|Home | About | Journals | Submit | Contact Us | Français|
The Sinorhizobium meliloti Rm1021ΔglnD-sm2 mutant, which is predicted to make a GlnD nitrogen sensor protein truncated at its amino terminus, fixes nitrogen in symbiosis with alfalfa, but the plants cannot use this nitrogen for growth (S. N. Yurgel and M. L. Kahn, Proc. Natl. Acad. Sci. U. S. A. 105:18958-18963, 2008). The mutant also has a generalized nitrogen stress response (NSR) defect. These results suggest a connection between GlnD, symbiotic metabolism, and the NSR, but the nature of this connection is unknown. In many bacteria, GlnD modifies the PII proteins, GlnB and GlnK, as it transduces a measurement of bacterial nitrogen status to a cellular response. We have now constructed and analyzed Rm1021 mutants missing GlnB, GlnK, or both proteins. Rm1021ΔglnKΔglnB was much more defective in its NSR than either single mutant, suggesting that GlnB and GlnK overlap in regulating the NSR in free-living Rm1021. The single mutants and the double mutant all formed an effective symbiosis, indicating that symbiotic nitrogen exchange could occur without the need for either GlnB or GlnK. N-terminal truncation of the GlnD protein interfered with PII protein modification in vitro, suggesting either that unmodified PII proteins were responsible for the glnD mutant's ineffective phenotype or that connecting GlnD and appropriate symbiotic behavior does not require the PII proteins.
The Gram-negative alphaproteobacterium Sinorhizobium meliloti is a soil bacterium able to form a symbiotic association with legumes such as Medicago, Melilotus, and Trigonella spp. The bacteria elicit the formation of specialized nodules on the roots of the host plant, in which they undergo differentiation into a microaerophilic bacteroid state (19, 30). Bacteroids reduce atmospheric dinitrogen (N2) to ammonia, which is supplied to the host plant in exchange for carbon compounds. A key feature of symbiotic N2 fixation is the uncoupling of N2 fixation from bacterial nitrogen stress response (NSR) metabolism so that bacteria generate “excess” ammonia and release nitrogen to the plant. The primary sensor in many bacterial NSR systems is the GlnD protein, a large multidomain uridylyltransferase and uridylyl cleavage enzyme whose primary uridylylation targets are the PII proteins, GlnB and GlnK (Fig. (Fig.1,1, mechanism 1). The uridylylation state of these proteins is regulated by α-ketoglutarate and glutamine (24, 25, 26) so that when GlnD senses that there is not enough nitrogen available to synthesize adequate glutamine, it activates the NSR by uridylylating the PII proteins.
In enteric bacteria grown on good nitrogen sources, GlnK is not expressed, while GlnB is expressed constitutively but is not modified. The ammonium-assimilatory enzyme glutamine synthetase (GS) is expressed at a basal level, but since unmodified GlnB stimulates adenylylation, this GS is present in the cells as the inactive adenylylated form. Unmodified PII proteins also lead to the dephosphorylation of NtrC, a response regulator transcriptional activator, and prevent the transcription of genes, like glnA, that are involved in nitrogen acquisition (3, 29, 33). When enteric bacteria become nitrogen limited, GlnB is rapidly uridylylated, resulting in NtrC phosphorylation and an increase in GS expression (7). At the same time, GlnB-UMP stimulates deadenylylation and activation of GS enzymatic activity. In steady-state nitrogen-limited growth, GlnK becomes more abundant and, due to its ability to form heterotrimers with GlnB, plays an important role in nitrogen stress response regulation (18, 44). Both GlnB-UMP and GlnK-UMP proteins can interact with the GlnE adenyltransferase to deadenylylate GS, stimulating its activity. However, several observations suggest that GlnB and GlnK differ in their functions (6, 8, 33). For example, although both PII proteins participate in regulating NtrC phosphorylation, only GlnB can repress NtrC-dependent GS synthesis (6). Additionally, GlnK directly inhibits ammonium uptake by binding to the AmtB ammonium transporter (3, 32), an interaction that is blocked by the uridylylation of GlnK that occurs when nitrogen is limited (3, 15, 21, 25).
In Rhizobium leguminosarum, GlnB was found to be essential for dephosphorylation of NtrC, resulting in constitutive GSII expression in a glnB mutant strain grown in sufficient nitrogen (2). In contrast to the case for enteric bacteria and R. leguminosarum, uridylylated GlnB is required in S. meliloti strain GMI708 for NtrB-dependent activation of NtrC by phosphorylation (4). Additionally, in the absence of GlnB, GSI is constitutively deadenylylated (4), while in enteric bacteria, glutamine-stimulated GlnE adenylyltransferase (ATase) activity is sufficient for GS adenylylation in nitrogen-limited cells without either GlnK or GlnB (6, 33). Thus, while the elements of the NSR circuit are similar, the actual operations of the circuit may differ in important ways.
