ABSTRACT
Bradyrhizobium japonicum USDA 110 has five polyhydroxyalkanoate (PHA) synthases (PhaC) annotated in its genome: bll4360 (phaC1), bll6073 (phaC2), blr3732 (phaC3), blr2885 (phaC4), and bll4548 (phaC5). All these proteins possess the catalytic triad and conserved amino acid residues of polyester synthases and are distributed into four different PhaC classes. We obtained mutants in each of these paralogs and analyzed phaC gene expression and PHA production in liquid cultures. Despite the genetic redundancy, only phaC1 and phaC2 were expressed at significant rates, while PHA accumulation in stationary-phase cultures was impaired only in the ΔphaC1 mutant. Meanwhile, the ΔphaC2 mutant produced more PHA than the wild type under this condition, and surprisingly, the phaC3 transcript increased in the ΔphaC2 background. A double mutant, the ΔphaC2 ΔphaC3 mutant, consistently accumulated less PHA than the ΔphaC2 mutant. PHA accumulation in nodule bacteroids followed a pattern similar to that seen in liquid cultures, being prevented in the ΔphaC1 mutant and increased in the ΔphaC2 mutant in relation to the level in the wild type. Therefore, we used these mutants, together with a ΔphaC1 ΔphaC2 double mutant, to study the B. japonicum PHA requirements for survival, competition for nodulation, and plant growth promotion. All mutants, as well as the wild type, survived for 60 days in a carbon-free medium, regardless of their initial PHA contents. When competing for nodulation against the wild type in a 1:1 proportion, the ΔphaC1 and ΔphaC1 ΔphaC2 mutants occupied only 13 to 15% of the nodules, while the ΔphaC2 mutant occupied 81%, suggesting that the PHA polymer is required for successful competitiveness. However, the bacteroid content of PHA did not affect the shoot dry weight accumulation.
INTRODUCTION
Rhizobia are soil bacteria that may live in two states: as free-living bacteria in the soil or as symbiotic bacteroids inside legume nodules. Soybean plants form determinate nodules, which are round and maintain the oldest infected cells at their center (1, 2). Bacteroids inside these determinate nodules synthesize polyhydroxyalkanoates (PHAs), which may serve as carbon- and energy-reserve polymers that are stored in complex cytoplasmic granules. In addition, bacteroids from determinate nodules are able to dedifferentiate back to free-living rhizobia once the nodule senesces (3).
PHA is of great interest due to its potential use as a biodegradable plastic (4). The cellular functions of PHA are related to the cell's metabolic balance, and PHA is especially useful to the cell as an energy and reductant reserve (5). In rhizobia, where PHA contributes to reproduction and survival (6), this polymer is synthesized in three steps (see Fig. S1 in the supplemental material): the condensation of two acetyl coenzyme A (acetyl-CoA) molecules to synthesize acetoacetyl-CoA is catalyzed by β-ketothiolase (PhaA), the reduction of acetoacetyl-CoA to hydroxyacyl-CoA in an NADPH-dependent reaction is catalyzed by NADP–acetoacetyl-CoA reductase (PhaB), and the polymerization of hydroxyacyl-CoA into PHA is catalyzed by PHA synthase (PhaC). The last enzyme is essential for PHA synthesis in all rhizobial species (5).
Although most sequenced rhizobia have one or two phaC homologs in their genomes, five putative phaC open reading frames (ORFs) have been detected in Bradyrhizobium japonicum USDA 110 (7) and are named bll4360, bll6073, blr3732, blr2885, and bll4548 (henceforth phaC1, phaC2, phaC3, phaC4, and phaC5, respectively). However, experimental evidence about the expression of these genes is available only for phaC1 and phaC2. In a proteomic study of soybean nodule bacteroids, Sarma and Emerich (8) detected PhaC2. This observation is consistent with that from another, transcriptomic study, in which the phaC2 transcript was observed to be significantly expressed in nodules and in microaerobiosis (9), as well as with the observation that phaC2 expression is promoted by FixK2, which is a transcription factor that is regulated by O2 levels (10, 11). Pessi and coworkers (9) also observed that in contrast to phaC2, phaC1 is downregulated in bacteroids in comparison to its regulation in aerobic free-living bacteria. In agreement with the last observation, Franck and coworkers (12) reported that phaC1 is upregulated in free-living chemoautotrophic bacteria, while phaC2 is downregulated in the same state. However, more recent proteomic and transcriptomic studies have indicated that both phaC1 and phaC2 are present in nodule bacteroids (13, 14). In a different study, Aneja and coworkers attempted to complement a phaC-defective Ensifer (formerly Sinorhizobium) meliloti mutant with a B. japonicum cosmid library in order to isolate genes encoding functional PHA synthases in this rhizobial species and were successful only with phaC1 (15).
These studies suggest that phaC1 and phaC2 might be specifically required in the free-living and bacteroid states, respectively. However, this suggestion is based only on expression studies, since PhaC1 is the only paralog for which proof of biological activity is available (15). In addition, there are no data on the possible functionality of the other three putative phaC paralogs mentioned above. To get a better understanding of the possible role of these different PHA synthases in the survival of the free-living state and the symbiotic efficiency of this species, we studied their requirement for PHA production, as well as the impact of this reserve polymer on growth kinetics, survival, competitiveness for nodulation, and plant growth-promoting activity.
