|Home | About | Journals | Submit | Contact Us | Français|
The symbiotic nitrogen-fixing bacterium Sinorhizobium meliloti 1021 encodes only one predicted aconitase (AcnA) in its genome. AcnA has a significant degree of similarity with other bacterial aconitases that behave as dual proteins: enzymes and posttranscriptional regulators of gene expression. Similar to the case with these bacterial aconitases, AcnA activity was reversibly labile and was regained upon reconstitution with reduced iron. The aconitase promoter was active in root nodules. acnA mutants grew very poorly, had secondary mutations, and were quickly outgrown by pseudorevertants. The acnA gene was stably interrupted in a citrate synthase (gltA) null background, indicating that the intracellular accumulation of citrate may be deleterious for survival of strain 1021. No aconitase activity was detected in this mutant, suggesting that the acnA gene encodes the only functional aconitase of strain 1021. To uncover a function of AcnA beyond its catalytic role in the tricarboxylic acid cycle pathway, the gltA acnA double mutant was compared with the gltA single mutant for differences in motility, resistance to oxidative stress, nodulation, and growth on different substrates. However, no differences in any of these characteristics were found.
Aconitases (EC 188.8.131.52) are key enzymes in the tricarboxylic acid (TCA) cycle in both prokaryotes and eukaryotes, catalyzing the reversible isomerization of citrate into isocitrate via the intermediate cis-aconitate. The active site contains a [4Fe-4S] cluster that is required for enzymatic activity. One of the four iron atoms in the cluster is easily lost under conditions of low iron availability. In mammalian cells, loss of this labile iron atom produces a cytosolic form of aconitase that has no catalytic activity but acquires the ability to bind to mRNA hairpin structures, affecting the translation of the corresponding mRNA (3). The ability of aconitases to bind mRNA and/or alter gene expression in an iron-dependent manner has also been found in bacteria, such as Escherichia coli (38, 40), Salmonella enterica (39), Bacillus subtilis (1), Mycobacterium tuberculosis (2), Xanthomonas campestris (44), and Pseudomonas aeruginosa (36).
Among symbiotic nitrogen-fixing rhizobia, only aconitases in Bradyrhizobium japonicum have been studied (41). The B. japonicum acnA gene was cloned and mutated. The acnA mutant showed impaired growth despite retaining 30% of the total aconitase activity. In Sinorhizobium meliloti 1021, acnA is the only aconitase-encoding gene predicted to be present in the genome (13). No acnA mutants have been described for S. meliloti. However, in S. meliloti strain 104A14, mutations in the metabolically upstream citrate synthase gene (gltA) or in the metabolically downstream isocitrate dehydrogenase gene (icd) are viable and produce ineffective nodules in plants (16, 25, 27). Interestingly, in the icd background, spontaneous mutants lacking citrate synthase activity appeared (25). Similar viable phenotypes have been obtained for other enzymes involved in the TCA cycle, such as 2-oxoglutarate dehydrogenase and succinate dehydrogenase, as well as malic enzymes (8-11, 14).
In this work, we demonstrated that the acnA gene of S. meliloti encoded the only functional aconitase in the genome. We found that under the conditions used, stable acnA mutants could be obtained only in a background where gltA was also mutated. The acnA gltA double mutant did not show differences in various phenotypes compared to the gltA single mutant.
Strains and plasmids used in this work are indicated in Table Table1.1. E. coli strains were grown at 37°C in LB medium. S. meliloti strains were grown at 30°C in TY (4), yeast extract-mannitol (YMB) (37), or mannitol minimal medium with ammonia (MinManNH4) (37). l-Arabinose was added to a final concentration of 0.1% (for MinManNH4 medium) or 0.25% (for YMB medium) for routine growth of gltA and acnA derivatives. Minimal salts medium identical to MinManNH4 but without mannitol or NH4 was supplemented with nitrogen and carbon sources as specified.
