Regulation of the transcriptional activators AnfA and VnfA by metals and ammonium in Azotobacter vinelandii

Abstract

Transcription of the genes encoding molybdenum (Mo)-independent nitrogenases 2 and 3 of Azotobacter vinelandii requires the activators VnfA and AnfA, respectively. The effect of NH⁺₄, Mo, or V (vanadium) was tested on the expression of vnfA-lacZ and anfA-lacZ transcriptional fusions. Mo repressed expression of both fusions whereas NH⁺₄ and V repressed the anfA-lacZ fusion, but not the vnfA-lacZ fusion. Thus the repressive effect on transcription of the anfHDGKOR operon by NH⁺₄, Mo, or V is mediated through their effect on transcription of anfA and the repressive effect of Mo on the vnfHFd and vnfDGK operons is mediated through Mo repression of vnfA transcription. Mo-dependent repression of anfA transcription is influenced but not entirely mediated by the Mo-responsive regulator ModE.

Azotobacter vinelandii, Nitrogen fixation, Regulation, anfA, vnfA, modE

1 Introduction

The diazotroph Azotobacter vinelandii possesses three nitrogenases that are expressed in response to the presence or absence of molybdenum (Mo) or vanadium (V) under diazotrophic growth conditions (for review see [1]). The synthesis of all three nitrogenases is repressed by NH⁺₄. It has been shown that NH⁺₄ regulates nitrogenase 2 at the posttranscriptional level as well as the transcriptional level while nitrogenases 1 and 3 are repressed at the transcriptional level [2]. Nitrogenase 1 is found in cells grown in the presence of Mo and nitrogenase 2 is expressed in the presence of V and in the absence of Mo whereas nitrogenase 3 is only synthesized in the absence of Mo and V [1]. The structural genes for nitrogenase 1 and nitrogenase 3 are arranged in a single transcriptional unit, nifHDK and anfHDGK, respectively, whereas those for nitrogenase 2 are split into two operons, vnfHFd and vnfDGK[1]. This permits independent expression of dinitrogenase reductase 2 (product of vnfH) and dinitrogenase 2 (product of vnfDGK). Unlike dinitrogenase 2, dinitrogenase reductase 2 is not only expressed under diazotrophic conditions in the presence of V, but also in the absence of Mo and V, conditions under which nitrogenase 3 is expressed [1, 3]. Dinitrogenase reductase 2 has been shown to be required for transcription of the anfHDGKOR operon [4].

The regulatory genes nifA, vnfA, and anfA encode transcriptional activators which are required for the transcription of nif, vnf, and anf operons, respectively [1, 5]. VnfA is also directly or indirectly involved in the repression of nitrogenase 1 in cells growing diazotrophically in Mo-deficient medium with or without V [5]. VnfA seems also to be involved, albeit indirectly, in mediating repression of nitrogenase 3 by V [5–7]. All three transcriptional activators, NifA, VnfA, and AnfA, activate transcription from the nifB promoter [8]. The activator proteins have a three domain structure [5]: an N-terminal domain thought to interact with a transmitter of environmental signals [9]; a central domain proposed to interact with RNA-polymerase, the sigma factor or both; and a C-terminal DNA-binding domain. vnf promoters adjacent to both the vnfHFd and vnfDGK operons contain duplicated inverse repeat sequences which are required for efficient transcriptional activation by VnfA, presumably as VnfA-binding sites [10]. The N-terminal domains of AnfA and VnfA differ from the N-terminus of NifA in that they have a cysteine cluster, reminiscent of the FNR family of redox-sensitive regulatory proteins [5, 11]. In the case of AnfA, cysteines-21 and -26 are essential for AnfA function [11]. Using chimeric transcriptional activators generated from VnfA and AnfA of A. vinelandii, Frise et al. [12] have shown that promoter specificity is determined by the C-terminal region of AnfA and that the requirement of dinitrogenase reductase 2 for transcriptional activation of the anfHDGKOR operon [4] by this protein is a function of its N-terminal domain. Austin and Lambert [13] demonstrated that an N-terminally truncated AnfA with the central and C-terminal DNA-binding domains intact could activate transcription in an in vitro transcription system. They also showed binding of AnfA to sites between 200–300 base pairs upstream of the anfH promoter [13].