Our recent data showed that mutations altering GlnD, the major bacterial nitrogen sensor protein, can produce a symbiosis that fixes nitrogen but does not stimulate plant growth (Nif+ Eff−) (45). This implies that some genes regulated by GlnD are needed for effective symbiosis. Our analysis of these S. meliloti glnD mutants also showed that these mutants also disabled many parts of the S. meliloti NSR. However, an isogenic mutant missing one of the two PII proteins, GlnB, did not exhibit significant defects in NSR regulation or in symbiosis. A possible explanation for these results with the glnB mutant was that in the absence of GlnB, the second PII protein, GlnK, might substitute for GlnB to some extent. Another explanation could be that the altered GlnD protein acts on the symbiosis in a way that does not require GlnB or GlnK as intermediaries. The main goal of the research described in this paper was to analyze the role of PII proteins in regulating nitrogen metabolism signaling in both free-living cells and S. meliloti Rm1021 symbiotic bacteroids.
The bacterial strains and plasmids used in the study are listed in Table Table1.1. S. meliloti strains were grown at 30°C on Luria-Bertani (LB) medium (36), on minimal (Min) mannitol-NH4+ (MM-NH4+) or on YMB medium (42). Escherichia coli strains were grown at 37°C on LB medium. Antibiotics for S. meliloti were added at 200 μg/ml (streptomycin), 200 μg/ml (neomycin), or 10 μg/ml (tetracycline [Tc]). For E. coli, antibiotics were added at 10 μg/ml (tetracycline) or 40 μg/ml (kanamycin [Km]). Sigma-Aldrich agar was used for medium preparation.
DNA manipulations were carried out according to standard procedures (36). The deletion mutants were constructed using an insertion/excision strategy (47). To construct S. meliloti mutants, PCR was used to amplify 0.5-kb and 0.6-kb chromosomal regions flanking regions to be deleted using the primers listed in Table Table2.2. The mutations were confirmed by sequencing.
PII uridylylation was examined according to the method of Colonna-Romano et al. (14) with some modifications. Strains were grown in Min-mannitol medium containing 0.04% glutamate as a nitrogen source. For every strain, three aliquots were taken in the exponential growth phase, incubated in 270 μl of 50 mM 2-methyl-imidazole buffer, pH 7.6, containing 200 mM KCl, 10 mM MgCl2, 0.1 mM ATP, and 30 μg/ml of cetyl trimethylammonium bromide (CTAB), and incubated on ice for 20 min. Labeling at 30°C was started by adding 1 μl of [α-32P]UTP (800 Ci/mmol) and 30 μl of 180 mM α-ketoglutarate or 200 mM glutamine. All samples were incubated for 2 h, reactions were stopped on ice, and cells were harvested and lysed by sonication or by adding gel loading buffer containing SDS. After proteins were separated on a 15% SDS-PAGE gel, the gel was dried and radioactivity was detected by autoradiography.
A translational gusA reporter gene was inserted into pFAJ1700, a stable plasmid RK2 derivative (17). The gusA gene, lacking a start codon, was amplified by PCR from pMK2030 (22) by using primers GUS2-F-2 and GUS2-R-2 (Table (Table2).2). The PCR fragment was digested with HindIII-XbaI and cloned into HindIII-XbaI-digested pFAJ1700, resulting in plasmid pJ1700G (Table (Table1).1). A 531-bp S. meliloti glnII promoter fragment was amplified from Rm1021 chromosomal DNA by PCR using primers glnII-P-F1 and glnII-P-R (Table (Table2).2). The PCR fragment was digested with BamHI and EcoRI restriction enzymes and cloned into BamHI/EcoRI-digested pK19mob, resulting in plasmid pK19mob-pGlnII. The fragment was then cloned into BamHI/EcoRI-digested pJ1700G, resulting in the plasmid pF1700G-pGlnII (Table (Table1).1). A similar approach was used for construction of translational fusions to the glnK and glnB promoters. Primers used for these constructions are listed in Table Table22.
The strains were grown in 3 ml MM-NH4+ medium at 30°C for 48 h. Cells were diluted 1/20 into fresh Min-mannitol medium containing 0.1% NH4+ or 0.04% glutamate and grown at 30°C for 16 h. β-d-Glucuronidase assays and estimation of β-d-glucuronidase activities (in units per cell optical density at 600 nm [OD600]) were carried out according to the protocol devised by Miller (28).