The acnA open reading frame was amplified from plasmid pESMc03846 (34) with primers acnORF-Fwd (5′-ATCTCTAGACATATGTCCAAATCCCTAGACAG-3′ [bases added to create XbaI and NdeI adaptor sites are shown in italics; the NdeI site is underlined]) and acnORF-Rev (5′-ATCGGATCCTCAGGCGGCGAGATCGCGCA-3′ [BamHI adaptor indicated in italics]), digested with NdeI and BamHI, and cloned in frame into pET-14b (Novagen). The resulting plasmid (pET14acnA) was used for expression of His-tagged AcnA (His-AcnA) in E. coli BL21(DE3)(pLysS) in LB medium at 37°C, and His-AcnA was purified from extracts with Ni-nitrilotriacetic acid agarose (Qiagen).
Direct activity assays were performed for the recombinant protein by measuring the production of cis-aconitate from isocitrate at 240 nm as described by Kennedy et al. (22). Aconitase activity in crude extracts was determined by the coupled assay (32). One unit is defined as 1 μmol of cis-aconitate produced per minute. Isocitrate dehydrogenase activity was measured by means of detecting the appearance of NADPH at 340 nm in a buffer containing Tris-HCl (50 mM; pH 7.4), MnCl2 (0.6 mM), NADP+ (0.2 mM), and sodium isocitrate (1 mM). One unit is defined as 1 μmol of NADPH produced per minute. Protein concentrations in crude extracts were determined using a bicinchoninic acid kit (Sigma-Aldrich).
Plasmid pK18acnA was constructed by inserting a 1.35-kb HindIII/PstI fragment of the acnA open reading frame (corresponding to positions 3512129 to 3513478 in the strain 1021 genome) from pESMc03846 (34) into the HindIII/PstI sites in pK18mobsacB (33). Plasmids pFL237, pFL3738, and pFL2240 (5) were kindly provided by T. Finan.
Suicide plasmids in the donor strain DH5α or XL-1 Blue were mobilized into S. meliloti strains by triparental mating using DH5α(pRK2013) as a helper strain (12). When performing the selection of transconjugants in minimal medium, the suicide plasmids were mobilized by biparental mating using strain S17-1 (35). Matings with negative results (less than 10−7 CFU per recipient cell) were repeated at least twice, and positive controls with other suicide plasmids were performed in all cases. S. meliloti strain 2AX with pESMc03846 inserted into the acnA gene showed kanamycin resistance (100 μg/μl) but neomycin (Nm) sensitivity (50 μg/μl), which allowed selection of pK18acnA insertion by Nm selection.
Genomic DNA from S. meliloti strains was digested with BamHI and fractionated by electrophoresis through a 1% agarose gel. DNA was then transferred to a nylon membrane and hybridized at 65°C with a biotinylated probe (NEBlot Phototope kit; New England Biolabs) consisting of a 927-bp XhoI fragment of the acnA open reading frame (ORF) from pET14acnA (see below; corresponding to positions 3511499 to 3512425 in the genome). Hybridization, washes, and detection with a Phototope-Star detection kit for nucleic acids (New England Biolabs) were performed according to the instructions of the manufacturer.
Deletion of the gltA region was done by a modification of the FLP recombination target (FRT) recombination strategy described by House et al. (18), where pMK2030 derivatives (Tetr) (19) substitute for pMK2017 derivatives and pMK2016-2 derivatives (Spcr) substitute for pMK2016 derivatives. The pMK2016-2 plasmid is similar to pMK2016 except that the orientations of the FRT sites are reversed (M. W. Mortimer and M. L. Kahn, unpublished data). pRG1SMc02089 is a pMK2030 derivative in which the SMc02089 ORF replaces the ccdA/Cmr cassette of pMK2030 (19). Plasmid pD2SMc02086, in which the SMc02086 ORF replaces the ccdA/Cmr cassette from pMK2016-2, was constructed by in vivo lambda integrase-mediated transfer from pESMc02086 (34) to pMK2016-2, as described by House et al. (18) for obtaining pMK2016 derivatives. After integration of pRG1SMc02089 and pD2SMc02086, the region between SMc02086 and SMc02089 (including gltA [SMc02087] and SMc02088) was then deleted as described previously (18).
