Purification and in vitro activity of a truncated form of ANFA. Transcriptional activator protein of alternative nitrogenase from Azotobacter vinelandii.

The ANFA protein is the transcriptional activator of the sigma 54-dependent anfHDGK operon, which codes for the structural genes of the third nitrogenase system in Azotobacter vinelandii. We have purified, in soluble active form, an N-terminally truncated form of the protein, delta ANFA, which activates transcription from the anfH promoter and other sigma 54-dependent promoters in a purified transcription system. Sequences upstream of the anfH promoter and the presence of the integration host factor protein stimulate transcription, and we have shown that delta ANFA binds to sites situated between 200 and 300 base pairs upstream of the anfH promoter. In common with other sigma 54-dependent activators, ANFA has a highly conserved ATP binding motif in its central domain, and we have demonstrated that ATP or GTP is required for productive complex formation and that the purified truncated protein has a constitutive ATPase activity, which is presumably required to drive open complex formation.

nitrogenase, and an alternative nitrogenase that contains neither vanadium nor molybdenum. Expression of all three nitrogenase systems is repressed by high concentrations of fixed nitrogen, but the availability of metals regulates the expression of the individual systems (reviewed in Bishop and Premakumar (1992)). Each system has its own set of nitrogenase structural genes, and a different specific activator protein is required for the expression of these genes: NIFA for nijTlDK, VNFA for vnfTIDGK, and ANFA for anfHDGK (Joerger et al., 1989a). Azotobacter NIFA is homologous to the NIFA protein from Klebsiella pneumoniae, and a conserved NIFA binding motif has been identified upstream of the nif promoters in A. uinelandii. All three sets of structural gene promoters have well conserved -12 to -24 sequences characteristic of d4-dependent promoters and require the RNA polymerase d4 holoenzyme ( E d 4 ) to activate transcription (reviewed in Merrick (1992)). NIFA, VNFA, and ANFA all have the three-domain structure characteristic of the family of d4-dependent activators . The N-terminal domains believed to be involved in environmental sensing are not highly homologous, but both VNFA and ANFA have a conserved motif (Cys-X-Cys-X,-Cys in VNFA and Ser-X-Cys-X,-Cys in ANFA), which may be involved in the interaction of these proteins with metals to form * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ''advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
f To whom correspondence should be addressed. Tel.: 44-273-678290; a redox-sensitive site (Joerger et al., 1989a). Active vnfH or nifH gene products, the vanadium and molybdenum nitrogenase iron proteins, have been shown to be required for anf-HDGK expression, and it has been suggested that transfer of electrons from the vanadium iron protein to ANFA activates ANFA by reduction of a redox site in the N terminus (Joerger et al., 1991). The central domains of NIFA, VNFA, and ANFA are highly homologous and contain the ATP binding motif characteristic of other d4-dependent activators (Joerger et al., 1989a;Ronson et al., 1987). The C-terminal domains of these proteins contain a helix-turn-helix motif involved in DNA binding. The sequence of the proposed recognition helix in A. vinelandii NIFA is homologous to that conserved in other NIFA proteins and reflects a common DNA binding site (upstream activator sequence (UAS)).' VNFA and ANFA do not contain the same recognition helix and are predicted to recognize different upstream sequences (Joerger et al., 1989a). Recently a UAS different from the conserved NIFA binding site has been proposed for VNFA (Kennedy et al., cited in Merrick (199311,' but that for ANFA remains to be identified. Zn vitro studies with NIFA proteins from a number of organisms have been hindered by the very insoluble nature of these proteins when overproduced (Tuli and Merrick 1988;Austin et al., 1990). Although the predicted amino acid sequences of the alternative NIFAs are homologous to the family of NIFA proteins, particularly in the C-terminal region of the protein, we considered that there might be sufficient differences to change the solubility characteristics of these proteins compared with NIFA. I n vivo in an enteric background, ANFA activation of the anfN promoter is dependent on the presence of the iron protein, but removal of the N-terminal domain of ANFA alleviates this requirement and produces a protein that is more transcriptionally active than the fulllength form of t h e p r~t e i n .~ Thus, it appears that in the absence of the iron protein the N-terminal domain has a negative role, and removal of this gives rise to a protein that has an enhanced ability to activate transcription. We have purified the N-terminally truncated ANFA in a soluble active form and characterized the requirements for transcriptional activation with d4dependent promoters using i n vitro transcription and DNA binding assays.  FIG. 1. a diagrammatic representation of the domain structure of ANFA, b, position of fusion junction between lac2 and anfA. The first seventeen codons of lac2 are fused to the SphI site in the Q-linker region in &A. The coding sequence of anfA starts with the alanine residue a t position 209 in the protein. c, DNA sequence of the anfH promoter and the upstream regulatory region. The sequence shown (Joerger et al., 1989b) is that spanning the two HincII sites from +27 to -405 with respect to the predicted transcription start site ( + I ) , The potential IHF binding site are underlined.