To evaluate growth on various carbon substrates, S. meliloti strains were first grown on YMB plates for 48 h. Cells were suspended in Min-salt solution (minimal medium without either a carbon or nitrogen source) to an OD600 of 0.5. Two microliters of the cell suspension was added to 200 μl MM medium that contained 0.2 g/liter of the indicated nitrogen compound. Growth at 30°C was monitored for 48 h by measuring the absorbance at 600 nm every 30 min by using a SPECTRAmax 250 microplate spectrophotometer system (Molecular Devices).
Alternatively, the cells were suspended in Min-salt solution to an OD600 of 0.5, and the cell suspensions were diluted with Min-salt solution 1 to 10, 1 to 100, 1 to 1,000, 1 to 10,000, and 1 to 100,000 in a 96-well microplate. Aliquots of these were then transferred using a sterile bolt replicator onto plates containing solid MM medium to which 0.2 g/liter of the indicated nitrogen compound had been added. After 3 to 7 days, the sizes of colonies were scored. The set of nitrogen sources tested included alanine, proline, histidine, glycine, γ-aminobutyric acid (GABA), glutamate, glutamine, ammonium, and nitrate. The nitrogen content of the agar was low, as indicated by the lack of growth unless a suitable nitrogen source had been added.
Crude extracts were prepared as described previously (45) from cultures grown for 24 h on MM-NH4+ (high-nitrogen), MM-glutamine (intermediate-nitrogen), or MM-glutamate (low-nitrogen) medium. Protein concentrations were determined by using a modified Bradford assay (Bio-Rad). Total protein (~50 μg) was separated on 10% SDS-PAGE gels (Invitrogen) and transferred to nitrocellulose, and the separated proteins were visualized immunologically by using SuperSignal West Pico chemiluminescent substrate (Thermo Scientific) (10). Rhodospirillum rubrum anti-GSI (27, 49) and S. meliloti anti-GSII (40) antisera were used to visualize the proteins.
Total GS activity was measured from cells grown on MM medium containing a good N source (0.5% NH4Cl) or a poor one (0.04% sodium glutamate) by using the incorporation of hydroxylamine into γ-glutamylhydroxamate via the transferase assay, which was carried out in the presence of manganese and is an index of the total amount of GSI (42).
Alfalfa (Medicago sativa cv. Champ) was used for all nodulation studies. The plant tests were performed as described previously (46). Four to five weeks after inoculation, plants were harvested; shoot dry mass and root nodule formation were examined.
To examine the role of the GlnK protein in S. meliloti nitrogen metabolism regulation, we generated an in-frame deletion of the entire glnK gene in strain Rm1021. The mutant, Rm1021ΔglnK, had a free-living phenotype that was very similar to that of Rm1021 and grew well on all nitrogen sources listed in Materials and Methods. An example of the growth of the mutant on glutamine and proline as the nitrogen source is presented in Fig. Fig.2.2. A similar phenotype was observed in Rm1021ΔglnB mutants (45). These data indicate that deletion of either the GlnK PII protein or the GlnB PII protein does not substantially affect NSR regulation in free-living cells.
In enteric bacteria, GlnK is crucial for the ability of cells to survive nitrogen starvation and resume rapid growth when fed ammonia (11). This was not the case in S. meliloti; Rm1021 lacking GlnK survived nitrogen starvation as well as wild-type Rm1021. After 72 h of nitrogen starvation, the growth of Rm1021ΔglnK on ammonium plates was slightly decreased, but the effect of nitrogen starvation was far less dramatic than was reported for E. coli (data not shown).
To examine the NSR and symbiotic phenotypes of a mutant missing both PII proteins, we performed an in-frame deletion of the glnB gene in Rm1021ΔglnK by integrating into the chromosome a plasmid that conferred sucrose sensitivity and contained the ΔglnB mutation. Selection for sucrose resistance in the merodiploid would lead to recombinational exchange of the mutant glnB allele into some of the resistant isolates. Sucrose-resistant colonies were visible after 3 to 4 days of growth on YMB agar supplemented with 5% sucrose (47), but PCR analysis of several dozen of these colonies showed that none contained the ΔglnB mutation. However, tiny colonies began to appear 7 to 10 days after the original cells were plated on sucrose medium. PCR analysis showed that these slow-growing cells were missing both glnK and glnB. Three independent colonies with double glnK and glnB deletions (strain Rm1021ΔglnKΔglnB, lines 1, 3, and 5) were selected for future experiments. We tested all three lines of Rm1021ΔglnKΔglnB in the experiments described below; in all tests, the strains had identical phenotypes both under free-living conditions and in symbiosis.