Plant assays were performed as described by Platero et al. (29). Strain ARG1 was obtained by integration of pRG1SMc03846 into strain 1021 as described previously (19). In this strain, the β-glucuronidase gene is under the control of the acnA promoter. Alfalfa (Medicago sativa cv. creola) seeds were inoculated with this strain as described previously (29). β-Glucuronidase staining assays were performed as described by Jefferson et al. (20).
AcnA of strain 1021 is highly similar to other bacterial aconitases known to have an iron-dependent activity. Therefore, we hypothesized that 1021 AcnA activity would be reversibly affected by iron availability. To test this hypothesis, we expressed and purified an AcnA N-terminal His tag fusion protein, His-AcnA.
In two independent purifications, the initial activities obtained using a direct aconitase assay were 13 and 28 U mg−1. After reconstitution with 1 mM Fe(NH4)2(SO4)2 and 5 mM dithiothreitol for 20 min, the corresponding activities rose to 32 and 38 U mg−1, indicating that His-AcnA possesses a labile [4Fe-4S] cluster. The specific activity of reconstituted purified His-AcnA was similar to previously reported data for purified aconitases from various organisms (23, 31, 42).
To determine if AcnA plays a role in gene regulation, the acnA gene was interrupted with the suicide plasmid pK18acnA. This plasmid carries a 1.35-kb HindIII/PstI fragment internal to the acnA ORF (Fig. (Fig.1).1). After 8 days, small colonies became visible on YMB (37) plus 0.25% arabinose but not in other media tested, including MinManNH4 (37) plus 0.1% l-arabinose, YMB, TY (4) plus 0.25% l-arabinose, and TY plus 0.25% l-glutamate (Table (Table2;2; also data not shown). In S. meliloti, arabinose can be transported and converted to 2-oxoglutarate, thus providing a substrate needed for glutamate biosynthesis that would not be produced in mutants defective in the decarboxylating leg of the TCA cycle (30). Some of these colonies were able to grow when restreaked, but growth was extremely slow. They were glutamate auxotrophs, as expected for aconitase mutants. Disruption of acnA with pK18acnA was confirmed by Southern blot analysis in one such colony (AK18U) (data not shown). To confirm that these strains were bona fide acnA mutants, the excision of pK18acnA was selected with 5% sucrose. However, the Nm-sensitive glutamate prototrophs that we obtained showed reduced growth on YMB and TY media compared to the wild type, suggesting that secondary mutations not involving acnA were still present after loss of pK18acnA. Therefore, AK18U was in fact a mutant containing a pseudoreversion.
Mutagenesis was also attempted with the suicide plasmid pFL237, which contains a 0.94-kb fragment internal to the acnA ORF, and with the suicide plasmids pFL3738 and pFL2240, which leave an intact copy of the acnA gene after recombination, as controls (5) (Fig. (Fig.1).1). Like pK18acnA, pFL237 produced only very small pseudorevertants after 8 days, while pFL3738 and pFL2240 gave rise to normal-size transconjugants after 4 days (Table (Table2).2). Insertions of plasmids pFL237 and pFL3738 into the acnA gene were confirmed by Southern blotting (data not shown).
To show that the acnA locus was not a “cold spot” for recombination, the suicide plasmid pESMc03846, containing the complete acnA ORF, was used. After recombination, this plasmid creates a duplication of the acnA ORF. One copy would lie downstream of the native promoter and should therefore be functional. The other copy would lack the promoter region and therefore would be nonfunctional. The resulting transconjugants were named 2AX and were subjected to mutagenesis by conjugation with the suicide plasmid pK18acnA. Ten of these transconjugants were analyzed by Southern blotting to determine in which copy of acnA pK18acnA was inserted. Insertion within the nonfunctional copy of the gene was confirmed by Southern blotting in 9 out of 10 colonies, and in the remaining colony, the insertion occurred within the pESMc03846 plasmid backbone (data not shown). No insertions were present in the functional copy of acnA (Table (Table22).
