In vitro reconstitution and characterization of the Rhodobacter capsulatus NtrB and NtrC two-component system.

Enhancer-dependent transcription in enteric bacteria depends upon an activator protein that binds DNA far upstream from the promoter and an alternative sigma factor (sigma 54) that binds with the core RNA polymerase at the promoter. In the photosynthetic bacterium Rhodobacter capsulatus, the NtrB and NtrC proteins (RcNtrB and RcNtrC) are putative members of a two-component system that is novel because the enhancer-binding RcNtrC protein activates transcription of sigma 54-independent promoters. To reconstitute this putative two-component system in vitro, the ReNtrB protein was overexpressed in Escherichia coli and purified as a maltose-binding protein fusion (MBP-RcNtrB). MBP-RcNtrB autophosphorylates in vitro to the same steady state level and with the same stability as the Salmonella typhimurium NtrB (StNtrB) protein but at a lower initial rate. MBP-RcNtrB autophosphorylates the S.typhimurium NtrC (St-NtrC) and RcNtrC proteins in vitro. The enteric NtrC protein is also phosphorylated in vivo by RcNtrB because plasmids that encode either RcNtrB or MBP-Rc-NtrB activate transcription of an NtrC-dependent nifL-lacZ fusion. The rate of phosphotransfer to RcNtrC and autophosphatase activity of phosphorylated RcNtrC (RcNtrC---P) are comparable to the StNtrC protein. However, the RcNtrC protein appears to be a specific RcNtrB P phosphatase since RcNtrC is not phosphorylated by small molecular weight phosphate compounds or by the StNtrB protein. RcNtrC forms a dimer in solution, and RcNtrC - P binds the upstream tandem binding sites of the g1nB promoter 4-fold better than the unphos-phorylated RcNtrC protein, presumably due to oligomerization of RcNtrC -P. Therefore, the R. capsulatus NtrB and NtrC proteins form a two-component system similar to other NtrC-like systems, where specific Rc- NtrB phosphotransfer to the RcNtrC protein results in increased oligoinerization at the enhancer but with subsequent activation of a sigma 54-independent promoter.

The NtrB and NtrC proteins of enteric bacteria form a twocomponent signal transduction system that has been extensively characterized genetically and biochemically (see Ref. 1 for review). Under conditions of nitrogen limitation, the NtrB sensor kinase autophosphorylates on a specific histidine residue (2)(3)(4)(5) and transfers the phosphate to the NtrC response regulator protein on a specific aspartate residue (2,4,6). Phos-phorylated NtrC (NtrCϳP) 1 is a transcriptional activator of genes involved in nitrogen metabolism such as glnA (glutamine synthetase). NtrCϳP has enhanced DNA binding activity (7), presumably due to increased oligomerization on the DNA template (8,9), and an ATPase activity (10,11), which may also be due to the oligomerization of the NtrC phosphoprotein (12,13). These properties of oligomerization and ATPase activity are essential for transcriptional activation in vitro (14) and in vivo (15,16,17).
Members of this class of proteins share certain properties. (a) They bind to DNA at tandem sites far upstream (Ͼ100 bp) of the promoters that they activate (see Ref. 18 for review); (b) they contain an ATP binding motif, and possess ATPase activity (11); and (c) they require a specific factor, called 54 , that binds with the core RNA polymerase at highly conserved promoters (see Refs. 19 -21 for reviews). The NtrC protein binds to sites over 100 bp upstream of the glnA promoter (22,23), and DNA looping occurs between the NtrC protein bound at the enhancer and the 54 /RNA polymerase holoenzyme (which forms a stable closed complex) bound at the promoter (24 -27). Interaction between the activated NtrC protein and the 54 / RNA polymerase holoenzyme, in combination with the ATPase activity of the NtrC protein results in a dramatic stimulation of the expression of the glnA gene (12,14).
