How Posttranslational Modification of Nitrogenase Is Circumvented in Rhodopseudomonas palustris Strains That Produce Hydrogen Gas Constitutively

Abstract

Nitrogenase catalyzes the conversion of dinitrogen gas (N₂) and protons to ammonia and hydrogen gas (H₂). This is a catalytically difficult reaction that requires large amounts of ATP and reducing power. Thus, nitrogenase is not normally expressed or active in bacteria grown with a readily utilized nitrogen source like ammonium. nifA* mutants of the purple nonsulfur phototrophic bacterium Rhodopseudomonas palustris have been described that express nitrogenase genes constitutively and produce H₂ when grown with ammonium as a nitrogen source. This raised the regulatory paradox of why these mutants are apparently resistant to a known posttranslational modification system that should switch off the activity of nitrogenase. Microarray, mutation analysis, and gene expression studies showed that posttranslational regulation of nitrogenase activity in R. palustris depends on two proteins: DraT2, an ADP-ribosyltransferase, and GlnK2, an NtrC-regulated P_II protein. GlnK2 was not well expressed in ammonium-grown NifA* cells and thus not available to activate the DraT2 nitrogenase modification enzyme. In addition, the NifA* strain had elevated nitrogenase activity due to overexpression of the nif genes, and this increased amount of expression overwhelmed a basal level of activity of DraT2 in ammonium-grown cells. Thus, insufficient levels of both GlnK2 and DraT2 allow H₂ production by an nifA* mutant grown with ammonium. Inactivation of the nitrogenase posttranslational modification system by mutation of draT2 resulted in increased H₂ production by ammonium-grown NifA* cells.

INTRODUCTION

Many bacteria and archea can convert N₂ to ammonia, a biologically usable form of nitrogen that sustains life on earth. The energetically demanding reaction of reducing the triple bond of N₂ to ammonia is accomplished by the highly conserved enzyme molybdenum nitrogenase (5, 16) according to the equation N₂ + 8e⁻ + 8 H⁺ + 16 ATP → 2NH₃ + H₂ + 16 ADP (5, 16).

Nitrogen fixation is complex. In an initial phase of the reaction, two protons are reduced to H₂, a catalytic event that is hypothesized to prepare the active site of nitrogenase for subsequent reduction of N₂. H₂ is an obligate product of the nitrogen fixation reaction, and its production cannot be excluded, even under 50 atm of N₂ (5, 8, 16, 36). Each of the eight electrons that participate in the overall reaction resulting in H₂ and NH₃ production is delivered to the active site individually with accompanying hydrolysis of two ATPs. Thus, the overall rate of catalysis by nitrogenase is very slow (a catalytic turnover of about 5 s⁻¹), necessitating the synthesis of large amounts of enzyme (7, 12). Nitrogenase also contains several complex metal cofactors, whose synthesis and insertion into apoenzyme require the coordinated activities of multiple accessory proteins (34). Because nitrogen fixation is an energetically demanding and slow process and because nitrogenase is a complicated enzyme to make, its synthesis and activity tend to be strongly repressed by ammonium, the most readily usable form of nitrogen for bacteria.

Although the main function of nitrogenase is the production of ammonia, it has long been known that in the absence of N₂, nitrogenase reduces protons exclusively, forming pure H₂, a process that can potentially be developed for the biological production of hydrogen fuel (17, 25). We have studied the phototrophic bacterium Rhodopseudomonas palustris as a model organism for biological H₂ production via nitrogenase. R. palustris can generate the ATP needed for this energy-intensive process from the abundant resource of sunlight, and it can degrade structurally diverse organic compounds, including lignin monomers as a source of electrons for H₂ production (3, 13, 32). It can also use the inorganic compound thiosulfate as an electron source for both nitrogen fixation and H₂ production (14). From a bioengineering standpoint as well, R. palustris is a hardy organism that can produce H₂ continuously for months (10).

In initial work, we sought to identify R. palustris mutants that synthesized nitrogenase constitutively, even when ammonium was present. Such mutants would be required in an applied situation since ammonium-containing agricultural or industrial wastes would most likely be used as organic feedstocks for an H₂ production process (17). We obtained four mutants with different single-nucleotide changes in nifA, encoding the transcriptional regulator of nitrogenase genes. The responsible mutations caused four different single amino acid changes that probably locked NifA into a structural conformation that allowed it to activate nitrogenase gene expression in the presence of ammonium. We named these NifA* mutants. These mutants synthesize large amounts of H₂ when grown with ammonium as a nitrogen source, whereas the wild-type parent grown under the same conditions does not produce H₂ (32).