MATERIALS AND METHODS
Bacterial strains and culture conditions.Strains and plasmids are summarized in Table 1. B. japonicum was grown in Götz minimal medium with mannitol as the sole carbon source (16). The total biomass was estimated by measurement of the optical density at 500 nm (OD500) and the number of viable bacteria on the basis of the number of CFU on yeast extract-mannitol (17) agar (YMA) plates. For conjugation, peptone salts-yeast extract (PSY) medium (18) was employed. Escherichia coli was grown in Luria-Bertani (LB) medium (19). Antibiotics were added to the media at the following concentrations (μg ml−1): streptomycin (Sm), 400 (B. japonicum) or 100 (E. coli); spectinomycin (Sp), 200 (B. japonicum) or 100 (E. coli); kanamycin (Km), 150 (B. japonicum) or 25 (E. coli); ampicillin (Ap), 200 (E. coli); gentamicin (Gm), 100 (B. japonicum) or 10 (E. coli); tetracycline (Tc) 20 and 50 (B. japonicum liquid and solid cultures, respectively) or 10 (E. coli); and chloramphenicol (Cm), 20 (B. japonicum).
Plasmids and bacterial strains used in this study
Genetic techniques and DNA manipulation.Cloning procedures, including DNA isolation, restriction digestion, ligation, and transformation, were performed as described previously (19). Tri- or biparental matings were performed with the E. coli DH5α or S17-1 strain, as has been previously described (20). Electroporation was performed with a Gene Pulser system (Bio-Rad, Hercules, CA) at 1.5 V, 25 μF, and 200 Ω in a 0.1-cm-gap-width electroporation cuvette.
Oligonucleotide primers were purchased from Life Technologies (Buenos Aires, Argentina). DNA amplification was performed by PCR using Taq DNA polymerase (Life Technologies, Buenos Aires, Argentina) for routine PCR or KAPA HiFi hot-start (HS) DNA polymerase (Kapabiosystems, Woburn, MA) for the amplification of targets longer than 1,000 bp in a Bioer Life Express thermocycler (Hangzhou, China). DNA sequencing was performed at Macrogen Corp. (Seoul, South Korea).
To construct the B. japonicum phaC mutants, primers specific for bll4360 (phaC1), bll6073 (phaC2), blr3732 (phaC3), blr2885 (phaC4), and bll4548 (phaC5) were designed (see Table S1 in the supplemental material). For phaC1 and phaC2, fragments of B. japonicum USDA 110 genomic DNA were generated from both sides of the target coding sequences. For phaC1, an upstream fragment of 310 bp and a downstream fragment of 383 bp were amplified using primers Fw1/Rv1 (fragment 1) and Fw2/Rv2 (fragment 2), respectively. These PCR fragments were cloned into pGEM-T Easy separately, creating pIQ16 and pIQ17, respectively. pIQ16 was then digested with EcoRI and subcloned into pBBR1MCS5, creating pIQ20. pIQ17 was digested with SphI and PstI and subcloned into pK18mob, creating pIQ22. The pIQ22 plasmid was digested with XbaI and KpnI, and the 402-bp XbaI-KpnI-digested fragment 1 product from pIQ20 was cloned at this site, generating pIQ24. The pIQ24 vector was subsequently digested with SmaI (resulting in a cut between fragments 1 and 2) and ligated with the SmaI-digested Ω-Sm-Sp interposon from pHP45, generating pIQ26. Gene replacement was performed by introducing pIQ26 into the wild-type strain B. japonicum USDA 110 by biparental mating, selecting by growth on Sm-Sp-supplemented YMA, and screening for Km sensitivity in order to select for the double-crossover event. The resulting strain was designated LP 4360 (ΔphaC1); this strain carries the Ω-Sm-Sp-interposon in place of the genomic DNA from base 4820583 to base 4821613, thus removing 1,031 bp from the center of the 1,803-bp ORF bll4360 (7).
For phaC2, an upstream fragment of 379 bp and a downstream fragment of 330 bp were amplified using primers Fw3/Rv3 (fragment 3) and Fw4/Rv4 (fragment 4), respectively. These PCR fragments were cloned into pGEM-T Easy separately, creating pIQ18 and pIQ19, respectively. pIQ18 was digested with EcoRI and subcloned into pBBR1MCS5, creating pIQ21. The pIQ19 vector was digested with EcoRI and subcloned into pG18mob2, creating pIQ23. The pIQ23 plasmid was then digested with HindIII and PstI, and fragment 3 (411 bp) from pIQ21 was cloned at this site, generating pIQ25. The pIQ25 vector was subsequently digested with SalI (resulting in a cut between fragments 3 and 4) and ligated with the SalI-digested nptI cassette from pUC4-K, generating pIQ27. Gene replacement was performed by introducing pIQ27 into the wild-type strain B. japonicum USDA 110 by triparental mating using the pRK2013 helper plasmid, selecting by growth on Km-supplemented YMA, and screening for Gm sensitivity in order to select for the double-crossover event. Strain LP 6073 (ΔphaC2) carries the nptI cassette between bases 6689567 and 6690793 of the genomic DNA, thus replacing 1,226 bp of the 1,842-bp ORF bll6073.
The double mutant LP 6060 (ΔphaC1 ΔphaC2) was constructed as described above, introducing pIQ27 in the background of LP 4360.
