Fnr Is Involved in Oxygen Control of Herbaspirillum seropedicae N-Truncated NifA Protein Activity in Escherichia coli

Abstract

Herbaspirillum seropedicae is an endophytic diazotroph belonging to the β-subclass of the class Proteobacteria, which colonizes many members of the Gramineae. The activity of the NifA protein, a transcriptional activator of nif genes in H. seropedicae, is controlled by ammonium ions through its N-terminal domain and by oxygen through mechanisms that are not well understood. Here we report that the NifA protein of H. seropedicae is inactive and more susceptible to degradation in an fnr Escherichia coli background. Both effects correlate with oxygen exposure and iron deprivation. Our results suggest that the oxygen sensitivity and iron requirement for H. seropedicae NifA activity involve the Fnr protein.

Herbaspirillum seropedicae is an endophytic diazotrophic bacterium found in tissues of important agricultural members of the Gramineae, such as wheat, rice, and maize (4). Interest in this bacterium has increased because of its potential as a biofertilizer for such important crops (4). Infection by H. seropedicae occurs by binding of bacteria to the root surface and proliferation on the secondary roots and at sites of root surface damage; this is followed by penetration and aggregation of the bacteria in intercellular spaces and vascular bundles and then by colonization and establishment in the xylem vessels (28, 16, 29). Nitrogen fixation in this organism occurs under microaerobic conditions and is regulated at both the level of synthesis and the level of activity of the NifA protein, the nif-specific transcriptional activator, in response to the levels of fixed nitrogen and oxygen (8). In members of the gamma subclass of the class Proteobacteria (gamma-proteobacteria), regulation of the transcriptional activity of the NifA protein by these two effectors involves the NifL protein, which forms an inactive complex with NifA in the presence of high levels of ammonium or oxygen. On the other hand, in H. seropedicae and Azospirillum brasilense (beta- and alpha-proteobacteria, respectively) the NifA protein is directly inactivated in response to increased levels of fixed nitrogen and oxygen (2, 32). Although the mechanisms of NifA activity control in these two groups of microorganisms differ, the signaling pathways leading to the ammonium response have similarities. In Klebsiella pneumoniae the GlnK protein is required to relieve the inhibitory effect of NifL on NifA under nitrogen-fixing conditions (1, 12, 15), whereas in Azotobacter vinelandii the GlnK protein is required for ammonium-dependent inhibition of NifA by NifL (20, 30, 21). In A. brasilense and H. seropedicae, which do not contain NifL, the PII protein, the product of the glnB gene, is necessary to relieve autoinhibition and ammonium control of NifA activity by its N-terminal domain (2, 5, 32).

The signaling pathway for control of NifA activity in response to oxygen is even less clear. In gamma-proteobacteria, the NifL protein is reversibly oxidized, and in this state it forms a transcriptionally inactive complex with NifA (13). On the other hand, sodium dithionite can reduce NifL of A. vinelandii in vivo, resulting in a protein unable to complex NifA. Although the in vivo NifL-reducing and -oxidizing species have not been defined yet (10, 18, 23), it has been suggested that a heme protein may be involved (13). The NifA proteins from rhizobia, A. brasilense, and H. seropedicae are not active in the presence of high oxygen concentrations and require iron for in vivo activation of nif gene promoters, suggesting that the NifA proteins of this class may sense oxygen directly (2, 32, 9, 25). It has been suggested that the oxygen sensitivity of these NifA proteins involves a cysteine motif located at the end of the central domain and a linker region for the C-terminal domain, which resembles an iron-sulfur cluster-binding motif (9).

An alternative possible iron-containing signal transducer is the Fnr protein. This transcriptional regulator is responsible for the switch from aerobic metabolism to anaerobic metabolism, and it responds to molecular oxygen (34). In this paper we show that the Fnr protein is essential for the activity of the N-truncated NifA protein of H. seropedicae in an Escherichia coli background. A lack of Fnr also makes an NifA protein more susceptible to degradation, possibly through proteolysis, an effect that correlates with the inhibitory effects of oxygen exposure and iron deprivation. The data suggest that the oxygen sensitivity of the N-truncated NifA protein of H. seropedicae and its iron requirement may in part be related to the Fnr protein activity.