Nitrogenase regulation in response to ammonium has been studied in detail in several different bacteria and archaea, and such studies allow one to infer the broad outlines of major regulatory control features in a particular organism by inspection of its gene content. Proteomics experiments indicate that R. palustris likely senses ammonium availability, as do most bacteria and archaea, using P_II signal proteins (6, 21). These proteins are uridylylated/deuridylylated by GlnD a bifunctional uridylyltransferase/uridylyl-removing enzyme. When intracellular glutamine (a signal of nitrogen sufficiency) is low, GlnD adds uridylyl groups to P_II proteins to form P_II-UMP. This alters the conformation of P_IIs and changes their ability to interact with target proteins (9, 21, 27). We predict that nitrogenase synthesis and activity are regulated at three levels in R. palustris based on its genome sequence (20) (Fig. 1). These include regulation by (i) the NtrBC two-component system, which responds to fixed nitrogen limitation to control gene expression in many proteobacteria (27). In some bacteria, this includes expression of nifA. (ii) The second level includes NifA, the master transcriptional activator protein of nif genes encoding nitrogenase and accessory proteins. In Rhodospirillum rubrum, the NifA protein is activated posttranslationally by P_II-UMP (39). Finally, (iii) the third level of regulation includes DraT and DraG, enzymes that control nitrogenase activity posttranslationally by adding or removing ADP-ribosyl groups to its dinitrogenase reductase subunit (NifH) (22, 30). This third level of regulation is also policed by P_II proteins. Where studied, one or more P_II paralogs interact with DraT to activate its activity (15, 38). nifA* mutants produce a variant of the NifA protein that is able to activate nif gene expression even in ammonium-grown cells (32), but this does not explain how amino acid changes in NifA alone could also allow R. palustris cells to circumvent predicted regulation of nifA synthesis by NtrBC and nitrogenase inactivation by ADP-ribosylation. Here we address these questions by defining the NtrBC regulon and by characterizing the R. palustris DraT posttranslational modification system.

Fig 1.

R. palustris nitrogenase synthesis and activity are predicted to be regulated at three levels in response to fixed nitrogen availability. The predicted activities of regulatory proteins NtrBC, NifA, and DraT in ammonium-grown cells are shown on the left side, whereas the predicted activities of NtrBC, NifA, and DraG in cells that are starved for fixed nitrogen (nitrogen-fixing cells) are shown on the right side. As described in the text, the activities of these proteins are influenced by interactions with P_II proteins, which are either unuridylylated or uridylyated, depending on cellular nitrogen status.

MATERIALS AND METHODS

Bacterial growth.

R. palustris was grown anaerobically in sealed tubes in nitrogen-fixing (NF) medium, a nitrogen-free mineral-based minimal medium with N₂ in the headspace, for nitrogen-fixing conditions (28). When a fixed source of nitrogen was used, ammonium sulfate was added to 0.1% (photosynthetic medium [PM]) (18). Cells were grown with 10 mM sodium succinate for transcriptome experiments and for the initial growth experiments described in Table 2. For all other experiments, cells were grown with 20 mM sodium acetate, which supports a slightly faster rate of growth. In some experiments, strains were grown in NF medium or PM supplemented with 0.3% yeast extract and succinate in an argon (Ar) atmosphere, as indicated. R. palustris strains were grown anaerobically in light at 30 ± 2°C. For mutant selections, R. palustris was grown aerobically on CA plates (NF medium, 4 g/liter yeast extract, 5 g/liter Casamino Acids, 1.5% Bacto agar). Escherichia coli cells were grown in Luria-Bertani (LB) medium at 37°C with shaking at 250 rpm. Where indicated, R. palustris was grown in the presence of 70 or 100 μg gentamicin (Gm) sulfate per ml or 100 μg kanamycin sulfate (Km) per ml. E. coli was grown with 100 μg/ml ampicillin (Ap) or 20 μg/ml Gm.

Table 2.

Growth rates and nitrogenase activities of various R. palustris strains

Strain	Genotype	Growth ratea (doubling time in h)			Nitrogenase activityb (nmol C2H4/min/mg protein)
		PM + succinate	NF + succinate	NF + succinate + yeast extract
CGA010	Wild type	7.3 (0.4)	11.9 (0.8)	12.4 (0.3)	104 (14.2)
CGA805	ΔntrBC	7.3 (0.3)	105.4 (11.0)	13.4 (0.3)	13.1 (1.8)
CGA552	ΔnifA	7.3 (0.2)	NGc	8.7 (0.8)	<1

^aThe data shown are the average of three measurements, and standard deviations are shown in parentheses. PM contains ammonium sulfate, whereas NF medium does not.

^bStrains were grown on NF medium plus succinate and yeast extract under an Ar gas atmosphere. The data shown are the average of three measurements, and standard deviations are shown in parentheses.

^cNG, no growth.

Strain constructions.

All R. palustris strains described in this work are either wild-type strain CGA009 or derivatives thereof (Table 1). The ntrBC and nifA mutant strains used for microarray analysis were derived from CGA010, a derivative of CGA009 in which a hupV frameshift mutation was repaired (33). Gene deletions or replacements were done via homologous recombination. For glnB, glnK1, and draT2 mutations, DNA sequences of approximately 1 kb in size that flanked either side of the desired deletion were cloned into pUC19. A Km resistance cassette from plasmid pBSL15 was cloned between the two flanking regions. This construct was then cloned into the suicide vector pJQ200SK (31). For ntrBC, nifA, glnK2, and draT1 mutations, flanking regions were cloned into pJQ200SK, but no Km resistance gene was added. The ntrBC, nifA, glnB, glnK2, draT1, and draT2 deletion suicide vectors were transferred to R. palustris from E. coli S17-1 by conjugation. The glnK1 deletion suicide vector was electroporated into R. palustris as described previously (29). Single recombinants were selected on CA plates containing Gm. Double recombinants were selected on PM-succinate plates containing 10% sucrose (wt/vol) for ntrBC, nifA, glnK2, and draT1 or 10% sucrose (wt/vol) and Km for draT2, glnB, and glnK1. The sucrose was included to counterselect against the donor plasmid as described previously (32). Complementation vectors were constructed using pBBR1MCS-5 (19), and constructs were amplified in and purified from E. coli strain DH5α before being either transferred into E. coli S17-1 as a conjugation donor strain or directly electroporated into R. palustris. In all cases, the gene was cloned behind the plasmid promoter.

