Strains and plasmids used are listed in Table 1. S. meliloti was grown at 30°C minimal mannitol NH₄Cl or YMB liquid medium or medium that had been solidified with 1.5% agar (Sigma) (Somerville and Kahn, 1983). For assays using high or low nitrogen and/or phosphate, Min MOPS (Al-Niemi et al., 1997) liquid media containing 0.4% NH₄Cl (nitrogen-sufficient) or 0.04% glutamate (nitrogen-deficient) was used. KH₂PO₄ was added at 2 mM for phosphate sufficient conditions and 50 μM for phosphate deficient conditions (Bardin et al., 1996). E. coli strains were grown at 37°C on LB media (Sambrook, 2001). Neomycin was added at 200 μg/mL and tetracycline at 10 μg/mL for S. meliloti. For E. coli, kanamycin was added at 40 μg/mL. LBMC was used for transduction (Finan et al., 1984). 5-Bromo-4-chloro-3-indolyl phosphate (X-Phos) was added to solid media at a concentration of 50 μg/mL.

Standard procedures were used for DNA manipulations (Sambrook, 2001). Tn5 transposon mutants were generated as described by Yurgel and Kahn (2008) using the non-replicating suicide plasmid pSUP5011 to deliver the transposon to S. meliloti. Arbitrary PCR was performed to identify the location of the transposon insertions (Yurgel and Kahn, 2008). Plasmids pJ1700G-pGlnA and pJ1700G-pPhoX were constructed in a manner similar to pJ1700G-pGlnII as described by Yurgel et al. (2010).

Replacement of a wild type allele of pstC was carried out through a cloning, conjugation, and transduction method (Labes et al., 1993). In Rm2011, a 438 base pair intergenic segment of DNA (chromosomal position 560651 to 561089) was amplified from an intergenic region upstream of pstC (SMc02145) between SMc02151 and SMc02150. The amplified DNA was ligated into plasmid pK19mobsacB to give p2011pstC. Using conjugation, the plasmid was integrated into the Rm2011 chromosome between SMc02151 and SMc02150, marking this region with a neomycin resistance gene contained within the plasmid. Using μ12 bacteriophage-mediated transduction, the integrated plasmid and nearby pst operon were delivered from Rm2011 to the Rm1021 strains of interest. Neomycin resistant colonies were selected and streaked for isolation from phage μ12. Visual confirmation of the repair of the pstC+ allele was observed on plates containing X-Phos, a substrate for AP that is an indirect method for observing PSR activation that results from the defect in pstC. Because of the phosphate transport defect caused by the pstC mutation, Rm1021 is normally blue on X-Phos media. White colonies were chosen after transduction. Phage-free colonies were then plated on YMB containing 5% sucrose to select for loss of the p2011pstC plasmid by recombination, leaving behind the pst operon and surrounding region from Rm2011. The repaired pstC+ gene was confirmed via DNA sequencing.

Phage-mediated transduction was performed as described by Finan et al. (1984). Phage μ12 was used to deliver DNA into S. meliloti strains. Phage lysates were filtered with a 0.2 μ syringe filter to remove bacteria before infecting new cultures. Resulting colonies were selected on the appropriate antibiotic and streaked out until the phage was eliminated (Yurgel et al., 2012).

Electroporation of S. meliloti was based on a protocol by Weigel and Glazebrook (2006) for transformation of Agrobacterium. Electrocompetent S. meliloti were made by growing 250 mL of cells in LB broth to an A₆₀₀ of ∼0.2. Cells were spun down and washed with cold 10% glycerol four times, each time concentrating the cells in one fifth the volume of glycerol. Cells were resuspended in a final volume of 1 mL and flash frozen with liquid nitrogen before storage at -80°C. 50 μL of electrocompetent S. meliloti were incubated for 10 min with ∼2500 ng of plasmid before being electroporated at 2.0 kV in a 2 mm gap length cuvette. Cells were allowed to recover in 1 mL LB broth and grown overnight at 30°C with agitation. The recovered cells were plated on LB media containing the appropriate antibiotic for selection of S. meliloti containing the desired plasmid.