Molybdenum-dependent transcriptional regulation of genes encoding the high-affinity molybdenum transporter, and the Escherichia coli moaA operon is mediated by the regulatory protein ModE [14–16]. In contrast to A. vinelandii and E. coli, Rhodobacter capsulatus has two genes in the mod (molybdenum transport) locus that encode ModE-like proteins, called MopA and MopB. Deletion of these genes results in constitutive expression of anfA[17].

It was of interest to see if the repressive effects of NH⁺₄, Mo, and V on nitrogenases 2 and 3 are brought about through their effect on VnfA and AnfA, respectively. The effect of ModE on molybdenum repression of A. vinelandii vnfA and anfA was also investigated. To this end, we have constructed strains carrying promoter-lacZ fusions for anfA and vnfA and we have investigated the effect of Mo, V, or NH⁺₄ on the expression of β-galactosidase activity by these fusion strains.

2 Materials and methods

2.1 Bacterial strains and media

The bacterial strains and plasmids used in this study are listed in Table 1. The manipulation of A. vinelandii strains and the removal of contaminating Mo from Mo- and V-deficient Burk medium (designated −Mo and −V medium) were done as described previously [4, 11]. When required, Na₂MoO₄ and V₂O₅ were added to −Mo,−V medium at the desired concentration. Fixed N was added as ammonium acetate (final concentration, 28 mM). When needed, kanamycin (10 μg ml⁻¹), ampicillin (50 μg ml⁻¹), and spectinomycin (20 μg ml⁻¹) were included in A. vinelandii growth medium. E. coli DH5α and pUC18 were purchased from Gibco-BRL, Gaithersburg, MD. E. coli strain DH5α was cultured at 37°C in Luria-Bertani medium supplemented with the appropriate antibiotics (kanamycin 50 μg ml⁻¹, ampicillin 100 μg ml⁻¹, spectinomycin 20 μg ml⁻¹, tetracycline 20 μg ml⁻¹). For β-galactosidase assays, promoter-lacZ fusion strains of A. vinelandii were transferred at least three times on Mo-deficient agar medium supplemented with NH⁺₄ and kanamycin before transfer to −Mo,−V medium of the same composition. After overnight growth, 40 ml of −Mo,−V medium supplemented with NH⁺₄ was inoculated to 10–12 Klett units (1 Klett unit=2×10⁶ CFU ml⁻¹) and growth was monitored with a Klett-Summerson photoelectric colorimeter equipped with a no. 66 filter. When the density of the culture reached 90–100 Klett units, the cells were collected by centrifugation at 5000×g for 5 min at 4°C, washed with 5 ml of −Mo,−V medium and resuspended in 40 ml of −Mo,−V medium. 10-ml aliquots were transferred to 50-ml flasks containing either NH⁺₄ (28 mM), Na₂MoO₄ (1 μM), V₂O₅ (1 μM) or no additions. Derepression was conducted by shaking at 30°C for 14 h for anfA-lacZ fusion strains and 6 h for vnfA-lacZ fusion strains. β-Galactosidase assays were performed on the derepressed cells as described previously [4, 11].