Table 1.

Strains and plasmids used in this study

Strain, plasmid, or primer	Genotype/phenotype or sequence	Source or reference
Strains
R. palustris
CGA009	Wild-type strain; spontaneous Cmr derivative of CGA001	18
CGA010	CGA009 hupS+ derivative	33
CGA552	ΔnifA in CGA010 background	This study
CGA676	NifA* nifAΔ48 bp, constitutive mutation	26
CGA653	ΔglnB::Kmr in CGA009 background	This study
CGA720	ΔglnK1::Kmr in CGA009 background	This study
CGA721	ΔglnK2 in CGA009 background	This study
CGA722	ΔdraT1::Kmr (RPA1431) in CGA009 background	This study
CGA723	ΔdraT2::Kmr (RPA2405) in CGA009 background	This study
CGA724	ΔdraT2 in CGA676 (NifA*) background	This study
CGA725	ΔglnK2 in CGA676 (NifA*) background	This study
CGA726	ΔntrBC in CGA676 (NifA*) background	This study
CGA805	CGA010 hupS+ ΔntrBC	This study
CGA806	CGA009 ΔntrBC	This study
E. coli
DH5α	F− λ−recA1 Δ(lacZYA-argF)U169 hsdR17 thi-1 gyrA96 supE44 endA relA1 ϕ80dlacZΔM15	GIBCO-BRL
S17-1	thi pro hdsR hdsM+recA; chromosomal insertion of RP4-2 (Tc::Mu Km::Tn7)	35
Plasmids
pJQ200SK	Gmr, sacB suicide vector	31
pUC19	Apr high-copy-no. cloning vector	37
pBSL15	Kmr cassette-containing plasmid	2
pJQntrBC	pJQ200SK containing ntrBC in-frame deletion construct	This study
pJQnifA	pJQ200SK containing nifA in-frame deletion construct	This study
pJQglnB	pJQ200SK containing glnB::Kmr deletion construct	This study
pJQglnK1	pJQ200SK containing glnK1::Kmr deletion construct	This study
pJQglnK2	pJQ200SK containing glnK2 in-frame deletion construct	This study
pEH012	pJQ200SK containing draT1 in-frame deletion construct	This study
pEH016	pJQ200SK containing draT2::Kmr deletion construct	This study
pBBR1MCS-5	Broad-host-range expression vector; Gmr	19
pEH023	pBBR containing glnK2 as well as its predicted promoter region	This study
pEH024	pBBR containing coding region of draT2	This study
pEH025	pBBR containing SDa consensus and coding region of glnK2	This study
pEH026	pBBR containing SD consensus and coding regions of glnK2 and draT2	This study
Primers for qRT-PCR
EH219	TGACGGAAGTGAAGGGATACG	glnK2 forward
EH220	CAGCGCCGCGGTAGATT	glnK2 reverse

^aSD, Shine-Dalgarno.

Nitrogenase assays.

Switch-off of nitrogenase activity was measured by monitoring the production of ethylene from acetylene or by monitoring the accumulation of H₂ in the culture headspace. For either assay, cultures of R. palustris were grown on acetate to an optical density at 660 nm (OD₆₆₀) of 0.35 to 0.45. For the acetylene reduction assay, 10 ml of cells was harvested anaerobically and resuspended in 10 ml of 25 mM sodium phosphate and potassium phosphate buffer (phosphate buffer), pH 7.0. Resuspended cells were transferred to 16-ml rubber septum-sealed tubes filled with Ar. Sodium acetate was added to 10 mM, and cells were allowed to recover at 30°C in light for 60 min. Then, 250 μl of 100% acetylene gas was added to the headspace of tubes to initiate the assay. Ethylene produced by nitrogenase activity was measured over time by gas chromatography as previously described (28). After linear ethylene production had been established (about 20 min after acetylene injection), either sodium chloride or ammonium chloride was injected into the assay tube to a final concentration of 100 μM. Stocks of 10 mM sodium chloride and ammonium chloride were prepared and stored anaerobically. Ethylene production was measured for at least 40 min after addition of sodium chloride or ammonium chloride. For the H₂ production assay, 60 ml of cells was harvested at an OD₆₆₀ of 0.35 to 0.45 and resuspended in 10 ml of 25 mM phosphate buffer before the addition of sodium acetate to 10 mM. Cells were transferred to sealed 27-ml rubber septum-sealed tubes in an Ar atmosphere and allowed to recover for 60 min before H₂ was measured by gas chromatography as described previously (14). The total protein content of cell suspensions was estimated from the OD₆₆₀ of the suspension compared with a standard curve prepared with whole R. palustris CGA009 cells grown under nitrogen-fixing conditions. To generate this curve, the Bio-Rad protein assay was used to measure total protein from NaOH-lysed cells (4).

Protein modification assay.