Taken together, these observations showed that under the culture conditions used, the acnA mutation could not be maintained free of compensating secondary mutations. This is reminiscent of mutations in other TCA cycle genes, such as mdh, sucC, and sucA, which cannot be interrupted (11) without severe growth defects.
In S. meliloti strain 104A14, spontaneous mutants lacking citrate synthase activity appeared when isocitrate dehydrogenase was mutated (25). Accumulation of citrate has been proposed to be the cause for the low growth rate that aconitase mutants usually display in E. coli, Streptomyces coelicolor, and B. subtilis. In these species, either intentional or spontaneous mutations arising in the citrate synthase genes helped overcome the growth retardation and other defects of the aconitase mutants (6, 15, 43). To test whether citrate accumulation was responsible for the instability of acnA mutants in strain 1021, strain GD1, carrying a deletion of the region comprising gltA and the neighboring putative ORF SMc02088, was made using a modification of the FRT-mediated deletion method (18). GD1 was a glutamate auxotroph. Deletion of acnA was also attempted using a similar strategy with negative results. However, acnA was easily interrupted in the GD1 background, and transconjugant colonies appeared in YMB with or without 0.25% l-arabinose after 4 days at 30°C (Table (Table2).2). Plasmid pFL237 or pK18acnA was used in this mutagenesis, and the resulting mutant strain, GDA237 or GDAK18, respectively, was confirmed by Southern blotting (data not shown). As expected, no aconitase activity was detected in coupled assays (32) of GDA237 extracts, while isocitrate dehydrogenase activity remained unchanged (Table (Table3).3). This observation supports the conclusion that acnA encodes the only aconitase in the 1021 genome, as predicted from in silico analysis, and partially accounts for the severity of acnA mutations in 1021. Aconitase mutants found in other genera, such as B. japonicum (41), the E. coli acnA acnB double mutant (15), and S. coelicolor (43), either retained some residual aconitase activity or were highly unstable. Surprisingly, the GD1 mutant strain had only about 50% of the aconitase activity present in the wild-type strain, which could be related to diminished gene expression in the absence of citrate production.
The fact that acnA mutations could be stably maintained in a gltA deletion background strongly suggested that accumulation of citrate hinders growth in the absence of aconitase activity. Accumulation of intracellular citrate is deleterious in a number of ways, which include chelation of divalent cations and acidification of the intracellular milieu (24).
Interestingly, icd mutants of S. meliloti strain 104A14 do not require secondary mutations for growth (25). The main difference between icd and acnA mutants may lie in the fact that the loss of icd blocks the TCA cycle but leaves a viable glyoxylate pathway through which citrate can be metabolized. In contrast, lesions in the acnA gene block both pathways. Furthermore, isocitrate lyase activity, the first reaction that is specific to the glyoxylate pathway, is substantially increased in 104A14 icd mutants (25).
To the best of our knowledge, the only aconitase mutant previously reported for rhizobia is the acnA mutant of the phylogenetically distant B. japonicum 110spc4 (41), which also encodes only one aconitase in its genome (21). These authors described that they were initially unable to obtain acnA insertion mutants and suspected the gene was essential, but they were eventually able to obtain two mutant strains. As the authors themselves concede, it is possible that their approach forced the selection of pseudorevertants.
Growth of gltA and gltA acnA double mutants was evaluated using several solid media. All these mutant strains were able to grow on MinManNH4 plus 0.1% l-arabinose, YMB, and YMB plus 0.25% l-arabinose. Growth of these strains was as good as that of the wild-type strain in YMB plus 0.25% l-arabinose and MinManNH4 plus 0.1% l-arabinose after 5 days, although poorer growth was evident at earlier time points. As expected, double mutants were also glutamate auxotrophs in MinManNH4 medium. l-Arabinose, glutamate, and to a lesser extent 2-oxoglutarate were able to rescue their growth under these conditions. These results were similar to those reported by Mortimer et al. (27) for gltA mutants of 104A14. The lack of differences in growth phenotypes between gltA and gltA acnA mutants in the media tested indicates that acnA does not appear to have a significant effect on the metabolism of the cell beyond its function in the TCA cycle.