To test growth rate and colony size on Min MOPS media, replication tests were performed as described by Yurgel et al. (2010). Briefly, cells were resuspended to an A₆₀₀ of 0.05. 100 μL of resuspended cells were added to one row of wells of a 96 well plate. Serial 10-fold dilutions were made ending with a 1:10⁶ dilution. A sterile bolt replicator was used to transfer aliquots of each dilution to solid media. Plates were then incubated at 30°C and checked at various times to compare growth between the strains.

Western blot procedures were carried out according to Yurgel and Kahn (2008) with a few modifications. Cells were grown in high (containing 0.4% NH₄Cl) or low (containing 0.04% glutamate) nitrogen media for 18 h before being spun down and resuspended in 1 M TE buffer. SDS loading dye was added and proteins were denatured by incubating at 95°C for 20 min. Proteins were separated on a 10% bis-tris acrylamide gel (Invitrogen) then transferred to a nitrocellulose membrane (Whatman) by electrophoresis. GSII was detected using a S. meliloti anti-GSII antibody (Shatters et al., 1989). The antibody/protein complex was detected using the Supersignal West Pico Chemiluminescence kit (Thermo Scientific).

Glutamine synthetase activity was measured by performing the γ-glutamyltransferase assay. Approximate GS activity was quantified based on the formation of γ-glutamylhydroxamate from glutamine and hydroxylamine. Cultures were grown on Minimal MOPS plates (Bardin et al., 1996) with 50 μM KH₂PO₄ (low Phosphate) or 2 mM KH₂PO₄ (high Phosphate) containing 0.4% NH₄Cl (nitrogen-sufficient) or 0.04% glutamate (nitrogen-deficient) with succinate as a carbon source for 60 h and then colonies were resuspended in broth of the same composition. GS activity was then performed as described by Somerville and Kahn (1983). Activity units are calculated using A₅₄₀/(mg protein^∗min).

Cultures were grown for 48 h in minimal MOPS media with 50 μM KH₂PO₄ (low phosphate) or 2 mM KH₂PO₄ (high phosphate) containing 0.4% NH₄Cl (nitrogen-sufficient) or 0.04% glutamate (nitrogen-deficient). Cultures were diluted 1:40 and grown for 18 h. Expression reactions were run as described by Miller (1972) using 4-nitrophenyl-β-glucuronide (ONPGluc) as the colorimetric substrate. Culture absorbance at 600 nm was measured as well as absorbance at 415 nm, which was due to release of o-nitrophenol. Approximate β-D-glucuronidase activity was calculated using the formula A₄₁₅/(A₆₀₀^∗min).

Alkaline phosphatase activity was measured as described by Yurgel et al. (2000). Briefly, cultures were grown on minimal MOPS plates with sufficient or deficient nitrogen and/or phosphate for 72 h at 30°C. Cells were resuspended and washed in 1 M Tris HCl, pH 8 and resuspended again in 1 ml of 1 M Tris HCl. The reaction was started when 0.1 mL of p-nitrophenyl phosphate (pNPP) substrate (Sigma) was added to 0.9 mL of cells and incubated at 30°C for 60 min. Reactions were stopped by addition of 0.1 mL of 1 M KH₂PO₄. Absorbance of cell cultures was measured at 600 nm and absorbance of the reaction mix was measured at 420 nm. Activity units are A₄₂₀/(A₆₀₀^∗min).

One of the most striking phenotypes of the BK deletion strain was its very slow growth rate and associated small colony size (Yurgel et al., 2010). S. meliloti BK was mutagenized with transposon Tn5, allowing the transposon to integrate randomly throughout the genome. Colonies were selected that were larger in size and seven distinct Tn5 insertion locations were identified in these colonies. Phage-mediated transduction was used to determine if the transposon was associated with faster growth by moving the neomycin resistant transposon into slow-growing BK cells. Of the seven identified locations, only one, containing a Tn5 insertion in phoB, was linked to improved growth of the BK strain, indicating that faster growth by the other six was due to unlinked mutations. The phoB::Tn5 mutation was in the DNA binding/sigma70 interacting C-terminal domain of PhoB, the response regulator in the PSR (Figure 1C) (Blanco et al., 2011).