To construct B. japonicum phaC3, phaC4, and phaC5 gene disruption mutants, blunt-end fragments of 312, 212, and 230 bp were amplified using primers Fw5/Rv5, Fw6/Rv6, and Fw7/Rv7, respectively, with KAPA HiFi HS DNA polymerase (Kapabiosystems, Woburn, MA). These PCR fragments were cloned into the SmaI site of pG18mob2 separately, creating plasmids pIQ32, pIQ33, and pIQ34, respectively. Conjugation was performed by introducing these plasmids into the wild-type strain B. japonicum USDA 110 by biparental mating using E. coli S17-1 as the donor and selecting by growth on Cm- and Gm-supplemented YMA in order to select for single homologous recombination events. The single mutant strains carrying phaC insertions have the catalytic triad interrupted by the plasmid vector and were named LP 3732 (ΔphaC3), LP 2885 (ΔphaC4), and LP 4548 (ΔphaC5). The plasmids were inserted at bases 3179636, 4142117, and 5042196 of the genomic DNA, respectively.
The double mutant LP 7337 (ΔphaC2 ΔphaC3) was constructed as described above, introducing pIQ32 in the background of LP 6073.
All the mutations were confirmed by PCR using external primers (see Table S1 in the supplemental material), as described previously (20), and by DNA sequencing.
Complementation experiments were performed by integrating the complete sequences of phaC1 or phaC2 (amplified with primers Fw1c/Rv1c and Fw2c/Rv2c, respectively) and their putative native promoters (inferred by the BPROM program) into a replicative vector. Briefly, the 2,110-bp phaC1 and the 2,379-bp phaC2 target sequences were amplified from B. japonicum USDA 110 chromosomal DNA and cloned into the SmaI site of pBBR1MCS2 to create pIQ28 and pIQ29. Then, pIQ28 was digested with BamHI and KpnI and pIQ29 was digested with XbaI and KpnI. Each fragment was inserted in the replicative broad-host-range vector pFAJ1708, generating plasmids pIQ30 (phaC1) and pIQ31 (phaC2). Recombinant plasmids were cloned with phaC genes under the strong, constitutive nptII promoter (21). These constructions were confirmed by sequencing. Then, each plasmid harboring the complete phaC1 or phaC2 gene was transferred into the desired B. japonicum strain by biparental mating.
Bioinformatics and in silico characterization of PHA synthases.The multiple-sequence alignment was performed using T-COFFEE software (22). The phylogenetic analysis was performed using the maximum likelihood method and the PhyML (version 3.0) program (23) with bootstrap analysis (1,000 iterations). The tree was constructed using MEGA (version 5.1) software (24). The occurrence of conserved residues in the catalytic triad and the lipase box was checked manually.
To estimate the PhaC three-dimensional structure, fold assignment was performed with the FFAS03 (25) and HHpred (26) programs and ab initio fold prediction was performed with the ITASSER program (27). The quality of the models was evaluated with the PROSAII program (28).
qRT-PCR.Fifteen milliliters of B. japonicum liquid cultures was centrifuged at 13,500 × g for 40 min and washed twice with 1 M NaCl. Then, the cells were disrupted with lysozyme in TE (Tris-EDTA) buffer, pH 8.0 (30 min, 37°C). Total RNA was extracted using the TRIzol reagent (Life Technologies, Buenos Aires, Argentina), following the manufacturer's instructions, and its quantity and quality were determined using a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE). Aliquots (0.5 μg) were treated with DNase I (30 min, 37°C), and cDNA was synthesized using random hexamer primers with Moloney murine leukemia virus reverse transcriptase (Life Technologies, Buenos Aires, Argentina), following the manufacturer's instructions. To check the quality of the cDNA preparation, PCRs were performed using primers Fw2/Rv2, Fw4/Rv4, Fw5/Rv5, Fw6/Rv6, and Fw7/Rv7 for phaC1, phaC2, phaC3, phaC4, and phaC5, respectively (see Table S1 in the supplemental material). The absence of contaminating DNA was demonstrated by the lack of PCR amplification in an RNA sample that was not subjected to reverse transcription (RT). The sigA primers were used as a positive control (29). Quantitative real-time RT-PCR (qRT-PCR) amplification was performed with primers Fw2/Rv2, Fw4/Rv4, Fw5-Rv5, Fw6/Rv6, and Fw7/Rv7 (see Table S1 in the supplemental material) in a Line-Gene instrument (Bioer, Hangzhou, China), and the amplicons were analyzed with a Line-Gene K fluorescence quantitative detection system (version 4.0.00 software). The ready-to-use iQ SYBR green Supermix (Bio-Rad, Hercules, CA) was used for all of the reactions, according to the manufacturer's instructions. Standard 10-fold serial dilutions of the cDNA samples were used to calculate the PCR efficiency. Standard curves were generated from the purified amplicons for absolute quantification. Normalized expression values were calculated as the ratio of the absolute quantities of the gene of interest to the absolute quantities of the housekeeping gene sigA.
Carbon starvation survival assay.Fifty-milliliter bottles containing 15 ml of Götz medium without a carbon source were inoculated with 19-day-old B. japonicum cells. To prepare the inocula, the cultures were pelleted and washed twice with sterile 0.5% NaCl at a final OD500 of 0.2. These cell suspensions were maintained at 28°C with shaking at 180 rpm, and the numbers of CFU were periodically counted from 0 to 60 days.