We also made several attempts to generate a triple ΔglnB ΔglnK glnD-sm2 mutant by introducing the glnD-sm2 deletion (45) into strain Rm1021ΔglnBΔglnK by using the recombination technique described above. These were unsuccessful, suggesting that adding this glnD mutation to the glnB and glnK deletions has a more severe effect in S. meliloti than in E. coli (6). This also consistent with the observation that some interruptions of glnD are lethal in S. meliloti (35).
We tested Rm1021ΔglnKΔglnB for its ability to use a set of nitrogen-containing compounds as the sole nitrogen source. Growth was monitored using a microplate reader (data not shown) or on solid agar plates (Fig. 2A and B), with the potential nitrogen sources added at 0.2 g/liter. Rm1021ΔglnKΔglnB grew very slowly on defined media containing alanine, proline, histidine, glycine, GABA, glutamate, glutamine, ammonium, or nitrate as nitrogen sources and also grew poorly on LB and YMB media. Supplementation of the media with 0.5, 1, or 2 g/liter glutamine and/or glutamate did not improve the growth of the Rm1021ΔglnKΔglnB mutants. This is in contrast to Rm1021ΔglnD-sm2, which grew much more normally on a favorable nitrogen source, such as glutamine (45). The growth defect of enteric bacteria lacking GlnB and GlnK proteins resulted from inhibition of serine biosynthesis (12). This defect could be rescued by supplementing the growth media with serine or glycine in combination with low levels of isoleucine, leucine, and valine. A similar mixture of these amino acids did not rescue the slow growth of Rm1021ΔglnKΔglnB.
S. meliloti contains three glutamine synthetase (GS) genes, glnA, glnII, and glnT, encoding GSI, GSII, and GSIII, respectively (16). Expression of GSII, and to some degree GSI, is NtrC dependent (5, 16), whereas the GSIII gene, glnT, is cryptic and is not usually expressed in the presence of GSI or GSII (39). To better understand the nature of changes in nitrogen response regulation in Rm1021ΔglnKΔglnB, we measured total manganese-dependent GS transferase activity in various mutants grown on ammonium or glutamate as the nitrogen source (Table (Table3).3). The GS activities of the mutants were measured in experiments that were published previously at the same time that the GS activities of Rm1021, Rm1021ΔglnD-sm2, and Rm1021ΔglnB were determined (see reference 45). Our results show that total GS activities in Rm1021ΔglnK strains grown under both high- and low-nitrogen conditions were similar to the levels in the parental strain Rm1021, indicating that GlnB can regulate GS activity in free-living S. meliloti cells and this regulation does not require GlnK (Table (Table3).3). A surprisingly high level of total GS activity was seen in Rm1021ΔglnKΔglnB cells grown in the presence of ammonium (Table (Table3).3). Rm1021ΔglnB also had an increased level of GS activity in cells grown under high-nitrogen conditions (Table (Table3)3) (45). These data indicate that the presence of GlnB attenuates GS expression/activity in normal S. meliloti Rm1021 metabolism and that GlnK cannot replace it for this function. This result prompted us to investigate GS production in the mutants more directly.
We showed previously that S. meliloti Rm1021 expresses GSI under conditions of both high and low nitrogen stress and that GSI was subject to significant posttranslational modification as a function of the quality of the nitrogen source (45). Using Western blot analyses, we did not detect the presence of a significant amount of GSI protein in extracts prepared from ammonium- or glutamine-grown cultures of Rm1021ΔglnK or Rm1021ΔglnKΔglnB (data not shown). On the other hand, our analysis indicated that the strains grown on glutamate, a poor nitrogen source, produced a detectable amount of GSI (Fig. (Fig.3,3, panel A). However, under these conditions, posttranslational modification of GSI in Rm1021ΔglnK and Rm1021ΔglnKΔglnB differed from that in Rm1021. While most of the GSI in Rm1021 grown on glutamate was unmodified (Fig. (Fig.3,3, panel A) (45), Rm1021ΔglnK and Rm1021ΔglnB cells contained significant amounts of the low-mobility, adenylylated form of the enzyme (Fig. (Fig.3,3, panel A) and almost all GSI in Rm1021ΔglnKΔglnB cells was modified. Therefore, both PII proteins normally contribute to GSI deadenylylation under nitrogen stress conditions, resulting in highly adenylylated GSI in the absence of any PII protein (Fig. (Fig.11 and and2).2). The NSR appeared to be uncoordinated under these circumstances, since GSI synthesis was induced by the nitrogen stress but the specific activity of the protein was lowered by adenylylation, a phenotype expected of cells with sufficient nitrogen (Fig. (Fig.1).1). A similar functional overlapping of GlnB and GlnK was reported for enteric bacteria (6).