K C A G T A T T T T C M T T O A T T C J L I L T A G G C T G C T C O C R G T C C
positions of the -24 to -12 downstream promoter element and the

Methods
Purification of Proteins: Overexpression and Purification of Duncated ANFA-A 1.2-kilobase pair EcoRI-Hind111 fragment from pEF3213 carrying the anfA gene with 625 bp of the coding sequence deleted from the 5'-end and replaced by the first 17 codons of lacZ fused to the SphI site in the Q-linker region ( Fig. 1, a and b ) was cloned into the expression vector pTTQl8 (Amersham Corp.). This places the coding sequence under control of the tac promoter. The pEF321 construct had been previously shown to synthesize a transcriptionally active prot e i~-~.~ The overexpression construct (pJL2) directed the synthesis of a stable 45-kDa fusion protein when induced with IPTG. Purification was achieved from 1 to 21 cultures of ET8894(pJL2) grown aerobically in Luna broth and induced for 3-4 h with 1 mM IPTG. Crude cell extract was obtained by French pressure disruption and low speed centrifugation followed by ammonium sulfate precipitation to 55% saturation of the low speed supernatant. Chromatography was carried out in TGED buffer (10 mM Tris-C1, pH 8, 5% glycerol, 0.1 mM EDTA and 1 mM dithiothreitol). The ammonium sulfate precipitate was dissolved in TGED buffer and chromatographed on a 5-ml heparin agarose column with a linear salt gradient from 50 to 500 m~ NaCl. The AANFA eluted at 350-400 mM NaCl. The peak fractions were pooled and applied to a 1-ml Mono Q column (Pharmacia Biotech Inc.) where the protein eluted at 350 mM NaCl. Further purification was achieved by gel filtration on a Superose 12 column (Pharmacia), which had been calibrated using protein markers for gel filtration chromatography (Sigma). For long term storage, the protein was kept in small aliquots in storage buffer (TGED containing 50% glycerol and 50 mM NaC1) in liquid N, but was also stable in this buffer at -20 "C for several months. Purification stages were monitored by SDS polyacrylamide gel electrophoresis and Western blotting of column fractions using a rabbit polyclonal antibody to ANFA. Protein concentrations were determined using Coomassie Blue G-250 reagent (Pierce Chemical Co.) and also estimated from densitometric scanning of stained bands on SDS gels. The most highly purified protein preparations were judged to be at least 99% pure with no other contaminating bands by visual inspection of overloaded Coomassie Blue-stained SDS gels. Native gel electrophoresis and Western blotting of native gels was carried out on a PhastSystem using 8-25 or 10-15% nondenaturing gradient gels and the Phastransfer system (Pharmacia). Molecular masses were estimated by comparison with protein markers for nondenaturing gel electrophoresis (Sigma). The molar concentration of AANFA was calculated assuming the protein to be a dimer in solution.
Core RNA polymerase and d4 from K pneumoniae were purified as described previously (Whitehall et al., 1992). Escherichia coli integration host factor (IHF) protein was the kind gift of Howard Nash.