1

Bacterial strains and plasmids

Bacterial strain or plasmid	Relevant characteristics	Source or reference
E. coli
DH5α	deoR recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relAΔ(lacZYA-argFV169) (φ80 lacZΔM15), F−	Gibco-BRL
A. vinelandii
CA	wild-type	[21]
CA143	KanR; anfA143::lacZ-Km (LK) from pRP1	This study
CA144	KanR; anfA144::Km-lacZ (KL) from pRP2	This study
CA145	KanR; vnfA145::lacZ-Km (LK) from pMEVA2	This study
CA146	KanR; vnfA146::Km-lacZ (KL) from pMEVA3	This study
CA143.20	KanR, SpcR; anfA143::lacZ-Km (LK) modE::Ω	This study
CA145.20	KanR, SpcR; vnfA145::lacZ-Km (LK) modE::Ω	This study
Plasmids
pUC18	cloning vector	Gibco-BRL
pKOK6	AmpR CamR KanR; contains lacZ-Km cassette	[18]
pMJH3	AmpR; pUC9 containing anfA and most of anfH on a 3.8-kb EcoRI insert	[5]
pMEVA1	AmpR; pUC18 containing vnfA on a 2.1-kb insert	This study
pMEVA2	AmpR KanR; pUC18 containing lacZ-Km cassette from pKOK6 inserted in the Eco47III site of vnfA	This study
pMEVA3	Same as pMEVA2 except that the lacZ-Km cassette is in the reverse orientation	This study
pRP1	AmpR KanR; pUC9 containing lacZ-Km cassette from pKOK6 inserted in Bal1 deletion in anfA-anfH	This study
pRP2	Same as pRP1 except that the lacZ-Km cassette is in the reverse orientation	This study
pLAM 86441	Sm/SpcΩ from pHP45Ω inserted in ClaI deletion (291–826) in modE	[16]

2.2 Construction of lacZ fusion strains

Transformation of A. vinelandii with plasmid DNA and all DNA manipulations were done as described previously [11]. To construct the transcriptional lacZ fusion of vnfA, plasmid pMEVA1 DNA was linearized by digestion with Eco47III (nucleotide 1693 of the vnfA sequence in [5]), ligated to the Klenow filled 4.73-kbp BamHI fragment of pKOK6 [18] and transformed into E. coli DH5α. Kanamycin-resistant transformants pMEVA2 and pMEVA3 were selected. The orientation of the cartridge insertion was checked by digestion with EcoRI. In pMEVA2, the cartridge was inserted such that lacZ expression was driven by the vnfA promoter. This was designated vnfA::LK to differentiate it from pMEVA3 which contains the cartridge inserted in the opposite orientation (vnfA::KL).

The transcriptional lacZ fusion of anfA was constructed as follows. A 4.5-kb BalI fragment from pMJH3 (with a 2-kb deletion spanning the C-terminus of anfA (nucleotide 1468 of the anfA sequence in [5]) and the N-terminus of anfH (nucleotide 715 in [19])) was ligated to the 4.73-kb BamHI fragment from pKOK6, filled with Klenow enzyme and transformed into E. coli DH5α. Two Kan^R transformants containing plasmids, pRP1 (anfA::LK) and pRP2 (anfA::KL), with the cartridge inserted in opposite orientations were selected. A. vinelandii strain CA was transformed with the following DNA fragments to obtain the fusion strains: a 6.48-kb KpnI fragment from pRP1 and pRP2; a 6.83-kb SphI-KpnI fragment from pMEVA2 and pMEVA3; and a 8.4-kb SmaI fragment from pMEVA4 and pMEVA5. Kan^R, Amp^S transformants (indicative of double cross-over events) were selected. To construct a double mutant with an anfA-lacZ or vnfA-lacZ fusion and a modE deletion, strain CA143 (with an anfA-lacZ fusion) or CA145 (with a vnfA-lacZ fusion) was transformed with linearized pLAM 86441, containing an Sm/Sp Ω interposon from pHP45Ω in a ClaI deletion in modE, followed by selection of Kan^R, Spc^R transformants.