Quantification of PHA.B. japonicum cultures were twice centrifuged at 12,000 × g for 40 min. Then, the pellet was homogenized with sodium hypochlorite overnight at room temperature, washed with double-distilled water, precipitated with 1:1 alcohol-acetone, and resuspended in chloroform. Then, PHA was determined as chrotonic acid in H2SO4 (30).
To determine the PHA content of bacteroid cells, seven mature nodules (weight, 60 to 90 mg) were crushed and homogenized, suspended in 1 ml of double-distilled water, and centrifuged for 5 min at 5,000 × g. The supernatants were then processed for the determination of PHA levels, as has been previously described (16).
Plant experiments.DonMario 3810 soybean seeds, kindly provided by DonMario SA, Argentina, were surface sterilized and germinated as has been previously described (20). Plants were sown in sterile vermiculite pots containing modified N-free Fåhraeus plant nutrient solution (20) and grown for the indicated times in a greenhouse at temperatures of 26°C and 18°C during the day and at night, respectively, with uncontrolled relative humidity, which varied from about 60% during the day to 90% at night. The pots were irrigated with sterile distilled water every 4 days and provided one watering with 1:5-diluted nitrogen-free plant nutrient solution every 15 days.
To study competition for nodulation, we inoculated 10 vermiculite pots containing 3-day-old soybean plantlets with bacterial mixtures of strain USDA 110 and each mutant strain (ΔphaC1, ΔphaC2, or ΔphaC1 ΔphaC2 strain) in a 1:1 proportion, as has been described previously (20). Each mixed inoculum was freshly prepared by dilution of early-stationary-phase liquid cultures (9 days old). As controls, two pots were kept uninoculated and sets of two pots each were inoculated with each competitor strain and unmixed. After 21 days in the greenhouse, nodules were excised, surface sterilized, and crushed, and their contents were grown in YMA plates to obtain the occupying rhizobia, as has been previously described (20). Nodule occupation was evaluated on the basis of the antibiotic resistance of the strains that were recovered from the nodules, after the recovered bacteria were inoculated in replicate YMA plates containing selective antibiotics. The statistical analysis was carried out using the χ2 method, considering 45% nodule occupation by each strain alone and 10% double occupation to be the null hypothesis (31).
Plant growth parameters were assessed in 67-day-old plants. Shoots or nodules from each plant were cut and dried in an oven at 60°C to a constant weight, and the materials from each plant were weighed individually. The statistical analysis of shoot dry weight, nodule number, and nodule dry weight per plant was performed by a one-way analysis of variance (ANOVA), with significance being a P value of <0.05.
Microscopy.Bacteria from liquid cultures and nodules excised from plants (each one transversally cut in halves) were collected and fixed in 2% (vol/vol) glutaraldehyde, as has been described previously (20). Next, the samples were dehydrated, infiltrated with epoxy resin, sectioned, and stained, as has been previously described (20). Transmission electron microscopy of the glutaraldehyde-fixed samples was performed as has been described previously (20). The grids were viewed at 80 kV in a JEM1200 EX II transmission electron microscope (JEOL Ltd., Tokyo, Japan), and micrographs were obtained with an ES500W Erlangshen charge-coupled-device camera (Gatan Inc., Pleasanton, CA).
To measure the bacterial major longitudinal cell size in free-living bacteria, pictures of 200 cells of each strain taken as described above were evaluated using ImageJ software (32).
For light microscopy, glutaraldehyde-fixed 2-μm-thick sections of nodules were dried onto glass slides, stained with a saturated solution of toluidine blue, and analyzed using a Nikon Eclipse E200 microscope (Melville, NY).
RESULTS
In silico analysis of the putative PHA synthases.The B. japonicum USDA 110 genome contains five ORFs encoding putative PHA synthases. All of them seem to be monocistronic and are at different locations in the genome, and none of them has neighboring genes related to PHA metabolism. To further characterize these ORFs, we constructed a tree based on a multiple-sequence alignment using a maximum likelihood method with 1,000 bootstraps (see Fig. S2 in the supplemental material). While PhaC1 and PhaC3 belonged to class I PHA synthases, PhaC4 grouped with class IV PHA synthases, PhaC5 grouped with class III/IV PHA synthases, and PhaC2 grouped with five unclassified sequences (four from the betaproteobacteria and one from the alphaproteobacteria), which indicated a considerable diversity of these ORFs within the B. japonicum USDA 110 genome. Interestingly, PhaC1 and PhaC2 were closely related to the only two PhaC paralogs of Rhodopseudomonas palustris.
The phaC1, phaC2, and phaC3 genes encode proteins of 600, 613, and 572 amino acids, respectively, while the phaC4 and phaC5 genes encode proteins of 360 and 371 amino acids, respectively. A multiple-sequence alignment of all these ORFs indicated that the C-terminal portions are conserved, while the N-terminal portions in PhaC1, PhaC2, and PhaC3 are longer than those in PhaC4 or PhaC5 (see Fig. S3 in the supplemental material). The C-terminal portions of the five ORFs contain a catalytic triad (C339, D491, and H519 in PhaC1), a tryptophan essential for dimerization (W437 in PhaC1), and the conserved lipase box GXCXG (33). In particular, in PhaC2 the lipase box is GXCLG, which is conserved with the other five sequences of this unclassified PHA synthase group.