When GSII antibodies were used to stain Western blots of extract proteins, a band at approximately 36 kDa, the size of GSII monomer, was visible in glutamate-grown cultures of Rm1021, Rm1021ΔglnB, Rm1021ΔglnK, and Rm1021ΔglnKΔglnB but was not seen in Rm1021ΔglnD-sm2 (Fig. (Fig.3,3, panel B) (45), Rm1021ΔrpoN, Rm1021ΔntrB, and Rm1021ΔntrC (data not shown). When Rm1021 and Rm1021ΔglnK were grown on NH4 as a nitrogen source, this band disappeared (Fig. (Fig.3,3, panel C). Crude extracts of both Rm1021ΔglnB and Rm1021ΔglnKΔglnB grown on MM-NH4 medium still produced a band corresponding to the GSII monomer (Fig. (Fig.3,3, panel C) (45). These data were consistent with the idea that GSII synthesis was activated by the NtrB/C two-component regulatory system acting through the RpoN sigma factor and that GSII synthesis was significantly decreased in the absence of nitrogen stress but only when GlnB was present (Fig. (Fig.11 and and3).3). Since phosphorylated NtrC is thought to activate transcription of genes regulated by NtrB/NtrC, these results suggest that the GlnB protein pushes the regulation in the direction of unphosphorylated NtrC but that without GlnB, sufficient phosphorylated NtrC was present to activate glnII expression (Fig. (Fig.11 and and3).3). GSII production can be regulated posttranscriptionally (43), and it was also possible that GSII inactivation or degradation was accelerated when nitrogen stress was low and high levels of GSII were not needed (31, 34).
To see if the observed accumulation of GSII protein resulted from glnII overexpression, we analyzed the expression of β-glucuronidase from a glnII::gusA fusion in S. meliloti mutants missing GlnB (Table (Table4).4). As we expected, both wild-type strains, Rm1021 and GMI708, grown in the presence of glutamate as a nitrogen source had higher glnII expression than ammonium-grown cells. Strain GMI3107 (GMI708 missing the glnB gene) did not substantially induce glnII expression under nitrogen stress (Table (Table4),4), an effect of the glnB deletion on GSII expression that was reported earlier (3). On the other hand, Rm1021ΔglnB had a high level of GSII expression, regardless of the availability of nitrogen in the media (Table (Table4).4). As expected from the GSII antibody data, deletion of glnK did not affect GSII expression (Table (Table4).4). The data presented here suggest that unmodified GlnB acted to repress glnII expression under high-nitrogen-status conditions in Rm1021 cells and, in the absence of GlnB, cells expressed GSII in a way that resembled induction by nitrogen stress (Fig. (Fig.11 and and33).
GSII expression and GSI modification responded very differently to deletion of glnB in GMI708, a derivative of S. meliloti strain Rm2011. In GMI708, GlnB acted as a positive regulator of glnII expression and site-specific mutagenesis of GlnB that removed the uridylylation site showed that the uridylylated form of GlnB was required to generate NtrC-phosphate, either for NtrB-dependent phosphorylation of NtrC or possibly for expression of the ntrC gene itself (Table (Table4)4) (3). Arcondéguy et al. (3) concluded that the glnB deletion in GMI708 signals nitrogen sufficiency for ntr regulon expression and results in constitutive dephosphorylation of NtrC. Arcondéguy et al. (3) also showed that in strain GMI3146, a derivative of GMI708 with a glnB deletion and a glnII::Tn5 mutation, GSI adenylylation is uniformly low regardless of the cells' nitrogen status. They concluded that the absence of GlnB signals nitrogen deficiency for regulation of GSI specific activity and results in constitutive GSI deadenylylation.
In our previous report, we showed that Rm1021 and GMI708, which were derived from the same parent about 30 years ago, have significant differences in their NSR regulations (45) and the data presented here reinforced this conclusion. To determine whether GMI708 inherited this altered NSR regulation from its immediate parental strain, Rm2011, or acquired it during further independent cultivation, we analyzed GSII production in Rm2011 and its derivative Rm2011ΔglnB. Western blot visualization of GSII in the strains grown on ammonium and glutamate showed that, in contrast to Rm2011, crude extracts of Rm2011ΔglnB grown on MM-NH4 medium still produced a band corresponding to the GSII monomer (data not shown), indicating that the regulation of GSII expression in Rm2011 was similar to this regulation in Rm1021 and suggesting that there was a difference between Rm2011 and GMI708.