3 Results and discussion

In A. vinelandii nitrogenase 3 proteins are synthesized as a result of AnfA-activated transcription of anfHDGK genes [7]. Transcription of the anfHDGK genes is repressed by NH⁺₄, Mo, or V [1]. In order to determine if this repression was due to the repressive effect of NH⁺₄, Mo, or V on the synthesis of AnfA, an anfA-lacZ fusion strain (CA143) of A. vinelandii was constructed. β-Galactosidase activities obtained with strain CA143 under different conditions is shown in Table 2. The activities obtained in the presence of NH⁺₄, Mo, or V were 13%, 4%, and 27%, respectively, of the β-galactosidase activity in the absence of these agents. Strain CA144, which has the lacZ-kan cartridge insertion in the opposite orientation to that in strain CA143, had 10% of the activity observed for strain CA143 under −N,−Mo conditions. Furthermore, this activity was not affected by Mo or NH⁺₄ (Table 2). This indicates that repression of anfA transcription may account for the observed repression of nitrogenase 3 by NH⁺₄, V, or Mo. In Rhodobacter capsulatus repression of the alternative nitrogenase by Mo and NH⁺₄ also occurs at the level of transcription of anfA[17].

2

Expression of the anfA-lacZ transcriptional fusion in the presence and absence of Mo, V and NH⁺₄

Strain (genotype)	β-Galactosidase activity (Miller units)a
	No addition	Mo	V	NH+4
CA143 (anfA143::lacZ-Km)	586	23	158	76
CA144 (anfA144::Km-lacZ)	59	94	Ndb	88
CA143.20 (anfA::lacZ, modE::Ω)	639	123	182	94

^aCells were grown in −Mo medium containing 28 mM NH⁺₄ acetate and later derepressed for 16 h in the presence of 1 μM Na₂MoO₄ (Mo), 1 μM V₂O₅ (V), 28 mM NH⁺₄ acetate (NH⁺₄) or no additions. Each value represents the average of two experiments.

^bNot determined.

ModE has been reported to be a repressor of the modABCD (molybdate transport) operon in E. coli[20] and in A. vinelandii[14]. In E. coli ModE has also been shown to mediate Mo-dependent activation of the moa operon [15]. It was, therefore, of interest to determine the effect of Mo on the anfA-lacZ fusion in a modE mutant background (CA143.20). The β-galactosidase activity of CA143.20 in the presence of Mo was 21% of its activity under −N,−Mo conditions (Table 2). The β-galactosidase activities of strain CA143.20 in the presence of V or NH⁺₄ were not significantly different from those of CA143 under similar conditions (Table 2).

The above experiment was done at a Mo concentration of 1 μM in the medium. Since concentrations as low as 100 nM Mo are sufficient to cause repression, an experiment to determine the effect of Mo concentration on β-galactosidase activity was conducted with strains CA143 and CA143.20. The level of activity observed with CA143.20 remained at about 25% of the activity in the absence of Mo irrespective of the Mo concentration (Table 3).

3

Effect of Mo concentration on the expression of transcriptional anfA-lacZ fusions in the presence and absence of modE

Concentration of Mo in the medium	β-Galactosidase activity (Miller units)a
	CA143 (anfA143::lacZ-Km)	CA143.20 (anfA143::lacZ-Km, modE::Ω)
0	400	481
100 nM	40	125
500 nM	16	115
1.0 μM	20	111
100 μM	NDb	115
1.0 mM	ND	125

^aA. vinelandii cells were grown in −Mo medium supplemented with 28 mM NH⁺₄ acetate and derepressed for 16 h in −N,−Mo medium to which was added the desired concentration of Na₂MoO₄. Each value represents the average of two experiments.

^bNot determined.

Although the molecular details of how molybdenum might mediate repression are not currently known, the above results indicate that ModE plays a role in repression of anfA expression by molybdenum. However, relief of molybdenum repression in a modE mutant background is only partial and is unrelated to molybdenum concentration, thus other genes are likely to be involved in what might be a rather complex regulatory circuit for molybdenum repression of anfA transcription.