PhaC1, PhaC2, and PhaC3 may indeed have two domains. Using fold assignment techniques, we found a carboxy-terminal domain starting at residue 220 of PhaC1 that probably adopts a fold similar to that of a lipase from Sulfolobus solfataricus (Protein Data Bank [PDB] accession number PDB 2rau; FFAS03 score, −41.2; HHpred E value, 1.4 10−27; both the FFAS03 score and HHpred E value are statistically significant). With ab initio fold prediction, we obtained tertiary models with two domains, mainly based on the structure of epoxide hydrolase 2 from Homo sapiens (PDB accession number 1s8o; Z score, −6.33). The N-terminal domain has an alpha-helix fold, and the C-terminal domain is similar to that obtained with fold assignment techniques (see Fig. S4 in the supplemental material). Since the main function of the N-terminal domain might be dimer stabilization (34), dimerization of PhaC4 and PhaC5, which have only one domain with a fold assignment similar to that of the C-terminal domain of the other three putative PHA synthases, might be disfavored. It was reported that full functional activity of class III and IV PHA synthases requires an additional polypeptide that noncovalently associates with the catalytic subunit in place of the missing N-terminal domain (33). However, homologs to such a polypeptide, named PhaE or PhaR, were not found in the B. japonicum USDA 110 genome.
Expression and functionality of phaC genes.The expression of phaC1 and phaC2 has been documented previously (9). To extend this knowledge to the other phaC paralogs, we performed real-time PCR (qRT-PCR) of the five phaC genes with B. japonicum USDA 110 cultures of two ages: 6 days of growth (OD500 = 0.3, corresponding to logarithmic phase) and 18 days of growth (OD500 = 1.5, corresponding to stationary phase). As shown in Fig. 1, only phaC1 and phaC2 were expressed at a significant rate in relation to the level of expression of sigA, used as a constitutive control. In particular, phaC2 had a high expression level in young cultures. Nevertheless, it was reported that small amounts of PHA synthase are sufficient for PHA synthesis and accumulation, while 5-fold increases in intracellular PHA synthase did not lead to a considerable increase of PHA in Pseudomonas oleovorans (35). Therefore, to better evaluate the significance of these different phaC expression levels, we obtained single mutants with mutations disrupting the catalytic triad in each of these genes and measured their PHA production. In agreement with the very low level of phaC3, phaC4, and phaC5 expression, mutation of these genes did not cause significant differences of PHA contents with respect to that of the wild type (not shown). Meanwhile, the phaC1 mutant was severely impaired for PHA production, and on the contrary, the phaC2 mutant produced different PHA levels than the wild type (see below).
Quantification of phaC paralog mRNAs in free-living B. japonicum USDA 110 cells at logarithmic and stationary phases. mRNAs were obtained at logarithmic phase (6 days old) or stationary phase (18 days old). Relative mRNA levels are normalized as the ratio of phaC mRNA levels to constitutive sigA mRNA levels. Results are representative of three different experiments. Each bar represents the mean ± SEM.
In view of these results, we performed a more detailed analysis of the phaC1 and phaC2 mutant strains in both the free-living and symbiotic states to assess the effects of different PHA accumulation levels on growth, survival, nodulation, and plant growth promotion. In order to unambiguously refer to the specific deletions analyzed here, these strains were named LP 4360 and LP 6073 (the numbers make reference to the ORFs altered). We also included in the following analysis a double mutant, the ΔphaC1 ΔphaC2 mutant, which was named LP 6060.
Growth of phaC mutants.The growth kinetics of LP 4360 (ΔphaC1) and LP 6073 (ΔphaC2) were analyzed in detail by recording the total biomass as the OD500 measurement and the number of viable cells as the number of CFU in YMA (Fig. 2). At early stationary phase (6 to 10 days), LP 4360 exhibited higher viable cell counts but a lower OD500 than the wild-type strain and LP 6073 (Fig. 2B). Accordingly, at the same growth state, the LP 4360 and LP 6060 cells were shorter (1.79 ± 0.38 μm and 1.54 ± 0.47 μm, respectively) than the wild-type cells (3.26 ± 0.88 μm). At late stationary phase, LP 4360 and LP 6060 had lower viable cell counts than the wild type but similar OD500 measurements (Fig. 2A), which might indicate a higher proportion of dead cells in those cultures. The ΔphaC2 strain LP 6073 behaved similarly to the wild type during the growth period examined, except in late stationary phase, when it had more viable cells than the wild type.
Growth kinetics of wild-type B. japonicum USDA 110 and ΔphaC1, ΔphaC2, and ΔphaC1 ΔphaC2 mutants. Growth was evaluated as the OD500 measurement (A) or the number of CFU ml−1 ± SD (B). When errors bars are absent, they are smaller than the symbol.
PHA accumulation in phaC1 or phaC2 mutants in the free-living state.The wild-type strain B. japonicum USDA 110 as well as the ΔphaC2 mutant LP 6073 produced typical refringent PHA granules at late stationary phase (18 days), as observed by transmission electron microscopy (Fig. 3). Interestingly, the granules produced by LP 6073 were larger than those produced by the wild type. In contrast, neither the ΔphaC1 LP 4360 single mutant nor the ΔphaC1 ΔphaC2 LP 6060 double mutant produced PHA granules.