MATERIALS AND METHODS

In order to determine if the N-truncated NifA protein (a protein with 202 amino acid residues deleted from the N-terminal domain and containing only the central and C-terminal domains) was active in an E. coli fnr background (strain JRG1728), we overexpressed it from a lac promoter and from a tac promoter (plasmids pRAM7 and pRAM8, respectively) and analyzed its ability to activate transcription of a nifHDK::lacZ fusion (pIMA217). The effect of iron and EDTA on the stability and transcriptional activity of the N-truncated NifA protein was also determined. Immunoblot analyses were carried out by using polyclonal antibodies against the central and C-terminal domains to demonstrate that the protein was present under all conditions tested.

Plasmids and strains.

The E. coli strains and plasmids used are shown in Table 1.

TABLE 1.

E. coli strains and plasmids

Strain or plasmid	Phenotype and/or genotype	Reference
E. coli strains
M182	Δ(lacIOPZYA)X74 galU galK rpsL Δ(ara-leu) Smr	33
JRG1728	Derived from M182 Δ(tyrR fnr rac trg)17 zdd-30::Tn9 Cmr	33
Plasmids
pRAM6	H. seropedicae His-tagged N-terminal domain of NifA protein expressed from pTZ18R vector (lac promoter)	26
pRAM7	H. seropedicae His-tagged N-truncated NifA expressed from pTZ18R vector (lac promoter)	26
pRAM8	H. seropedicae His-tagged N-truncated NifA expressed from pDK6 vector (tac promoter)	This study
pNH11	ptac-nifA (K. pneumoniae) Ampr	3
pIMA217	nifHDK(H. seropedicae)::lacZ Tcr	22

Plasmid construction.

The XbaI/HindIII fragment of pRAM7 (26) encoding an N-truncated NifA protein was subcloned into pDK6 (17), generating plasmid pRAM8.

Transcription activation of a nifHDK::lacZ fusion.

E. coli strains containing pIMA217 and different plasmids were grown in Luria-Bertani medium overnight at 30°C and then diluted to an optical density at 600 nm of 0.2 in NFDM medium (7) containing 0.5 mmol of isopropyl-β-d-thiogalactopyranoside (IPTG) per liter and antibiotics. After incubation for 8 h at 30°C under air or N₂, cell cultures were analyzed for β-galactosidase activity as described previously (24).

Effect of iron on the N-truncated NifA protein.

For analyses of the effect of iron and EDTA on the stability and transcriptional activity of the N-truncated NifA protein, E. coli strain M182 harboring the plasmids indicated above was grown in NFDM medium as described above, except that iron was not added. In some experiments (see below), EDTA (0.5 mmol/liter) was added to the iron-free NFDM medium.

Protein analyses.

Protein expression was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (19). Immunoblot analyses were carried out as described previously (6) by using a chemiluminescence assay (ECL; Amersham Biosciences). A densitometric analysis was performed with a Molecular Dynamics PSI densitometer. β-Galactosidase activity was determined by using o-nitrophenyl-β-d-galactopyranoside as described by Miller (24) and was expressed either in Miller units or as total activity as indicated below.

RESULTS AND DISCUSSION

E. coli strains harboring plasmids pIMA217 and pRAM7 were analyzed for transcriptional activation of the nifHDK::lacZ fusion. N-truncated NifA (with the first 202 amino acid residues deleted and without the N-terminal domain and the Q-linker) activated transcription of the H. seropedicae nifH promoter in E. coli under low-oxygen conditions (Table 2) regardless of the ammonium levels, as shown previously (25). However, in fnr strain JRG1728 of E. coli, the N-truncated NifA protein failed to activate the H. seropedicae nifH promoter either in the presence or in the absence of oxygen. A wild-type behavior was observed when strain JRG1728 was transformed with a plasmid carrying the fnr gene (31) expressed from its own promoter. We could not use the full-length NifA protein in these experiments since it has no activity in E. coli (25, 26). On the other hand, as expected, the K. pneumoniae NifA protein was fully active under all conditions tested, including in the fnr mutant (Table 2), since it is not directly sensitive to oxygen or directly dependent on fnr (10). Immunoblot analysis (23) of the N-truncated NifA protein produced from pRAM7 in extracts of cells grown under the same conditions showed that the lack of NifA activity in the fnr strain JRG1728 corresponded to the absence of the N-truncated NifA protein (Fig. 1). A 95% decrease in the N-truncated NifA protein content was observed in the fnr strain compared to the content in the wild-type strain M182 under anaerobic conditions. We also expressed the N-terminal domain of the NifA protein produced from pRAM6 (lac promoter) in the fnr mutant, and it was easily detected in the presence or absence of O₂ (data not shown), indicating that the absence of the N-truncated NifA protein in the fnr mutant was not due to inactivation of the lac promoter in this mutant and also that the observed effect was specific for the central and C-terminal domains of NifA, which show sensitivity to oxygen.