Protein for visualization of NifH posttranslational modification was prepared from cells that had been grown under nitrogen-fixing conditions with acetate and exposed to either 100 μM sodium chloride or 100 μM ammonium chloride for 30 min prior to harvest. The NIfA* strain was grown in PM-acetate. To arrest cell metabolism, 5 ml of treated cell suspension was placed in an ice water bath for 5 min and then centrifuged for 7 min at 4°C. The supernatant was decanted, and the cell pellet was frozen quickly in liquid nitrogen and stored at −20°C. Frozen cell pellets were thawed on ice and resuspended in 300 μl lysis buffer (25 mM Tris-HCl [pH 7.8], 150 mM NaCl, 1 mM EDTA, 1 mM fresh dithiothreitol [DTT], 0.5% Tween, 10% glycerol, 1× HALT protease inhibitor cocktail [Thermofisher]). Cells were lysed by sonication, and cell debris was removed by centrifugation at 4°C. Cell extracts were measured for total protein content by the Bio-Rad protein assay. Approximately 1 or 2 μg of protein was loaded per SDS-PAGE well. Low-cross-linker SDS-PAGE gels were used to resolve modified from unmodified protein (acrylamide/bisacrylamide ratio, 171:1). Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane and incubated with rabbit antiserum prepared against NifH protein purified from Azotobacter vinelandii. Anti-rabbit horseradish peroxidase-conjugated secondary antibody was hybridized to the primary antibody, and Pierce/ThermoFisher ECL Femto substrate was used for visualization.

Transcriptome analysis.

RNA was isolated from cells as described previously, and gene expression profiles were analyzed by Affymetrix GeneChip (32) and glass-spotted microarrays (28). For Affymetrix GeneChip analysis, the mRNA abundance of a gene was considered significantly up- or downregulated if the fold change was greater than or equal to 2, the P value was <0.001, and the posterior probability of differential expression (PPDE) threshold was >0.97. For glass-spotted microarray analysis, genes whose wild type/mutant expression ratios were ≥2 and whose scores (the probability that the logarithmic wild-type/mutant ratio of expression level is <0) were <0.025 were considered to be the genes that were expressed at higher levels in the wild-type strain. Genes whose expression ratios were ≤−2 and whose scores were >0.975 were considered to be the genes that were expressed at lower levels in the wild-type strain.

qRT-PCR.

For quantitative reverse transcription real-time PCR (qRT-PCR), cDNA was synthesized as in the Affymetrix GeneChip experiment [without adding a eukaryotic poly(A) RNA control] and purified with the QIAquick PCR purification kit (Qiagen). Primers were designed using the Primer Express version 2.0.0 software (Applied Biosystems). PCR mixtures included 1 ng of cDNA and primers at a concentration of 500 nM in 20 μl of SsoFast Evagreen PCR amplification Supermix (Bio-Rad). Genomic DNA was used as a standard. The temperature profile was as follows: 98°C for 2 min and then 45 cycles at 98°C for 5 s (denaturation) and 60°C for 10 s (annealing and extension). Amplification and the amount of PCR products were monitored by using the Bio-Rad CFX96 real-time PCR system. Each reaction was performed in triplicate, and average data from at least two experimental replicates are reported.

Microarray data accession number.

The microarray data have been deposited at http://www.ncbi.nlm.nih.gov/geo/ under accession no. GSE32292.

RESULTS

The NtrBC and NifA regulons.

NtrBC is a two-component regulator responsible for activating expression of nitrogen starvation genes in most proteobacteria. Where studied, the histidine kinase NtrB in combination with a bound nonuridylylated P_II protein stimulates dephosphorylation of the transcription factor NtrC. In glutamine-depleted cells, NtrB is not bound by P_II. This stimulates its kinase activity and phosphorylation of NtrC (27). In the phototrophic bacterium Rhodobacter capsulatus, NtrC-P activates expression of nifA. NifA in turn directly activates nif gene expression if intracellular glutamine is low. Consistent with its suspected role in controlling nitrogen fixation, an R. palustris ntrBC mutant grew extremely slowly under nitrogen-fixing conditions, but grew as the wild type when ammonium was supplied as the nitrogen source (Table 2). An nifA mutant failed to grow under nitrogen-fixing conditions. Addition of 0.3% yeast extract to nitrogen-fixing medium permitted growth of both the ntrBC and nifA mutants. Wild-type cells grown in nitrogen-fixing medium with 0.3% yeast extract and 10 mM succinate with an Ar headspace had levels of nitrogenase activity that were similar to that of cells grown in nitrogen-fixing medium with a N₂ headspace without yeast extract (28). Thus, we used medium supplemented with yeast extract to grow ntrBC, nifA, and wild-type strains and to compare their gene expression profiles. The nitrogenase structural genes nifDK were expressed at 50- to 100-fold-higher levels in wild-type cells grown with yeast extract compared to cells grown with ammonium (Table 3). Comparison of the gene expression patterns of the nifA mutant and the wild type confirmed, as expected, that expression of nifDK and nif accessory genes (RPA4602 to -4631) is tightly regulated by the NifA transcription factor (Table 3; see Table S1 in the supplemental material). The third structural gene, nifH, was not printed on the glass microarray slides, so we were unable to analyze its expression levels. Comparison of the gene expression profile of the ntrBC mutant with that of the wild type indicates that the NtrBC regulon includes about 350 genes (see Table S1). Many of these encode predicted amino acid transport systems that are likely involved in nitrogen acquisition under nitrogen starvation conditions. NtrBC also controls expression of glutamine synthetase genes that could be important for ammonium assimilation under nitrogen-fixing conditions (Table 3). As expected, the NtrBC regulon includes nifA, which was expressed at about a 3-fold-lower level in the ntrBC mutant. Most of the nif genes in the R. palustris nif gene cluster (RPA4602 to -4631) were expressed in the ntrBC mutant, but at much lower levels than in fully induced wild-type cells. Basal expression of nifA in the ntrBC mutant is likely responsible for the low but measurable levels of NifA regulon expression in this background (Table 3). Inspection of previously determined gene expression profiles of NifA* strains grown with ammonium showed that the NtrBC regulon was not turned on (32). The nifA gene was not induced in NifA* strains and was expressed at basal levels. However, basal levels of NifA* protein are sufficient to activate expression of nifHDK and other nif genes in NifA* strains, apparently because this protein is synthesized in a form that is active in ammonium-grown cells. In fact, the nif gene cluster in NifA* strains grown with ammonium is expressed at about 3 times higher levels than in wild-type cells grown under nitrogen-fixing conditions (32).