The NtrC protein from the photosynthetic bacterium Rhodobacter capsulatus (RcNtrC) is a novel enhancer-binding protein that does not require the 54 factor to activate transcription of the R. capsulatus nifA1, nifA2, and glnB genes (28 -32). The promoters of these genes have been defined by lacZ translational fusions and primer extension analysis; they are expressed in strains lacking 54 and have no sequence homology to 54 promoters. The proteins encoded by the nifA1 and nifA2 genes are themselves transcriptional activators that induce nitrogen fixation (nif) gene expression, using the 54 RNA polymerase under conditions of nitrogen and oxygen limitation (29,33). The glnB gene is part of a glnBA operon; the GlnB protein putatively acts to repress R. capsulatus NtrB (RcNtrB) function under conditions of nitrogen excess (34, and see 28 and 35 for reviews). The RcNtrC protein also binds to sites on the DNA greater than 100 bp upstream from the promoters that it activates. The RcNtrC binding sites have been characterized by extensive deletion analysis of the nifA1 and nifA2 promoters (29,31). In vitro, DNase I footprinting directly demonstrates that RcNtrC binds to tandem sites of dyad symmetry at the nifA1, nifA2, and glnB upstream regions (31,32). In addition, the RcNtrC protein has an ATP binding motif, which by homology with other ATP-binding proteins is predicted to bind and hydrolyze ATP (36); mutations in this motif prevent transcriptional activation by the RcNtrC protein in vivo (31).
The RcNtrB and RcNtrC proteins are putative members of a two-component system based on sequence homology to the enteric NtrB/NtrC proteins, especially in the regions that are highly conserved in other two-component systems (37). Genetic evidence demonstrates that the R. capsulatus ntrB and ntrC genes are members of an operon, and both genes are essential for transcriptional activation of nif genes in vivo (38,39). The present study demonstrates that the RcNtrB and RcNtrC proteins comprise a two-component regulatory system, that RcNtrBϳP is a specific substrate for RcNtrC, and that the phosphorylated RcNtrC protein has increased DNA binding activity in vitro at RcNtrC tandem upstream binding sites. The R. capsulatus proteins are compared to their counterparts in enteric bacteria.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-All strains and plasmids are described in Table I. The maltose-binding protein fusion to the RcNtrB protein was made by polymerase chain reaction (PCR) of the R. capsulatus ntrB gene in plasmid pDQ2013 using the upstream oligonucleotide 5Ј-CCGGATCCATGAACCTGCCCCCGCCCGGCATC-3Ј and the downstream oligonucleotide 5Ј-CCCCAAGCTTCAAAGCTCCTTCGG-GGCGAC-3Ј. The 1.2-kb PCR product was cut with BamHI and HindIII and cloned into the pmal-C2 vector (New England Biolabs, Beverly, MA) to create an in-frame malE-R. capsulatus ntrB fusion (pMBPRcB). The pETRcB plasmid that contains the R. capsulatus ntrB gene directly downstream of an inducible T7 promoter was made by PCR amplification of the R. capsulatus ntrB gene using the upstream oligonucleotide 5Ј-CATGCCATGGACCTGCCCCCGCCCGGCATC-3Ј and the same downstream oligonucleotide for pMBPRcB. The PCR fragment that contained R. capsulatus ntrB was cut with NcoI and HindIII and cloned into pET21B (Novagen, Madison, WI). These PCR-generated genes have been shown to complement R.capsulatus ntrB mutants. The plasmid pglnBP12 was constructed by cutting pRGK1218 (that contains the glnBA genes) with SalI, excision of the 300-bp R. capsulatus glnB upstream region, and ligation into pUC118.
Protein Purification-The malE-ntrB fusion on plasmid pMBPRcB was overexpressed in Escherichia coli strain TB1 by the addition of 1 mM IPTG for 3 h at 37°C. Cells were sonicated in lysis buffer (10 mM NaPO 4 , pH 7, 30 mM NaCl, 10 mM EDTA, 10 mM EGTA, 1 mM DTT, 0.25% Tween 20, and 0.5 mM PMSF) to disrupt cell membranes and centrifuged at 53000 g for 1 h. Supernatant from the ultracentrifugation was diluted in column buffer (10 mM NaPO 4 , pH 7, 500 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, and 0.5 mM PMSF) and loaded onto an amylose column (New England Biolabs, Beverly, MA). The column was washed with 20 volumes of column buffer, and the MBP-RcNtrB protein was eluted by the addition of column buffer and 10 mM maltose. The RcNtrC protein was purified as described (31) with the addition of an ion exchange step using a DEAE-cellulose column (Sigma) before gel filtration chromatography.