Electron micrographs of free-living (18-day-old) B. japonicum cells showing refringent PHA granules. Dark spots resembling polyphosphate granules. (A) USDA 110; (B) ΔphaC2 mutant; (C) ΔphaC1 mutant; (D) ΔphaC1 ΔphaC2 mutant. Bars, 0.5 μm.
To better analyze the PHA production by USDA 110 and LP 6073, we quantified their PHA content throughout their growth period in Götz minimal medium (Fig. 4A). In USDA 110, PHA accumulated with culture age, as has been previously observed (16). Meanwhile, LP 6073 (ΔphaC2) produced 50% less PHA than the wild-type strain in logarithmic phase (6 days), but later, PHA production sharply increased, reaching 228% more PHA than that in the wild type at the beginning of stationary phase (9 days) and then stabilizing to 35% more PHA than that in the wild type in old cultures (41 days), in agreement with the larger PHA granules observed in this strain under the microscope. The strains carrying the phaC1 deletion (LP 4360 and LP 6060) produced a very low level of chrotonic acid during PHA determination, in agreement with their lack of PHA granules.
PHA levels in free-living cells of wild-type B. japonicum USDA 110 and phaC mutant cultures. PHA quantities (μg/ml) are normalized in relation to culture biomass (OD500). (A) PHA content of the wild type and the ΔphaC2 mutant harvested at different time points along the growth period; (B) average PHA content (±SD) of the wild type and ΔphaC2, ΔphaC1, and ΔphaC1 ΔphaC2 mutants at logarithmic growth phase (6 days old) and the same mutant strains complemented in trans with pFAJ1708 carrying either phaC1 or phaC2 wild-type genes; (C) PHA levels of wild type and different phaC single and double mutants at stationary growth phase (15 days old).
The results presented above suggested that only phaC1 is required for PHA synthesis in B. japonicum USDA 110. To corroborate this finding, we complemented each mutant with phaC1 or phaC2 in logarithmic cultures (6 days old). As expected, the absence of PHA in phaC1 deletion mutants LP 4360 and LP 6060 could be reversed by trans complementation with phaC1 but not with phaC2 (Fig. 4B), although PHA levels in the complemented cells were higher than those in the wild type. Regarding the phaC2 deletion mutant LP 6073, which at this growth state produced less PHA than the wild type, trans complementation with phaC2 restored the intracellular PHA amounts to wild-type levels. However, the complementation of LP 6073 with phaC1 significantly increased the PHA content, as in the phaC1-complemented ΔphaC1 mutants, suggesting that PhaC1 overproduction from the plasmid led to increases in PHA synthesis in the three mutants. We were unable to perform these complementation experiments at stationary phase due to the poor growth for a prolonged time in the presence of tetracycline.
In Rhodospirillum rubrum, which has three phaC paralogs, deletion of phaC1 and phaC3 leads to an increase of PhaC2 (36). To assess whether deletion of phaC2 in LP 6073 caused a differential expression of other phaC paralogs in B. japonicum that could explain the higher accumulation of PHA observed by us (Fig. 3 and 4A), we performed qRT-PCR of all phaC genes in LP 6073. As shown in Fig. 5, the expression of phaC3 in logarithmic phase was significantly higher than that in the wild type (Fig. 1). In parallel with this change in expression, PHA levels were restored to the wild-type levels in the LP 7337 ΔphaC2 ΔphaC3 double mutant (Fig. 4C), suggesting that PhaC3 is active in the LP 6073 mutant background.
Quantification of phaC paralog mRNAs in logarithmic and stationary phases of free-living B. japonicum LP 6073 (ΔphaC2) cells. mRNAs were obtained at logarithmic phase (6 days old) or stationary phase (18 days old). Relative mRNA levels are normalized as the ratio of phaC mRNA levels to constitutive sigA mRNA levels. *, deleted phaC2 gene. Results are representative of three different experiments. Each bar represents the mean ± SEM.
Survival of B. japonicum USDA 110 and ΔphaC1 and ΔphaC2 strains.To determine whether PHA levels affect cell survival, we grew the wild type and ΔphaC1 or ΔphaC2 mutant strain to late stationary phase (20 days) in Götz minimal medium, transferred the growth to carbon- and nitrogen-free Fåhraeus mineral solution (37), and continued the incubation for an additional 60 days at the same temperature and agitation. Similar to the wild type, all the B. japonicum phaC mutants analyzed showed a small but statistically nonsignificant increase in CFU numbers in this carbon- and nitrogen-free solution (see Fig. S5 in the supplemental material). This increase might simply reflect the completion of cell cycles initiated in the previous growth in culture medium using raw materials obtained by recycling of cellular components. Nevertheless, this result indicates that B. japonicum does not depend on PHA storage for its long-term survival.
PHA accumulation in ΔphaC1 or ΔphaC2 mutant bacteroids.Macroscopically, the nodules produced by all of the mutant strains appeared to be normal, and the central region of these nodules was uniformly infected with bacteria (data not shown).