Dunscription Assays-Single round transcription assays with purified proteins were carried out in either acetate buffer (50 mM Tris acetate, pH 7.8,lOO mM potassium acetate, 8 mM magnesium acetate, 27 mM ammonium acetate, 1 mM dithiothreitol) or chloride buffer (50 mM Tris-HC1, pH 7.8, 10 mM magnesium chloride, 100 mM potassium chloride, 0.1 mM EDTA, 0.1 mM dithiothreitol, 50 pg/ml bovine serum albumin). Template DNA (10 IN) was preincubated at 37 "C for 20 min with core RNApolymerase (75 IN); d4 (230 m); ATP, GTP, and CTP (0.4 mM); and AANFA at the concentrations indicated. Transcript synthesis was initiated by adding heparin (100 pg/ml) and either [36SlUTPaS or [32PlUTP and unlabeled UTP (final UTP concentration,12 p~) in a reaction volume of 50 pl. After 10 min, the reaction was terminated by adding an equal volume of stop mix (5 M ammonium acetate, 100 mM EDTA). RNA transcripts were phenol extracted, ethanol precipitated, and then run on 6% polyacrylamide sequencing gels. The radioactive bands were identified by autoradiography of the dried gel. Transcripts were quantitated either by densitometrically scanning the bands corresponding to full-length transcripts or by excision of radioactive bands containing full-length transcripts and counting in Cocktail T (BDH) in a LKl3 Wallac scintillation counter.
DNA Templates-For the transcription assays, the plasmid containing the full-length anfH promoter (pSA8) was constructed by cloning a 431-bp HzncII fragment carrying the downstream promoter element and the region upstream of this to -405 (see Fig. IC) into the HincII site of the transcription vector pTElO3 (Elliot and Geiduschek, 1984). pSA9 was derived from this by deleting the upstream region from the NruI site at -53 to the BamHI site in the polylinker in pTElO3. Where indicated, supercoiled template DNA was linearized by digestion with ScaI. Other transcription templates (pRD580, pRD581, and pTES74) were as described previously (Austin et al., 1987;Whitehall et al., 1992). DNA templates for the footprinting reactions were constructed by cloning the 450-bp BamHI-Hind111 fragment from pSA8 into pTZ18 to give pSAlO and pTZ19 to give pSA11. Single-stranded template DNA from these constructs was extended with 32P-5'-end labeled primers to produce duplex DNA templates for dimethyl sulfate and DNasel footprinting. Oligonucleotide primers were 5'-CGCAAGTATGGCCGW-3' from -404 to -389 for pSA10, and 5'-CCAGGTAACCAGGAGAGC-3' from -116 to -132 for pSA11.
Footprinting-Footprinting reactions were carried out in acetate buffer as used for the transcription assays in a final volume of 50 1. 11. Template DNA and AANFA were incubated together at the concentrations indicated in the figure legends for 20 min at 30 "C before adding the footprinting reagent. Dimethyl sulfate and DNasel footprints were carried out as described by Buck and Cannon (1992).
ATPase Assays-Measurement of ATP hydrolysis was carried out as described by Austin and Dixon (1992).

RESULTS
Purification of N-terminally Zkuncated ANFA-Attempts to purify the full-length form of the ANFA protein from A. vinelandii were unsuccessful due to the tendency of the overproduced protein to precipitate when exposed to conditions necessary for chromatography. As the in vivo data indicated that the removal of the N-terminal domain of the protein enhanced its ability to activate transcription and eliminated its dependence on the presence of the iron p r~t e i n ,~ we considered that an N-terminally deleted form of the protein might be more amenable to purification and characterization in vitro. We constructed a plasmid in which the N-terminal domain of approximately 20 kDa was deleted by fusion of the first 17 codons of When placed under the control of the tac promoter, this construct directs the synthesis of a fusion protein with a subunit molecular mass of approximately 45-kDa after induction with IPTG. The protein remained in the cell supernatant on low speed centrifugation after cell lysis (Fig. 2u, compare lanes 2 and 3) and was purified by ammonium sulfate fractionation of the crude extract followed by chromatography on heparin agarose, then on Mono Q, and finally on Superose 12 on the fast protein liquid chromatography system (see "Methods"). On gel filtration on Superose 12, AANFA chromatographed as a broad peak (Fig. 2, b and c). The peak fractions corresponding to molecular masses of approximately 200-90 kDa (frac-tions 6-8) were pooled and dialyzed into storage buffer. Analysis of the Superose 12-pooled fractions by native gel electrophoresis confirmed that these contained species representing the tetramer to dimer forms of the protein (Fig. 2d). The activity of the purified protein before and after Superose 12 chromatography was roughly the same. Similar concentrations of protein were required to activate transcription and to detect DNA binding in footprinting experiments. The Superose 12 fractions were used in all future experiments. The identity of the protein was confirmed by N-terminal sequence analysis in that the first 9 N-terminal residues corresponded to those of the lac2 gene product. The truncated protein also cross-reacted with a polyclonal antibody raised against the full-length ANFA.