Table 3.

Transcriptome analysis of nitrogen regulatory mutants by microarray

RPA no.	Gene name	Annotation	Fold transcript change for:
			WT on YE/WT on NH4a	WT on YE/ΔntrBC mutant on YEb	WT on YE/ΔnifA mutant on YEb
0272	glnK1	Nitrogen regulatory protein PII	NCc	−3.6	NC
0273	amtB1	Ammonium transporter	NC	NC	NC
0274	glnK2	Nitrogen regulatory protein PII	49	22	−3.2
0275	amtB2	Ammonium transporter	28	24	NC
1374	vnfA	σ54-dependent, vanadium nitrogenase transcriptional regulator	9.8	4.4	5.1
1439	anfA	σ54-dependent, iron nitrogenase transcriptional regulator	2.8	NC	NC
1927		Hypothetical protein	75	4.5	39
1928		Ferredoxin-like protein [2Fe-2S]	47	4.2	22
2405	draT2	NAD+ ADP-ribosyltransferase	NC	NC	NC
2592	ntrB	Nitrogen regulatory histidine kinase	NC	Mutated	NC
2593	ntrC	Nitrogen regulatory response regulator	2.5	Mutated	NC
2966	glnB	Nitrogen regulatory protein PII	3.8	5.0	NC
2967	glnA	Glutamine synthetase I	2.3	2.7	NC
4209	glnA II	Glutamine synthetase II	56	26	−4.3
4602	fixX	Ferredoxin-like protein	68	3.9	59
4603	fixC	Nitrogen fixation protein	260	3.2	87
4604	fixB	Electron transfer flavoprotein alpha chain protein	260	2.3	9.2
4605	fixA	Electron transfer flavoprotein beta chain	372	4.4	70
4607	nifV	Putative homocitrate synthase	41	3.6	46
4612	fdxB	Ferredoxin 2[4Fe-4S] III	34	3.8	50
4618	nifK	Nitrogenase molybdenum-iron protein beta chain	110	4.3	81
4619	nifD	Nitrogenase molybdenum-iron protein alpha chain	68	4.0	93
4629	ferN	Ferredoxin 2[4Fe-4S]	46	2.0	6.3
4631	fer1	Ferredoxin 2[4Fe-4S]	95	3.2	NDd
4632	nifA	Mo/Fe nitrogenase-specific transcriptional regulator	5.7	2.8	Mutated

^aThe wild-type strain was grown on NF medium containing 0.3% yeast extract and 10 mM succinate (WT on YE) or PM containing 10 mM succinate (WT on NH₄). PM contains ammonium sulfate, whereas NF medium does not. Values are fold transcript change determined by Affymetrix GeneChip using 2 biological replicates for each sample.

^bBoth wild-type and mutant strains were grown on NF containing 0.3% yeast extract and 10 mM succinate. Values are fold transcript change determined by glass-spotted microarrays using 2 calibration and 2 (WT/ΔnifA) or 3 (WT/ΔntrBC) comparative experiments (with 2 slides for each experiment).

^cNC, no significant change.

^dND, not determined.

R. palustris nitrogenase activity is inhibited by ammonium, and NifH is posttranslationally modified.

A number of nitrogen-fixing bacteria posttranslationally regulate nitrogenase activity by ADP-ribosylation. This has been characterized primarily in the purple nonsulfur bacteria R. rubrum (30, 38) and R. capsulatus (24), as well as in Azospirillum brasilense (15). Transfer of an ADP-ribose moiety by dinitrogenase reductase ADP-ribosyltransferase (DraT) from NAD to a conserved arginine on dinitrogenase reductase (NifH) inactivates nitrogenase. This modification occurs when cells that have active nitrogenase are exposed to ammonium and is reversed by dinitrogenase reductase-activating glycohydrolase (DraG) in response to removal of ammonium. R. palustris contains genes that encode two predicted DraTs and one predicted DraG (20). Inhibition of nitrogenase activity in response to ammonium (also known as “switch off”) has been reported in R. palustris (1, 41). To verify that this regulation occurs in strain CGA009, we grew cells under nitrogen-fixing conditions, washed them, and assayed for inhibition of nitrogenase activity in response to addition of ammonium. We found that nitrogenase activity was inhibited by about 70% after addition of 100 μM ammonium chloride compared to addition of 100 μM sodium chloride (Fig. 2A). Addition of 100 μM sodium chloride resulted in slightly reduced nitrogenase activity, possibly due to a nonspecific response to salt or to introduction of trace amounts of oxygen. Immunoblot analysis showed that the NifH protein was modified when cells were exposed to ammonium, as evidenced by the appearance of a slower-migrating form of NifH on SDS-PAGE gels (Fig. 2B). The pattern of modification looks very similar to that previously reported for NifH modification by ADP-ribosylation in other bacteria (23, 40). Cells treated with sodium chloride did not show such an NifH protein modification.