We had no antibody that could detect the PII proteins. To see whether GlnB and GlnK were uridylylated in the glnD-sm2 mutant, we analyzed the ability of Rm1021ΔglnD-sm2 to uridylylate PII proteins in permeabilized cells by using [α-32P]UTP as a substrate. For a positive control, we analyzed Rm1021 to show protein uridylylation and Rm1021ΔglnKΔglnB, which lacks the PII proteins and would not be expected to contain a radioactive band after labeling. As shown in Fig. Fig.4,4, a band corresponding to uridylylated protein with an approximate size of 11 kDa was visible in Rm1021 exposed to α-ketoglutarate (lane 1), and no band was detected in Rm1021ΔglnKΔglnB (lane 2). We also could not detect the presence of any uridylylated PII proteins in the Rm1021ΔglnD-sm2 mutant (lane 3). No labeling was observed in Rm1021 exposed to glutamine, which is in agreement with the previous reports on the sensitivity of PII modification to the α-ketoglutarate/glutamine ratio (1; data not shown).
To test whether PII proteins were expressed in Rm1021ΔglnD-sm2, we analyzed the expression of β-glucuronidase from glnB::gusA and glnK::gusA fusions (Table (Table4).4). In contrast to glnK, which had a low level of expression in both ammonium- and glutamate-grown cells, glnB was highly expressed in Rm1021 in the presence of ammonium as the nitrogen source, and its expression was even higher in the presence of glutamate. The glnD-sm2 deletion significantly decreased glnB::gusA expression in the cells grown on glutamate. However, the level of glnB::gusA expression was still detectable in Rm1021ΔglnD-sm2 (Table (Table4).4). These results suggest that in Rm1021ΔglnD-sm2, GlnD protein lost its ability to uridylylate PII proteins, leading to constitutively deuridylylated PII proteins. In turn, this leads to the idea that unmodified GlnB in Rm1021ΔglnD-sm2 stimulates NtrC dephosphorylation and deactivates ntrC-dependent promoters (Fig. (Fig.1).1). This is in agreement with the lack of GSII in the glnD-sm2 mutant (Fig. (Fig.3).3). Rm1021ΔglnD-sm2 also produced less GSI, expression of which is partially NtrC-P dependent (Fig. (Fig.3).3). A similar effect of the glnD-sm2 mutation on PII modification and GSII expression was observed in R. leguminosarum (38).
A comparison of the symbiotic phenotypes of the mutants is shown in Table Table1.1. All tested strains formed large, pink nodules that were similar in appearance to those induced by parental strain Rm1021. Plants nodulated with Rm1021ΔglnB were a rich green and appeared to be healthy. The shoot dry masses of the plants nodulated with the ΔglnB and ΔglnK mutants were similar to those of the plants inoculated with Rm1021. The shoot masses of plants nodulated by Rm1021ΔglnD-sm2 were significantly lower than those of plants nodulated by Rm1021, and the leaves were yellow. These symbiotic phenotypes of Rm1021ΔglnD-sm2 and Rm1021ΔglnB were described previously (45). The plants infected by Rm1021ΔglnK did not exhibit a nitrogen starvation phenotype and were similar to plants infected by Rm1021 (Table (Table3).3). Because of the severe free-living phenotype of the Rm1021ΔglnKΔglnB mutant, we were surprised to find that plants nodulated with Rm1021ΔglnKΔglnB were green and healthy and the masses of these plants did not differ from those nodulated by Rm1021 (Table (Table3).3). All bacteria reisolated from the nodules formed by Rm1021ΔglnKΔglnB had a free-living phenotype identical to the original Rm1021ΔglnKΔglnB, indicating that symbiotic effectiveness of the mutant was not due to reversion or pseudoreversion.
The ΔglnD-sm2 mutation, which leads to an ineffective symbiosis, is thought to generate a GlnD protein with an N-terminal deletion that is probably unable to add UMP to a PII protein, thus creating a strain in which neither GlnB nor GlnK can be modified (45). However, the symbiotic competence of the Rm1021ΔglnB, Rm1021ΔglnK, and Rm1021ΔglnKΔglnB mutants clearly demonstrates that GlnB and GlnK were not needed to establish a normal symbiosis. However, this does not necessarily mean that GlnB and GlnK do not contribute to the ineffective phenotype of Rm1021ΔglnD-sm2, since it is possible that the unmodified form of GlnB or GlnK in Rm1021ΔglnD-sm2 represents a signal with consequences. As reported earlier, it does not seem to be possible to delete glnD (35), although a mutant with a transposon inserted in the middle of the gene is still viable (45).