Strikingly, none of the gltA mutant strains was able to grow in TY, and growth was extremely poor in TY plus 0.25% l-arabinose or TY plus 0.25% glutamate. The reason behind this growth characteristic of gltA mutants is unclear. However, it does not appear to be a characteristic of all 1021 glutamate auxotrophs, since a Tn5-induced glutamate synthase mutant was successfully isolated in regular TY medium (L. Hannibal and F. Noya, unpublished observation).
In order to uncover a possible role of AcnA as an iron sensor and a regulator of iron homeostasis in S. meliloti, we evaluated cell growth in iron-depleted liquid medium, YMB plus 0.25% l-arabinose in the presence of 100 μM ethylenediamine-N,N′-bis(2-hydroxyphenylacetic acid) (EDDHA). Although both GD1 and GDA237 grew poorly in the presence of EDDHA compared to the wild type, the two mutants showed similar growth (data not shown). The poor growth may be due to the fact that the biosynthesis of rhizobactin 1021, a citrate-based hydroxamate siderophore, is likely to be affected in the gltA mutant background. To evaluate the utilization of rhizobactin and other iron sources, we performed bioassays with 1021, GD1, and GDA237 in the presence of purified rhizobactin 1021, hemin, and FeCl3, as described previously (28). No significant differences in the growth halos formed by the three strains for each of the tested compounds were observed (Table (Table4),4), demonstrating that the internalization of siderophores, hemin, or iron was not compromised in these mutants. These data indicate that 1021 AcnA does not affect the internalization and use of iron compounds, limiting a potential role of AcnA in iron homeostasis.
In other species, aconitase is involved in the regulation of motility and in the response to oxidative stress (39, 40). To determine if aconitase plays a similar nonmetabolic role in S. meliloti, we investigated the growth of acnA and acnA gltA double mutants in the absence or presence of 300 μM paraquat, an agent that generates superoxide via redox cycling. After 48 h, growth under superoxide stress was similar in the wild type, GD1, and GDA237 (data not shown). Likewise, no significant defects in motility could be detected in any mutant strain by observing the halo surrounding streaks of strains in 0.3% agar medium (data not shown).
Finally, we studied the symbiotic phenotype of the gltA mutant and gltA acnA double mutant strains. As reported for 104A14 (27), inoculation of plants with GD1 resulted in delayed nodulation. The nodules induced by 1021 appeared 8 days postinoculation, whereas GD1 and GDA237 developed nodules 5 and 6 days later, respectively. GD1 and GDA237 developed white and ineffective nodules (data not shown). The use of a reporter construct acnA-GUS gene fusion in strain ARG1 (1021::pRG1SMc03846) (19), which is AcnA+, showed that the acnA promoter was active in the nodule, as evidenced by β-glucuronidase staining (Fig. (Fig.2).2). However, whether acnA is necessary for an efficient symbiosis with alfalfa remains to be confirmed, since both gltA and acnA gltA mutants induced delayed, white, ineffective nodules.
Thus, we were unable to find a function for 1021 AcnA beyond its expected catalytic role in the TCA cycle by the methods employed. Our results do not exclude AcnA from having a role in posttranscriptional gene regulation, although, should it exist, it must affect functions not yet described for other organisms (1, 3, 38, 39).
We thank T. Finan for kindly providing plasmids used in this study and R. Platero, F. Battistoni, F. Rosconi, V. Amarelle, J. Humann, S. Yurgel, and M. Mortimer for helpful discussions.
This work was supported by research grants from TWAS (03-403 RG/BIO/LA), Fondo Clemente Estable (FCE/9014), PEDECIBA, and Comisión Sectorial de Investigación Científica of the Universidad de la República Oriental del Uruguay (CSIC).
Published ahead of print on 9 October 2009.