TABLE 2.

Activation of H. seropedicae nifH::lacZ expression by the N-truncated NifA protein of H. seropedicae in wild-type and fnr mutant E. coli strains

Strain/plasmid	Relevant genotypeb	Expression of nifH::lacZa (Miller units)
		With O2	Without O2
M182	Wild type	5 ± 1	0
JRG1728	fnr mutant	3 ± 1	6 ± 2
M182/pNH11	Wild type, KpnifA	2,349 ± 100	2,443 ± 200
JRG1728/ pNH11	fnr mutant, KpnifA	2,412 ± 80	2,683 ± 150
M182/pRAM7	Wild type, HsΔNnifA	20 ± 5	410 ± 20
JRG1728/pRAM7	fnr mutant, HsΔNnifA	10 ± 2	15 ± 3
JRG1728/pRAM8	fnr mutant, ptac-HsΔNnifA	3 ± 1	2 ± 1
JRG1728/pRAM8/pGS24	fnr mutant, ptac-HsΔNnifA fnr	5 ± 2	177 ± 12

^aThe values are the means ± standard deviations for at least three independent assays. Cells were incubated under air (with O₂) or under N₂ (without O₂).

^bKpnifA, K. pneumoniae nifA; HsΔNnifA, H. seropedicae N-truncated nifA.

FIG. 1.

Expression of the H. seropedicae N-truncated NifA protein in wild-type E. coli strain M182 (+) or in an fnr mutant E. coli strain (−). Cells were incubated under an N₂ atmosphere (lanes 1 and 2) or under air (lanes 3 and 4) in the presence of 0.5 mmol of IPTG per liter for 8 h at 30°C. Lanes 5 and 6 contained cells without plasmids incubated under air. Lane 7 contained purified N-truncated NifA. Ten micrograms of cell extract was loaded into each lane. The immunoblots were developed by using the chemiluminescence assay.

These results indicate that the Fnr protein is necessary for NifA activity, possibly because it maintains the N-truncated NifA protein in a reduced state. The decrease in the NifA concentration in the fnr background suggests that the inactive form of the N-truncated NifA protein was degraded in vivo. These observations also suggest that there is either an indirect or an additional requirement for Fe ions for the activity of NifA, inasmuch as iron is required for the activity of the Fnr protein (11).

The N-truncated NifA protein was not detected when its expression was induced in the wild type under air, an effect similar to the effect in the fnr background under anaerobic conditions (Fig. 1). To confirm that NifA was degraded in the presence of high oxygen concentrations, cells harboring plasmid pRAM7 were induced with 0.5 mmol of IPTG per liter in the absence of oxygen for 4 h and then either exposed to air or kept under anaerobic conditions. In the presence of oxygen the N-truncated NifA protein content decreased by 6% in the first 30 min, by 43% after 1 h, and by 69% after 2 h, suggesting that there was protein degradation (Fig. 2A). After 4 h of exposure to oxygen, only 10% of the NifA was still present (data not shown). In contrast, the N-truncated NifA protein content in the absence of oxygen showed little variation. To determine if the increased protein degradation was due to nitrogen starvation, experiments were carried out in minimal medium containing 20 mmol of ammonium chloride per liter (Fig. 2B) or in Luria-Bertani medium (data not shown), and the same results were observed, indicating that the proteolysis was not due to nitrogen limitation. The requirement for O₂ for NifA protein degradation suggests that there is activation of proteolytic systems or exposure of susceptible proteolytic sites of NifA in the presence of O₂. Also, the transcriptional activity of the N-truncated NifA protein was inhibited by oxygen since the total β-galactosidase activity of the H. seropedicae nifH::lacZ fusion did not increase under air, suggesting that there was inactivation of NifA prior to degradation (Fig. 3A).

FIG. 2.

Effect of oxygen and EDTA on the stability of the N-truncated NifA protein (ΔN-NifA). E. coli strain M182 harboring plasmid pRAM7 was grown in NFDM medium (A), in NFDM medium containing 20 mmol of NH₄Cl per liter (B), or in iron-free NFDM medium containing 0.5 mmol of EDTA per liter (C) under anaerobic conditions with 0.5 mmol of IPTG per liter and then either exposed to air or kept under N₂ for different times. Ten micrograms of cell extract was loaded into each lane. N-truncated NifA was detected by immunoblotting and was quantified by densitometry.