The pattern of PHA granule formation in the nodule bacteroids of soybean plants at the V5 phenological stage was similar to the pattern in free-living cells that was described above. The USDA 110 and LP 6073 (ΔphaC2) bacteroids produced PHA granules (Fig. 6A and B), although those produced by LP 6073 were larger and more abundant, while the LP 4360 (ΔphaC1) and LP 6060 (ΔphaC1 ΔphaC2) nodule bacteroids were devoid of PHA granules (Fig. 6C and D). Furthermore, the analytical determination of PHA levels from crushed nodules obtained at the V6 stage corroborated these observations (Fig. 7).
Electron micrographs of soybean nodules inoculated with USDA 110 (A), the ΔphaC2 mutant (B), the ΔphaC1 mutant (C), and the ΔphaC1 ΔphaC2 mutant (D). In panels A and B, typical refringent PHA granules inside bacteroids can be observed; these granules are absent in panels C and D. Bars, 1 μm; bars in insets, 0.2 μm.
Average PHA content (±SD) from soybean nodules obtained from plants at the V6 stage that were inoculated with B. japonicum USDA 110 or the ΔphaC2, ΔphaC1, or ΔphaC1 ΔphaC2 mutant derivatives.
Accumulation of PHA enhances competitiveness for nodulation.PHA is an energy reserve which can be mobilized during infection and nodule invasion. Therefore, strains that produce PHA may have an advantage in competition for nodulation against strains that lack this polymer. To see whether the accumulation of PHA is related to the competitiveness of B. japonicum for soybean nodulation, we coinoculated late-log-phase cultures of the wild-type USDA 110 strain with each mutant strain on soybean plants. In addition, similar competition experiments were performed between the mutants.
Mutant strains LP 4360 (ΔphaC1) and LP 6060 (ΔphaC1 ΔphaC2), which cannot synthesize PHA, were less competitive for nodule occupancy than the wild-type strain (Table 2). In contrast, mutant strain LP 6073 (ΔphaC2), which synthesizes more PHA, was more competitive with both the wild-type strain and the PHA-defective mutants (Table 2), indicating a correlation between PHA accumulation and competitiveness.
Competition for nodulation of Bradyrhizobium japonicum USDA 110 and derived phaC mutants coinoculated at the indicated proportions on soybeana
Dry mass accumulation in soybean plants inoculated with different PHA mutants.Because the role of PHA in nitrogen fixation in different rhizobia is controversial (38–40) and there are no data available for B. japonicum, we decided to investigate whether the lack of PHA or excess PHA affects the accumulation of dry mass in soybean plants cultured in an N-free plant nutrient solution.
Soybeans were inoculated with USDA 110, LP 4360, LP 6060, or LP 6073 and grown for 70 days. The plants were then harvested, and the shoot dry mass, nodule dry mass, and nodule number per plant were recorded. As shown in Table S2 in the supplemental material, all of these parameters were statistically significantly similar among the plants that had been inoculated with the different bacterial strains.
DISCUSSION
The five phaC paralogs of B. japonicum USDA 110 were studied by bioinformatics analysis and site-directed mutagenesis. The bioinformatics analysis suggested that the encoded PHA synthases have a common C-terminal catalytic domain but only three of them have the N-terminal domain that might be involved in PhaC dimer stabilization. These paralogs seem to have evolved from different lineages, as PhaC1 belongs to class I PHA synthases and is similar to other rhizobial PHA synthases, PhaC2 is related to nonrhizobial unclassified PHA synthases from the betaproteobacteria, and PhaC3 belongs to well-known class I PHA synthases but is unrelated to rhizobial PHA synthases, while PhaC4 and PhaC5 belong to class III and IV PHA synthases that generally occur in taxa belonging to the Firmicutes, Cyanobacteriaceae, and Gammaproteobacteria. To our knowledge, this is the most remarkable PHA synthase diversity observed within a single bacterial genome. Despite this diversity, only phaC1 and phaC2 were significantly expressed in wild-type strain USDA 110, and in agreement with this expression profile, single mutations in these genes provoked alterations in PHA accumulation, while single mutations in phaC3, phaC4, or phaC5 did not produce appreciable changes in bacterial PHA contents.
Several other bacterial species have two or more phaC genes in their genomes. Among the best-studied species harboring multiple phaC genes, the most related to B. japonicum is Ralstonia eutropha H16. This strain has two phaC genes: phaC1 encodes an active class I PHA synthase that is more closely related to B. japonicum PhaC3 than to B. japonicum PhaC1, and phaC2 encodes an unclassified PHA synthase that belongs to the same group as B. japonicum PhaC2, has no catalytic activity in vitro, and is unable to complement a phaC1 mutation in R. eutropha. In addition, R. eutropha H16 phaC2 is not transcribed (41). In B. japonicum, the ΔphaC1 mutant was unable to synthesize PHA, while the ΔphaC2 mutant accumulated more PHA than the wild-type strain, both in liquid culture at stationary phase and in mature soybean nodules. Hence, similar to R. eutropha PhaC2, B. japonicum PhaC2 may be a catalytically inactive PHA synthase or a negative regulator of PHA biosynthesis, as has been suggested for Rhodospirillum rubrum PhaC1 (33). In B. japonicum USDA 110, phaC2 was transcribed at similar or even higher levels than phaC1, suggesting that this paralog may play a role in the PHA metabolism of this species.