Danscriptional Activation a t anfH Promoter in Vitro-The purified truncated ANFA (AANFA), was able to activate the A. vinelandii anfH promoter in a single round transcription assay. The anfH promoter region is shown in Fig. IC. The downstream promoter element contains the -12 to -24 recognition sequence for d 4 holoenzyme, and there is a potential IHF binding site (5'-TTTCAATTGATTG-3') between -89 and -100 with respect to the transcription start site. The IHF DNA binding consensus is 5'-(A/T)ATCAAN4TTNG-3', so this potential site contains only one mismatch. We used a supercoiled DNA template (pSA8) in which the anfH promoter and 400 bp of the region upstream of the promoter containing the putative ANFA UAS and IHF sites have been cloned upstream of the strong T7 transcriptional terminator in the transcription vector pTElO3. Transcripts initiating at the predicted anfH transcription start site and terminating at the T7 terminator would be 330 nucleotides in length. We compared the ability of AANFA to activate transcription in either a chloride-based or acetate-based buffer system. The use of acetate instead of chloride enhances DNAprotein interactions and facilitates open complex formation. In the presence of core RNA polymerase and d4, a full-length transcript of the expected length was observed in both buffers (Fig. 3c). Transcription was stimulated in acetate buffer by a t least 50% at limiting protein concentrations, compared with that in chloride buffer, and occurred at a lower protein concentration. Transcription was detectable in acetate buffer between 10-25 nM AANFA compared with 50 nM in chloride buffer. Therefore, acetate buffer was used in further experiments with the an@ promoter.
Transcriptional activation by AANFA at the anfH promoter was not sensitive to the incubation temperature of the assay. Performing the assays at 37 or 30 "C did not influence the level of transcript synthesized (data not shown). Formation of productive complexes at the anfH promoter required a nucleoside triphosphate. Productive complexes were formed when ATP, GTP, and CTP were present together in the assay before the heparin challenge. When these were added after the addition of heparin, no transcripts were made (Fig. 3d). Addition of either ATP or GTP individually supported formation of productive complexes, but CTP did not.