PhoB regulates AP and AP activation is often used as a measure of the PSR and phosphate stress. In order to determine if the phoB mutant was able to activate the PSR, the mutant was placed on LB media containing X-Phos, a substrate for AP that produces a blue pigment when X-Phos is hydrolyzed (Long et al., 1988). Strains containing the phoB::Tn5 mutation produced white colonies on X-Phos media whereas strains with the unrepaired phoB were blue. This indicated that the phoB::Tn5 mutant lacked the ability to induce a PSR. The other six faster-growing transposon mutants were blue on X-Phos, suggesting that the spontaneous mutations that resulted in faster growth were not due to mutations inactivating AP or the PSR. This data indicates that mutations in other pathways, not just the PSR, can alter the growth rate of strains deficient in the P_II proteins, GlnB and GlnK. Multiple attempts to delete phoB were unsuccessful and in the published literature, most phoB mutants in S. meliloti include a transposon or other mutation at the C-terminal end of the protein similar to the mutant isolated here (Krol and Becker, 2004). While the mutation in phoB was initially thought of as a knockout mutation, it may be a protein with altered function instead of a true knockout. The lack of a mutant described in literature as a complete gene deletion or disruption or mutation in the N-terminal region in S. meliloti suggests that this protein cannot be knocked out completely and that the truncated version may retain some sort of activity despite its inability to mediate the PSR.

In order to determine the impact of the phoB::Tn5 in a wild type background, the mutation was also transduced into Rm1021. A replication test of the various Rm1021 and BK strains discussed in this paper was performed on Min MOPS media with nitrogen and phosphate sufficient and deficient conditions. Figure 2 shows a representative plate on nitrogen and phosphate sufficient media after 3 days of growth. All media conditions showed similar growth patterns between strains, but it was most pronounced on the N and P sufficient media. BK grew much more slowly than Rm1021, as described previously (Yurgel et al., 2010). The BK strain containing the phoB::Tn5 mutation grew at a faster rate and produced larger colonies than the BK strain with wild type phoB. The BK strains had a smaller colony size than the Rm1021 strains for each pair of mutants (wild type, pstC+, phoB::Tn5), indicating that the absence of the P_II genes affected colony size and growth rate in all backgrounds. However, adding either pstC+ or phoB::Tn5 to the BK background improved growth, suggesting that activation of the PSR was inhibitory.

Another striking phenotype of the BK deletion strain observed previously was the altered abundance of GSII protein during both nitrogen sufficient and deficient conditions (Yurgel et al., 2010). Typically, GSII is only detected during nitrogen stress but its expression in the BK background was very high and actually higher when ammonium was present. To test whether GSII continued to be expressed in the BK strain containing the phoB::Tn5 mutation, the protein was visualized using Western blots (Figure 3A; Supplementary Figure S1). In Rm1021 phoB::Tn5, GSII was detected only during nitrogen stress, similar to the pattern of expression of GSII in Rm1021 without the phoB mutation. However, the BK phoB::Tn5 mutant did not have any detectable GSII protein when grown in either nitrogen condition, in contrast to what was observed in the BK strain and the wild type. GSII protein levels in pstC+ strains were similar to the parent strains (Figure 3B) in that GSII in Rm1021 and its pstC+ derivative were repressed by ammonium, while GSII was still expressed in BK and its pstC+ derivative in the presence of ammonium. These data suggest that GlnB/GlnK and PhoB are both involved in the expression of the GSII protein. Since the original expectation was that the ΔglnBΔglnK mutations in BK would block the NSR and expression of GSII, these results are consistent with the idea that expression of GSII in BK is anomalous and that the anomaly depends on PhoB.