FIG. 3.

Effect of oxygen and EDTA on the transcriptional activity and concentration of the N-truncated NifA protein (ΔN-NifA). E. coli strain M182 harboring plasmids pRAM7 and pIMA217 was grown under nitrogen in the presence of 0.5 mmol of IPTG per liter and then incubated under different conditions. Samples containing 0.5 mmol of EDTA per liter (+ EDTA) or no EDTA (− EDTA) were cultured in iron-free NFDM medium and kept under anaerobic conditions. The total number of units of β-galactosidase activity per 100 μl of cell culture (A) and the relative concentrations of the N-truncated NifA protein (B) were determined at different times. The data are representative of the data obtained in five independent assays.

Souza et al. (32) demonstrated that the NifA protein of H. seropedicae requires iron for activity. Sequestration of iron by addition of EDTA to E. coli M182 containing pRAM7 under anaerobic conditions also led to inactivation, followed by degradation of NifA (Fig. 2C and 3). The N-truncated NifA protein content had decreased by 84% 2 h after addition of 0.5 mmol of EDTA per liter, and the total β-galactosidase activity remained constant (Fig. 3), indicating that iron is necessary for the structural stability and function of the N-truncated protein. These results suggest that the NifA sensitivity to oxygen and iron depletion are related and that the fnr gene is essential for activity and stability of the N-truncated NifA protein from H. seropedicae in an E. coli background.

Inactivation of the NifA protein is a key event leading to regulation of nitrogen fixation in the presence of high oxygen concentrations. Two diverse mechanisms have been described for oxygen-mediated inactivation of NifA. In gamma-proteobacteria, the NifA protein is controlled by a second regulatory component, the flavoprotein NifL, which can be reversibly oxidized in vitro and can complex with NifA in the oxidized form (13). In the alpha- and beta-proteobacteria, the NifA protein has a cysteine motif resembling a metal-binding site, requires iron ions for activity, and does not require NifL for inactivation by oxygen (8). These observations have been interpreted as indications that there is direct sensitivity of NifA to oxygen (8). Here we show that the activity of the H. seropedicae N-truncated NifA protein in E. coli requires the Fnr protein and that either the lack of Fnr, oxygen exposure, or iron depletion leads to N-truncated NifA degradation. Since Fnr is a transcriptional activator which is active in the presence of low oxygen concentrations, these results suggest that NifA requires an activating protein whose expression is dependent on Fnr. It is not clear whether iron is also required for NifA activity, whether NifA can be directly inactivated by oxygen, or whether both effects involve Fnr in vivo. Although the bulk of the available data suggest that in alpha- and beta-proteobacteria NifA is destabilized in vivo in the presence of high oxygen concentrations, the purification under air of a partially active N-truncated form of H. seropedicae NifA (25) suggests that NifA inactivation is in part an in vivo phenomenon that is possibly related to the activity of the Fnr protein. Although we used a heterologous system, at least two fnr-like genes were identified in H. seropedicae by analysis of the bacterial genome (The Herbaspirillum seropedicae Genome Sequencing Project, GENOPAR, unpublished results), and it is possible that Fnr is involved in NifA activity also in H. seropedicae. Whether one or more of such gene products is involved in NifA activity in this organism remains to be tested.

Recently, Grabbe et al. (10) reported that fnr null mutants of E. coli failed to release NifL inhibition of K. pneumoniae NifA transcriptional activity under anaerobic conditions. Morret et al. (27) also observed a loss of Bradyrhizobium japonicum NifA activity in E. coli upon exposure to oxygen or iron depletion, followed by NifA degradation. Huala et al. (14) showed that the tolerance to oxygen of Sinorhizobium meliloti NifA in E. coli was related to decreased proteolysis. These results are consistent with our observations and support the hypothesis that an Fnr-like protein is required for NifA activity in both rhizobia and H. seropedicae, which do not contain the NifL protein. It is possible that the same pathway, which involves the Fnr protein, signals low oxygen concentrations to NifL of K. pneumoniae and to NifA of H. seropedicae and rhizobia. However, because of the global involvement of Fnr in the aerobic-to-anaerobic switch (34), it is premature to suggest that there is a direct interaction between Fnr and NifA activity.