In order to determine whether sequences upstream of the anfH promoter were required for transcriptional activation, we constructed a template in which the sequence from -53 to -405 with respect to the predicted transcription start site was deleted. The ability of AANFA to activate transcription from the full-length and deleted promoter was compared in acetate buffer and is shown in Fig. 3, a and b. Activation from the deleted promoter required a higher protein concentration, 50 nM compared with 10-25 nM for the full-length promoter (Fig.  3a). Transcription from the full-length promoter was maximal at 200 nM protein, and no further increase in transcription was observed at higher protein concentrations. Transcription from the deleted promoter increased more slowly with increasing ., template DNA (pSA9) was the autoradiography as described under "Methods." Transcript levels are arbitrary units of transcription obtained by densitometrically scanning the deleted promoter lacking the upstream sequences -53 to -405 with respect to the transcription start site. 6 , reaction conditions were the same as in Fig. 2a except that transcription was initiated using [32PlUTP (12 p~) and quantitated by determination of the radioactivity in excised bands corresponding to the full-length transcripts as described under "Methods." 0, template DNA was pSA8; ., template DNA was pSA9. c, comparison of transcription from the full-length anfH promoter in either acetate (0) or chloride (0) buffer (see "Methods"). Reaction conditions and transcript processing and quantitation were as for Fig. 2a. Incubation conditions, transcript processing, and autoradiography were as described in Fig. 2 u . e, in vitro transcription was carried out using the full-length anfH promoter (pSA8) in acetate buffer. Template DNA (10 nM) was linearized as described under "Methods." IHF was used at 50 nM. Incubation conditions transcript processing and densitometric analysis were as described in Fig. 2c. 0, -1HF; ., +IHF. f, reaction conditions were the same as in Fig. 3a except that transcription was initiated using [32PlUTP (12 PM) and quantitated by determination of the radioactivity in excised bands corresponding to full-length transcripts as described under " Methods." Lune 1, no AANFA, lanes 2,4,6, and 8,100 nM AANFA, lanes 3 , 5 , 7 ,  In the absence of IHF, transcription from the linear template was only detected above 100 nM N F A . Transcription was not stimulated when the deleted promoter, which lacks the potential IHF binding site as well as the ANFA UAS, was used as the DNA template. An example of the primary data is shown in Fig. 3f. DNA Binding by AANFA-DNase1 and dimethyl sulfate footprinting of the anfTI promoter was carried out using labeled DNA templates containing 400 bp of the region upstream of the predicted transcription start site and including the downstream promoter element. Fig. 4a shows the dimethyl sulfate protection data on both the top and bottom strand with increasing concentrations of AANFA. On the top strand (Fig. 4a, lanes Id), the major site of protection is the GG doublet at -216 and -217 with the two G residues at -225 and -226 being strongly hypermethylated. The DNasel protection data on the same strand confirm this region as a binding site for M A (Fig. 5). This shows protection in the region -205 to -217 and enhancement of DNase cutting around -220 in a similar pattern to the methylation protection data. Changes in DNasel protection and enhancement upstream of -250 indicate a possible second site for ANFAbinding, and this is more evident from the methylation protection pattern on the bottom DNA strand (Fig. 4a,  lanes 6-10). The major site of protection is the GG doublet at -253 and -254, and the two G residues at -262 and -263 are hypermethylated. The G at position -268 is also protected, while that at -270 is hypermethylated. Further weaker sites of protection and hypermethylation are observed upstream of -300. On the top DNA strand, the G residues at -302 and -313 protein concentration. By 500 nM protein, the level of transcription was the same from both DNA templates (Fig. 3b). Thus the region upstream of the anfH promoter contains sequences stimulatory for transcription, which were shown to include the ANFA UAS and potential IHF binding site.

5' T T C W W m T
We examined the effect of purified IHF protein on transcriptional activation from the a n m promoter. Using 50 n~ IHF, there was only a very small stimulation of transcription with the supercoiled template between 10 and 25 nM protein (data not shown). However, the same concentration of IHF strongly simulated transcription from a linear template containing the full-length promoter. The stimulation was greatest at low protein concentrations (Fig. 3e) Where indicated the template DNA was linearized as described under "Methods." Reaction conditions, transcript processing, and densitometric analysis of full-length transcripts were as for Fig. 2. Fig. 6u shows an autoradiograph of the transcripts synthesized from t h e g l d p 2 promoter, and 6b shows the densitometric analysis. 0, acetate buffer supercoiled template; W, chloride buffer supercoiled template; A, acetate buffer linear template.
are weakly protected, while on the bottom strand the two Gs at -305 and -306 are weakly hypermethylated. In both types of footprint, protection is observed at 250 nM protein, and the sites are fully protected above 750 nM AANFA. The methylation protection data are summarized in Fig. 4b and indicate that there are at least two and probably three binding sites for ANFA upstream of -200, which are located approximately 40 base pairs apart. Examination of the sequences protected from dimethyl sulfate by ANFA do not show any obvious dyad symmetry characteristic of DNA-binding proteins which bind as dimers. However, alignment of the protected regions shows some conservation of the sequences involved, and this is shown in Fig. 4c. There is a totally conserved B'-GGTA motif in each potential site within the following consensus: 5'-CnGG-(c\g)(c\g)nGGTA. There does not appear to be a corresponding symmetrical sequence in any of the three putative sites, although there is an inverted repeat of the GGTA motif in the most downstream site at position -225. Therefore, these may represent half-sites for ANFA binding.