The phenotypic differences in the phoB::Tn5 mutant between the Rm1021 and BK backgrounds suggest interactions between the nitrogen and PSRs. This hypothesis also suggests that phosphate availability may affect the NSR and nitrogen may affect the PSR. In addition to exploring the phosphate and nitrogen effects on the NSR and PSR, the phoB::Tn5 mutant strain was characterized further to better understand how GlnB/GlnK and PhoB may be interacting to jointly regulate the two stress responses.

To characterize the effects of nitrogen on the PSR, the PSR must be intact so that classical regulation and activation can be observed. Since the pstC mutation in Rm1021 partially activated the PSR, the strains were converted to pstC+ by transduction. Rm1021 without the repaired pstC gene was included in some experiments as a control (not shown) to ensure results were consistent with previously published data (Yurgel et al., 2010). The joint effects of nitrogen and phosphate availability on the NSR and PSR were explored in the pstC+ strains. Additionally, phoB::Tn5 mutant strains were included in some experiments to gather information on how PhoB may be involved in the regulation of the NSR or crosstalk with the P_II proteins. The pstC+ allele also reverses the slow growth characteristic of the BK strain (Figure 2), an effect that is consistent with the observed downregulation of the PSR in this strain as the result of improved phosphate uptake. While this change does not disrupt PhoB, it does lead to less PhoB-activated transcription.

Total GS enzymatic activity was used to determine whether the NSR was active during nitrogen limitation in the newly constructed strains and if the level of activity was influenced by phosphate stress. Most rhizobia are unusual because they contain two distinct GS proteins that are regulated by the NSR using circuits that involve the P_II proteins. The expression and overall activity of GSI and GSII are upregulated when nitrogen is limited and downregulated when nitrogen is sufficient. It was shown that, compared to wild type Rm1021, total GS activity increased in a strain missing the P_II proteins because, while GSII protein was present in Rm1021 only during nitrogen limitation, GSII protein was present in the BK deletion strain at a very high level when ammonium was available and at a lower level when the nitrogen source was glutamate (Yurgel et al., 2012). The observed increase in GSII protein may be related to the observation that, regardless of nitrogen source, all of the detected GSI in the BK strain was present as GSI-AMP, the inactive form of GSI. In Rm1021, GSI, but not GSI-AMP, was detected under nitrogen stressed conditions, reflecting the need for active enzyme (Yurgel et al., 2010). These results are consistent with the need for PII proteins to transduce the signal generated by GlnD that nitrogen is limited.

Glutamine synthetase activity was affected by phosphate in a wild type background, but not in a P_II deletion background. Figure 4 shows total GS activity in the pstC+ and phoB::Tn5 strains during high and low nitrogen and phosphate conditions. (The data used to generate Figures 4 and 5 are included in Supplementary Table S1 in order to permit other comparisons to be made.) When nitrogen was sufficient, phosphate stress did not have a significant effect on total GS activity in any strain. In contrast, when nitrogen was limited, phosphate deficient conditions repressed GS activity by ∼2-fold in Rm1021 pstC+ and Rm1021 phoB::Tn5 compared to phosphate sufficient conditions. This reduction in GS activity when phosphate availability was limited was not seen in the BK pstC+ strain. The BK phoB::Tn5 strain had a very low level of total GS activity regardless of the nitrogen or phosphate available, consistent with the idea that both the NSR and PSR were important in enzyme production or activation.

Glutamine synthetase I activity is primarily regulated through adenylylation whereas GSII is primarily regulated through transcription and subsequent translation of the glnII gene. GSI and GSII are the protein products of the glnA and glnII genes. Activity of the glnA and glnII promoters was examined to see if they responded to phosphate limitation in parallel with GS activity. If phosphate conditions had an effect on the GS activities by influencing transcription, this would be most obvious in the effect on glnII expression. Expression of glnA and glnII were assayed by measuring β-glucuronidase activity generated by the activity of the glnA and glnII promoters fused to a gusA reporter gene in plasmids pJ1700G-pGlnA and pJ1700G-pGlnII.