Table 4 gives the results of an experiment where β-galactosidase activities were measured under different conditions using strains CA145 (vnfA::lacZ) and CA145.20 (vnfA::lacZ, modE::Ω). β-Galactosidase was expressed in both strains under −N,−Mo and −N,+V conditions to approximately the same extent. Since dinitrogenase reductase 2 is present under −N,−Mo conditions [4], it is logical that vnfA is transcribed under these conditions. Ammonium did not repress the vnfA::lacZ fusion (Table 4). The lack of repression of the vnfA::lacZ fusion by NH⁺₄ is consistent with the observation that transcription of the vnfHFd and vnfDGK operons is only partially repressed by NH⁺₄[2]. Nitrogenase 2 subunits, however, are undetectable in extracts of cells derepressed for nitrogenase 2 in the presence of NH⁺₄[2]. This indicates that this nitrogen source acts at the posttranscriptional level as well as at the transcriptional level. In the case of VnfA, posttranscriptional repression by NH⁺₄ would not be expected since transcriptional activation of the vnfHFd and vnfDGK operons requires VnfA [5]. In the presence of Mo and V, both strains CA145 and CA145.20 gave β-galactosidase activities which were 41% and 19% of the activities obtained under −N,−Mo conditions when incubated for 6 h and 16 h, respectively. The fact that the β-galactosidase activities decreased as the time of exposure to Mo increased suggested that if the cells were pregrown in the presence of Mo before exposure to V in the presence of Mo, β-galactosidase expression might be repressed by a much greater extent. Results in Table 5 show that this is indeed the case, leading to the conclusion that Mo represses vnfA transcription.

4

Expression of the vnfA::lacZ transcriptional fusion in the presence and absence of Mo, V, and NH⁺₄

Strain (genotype)	Time of derepression (h)	β-Galactosidase activity (Miller units)a
		No addition	V	Mo+V	NH+4+V
CA145 (vnfA::lacZ-Km)	6	551	562	229	586
	16	527	550	107	643
CA145.20 (vnfA145::lacZ-Km; modE::Ω)	6	493	500	203	535
	16	693	711	137	736

^aCells were grown in −Mo medium containing 28 mM NH⁺₄ acetate followed by derepression for 6 or 16 h in the presence or absence of 1 μM Na₂MoO₄ (Mo), 1 μM V₂O₅ (V), and 28 mM NH⁺₄ acetate (NH⁺₄). Each value represents the average of two experiments.

5

Effect of Mo preloading on expression of the vnfA::lacZ fusion in strain CA145

Pregrowth conditions	β-Galactosidase activity (Miller units)a
	No addition	NH+4	V	Mo+V
NH+4	358	469	422	207
NH+4+Mo	16	67	16	15

^aCells were precultured in −Mo medium in the presence of 28 mM NH⁺₄ acetate (NH⁺₄) with or without 1 μM Na₂MoO₄ (Mo) followed by 6 h of derepression in the presence or absence of 1 μM Na₂MoO₄ (Mo), 1 μM V₂O₅ (V), and 28 mM NH⁺₄ acetate (NH⁺₄). Each value represents the average of four experiments.

The experiments described in this paper do not address the possibility of autoregulation by AnfA and VnfA, however, the transcriptional lacZ fusions of both anfA and vnfA seem to be repressed by molybdenum. Therefore, if autoregulation is coupled to regulation by molybdenum, its contribution is relatively minor. The small amounts of β-galactosidase activity (∼4%) observed in the presence of molybdenum for both the anfA::lacZ (Table 2) and vnfA::lacZ (Table 5) fusions are likely to be due to the persistence and stability of β-galactosidase.

In conclusion, our results show that: (a) anfA transcription is repressed by NH⁺₄, Mo, and V; (b) ModE plays a role in the repression of anfA transcription by Mo; (c) V is not required for the transcription of vnfA; and (d) Mo represses vnfA transcription but NH⁺₄ does not.

Acknowledgements

This was a cooperative study between the USDA Agricultural Research Service and the North Carolina Agricultural Research Service. This work was supported by U.S. Department of Agricultural Grant 92-37305-7722 and by NATO Travel Grant 880532.