Autophosphorylation Assays-The purified MBP-RcNtrB or StNtrB proteins (200 nM dimers) were preincubated at 37°C for 2 min in phosphorylation buffer (25 mM Tris acetate, pH 8.0, 0.1 mM EDTA, 50 mM KCl, 1 mM DTT, and 4% glycerol). 100 M unlabeled ATP, 0.2 M [␥-32 P]ATP (6000 Ci/mmol, Amersham Corp., final concentration except when noted), and 10 mM (MgCl 2 , final concentration) were then added to initiate reactions for indicated times at 37°C to autophosphorylate the MBP-RcNtrB and StNtrB proteins. Reactions were terminated by the addition of stop solution (50 mM Tris-HCl, pH 6.8, 10% glycerol, 1% SDS, 5 mM DTT, and 10 g of bromphenol blue) and were stored at Ϫ20°C. Reactions were heated briefly at 37°C to dissolve the SDS, run on a 10% SDS-polyacrylamide gel, and dried to Whatman paper. Phosphate incorporation was determined by excision of the dried radiolabeled gel slices and quantitation in a scintillation counter. For some experiments, reactions were blotted directly onto nitrocellulose (Amersham) washed in 50 mM Tris-HCl for 20 min and quantitated by scintillation counts. MBP-RcNtrBϳP was separated from ATP either by hydrolysis of ATP with Na,K-ATPase (1 unit, Sigma) for 5 min at 37°C, or separation of the MBP-RcNtrBϳP from ATP by an affinity amylose column; autophosphorylation reactions were prepared in volumes of 100 -500 l and applied to a 100-l amylose column; the column was washed in 2 ml of phosphorylation buffer to remove ATP and eluted with 100 l of phosphorylation buffer plus 10 mM maltose to recover the MBP-RcNtrBϳP. The phosphorylated MBP-RcNtrB was stored at Ϫ80°C (in phosphorylation buffer) for at least 2 weeks without any observed loss of activity.
Phosphorylation Assays-For heterologous phosphorylation assays, autophosphorylation reactions that contained MBP-RcNtrB (200 nM) were incubated for 10 min at 37°C (to reach a steady state level of MBP-RcNtrBϳP) and then added to StNtrC (1 M) for indicated times at 37°C. StNtrC (500 nM) was phosphorylated by incubation with StNtrB (200 nM, dimers) as described (3). For RcNtrB/RcNtrC phosphorylation assays, MBP-RcNtrB autophosphorylation reactions were modified (with 500 nM MBP-RcNtrB and 10 mM DTT) to obtain optimal phosphorylation of the RcNtrC protein. Reactions were also performed at 24°C due to inconsistent results at 37°C, possibly because the phosphorylation properties of the RcNtrC protein were shown to be heat-sensitive (above 50°C; data not shown), unlike the StNtrC protein. 2 Autophosphorylation reactions were incubated for 10 min at 37°C, equilibrated to 24°C, and added to the RcNtrC protein (500 nM) that was preincubated at 24°C for 2 min. Phosphorylation reactions were incubated for 15 min at 24°C and were terminated and quantitated as described above. Autophosphatase activities of StNtrCϳP and RcNtrCϳP proteins were determined by the incubation of 2 M NtrC protein with purified MBP-RcNtrB-phosphate (separated away from ATP as described above). Loss of NtrCϳP signal (quantitated by SDS-PAGE as described above) was used to determine the autophosphatase rate.
Phosphorylation of NtrC proteins by acetyl phosphate was carried out with a method modified from a procedure graciously supplied by Dr. Tracy Nixon (Pennsylvania State University, University Park, PA). Radiolabeled acetyl phosphate was prepared by the incubation of acetate kinase (2 units; Sigma) with 60 mM potassium acetate and 5 l of [␥-32 P]ATP (6000 Ci; Amersham) in 25 mM Tris-HCl and 10 mM DTT at 24°C. After 15 min the reactions were added to an equal volume of either RcNtrC or StNtrC proteins (500 nM). After either 10 or 40 min at 24°C, the reactions were stopped and loaded onto a 10% SDS-polyacrylamide gel for analysis of radiolabeled proteins.
