cause a significant reduction in the growth parameters analyzed and, as a consequence, plants could gradually undergo an adaptive process in response to mild or moderate stress without major changes in their growth (9, 21).
Tomato inoculation assays with the constitutive gusA strain (s7p-gusA) confirmed the ability of B. unamae to colonize seeds during germination and seedling roots, mainly at the radicle and stem base, as well as at the points of emergence of lateral roots. It has long been known that increased exudation of many organic compounds occurs in natural wounds, such as cracking of seed coats and damage by radicle and lateral root emergence (23). The colonization pattern and infection sites observed with B. unamae seem to be common features in plant-bacterium interactions; for instance, they occur with the endophytic bacteria G. diazotrophicus (45) and Azoarcus sp. strain BH72 (29) in association with gramineous plants and with Burkholderia sp. (currently B. phytofirmans) strain PsJN associated with Vitis vinifera (19). Although compared to the constitutive gusA strain a lower level of gusA expression was observed for the acdSp-gusA strain on germinating seeds and at the base of the radicle, the results presented here suggest that the concentration of ACC exuded was at least 0.5 μM (the minimum concentration required for induction of the acdS gene) just after the seed germinated, as well as in natural wound sites. Micromolar levels of ACC (range, 1 to 100 μM) were sufficient to induce expression of AcdS in R. leguminosarum bv. viciae 128C53K (52). Similarly, while gusA expression was observed along the root system (plants grown in sand) with the constitutive gusA strain, a reduced number of root sites showing gusA expression (which notably was highest at the base of the radicle and stem) were observed with the acdSp-gusA strain. It is known that root exudates differ in chemical composition and quantity, which vary at specific sites on roots (23). In addition, the impact of AcdS apparently is modest, and AcdS probably specifically affects local regulatory mechanisms, such as the mechanisms controlling root hair elongation in Arabidopsis inoculated with Rhizobium and Pseudomonas (20). Accordingly, this could explain the limited number of root sites where gusA expression was observed with the acdSp-gusA strain in a quantity of ACC sufficient to induce the acdS promoter induction at localized root sites. Even so, the localized effect produced by B. unamae MTl-641T was enough to improve seedling growth, as described above for the tomato inoculation experiments with wild-type strain MTl-641T, and the root length observed with the acdS::ΩKm(pFacdS+) strain.
Although the beneficial effect of ACC-expressing bacteria on plant growth, illustrated in this study with B. unamae, needs to be confirmed in field experiments, the widespread ACC deaminase activity in Burkholderia species and the common association of these species with plants suggest that this genus could be a major contributor to plant growth under natural conditions.
We thank L. Martínez-Aguilar for technical assistance and V. M. Hernández-Velázquez (CEIB-UAEM) for his valuable advice on statistical data analysis. We are grateful to Michael Dunn (CCG-UNAM) for reading the manuscript.
J.O.-L. is a Ph.D. student at the Doctorado en Ciencias Biomédicas-UNAM and acknowledges fellowships from the CONACYT and the DGEP-UNAM.
Published ahead of print on 21 August 2009.