Fig 2.

Switch-off nitrogenase activity in R. palustris. (A) Nitrogenase activities of wild-type cells grown under nitrogen-fixing conditions were measured before and after exposure to 100 μM ammonium chloride (open symbols) or sodium chloride (closed symbols). Only the ammonium-exposed cells lost the majority of their nitrogenase activity. Data are representative of at least four different experiments. (B) Posttranslational modification of NifH protein in wild-type (WT) cells by addition of ammonium chloride. (C) NifH protein from the NifA^* strain grown with ammonium is modified slightly. Cell proteins were separated on low-cross-linker polyacrylamide gels in order to resolve the modified and unmodified forms of NifH. NifH was visualized by immunoblotting.

The NifH protein from an ammonium-grown NifA* strain is slightly modified.

Our original nifA* mutants have single missense mutations that map to the Q-linker region of NifA between its N-terminal GAF domain and the central AAA+ domain that precedes its C-terminal helix-turn-helix DNA binding domain (32). These mutants revert to the wild-type phenotype with detectable frequency. Thus, we subsequently constructed a stable nifA* mutant that has a 16-amino-acid deletion in the Q-linker region of the protein (26). This NifA* strain (CGA676) produces H₂ when grown with ammonium and behaves as the previously described nifA* missense mutants, and we will refer to it as the NifA* strain here. When we harvested CGA676 cells grown in ammonium-containing medium and examined the NifH protein in immunoblots, we found that it was modified, but to a much lesser extent than NifH from nitrogen-fixing-grown wild-type cells that have been switched off by ammonium (Fig. 2C).

DraT2 and GlnK2 are each required for nitrogenase switch off by ammonium.

To understand why nitrogenase activity from the NifA* strain was relatively resistant to ammonium, we needed to investigate the specific mechanism governing posttranslational modification of nitrogenase in wild-type R. palustris. Toward this end, we constructed draT1 (RPA1431) and draT2 (RPA2405) mutants and tested their sensitivity to ammonium. We found that as with the wild type, nitrogenase activity in the draT1 mutant was inhibited by ammonium addition. In contrast, draT2 mutant nitrogenase activity was relatively insensitive to ammonium addition (Fig. 3A and C). This mutant phenotype was complemented when the draT2 gene was expressed in trans from a broad-host-range plasmid (Fig. 3B and C). Previous work in our laboratory indicates that draT2 is constitutively expressed in wild-type and NifA* strains. Levels of expression are the same in ammonium-grown wild-type and NifA* strains as well as in the wild type grown under nitrogen-fixing conditions (28, 32).

Fig 3.

Evidence that DraT2 ADP-ribosyltransferase is responsible for nitrogenase switch off in R. palustris. (A) Switch-off nitrogenase activities of draT1 (diamonds) and draT2 (triangles) mutants. Open symbols indicate ammonium chloride addition, and closed symbols indicate sodium chloride addition. draT1 cells exposed to ammonium chloride lost significant nitrogenase activity, whereas cells exposed to sodium chloride lost little activity. The draT2 mutant nitrogenase, in contrast, was not inactivated by addition of ammonium chloride. Data are representative of at least four independent experiments. (B) The inability of the draT2 mutant to inactivate nitrogenase activity in response to ammonium chloride was restored by provision of the draT2 gene in trans (ΔdraT2/pEH024 [circles] and ΔdraT2/vector [squares]). (C) Anti-NifH immunoblots indicate that in the absence of DraT2, NifH is not ADP-ribosylyated and that this phenotype can be complemented.

Bacteria that have the DraG/DraT system have two or three P_II paralogs (15, 24, 38). P_II proteins have been shown to interact with DraT to activate its ADP-ribosylation activity (15, 24, 38), but the P_II protein specificity for this varies depending on bacterial species. R. palustris contains genes that encode GlnB, GlnK1, and GlnK2 P_II proteins (20). We investigated the role of each of these by constructing mutants and assaying for nitrogenase inhibition in response to ammonium. Nitrogenase activities in glnB and glnK1 mutants were each sensitive to ammonium addition (Fig. 4A). The NifH proteins from each strain were also modified in response to ammonium (Fig. 4D). However, the nitrogenase activity of the glnK2 mutant was insensitive to ammonium addition, and its NifH protein did not become modified (Fig. 4B and D). This mutant phenotype was complemented by provision of the glnK2 gene in trans (Fig. 4C and D). These results suggest that GlnK2 is required to activate DraT2.

Fig 4.

Evidence that the P_II protein GlnK2 is essential for nitrogenase switch off in R. palustris. (A) Switch-off nitrogenase activities of glnK1 (triangles) and glnB (diamonds) mutants. Sodium chloride (closed symbols)-exposed cells lost little activity, whereas ammonium chloride (open symbols)-exposed cells lost significant nitrogenase activity. Data are representative of at least four independent experiments. (B) The glnK2 mutant (squares) did not lose nitrogenase activity after ammonium chloride addition, as did the wild type (WT) (circles). (C) The glnK2 phenotype was complemented by provision of the pEH023 in trans. (D) Anti-NifH immunoblots corroborate the results seen in the nitrogenase activity assays, showing that GlnK2 is required for NifH protein modification.

The glnK2 gene is expressed at low levels in ammonium-grown NifA* cells.