DNase I Footprinting-End labeled probe (5Ј) was prepared as described previously (31) using an end-labeled BamHI-HindIII fragment from the pglbBP12 template. DNA binding reactions with the unphosphorylated RcNtrC were carried out by the addition of end-labeled DNA (approximately 20,000 cpm) to binding buffer (50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 10 mM MgCl 2 , 50 mM KCl, 10 mM DTT, 0.5 mM PMSF) that contained various concentrations of RcNtrC, either 0 or 1 mM ATP, and 100 ng of poly(dI-dC) as a nonspecific competitor. Complexes were allowed to form for 10 min at 24°C in a total volume of 50 l, after which DNase I digestion and DNA purification were performed as described previously (31). DNA binding reactions with phosphorylated RcNtrC were performed by incubation of MBP-RcNtrB (1 M) in binding buffer that contained 1 mM ATP for 10 min at 37°C, cooling the reactions to 24°C, and then addition of RcNtrC for 15-45 min. Endlabeled DNA was added, and complexes were allowed to form for 10 min 2 S. Kustu, personal communication. at 24°C in a total volume of 50 l. Reactions were analyzed on an 8% sequencing gel.
Other Procedures-Ammonium and potassium phosphoramidate was prepared as described (40). Western analysis was performed using ECL detection agents (Amersham, Little Chalfont, United Kingdom). Seminative gels contained SDS only in the running buffer (0.1% SDS); no SDS was present in the sample buffer or the polyacrylamide gels themselves. Protein concentrations for all experiments were determined by BCA assays (Pierce).

RESULTS AND DISCUSSION
Purification and Autophosphorylation of the R. capsulatus NtrB Protein-For in vitro studies on the RcNtrB and RcNtrC two-component system, both proteins were purified. RcNtrB was purified as a maltose-binding protein fusion (MBP-Rc-NtrB) because overexpression of the RcNtrB protein alone using an inducible T7 overexpression system (Ref. 41; Novagen) produced only low levels of the RcNtrB protein (data not shown). Additionally, a previous report indicated that a MBPfusion to the E. coli NtrB protein did not affect the autophosphorylation or phosphate transfer activities of NtrB (42). The MBP-RcNtrB fusion was made as an N-terminal R. capsulatus ntrB fusion to the C terminus of the malE gene to create pMBPRcB. E. coli cells transformed with pMBPRcB were induced in liquid culture with IPTG to overexpress MBP-RcNtrB, which was observed as a doublet that migrates at approximately 80 kDa by SDS-PAGE analysis (Fig. 1, lane 3); the predicted molecular mass for the fusion protein is 78 kDa, and 38 kDa for the RcNtrB protein. After ultracentrifugation (Fig.  1, lane 4) of sonicated cell extracts, MBP-RcNtrB was purified by affinity chromatography to more than 95% purity by SDS-PAGE analysis (Fig. 1, lane 7). The MBP-RcNtrB protein had several major forms that are presumably degradation products of the MBP-RcNtrB protein. Attempts to prevent or limit this proteolysis were unsuccessful. Proteolysis occurs in vivo upon overexpression; however, the lower forms did not interfere with the activity of the RcNtrB protein in vivo or in vitro (see below).
To determine if the R. capsulatus NtrB protein is a histidine kinase capable of autophosphorylation, MBP-RcNtrB was incubated with [␥-32 P]ATP at 37°C. MBP-RcNtrB became labeled, indicating that it autophosphorylates (Fig. 2, lane 3). To compare the autokinase activity of RcNtrB to the enteric NtrB protein directly, purified S. typhimurium NtrB protein (St-NtrB) was also incubated with [␥-32 P]ATP (Fig. 2, lane 1). Time-course experiments demonstrated that the StNtrB protein became labeled more rapidly than the MBP-RcNtrB protein, but at maximal phosphorylation both proteins incorporated an equivalent level of label, indicating that both proteins were phosphorylated to the same degree (Fig. 3). MBP-RcNtrBϳP was separated from ATP and determined to be stable for over 2 h at 37°C, which is comparable to the enteric NtrBϳP  Tet r , K. pnemoniae nifL-lacZ gene (48) cific histidine residue in the conserved C terminus of the protein (6,44,45). The RcNtrB protein contains a histidine at position 214 that is completely conserved with the enteric NtrB proteins and other sensor proteins. To demonstrate that the RcNtrB protein is a histidine kinase, purified MBP-RcNtrBϳP was blotted directly onto nitrocellulose, washed in 50 mM Tris-HCl, and exposed to neutral, acidic, or basic conditions prior to radiodetection. MBP-RcNtrBϳP was sensitive to acidic conditions (a 13-fold loss of signal was observed in the presence of 1 N HCl) and stable in the presence of basic conditions (no loss of signal was observed in the presence of 0.5 N NaOH; data not shown). These results are consistent with the properties of histidinyl-phosphate residues (46,47), indicating that phosphorylation of the MBP-RcNtrB protein probably occurs on a histidine residue.