The symbiotic effectiveness of Rm1021ΔglnKΔglnB might also be explained by the fact that the strain overproduced GS and that, even though many other aspects of the NSR were not functional in this mutant, overexpression of GS was sufficient to maintain effectiveness. If this idea is correct, overexpression of GS in Rm1021ΔglnD-sm2 might restore symbiotic effectiveness, most probably leaving the NSR regulation in free-living cells broken.
To complement the lack of GS activity in Rm1021ΔglnD-sm2, we introduced a plasmid carrying a constitutively expressed copy of glnA, which codes for GSI. Western blot analysis with anti-GSI antiserum confirmed that crude extracts of Rm1021ΔglnD-sm2(pCPPGlnA) grown on MM-NH4 medium produced bands corresponding to GSI (Fig. (Fig.5,5, lanes 3 and 4), whereas these bands were absent in crude extracts of Rm1021ΔglnD-sm2 grown on MM-NH4 medium (Fig. (Fig.5,5, lanes 1 and 2). When GSI antibodies were used to stain the Western blot, in addition to the lower-mobility band corresponding to the adenylylated form of GSI in Rm1021 cells (Fig. (Fig.5,5, lanes 5 and 6), a band corresponding to unmodified GSI was visible in the extracts prepared from Rm1021(pCPPGlnA) (Fig. (Fig.5,5, lanes 7 and 8) and Rm1021ΔglnD-sm2(pCPPGlnA) cells (Fig. (Fig.5,5, lanes 3 and 4) grown on ammonium. This indicates that plasmid-encoded GSI did not undergo significant posttranslational modification and was present as the active unadenylylated form.
As expected if GSI remained unmodified and active, introduction of pCPPGlnA resulted in a significant increase of GS enzymatic activity in both Rm1021 and Rm1021ΔglnD-sm2 cells grown on either ammonium or glutamate (Table (Table3).3). Furthermore, the presence of pCPPGlnA appeared to slightly improve the growth rate of Rm1021ΔglnD-sm2 on nitrogen sources that had been shown to be poor substrates for the mutant (45). An example of the growth of Rm1021ΔglnD-sm2(pCPPGlnA) using alanine as a sole nitrogen source is presented in Fig. Fig.6.6. However, increased GSI activity did not result in improved symbiotic performance by Rm1021ΔglnD-sm2(pCPPGlnA). Plants inoculated with the strain were yellow, and the masses of these plants were similar to those of the plants inoculated with Rm1021ΔglnD-sm2 (Table (Table3).3). Therefore, decreased GS activity in Rm1021ΔglnD-sm2 could play a role in the failure of the strain to grow on a number of nitrogen sources, but there are other components of the nitrogen stress regulatory system affected in the glnD mutant that were not complemented by GS overexpression.
High GS activity can cause growth problems in a number of organisms under certain conditions (48). While Rm1021ΔglnB formed smaller colonies than Rm1021 on YMB plates (45), the ability of Rm1021ΔglnB to use a number of other nitrogen sources was not affected (45). We also found that introducing pCPPGlnA into Rm1021 did not affect growth of the strains on any tested medium (data not shown). This indicated that elevated GS activity could not explain the poor growth of Rm1021ΔglnKΔglnB and that nitrogen stress metabolism was affected at another level in this strain.
In this report, we describe the analysis of S. meliloti Rm1021 mutants missing one or both of the PII homologs, GlnB and GlnK, adding context to a previously published study on the role of GlnD uridylyltransferase and the GlnB PII proteins in S. meliloti NSR regulation and symbiotic competence (45). We show here that GlnB and GlnK play a critical role in the NSR regulation of free-living cells of S. meliloti and can partially substitute for each other in their function as regulators of nitrogen stress metabolism. GlnB or GlnK was required for deadenylylation of GSI under limited nitrogen and deletion of one of two PII proteins resulted in partial adenylylation of GSI in cells grown under nitrogen limitation (Fig. (Fig.1).1). We could not detect any deadenylylated form of GSI in nitrogen-starved Rm1021ΔglnKΔglnB, while GSI was completely deadenylylated in Rm1021 (Fig. (Fig.3).3). On the other hand, the proteins have distinct specificity in certain functions; unlike GlnK, GlnB acted to downregulate GSII expression in ammonium-grown culture (Fig. (Fig.11 and and3).3). Deletion of glnB, but not glnK, induced overproduction of GSII (Fig. (Fig.3)3) (45). Additionally, the mutants missing GlnB had increased total GS activity under nitrogen excess (Table (Table33).