The glnK2 gene is expressed at 30-fold-higher levels in N₂-grown, compared to ammonium-grown, wild-type cells (28). Previous transcriptome data showed that GlnK2 expression levels in the NifA* strains are not elevated in cells grown with ammonium, even though these strains have high levels of nitrogenase gene expression (32). We verified these results in the NifA* strain used in this study by qRT-PCR (Table 4). Low levels of GlnK2 protein in the NifA* strain would explain why its nitrogenase is resistant to posttranslational modification and inhibition when cells are grown with ammonium. Inspection of the ntrBC mutant microarray data provides an explanation for this regulatory pattern. Transcription of the NtrBC regulon is not activated in wild-type cells grown with ammonium (Table 3) or in NifA* cells (32), and glnK2 is part of the NtrBC regulon.

Table 4.

Expression of glnK2 genes determined by qRT-PCR

Strain	Genotype	Plasmid	Growth medium	Expression of glnK2 cDNA (pg)a
CGA009	Wild type		NF	589 (61)
CGA009	Wild type	pBBR1MCS-5 (control)	NF	409 (29)
CGA009	Wild type		PM	9 (2)
CGA009	Wild type	pBBR1MCS-5 (control)	PM	10 (1)
CGA676	nifA*		NF	289 (21)
CGA676	nifA*		PM	32 (6)
CGA676	nifA*	pBBR1MCS-5 (control)	PM	10 (2)
CGA676	nifA*	pEH025 (glnK2)	PM	650 (33)
CGA725	nifA* ΔglnK2		NF	<1
CGA725	nifA* ΔglnK2		PM	<1
CGA726	nifA* ΔntrBC		NF	11 (2)
CGA726	nifA* ΔntrBC		PM	7 (1)

^aThe data shown are the average picograms of glnK2 cDNA measured from at least two experiments compared to a genomic DNA standard curve. Standard errors of the mean are shown in parentheses. PM contains ammonium sulfate, whereas NF medium does not. Cultures had an N₂ atmosphere. Acetate was the carbon source for growth.

Lack of glnK2 expression in the NifA* strain does not fully explain its resistance to switch off.

To further probe our hypothesis that the incomplete-switch-off nitrogenase in ammonium-grown NifA* was due to a lack of adequate glnK2 expression, we expressed glnK2 from a constitutive promoter in trans and measured production of H₂ by ammonium-grown cells. We verified by qRT-PCR that glnK2 was well expressed (Table 4). Surprisingly, the amount of H₂ produced by the strain dropped only slightly compared to that in the vector control strain (Table 5). In complementary experiments, we found that the NifA* strain grown with N₂ as its sole nitrogen source (a condition which should upregulate GlnK2 expression) and subsequently exposed to exogenous ammonium chloride retained most of its nitrogenase activity. Thus, it appears that the resistance of the NifA* strain to posttranslational modification by ADP-ribosylation cannot be solely due to low expression levels of glnK2.

Table 5.

Expression of GlnK2 or DraT2 in trans reduces H₂ production by the ammonium-grown NifA* strain

Strain	Plasmid	Hydrogen production (μmol/mg total protein)a	% relative to NifA* (pBBR1MCS-5) strain
CGA009	pBBR1MCS-5 (control)	<1	0
CGA676	pBBR1MCS-5 (control)	81 (3.4)	100
CGA676	pEH025 (glnK2)	66 (2.6)	81
CGA676	pEH024 (draT2)	50 (3.4)	62
CGA676	pEH026 (glnK2 and draT2)	18 (2.1)	22

^aThe data shown are the average of at least three experiments. Standard errors of the mean are shown in parentheses.

We have previously reported that NifA* strains express nitrogenase genes at higher levels than wild-type cells grown under nitrogen-fixing conditions (32). In agreement with this, we found that the nitrogenase activity of the NifA* strain was three times higher than that of the wild type (not shown). Thus, a possible explanation for the inability of the NifA* strain expressing glnK2 to significantly switch off nitrogenase activity is that the amount of active DraT2 that it produces is insufficient to completely inactivate the increased amount of nitrogenase that it synthesizes. To test this, we expressed draT2 from a plasmid in the NifA* strain. The nitrogenase activity and amount of H₂ produced in this strain grown with ammonium were reduced relative to those of the vector control (Table 5). When we expressed both glnK2 and draT2 from the same plasmid in the NifA* strain, we observed a substantial reduction in the amount of H₂ produced relative to the vector control strain. Thus, inadequate expression levels of both glnK2 and draT2 in the NifA* strain contribute to the resistance of this strain to switch off its nitrogenase.

Introduction of a draT2 mutation into the NifA* strain results in increased production of H₂.

Our initial observation that nitrogenase from the NifA* strain grown with ammonium was partially modified, most likely by ADP-ribosylation (Fig. 2C), suggested there was some active DraT2 present. To test this, we constructed an nifA* draT2 double mutant. Protein immunoblots showed that this strain did not modify its nitrogenase protein (data not shown). We also observed that the nifA*draT2 double mutant produced more H₂ than the nifA* strain (Table 6).

Table 6.

H₂ production by ammonium-grown NifA* cells is increased when switch off is eliminated

Strain	Genotype	Hydrogen production (μmol/mg total protein)a	% relative to NifA* (pBBR1MCS-5) strain
CGA009	Wild type	<1	0
CGA676	nifA*	80 (3.6)	100
CGA724	nifA* ΔdraT2	107 (10.0)	134

^aThe data shown are the average of at least three experiments. Standard errors of the mean are shown in parentheses.