The RcNtrBϳP Phosphorylates the E. coli NtrC Protein in Vivo and the StNtrC Protein in Vitro-
We wanted to determine if the RcNtrB protein phosphorylates the enteric NtrC protein in vivo and to compare the activities of the RcNtrB and the MBP-RcNtrB proteins. The nitrogen-regulated nifL gene from Klebsiella pneumoniae, which is under control of the enteric NtrB/NtrC and is expressed to high levels under nitrogen limiting conditions (48), was used as a reporter for RcNtrB activity. Overexpression plasmids that contained the R. capsulatus ntrB gene (pETRcB), the R. capsulatus malE-ntrB gene (pMB-PRcB), the R. capsulatus ntrC gene (pETRcC), and the pmal-C2 and pET21B plasmids as controls were transformed into strains that contained the nifL-lacZ fusion on a compatible plasmid (Table I). Two E. coli strains were used in this study: TB1, to induce the expression of the malE and malE-R. capsulatus ntrB genes from the lac promoter, and JM109 (DE3), to induce expression in the pET21B, pETRcB and pETRcC plasmids. Colonies that contained both plasmids were picked onto nitrogen-rich plates to prevent activation by the endogenous E. coli NtrB protein. These plates also contained 0.5 mM IPTG to induce protein expression from both the T7 and the lac promoters (the MBP-RcNtrB protein is overexpressed to approximately 5% of the total cell protein (see Fig. 1 for MBP-RcNtrB), which is presumably sufficient to overcome inhibition by the E. coli GlnB protein of the RcNtrB kinase activity). 5-Bromo-4chloro-3-indol-␤-D-galactopyranoside at 50 g/ml was added to observe the level of lacZ expression from the nifL gene. The R. capsulatus ntrB gene as well as the R. capsulatus malE-ntrB gene induced expression from the nifL-lacZ fusion to comparable levels (Fig. 4, A and C, respectively), whereas plasmid pET21B, or plasmids that encode RcNtrC, or MBP alone did not induce expression (Fig. 4, B, D, and E, respectively). We conclude that the RcNtrB protein can phosphorylate the E. coli NtrC protein, the transcriptional regulator of the nifL gene, and that the RcNtrB and the MBP-RcNtrB show similar in vivo activities.
Previous experiments have shown that the enteric NtrC protein can be phosphorylated in vitro by a variety of sensor kinases, including the CheA protein (49). To determine if MBP-RcNtrBϳP could phosphorylate the enteric NtrC protein, MBP-RcNtrBϳP was incubated with the Salmonella typhimurium NtrC protein (StNtrC) in vitro. Label disappeared from both major forms of the MBP-RcNtrB protein and appeared in the StNtrC, indicating that the MBP-RcNtrBϳP is a substrate for the StNtrC protein (Fig. 2, lane 4). To directly compare the phosphotransfer properties of the RcNtrB to the enteric NtrB protein, both MBP-RcNtrBϳP and StNtrBϳP (at 200 nM dimers) were incubated with the StNtrC protein, and the transfer of phosphate was measured. Both MBP-RcNtrBϳP and StNtrBϳP labeled StNtrC to a comparable level (Fig. 2, compare lanes 2 and 4). The autophosphatase activity of StNtrCϳP was determined to be the same irrespective of phosphorylation by the MBP-RcNtrB or StNtrB proteins (see below). Thus, the StNtrC protein probably interacts with a highly conserved functional domain of both NtrB proteins (supported by amino acid sequence homology in the region of the conserved histidine).