These data show that S. meliloti Rm1021 nitrogen stress response regulation shares some similarities to NSR in enteric bacteria and R. leguminosarum. In those bacteria, GlnB is essential for dephosphorylation of NtrC and deletion of glnB resulted in constitutive GS expression even when cells were grown in sufficient nitrogen. However, this was not observed for strain GMI708, which is closely related to Rm1021, for which GlnB-UMP was shown to be necessary for NtrC phosphorylation. Additionally, in R. leguminosarum, a glnD mutation abolished GSII expression, suggesting that GlnD was required to promote NtrC phosphorylation (38). An effect similar to that of the ΔglnD-sm2 mutation was observed in S. meliloti strain Rm1021 (Fig. (Fig.3)3) (45). Rm1021ΔglnD-sm2 produced less GSI (Fig. (Fig.3)3) (45), which is consistent with the observation that GSI expression was, at least partially, ntrC dependent (Fig. (Fig.1).1). In the case of E. coli, glnD mutations result in an impaired ability to increase the level of GS in response to nitrogen limitation (13). In addition, the small amount of GS present in the cell was highly adenylylated, and the mutants were consequently unable to grow without glutamine, a phenotype similar to that of Rm1021ΔglnD-sm2 (Fig. (Fig.3)3) (45). However, there appear to be some additional differences in the function of the PII proteins in E. coli and S. meliloti. In contrast to enteric bacteria, for which GlnK was shown to be critical to survive nitrogen starvation and recovery on ammonia-rich media (11), GlnK does not seem to play a similar role in S. meliloti.
We cannot explain the free-living phenotype of Rm1021ΔglnKΔglnB by a single substrate-level abnormality due to NSR regulation. The growth defect of Rm1021 lacking both PII proteins was not simply a result of impaired serine biosynthesis, which was shown to be the case for E. coli (12), since the supplementation of growth media with serine did not rescue this growth phenotype. GS overexpression itself could not affect growth of Rm1021ΔglnKΔglnB for the reasons discussed above, for the experiments in which we overexpressed GSI in Rm1021. Rm1021ΔglnKΔglnB overexpressed GSII and probably other proteins corresponding to the ntr regulon, suggesting activation of a stress response similar to that in nitrogen-starved cells. If this were the only problem in Rm1021ΔglnKΔglnB metabolism, cultivation with a poor nitrogen source, like glutamate, would rescue the growth phenotype of the mutant. This occurs in Rm1021ΔglnB, which grows poorly on glutamine, a good nitrogen source, but well on glutamate. Nevertheless, Rm1021ΔglnKΔglnB did not grow well on either a good nitrogen source or a poor nitrogen source, which indicates much deeper involvement of PII proteins in the regulatory network or in nitrogen uptake and/or metabolism.
The finding that the double deletion mutant Rm1021ΔglnKΔglnB could form an effective symbiosis with alfalfa shows that regulatory circuits that do not employ GlnB or GlnK were sufficient for satisfactory symbiotic function. Since the Rm1021ΔglnD-sm2 strain is ineffective, this would suggest that GlnD has some role in the symbiosis other than modifying the PII proteins. Another possibility is that the PII proteins in Rm1021ΔglnD-sm2 are never modified and that these unmodified proteins interfere with symbiotic function. The free-living growth phenotypes of the Rm1021ΔglnD-sm2 mutant indicate that GlnD in Rm1021ΔglnD-sm2 always reports that the cell nitrogen status is good so that nitrogen-scavenging pathways are repressed. This is consistent with the PII proteins remaining unmodified, and this condition may disrupt nitrogen exchange in the Rm1021ΔglnD-sm2-alfalfa association. The PII protein mutants generated in this work remove all of the PII proteins but lead to an effective symbiotic phenotype.
This work was supported by the Agricultural Research Center at Washington State University and grant DE-FG03-96ER20225 from the Energy Biosciences Program at the United States Department of Energy.
We thank the WSU Laboratory for Biotechnology and Bioanalysis for sequencing support, Daniel Kahn for Rm2011 derivatives, Frans deBruijn for Rm1021 GS mutants, Stefan Nordlund and Paul Ludden for GSI antisera, and Jodi Humann for assistance at various stages of the research.
Published ahead of print on 19 March 2010.