The active PHA synthase of R. eutropha is homodimeric (42), while in R. rubrum, homo- and heterodimers are both active in vitro (33). Therefore, it might be speculated that in wild-type B. japonicum free-living and bacteroid cells, catalytically active PhaC12 and catalytically inactive PhaC22 homodimers, as well as a PhaC1-PhaC2 heterodimer that exhibits intermediate catalytic activity, are formed. Therefore, the combined amounts of PhaC1 and PhaC2 might determine the global PHA synthase activity in the cell, and as suggested by our complementation experiments, a relative increase of PhaC1 might lead to higher PHA accumulation. This hypothesis could explain why the phaC1 mutant was unable to produce PHA and the phaC2 mutant produced more polymer than the wild-type strain at stationary phase. In addition, at the onset of PHA synthesis, PhaC is soluble in the cytoplasm, but as the PHA granule grows, PhaC molecules bind to the surface of the granule, where they express their full catalytic activity (42). Since the PHA granules tend to be spherical, their surface/volume ratio diminishes as the PHA granules grow. Therefore, if the early increase in phaC2 transcript levels observed in this study results in a later increase in PhaC2 protein levels, this paralog would soon occupy the limited PhaC sites on the granule surface, thus disturbing the access of PhaC1, which might lead to a diminished biosynthetic activity, or it may preclude the access of a PHA depolymerase enzyme. The interplay between PhaC1 and PhaC2 might be relevant in nodules, where phaC2 expression is stimulated 6-fold by FixK2 (a transcription factor enhancer that is stimulated by low O2 levels) (10). Surprisingly, phaC3, which was poorly expressed in the wild type, was upregulated at logarithmic growth phase in the ΔphaC2 mutant. Such a differential accumulation of a PhaC paralog as a consequence of mutation in another phaC paralog was already observed recently in R. rubrum (36). We have no explanation for this emergent property, but mutation in phaC3 in the ΔphaC2 background restored the PHA levels to those of the wild type, indicating that in this background PhaC3 was in part responsible for PHA synthesis, in addition to PhaC1. Nevertheless, the results obtained with the ΔphaC1 ΔphaC2 double mutant as well as in the complementation experiments clearly indicate that PhaC1 is required for PHA synthesis, a function that cannot be fulfilled by PhaC3 alone. Further in vitro experiments with purified PhaC1, PhaC2, and PhaC3 are required to shed light on the functionality and association of these three proteins.
The competitiveness of the phaC mutants for nodulation was correlated with their PHA levels, in agreement with previous reports of lower competitiveness of PHA-deficient E. meliloti mutants in alfalfa (38, 40). The invasion of plant tissues toward the nodule requires the cells to traverse a stressful environment (43–46), and it was observed that B. japonicum cells that are deficient in α-ketoglutarate dehydrogenase produce nodule primordia that abort at higher rates. This observation suggests that energy must be efficiently produced by the rhizobia to complete root infection (47). Because PHA is a carbon- and energy-reserve polymer, higher levels of this polymer may help the cells to complete infection and nodule invasion in this metabolically unfavorable situation (5), thus explaining the observed correlation between PHA levels and nodulation competitiveness.
In contrast to the effects on competition for nodulation, PHA levels had no effect on the production of shoot or nodule biomass. These results are similar to those that have been obtained in other determinate nodule symbiotic systems. Soybean plants that have been nodulated by B. japonicum USDA 110 isocitrate dehydrogenase mutants, which have increased PHA levels in their nodules, produce similar shoot dry weight and acetylene reduction activity as those that have been nodulated by the wild type (48). Moreover, Phaseolus vulgaris plants that have been inoculated with a phaC-defective strain of Rhizobium leguminosarum 8401 carrying the pRL2JI (Rhizobium phaseoli) pSym plasmid produce similar shoot dry weight as plants that have been inoculated with the wild-type strain (49). In contrast, other studies reported increased nitrogen fixation activity and biomass production in P. vulgaris plants that had been inoculated with a phaC-defective mutant of Rhizobium etli CFN42 (39). These authors suggested that PHA synthesis may compete with nitrogenase for the reducing power in the nodule. If this hypothesis is true, these apparently contradictory results may be explained because, unlike R. etli CFN42, B. japonicum USDA 110 possesses the hydrogenase uptake (hup) system, which recycles the hydrogen produced in the nitrogenase reaction, thus supplying extra reducing power (50, 51). Although the hup system has been detected in several R. leguminosarum strains, it is unknown whether the 8401 strain possesses it.
ACKNOWLEDGMENTS
We thank Paula Giménez, Silvana Tongiani, Abel Bortolameotti, and Bernabé Castillo for their excellent technical assistance. We also thank Carolina Jaquenod De Giusti for advice and assistance in the qRT-PCR experiments and Susana Jurado (Servicio Central de Microscopía, Facultad de Ciencias Veterinarias, Universidad Nacional de La Plata, La Plata, Argentina) for assistance with electron microscopy observations.
This work was supported by the Agencia Nacional de Promoción de la Investigación Científica y Tecnológica, Argentina. J.P.-G. is a researcher at the Universidad Nacional de La Plata, La Plata, Argentina. E.J.M., G.P., A.R.L., and J.I.Q. are members of the scientific career of CONICET, Argentina.
FOOTNOTES
- Received 10 December 2012.
- Accepted 1 May 2013.
- Accepted manuscript posted online 10 May 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.02203-12.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