The MBP-RcNtrBϳP Protein Phosphorylates the RcNtrC Protein in Vitro-To determine if the RcNtrC protein can be phosphorylated by RcNtrB, purified MBP-RcNtrB was phosphorylated as described above and incubated with purified RcNtrC (Fig. 1, lane 8). As shown in Fig. 2 (lane 7), the addition of RcNtrC led to the disappearance of label from RcNtrB with its subsequent appearance in RcNtrC. Optimal transfer of phosphate from RcNtrB to RcNtrC was found to occur when RcNtrB was present at approximately one-half the concentration of RcNtrC (Fig. 5A). Increasing StNtrB to equimolar with StNtrC did not improve StNtrC phosphorylation (Fig. 2, lane  10). Optimal phosphorylation of RcNtrC also occurred when the DTT concentration was increased from 1 mM to 10 mM (Fig.  2, compare lanes 7 and 9), even though increasing the DTT concentration had no effect on the level of autophosphorylation of MBP-RcNtrB (data not shown). This result may reflect a higher sensitivity of RcNtrC than StNtrC to oxidizing conditions, since the StNtrC protein did not exhibit this increased phosphorylation with higher concentrations of reducing agents (data not shown).
The phosphotransfer reaction and autophosphatase activity of RcNtrC were studied under the optimal conditions described above. RcNtrCϳP was detectable within the first 30 s of incubation with MBP-RcNtrBϳP and reached a steady state maximal level after 15 min (Fig. 2, lane 7) that was stable for at least 1 h (Fig. 2, lane 8). RcNtrC alone did not label in the presence of [␥-32 P]ATP under any conditions (Fig. 2, lane 5). To directly compare the phosphorylation of the RcNtrC and St-NtrC proteins, we used the MBP-RcNtrBϳP protein to phosphorylate both proteins. The RcNtrC and StNtrC proteins were phosphorylated at approximately the same rate by the MBP-RcNtrB protein; the maximal level of RcNtrCϳP was comparable to the StNtrCϳP (Fig. 5B). Previous work demonstrated that the enteric NtrCϳP has an autophosphatase activity that recycles the protein to its unphosphorylated form. (Phosphate is also removed from RcNtrCϳP by a mechanism called regulated dephosphorylation, which requires both the NtrB and GlnB proteins (Ref. 3).) To determine if the RcNtrCϳP has an autophosphatase activity, RcNtrCϳP was formed and its decay was measured and compared to the decay for the StNtrCϳP protein. RcNtrCϳP had an autophosphatase activity with a half-life of 2-4 min (Fig. 6) similar to the 3-5 min observed for StNtrCϳP (Fig. 6, and see Ref. 3).
In order to address the specificity of the RcNtrC protein for phosphate donors (or kinase proteins), we tested the ability of RcNtrC to be phosphorylated by a variety of substrates. Small molecular weight high energy phosphate compounds (acetyl phosphate, carbamyl phosphate, and phosphoramidate) have been shown to phosphorylate the enteric NtrC protein (50) and other members of the response regulatory protein family in vitro (51). Radiolabeled acetyl phosphate was prepared and incubated with both the StNtrC and RcNtrC proteins for 10 and 40 min; StNtrCϳP was detected after 10 min and increased in concentration during the following 40 min; however, no label was incorporated into the RcNtrC protein after 40 min (data not shown). Unlabeled acetyl phosphate (Sigma), carbamyl phosphate (Sigma), and either ammonium or potassium phosphoramidate (each at 20 mM) did not inhibit the phosphorylation of RcNtrC by the MBP-RcNtrBϳP (data not shown), whereas these compounds can compete with the enteric NtrBϳP for enteric NtrC phosphorylation (52). Additionally, DNase I footprinting (see below) of the RcNtrC protein was not enhanced by any of these small molecular weight compounds. The enteric NtrB protein can phosphorylate other response regulatory proteins, including the CheY protein in vitro (49). To determine if the RcNtrC protein could be phosphorylated by the enteric NtrB protein in vitro, phosphorylated StNtrB protein was prepared and incubated in the presence of the RcNtrC protein. No loss of label was observed from the StNtrBϳP, and no label was incorporated into the RcNtrC protein (Fig. 2, lane  6). In this respect it is interesting that R. capsulatus ntrB mutants that are not polar on R. capsulatus ntrC are still Nif Ϫ (38), suggesting that cross-talk with other kinases or RcNtrC phosphorylation by small molecular weight compounds does not occur in vivo. This is in contrast with the enteric NtrC protein (53) and consistent with a hypothesis that the phosphorylation domain of RcNtrC may form a structure that makes it less accessible than the domain of the StNtrC protein for other substrate kinases.