DISCUSSION

Here we found that nitrogenase is posttranslationally modified by the DraT “switch-off” system in R. palustris, as was predicted from its genome sequence. We observed that switch off in wild-type cells resulted in a decrease in nitrogenase activity and the appearance of higher-molecular-weight bands in anti-NifH immunoblots. This is consistent with posttranslational modification by ADP-ribosylation. The substantial amount of unmodified NifH that is present in inactivated samples is in keeping with past studies in other bacteria showing that only one subunit of the NifH dimer is modified at any one time, presumably because modification at one subunit sterically blocks modification on the opposing subunit (11, 30).

A revised model for three-tiered regulation of nitrogenase expression and activity in R. palustris derived from our data is shown in Fig. 5. At the top level, in wild-type cells (Fig. 5A), NtrB responds to fixed nitrogen limitation by failing to interact with P_II protein, phosphorylating NtrC, and allowing the transcription of genes associated with nitrogen starvation, including nifA. In ammonium-grown cells, nifA, which is normally upregulated 2- to 3-fold by NtrBC, is not induced. At level 2, NifA is activated under low-nitrogen growth conditions, presumably by P_II-UMP, to become proficient to turn on nif gene expression. The NifA* variant (Fig. 5B) does not need to be activated by P_II-UMP. Moreover, basal levels of this protein are sufficient for it to turn on expression of nitrogenase genes during growth with ammonium. Thus, nitrogenase genes, but not NtrBC-regulated genes, including glnK2, are expressed in the NifA* strain grown with ammonium. At the third level, in wild-type cells, the constitutively expressed DraT switch-off system posttranslationally modifies and inactivates nitrogenase in response to exogenous addition of ammonium, when it interacts with GlnK2 (Fig. 5A). GlnK2 is not well expressed in ammonium-grown NifA* cells because the NtrBC regulon is not expressed. Furthermore, nitrogenase is expressed at elevated levels in the NifA* strain, overwhelming the switch-off apparatus. These two factors allow continued activity of nitrogenase and H₂ production in the presence of ammonium. Therefore, a single mutation in nifA is sufficient to bypass three levels of regulatory control and allow H₂ production by R. palustris in the presence of ammonium.

Fig 5.

A model for the regulation of nitrogenase in wild-type and NifA^* R. palustris strains. (A) In N₂-grown wild-type (WT) cells, NtrB trans-phosphorylates NtrC, which then upregulates genes of the NtrBC regulon, including glnK2 (encoding a PII isozyme) and nifA. NifA undergoes a conformational change which may be promoted by a P_II-UMP. This allows it to upregulate nif gene expression and the synthesis of nitrogenase. DraG, which is constitutively synthesized in R. palustris, ensures the continued activity of nitrogenase by removing any ADP-ribose moieties that may be attached. When wild-type cells are exposed to ammonium, GlnK2 binds to DraT2 (also constitutively synthesized) to activate its ability to ADP-ribosylate nitrogenase. (B) When the NifA^* strain is grown with ammonium as a nitrogen source, NtrB protein is inactivated by interaction with P_II protein (an unknown combination of GlnB, GlnK1, or GlnK2), and it cannot promote NtrC phosphorylation. As a consequence, neither the P_II gene, glnK2, nor nifA is upregulated. However, sufficient basal expression levels of NifA^* promote the overexpression of the nif genes. We assume that the NifA^* protein does not need to interact with a P_II protein to be active, but we have not shown this. Without adequate GlnK2 and with overexpression of nitrogenase, the majority of nitrogenase escapes ADP-ribosylation by DraT2, and as a result, H₂ is produced when the NifA^* strain is grown in the presence of ammonium.

We have shown that the R. palustris switch-off system depends on GlnK2, one of three P_II-proteins found in this species. It was not possible to predict based on homology to other systems which functions a given P_II protein will perform. Also, P_II paralogs in a single bacterium can have redundant functions. The use of only one P_II protein to activate switch off is unusual among purple nonsulfur bacteria. R. palustris is so far unique in that only GlnK2, which is part of the NtrBC regulon and is not expressed when the cells are grown in the presence of ammonium, can activate the nitrogenase switch-off system. When NifA* mutations were introduced in R. rubrum, it was necessary to also inactivate the draT gene in order to get H₂ production in the presence of ammonium because its activity is controlled by a constitutively expressed P_II protein (40). We also found that inactivation of the switch-off system by mutation of the ADP-ribosyltransferase gene draT2 in an NifA* R. palustris strain resulted in increased H₂ production in ammonium-grown cells. This is in keeping with our observation that nitrogenase expressed in NifA* cells was posttranslationally modified to some degree (Fig. 2C). However, it is not clear that the small amount of modified NifH that we observed is enough to account for the 30% increase in H₂ production seen when the modification was blocked (Table 6). It seems possible that DraT2 may regulate other functions, such as electron transfer to NifH.

The data presented here have helped to elucidate the regulatory hierarchy controlling nitrogenase expression and activity in R. palustris. We have shown that even when the components of a regulatory network have been well studied in other bacteria, understanding exactly how these components work together in a bacterium of interest can be important for determining how to engineer such a bacterium for a practical application—in this case, H₂ production. We have also solved a paradox posed by our previous work on the NifA* mutant and gained a better understanding of nitrogen regulation in general in R. palustris.