As expected, glnII expression was activated by nitrogen limitation and repressed by ammonium (Figure 5). However, the level of this response was regulated by phosphate availability. In Rm1021 pstC+, changing phosphate concentration did not lead to a significant change in glnII expression when nitrogen was limited, but adding phosphate significantly stimulated glnII expression during nitrogen sufficient conditions from an undetectable level to a low but measurable β-glucuronidase activity.

Expression of the glnII-gusA fusion was strikingly higher in the BK pstC+ background than in the Rm1021 pstC+ background, an observation consistent with GSII protein abundance. The level of expression in BK pstC+ ranged from ∼3-fold higher than in Rm1021 pstC+ under the “inducing” condition of low N and nearly 100-fold higher in high N where glnII expression in Rm1021 pstC+ was almost completely repressed. The BK pstC+ strain had reduced activity during phosphate limitation compared to sufficiency when nitrogen was limited, demonstrating a response to phosphate levels under conditions when glnII expression is typically induced. Introducing the phoB::Tn5 mutation substantially lowered glnII-gusA expression in both the wild type and BK backgrounds under each of the media conditions tested. The addition of phosphate raised the level of expression of glnII-gusA in both Rm1021 and BK with the phoB::Tn5 mutation indicating that GSII expression was induced by phosphate in the absence of a proper PhoB regulatory protein.

Unlike GSII, GSI is regulated post-translationally by adenylylation. In rhizobia, glnA transcription is not as significant in controlling the amount of GSI activity present and the effect of nitrogen concentration on glnA-gusA expression observed here was not as dramatic as the effect on glnII-gusA (Figure 5). Raising phosphate increased the expression of glnA-gusA for both Rm1021 pstC+ and BK pstC+ when nitrogen was available but only Rm1021 pstC+ had significantly higher expression when nitrogen was limiting. The presence or absence of the phoB::Tn5 mutation did not have a strong effect on the expression of glnA-gusA under most conditions but, for both 1021 phoB::Tn5 and BK phoB::Tn5, phosphate increased glnA-gusA expression by about twofold when phosphate was higher under similar nitrogen conditions.

The GS activity and expression results indicated that phosphate status affected aspects of the NSR. The hypothesis that nitrogen conditions would affect the PSR was also explored. As discussed previously, AP activity is an index of the PSR that is typically induced by low phosphate conditions. AP expression and activity were measured to determine if nitrogen limitation would influence the production of AP. The classical activation pattern was observed and displayed a very definite on/off activity pattern in the Rm1021 pstC+ strain when grown in low or high phosphate, respectively. When the cells were pstC+ in either wild type Rm1021 or the BK deletion strain, nitrogen limitation had no significant effect on the activity of AP (Figure 6). phoX, which codes for the major AP in Rm1021, is upregulated during phosphate limitation in a PhoB-dependent manner (Zaheer et al., 2009). Since phosphatases are typically inhibited by phosphate (Fernley and Walker, 1967), a gusA reporter fusion to phoX was constructed and introduced into the mutant strains to measure a reporter activity based on expression of the phoX promoter. The activity of β-glucuronidase from the phoX-gusA fusion mimicked the activity levels of AP for each strain and condition (Figure 7). The pstC+ strains showed increased expression during phosphate limitation and very minimal expression when the phosphate level was sufficient, but changing the nitrogen availability had no effect. The phoB::Tn5 mutant strains had minimal expression of the phoX-gusA, regardless of phosphate or nitrogen levels. These data show nitrogen status was not having an effect on AP expression or activity in addition to the known regulation by phosphate availability. When any of these strains carried the phoB::Tn5 transposon insertion, they had very low AP or phoX-gusA derived β-glucuronidase activity, regardless of nitrogen or phosphate conditions. This pattern of expression would be expected if the phoB gene was mutated, phoX was regulated by PhoB and the phoB::Tn5 mutation was not leaky in its control of the PSR. This implies that the phosphate responsiveness of both glnA and glnII expression was not due to some residual PhoB activity and that another level of phosphate regulation in S. meliloti appeared to be involved.