Phosphorylation of RcNtrC protein is predicted to occur on an aspartate residue within the conserved N terminus; RcNtrC has an aspartate at residue 53 that is highly conserved between other members of the NtrC class of enhancer-binding proteins (54). Using the same methods described for RcNtrBϳP, the RcNtrCϳP was determined to be sensitive to base (a 3-fold loss of signal was observed in 0.5 N NaOH) and stable in acid (less than 10% loss of signal was observed in 1 N HCl; data not shown), indicative of a serine, threonine, or aspartate residue (4,55,56).
DNase I Footprinting with the Unphosphorylated and Phosphorylated RcNtrC Protein in Vitro-During purification, the RcNtrC protein eluted from a gel filtration column in fractions that suggested a native molecular mass of dimers rather than the 54-kDa monomer (31). The enteric NtrC protein naturally forms a dimer in solution (57,58), and upon activation by phosphorylation, it oligomerizes at specific DNA binding sites (8). To more fully characterize the purified RcNtrC protein, a 95% pure preparation of the RcNtrC protein was loaded onto a gel filtration column (Sephacryl S200-HR, Pharmacia) calibrated by size standards to determine its native size. The majority of the purified RcNtrC protein elutes predominantly at 110 kDa, confirming that it behaves as a dimer (Fig. 7A). The RcNtrC protein was also observed as a dimer by Western blot analysis of semi-native polyacrylamide gels (see "Experimental Procedures"). Antibodies against RcNtrC were able to detect a monomer form (54 kDa), a dimer form (110 kDa), and oligomer forms (Ͼ150 kDa) of the RcNtrC protein (Fig. 7B). The dimer form was the predominant species when 20 nM, 100 nM, or 500 nM protein was loaded onto the gel (Fig. 7B). The addition of SDS (1%) to the loading buffer caused greater than 90% of the upper bands to shift to the 54-kDa band (see Fig. 2, lane 8). Additionally, cell extracts of R. capsulatus grown under nitrogen limiting conditions revealed the presence of both the monomer and dimer forms of RcNtrC by Western analysis (Fig. 7B,  lane 4). Thus, the RcNtrC protein forms a dimer in vitro and presumably in vivo.
The enteric NtrC protein binds in vitro and in vivo to specific sites on DNA far upstream of promoters (22,23,59). Phosphorylation of the enteric NtrC protein stimulates enhancer binding in vitro by 4 -20-fold by oligomerization at tandem binding sites (7,8,15). We wanted to analyze this property of the RcNtrC protein. DNase I footprinting was performed with the phosphorylated RcNtrC protein to determine if, like the enteric NtrC, RcNtrCϳP has enhanced DNA binding activity compared to unphosphorylated RcNtrC. RcNtrC was phosphorylated by the MBP-RcNtrBϳP under optimal conditions (various concentrations of RcNtrC were incubated for 15-45 min at 24°C with 1 M MBP-RcNtrB protein in the presence of 10 mM DTT) and allowed to bind to an end-labeled DNA probe containing the glnB promoter region. DNase I digests were performed as described under "Experimental Procedures," and the regions that were protected from digestion were compared to the regions protected by the unphosphorylated RcNtrC protein.
Complete protection was observed at the upstream binding sites of the glnB promoter at 160 nM unphosphorylated RcNtrC (Fig. 8A) but at 40 nM with phosphorylated RcNtrC (Fig. 8C). The binding of unphosphorylated RcNtrC was unaffected by the presence of ATP (Fig. 8, compare A and B). This increase in binding is also clearly indicated by the hypersensitive site induced by RcNtrC binding at the tandem sites (Fig. 8, see  arrowheads). We conclude that phosphorylation of RcNtrC increases DNA binding by approximately 4-fold, similar to the enteric NtrC system. Similar to the enteric NtrCϳP, RcNtrC may show enhanced DNA binding activity due to increased oligomerization at the enhancer (12). Shaded bars mark the strong tandem binding sites of RcNtrC (see text); numbers refer to the distance from the transcriptional start; large and small arrowheads mark areas of increased DNase I sensitivity. Brackets indicate the regions of protection described previously (32).