Tkanscriptional Activation a t glnAp2 and nifZp-The effectiveness of AANFA as a transcriptional activator was also examined a t other d4-dependent promoters. We compared the ability of the protein to activate the K. pneumoniae glnAp2 and nifz promoters and a mutant nifZ promoter, (nifZ741, with a G to T transition at -26, increasing the homology to glnAp2 and to the consensus E d 4 binding site. This mutation increases the formation of closed promoter complexes with E d 4 in the absence of an activator protein (Whitehall et al., 1992). In acetate buffer, all three promoters were activated by AANFA and gave the expected 420 (nifZ) or 440 (gZnAp2) nucleotide transcript when assayed in a single round transcription assay as described for the anfl promoter (Fig. 6). Transcripts were observed a t 100 nM protein, were d4-dependent, and required the presence of a nucleoside triphosphate for open complex formation. Transcription also occurred at the same protein concentrations on a linearglnAp2 promoter template in acetate buffer (Fig. 6, a and b). In chloride buffer transcript levels from the glnAp2 promoter were similar to those in acetate buffer, but no transcripts were made from the nifz promoter even at 400 nM protein. At the nifZ74 promoter, a low level of transcription was observed above 200 nM protein when assayed in chloride buffer (Fig. 6, c and d). The glnAp2 promoter forms a stable closed complex with E d 4 in the absence of the activator, while the nifi promoter interacts only weakly with E d 4 alone and requires presence of the activator to stabilize closed complex formation (Minchin et aZ., 1989). In acetate buffer, which enhances DNA-

In Vitro
Activity of Bacterial Danscription Factor 18147 protein interactions, AANFA can activate efficiently all three promoters. In chloride buffer, the requirements for open complex formation are more stringent, and the inability of the protein to activate the nifL compared with theglnAp2 promoter may reflect the difference in the promoter's ability to form a stable closed complex. Although the nifL74 mutation increases the promoter's homology t o glnAp2 and the Ed4 consensus binding sequence, it does not overcome the requirement for acetate buffer to stimulate transcription by AANFA at the concentrations tested. ATP Hydrolysis by MFA-ANFA, in common with all other d4-dependent activator proteins, has a highly-conserved central domain that is predicted to interact with Ed4 during transcriptional activation and contains a putative nucleotide binding site (Ronson et al., 1987). The NTRC activator protein has been shown to have a phosphorylation-dependent ATPase activity that is stimulated by site-specific DNA binding (Weiss et al., 1991;Austin and Dixon, 1992). To characterize the potential ATPase activity of AANFA, the amount of inorganic phosphate released from incubations containing [q2P]ATP and highly-purified M F A was measured under different conditions. We compared ATP hydrolysis in chloride and acetate buffer at 25 "C. In both buffers there were similar levels ofATP hydrolysis by the protein alone (Fig. 7, a and b). Addition of supercoiled plasmid DNA stimulated the ATPase activity of the protein in both buffers, but this presumably occurs by nonspecific DNA binding as the stimulation was observed irrespective of whether or not the DNA contained the anfTi promoter upstream region with the ANFA binding sites present (Fig. 7, a and b). The ATPase activity of the protein was linear with increasing protein concentration up to 1 VM protein at 3 mM ATP (Fig. 7c). Fig. 7d shows a time course of ATP hydrolysis that is linear for at least l h at 300 nM protein. DISCUSSION We have purified in soluble active form an N-terminally truncated form of the ANFA protein from A. uinelandii. The protein, although soluble, still demonstrated the tendency, common to a number of $4-dependent activators and in particular NIFA proteins, to aggregate and precipitate during purification. However, the truncated protein was significantly more amenable to purification than the full-length protein, which precipitated after chromatography on heparin agarose. The full-length protein may also require the presence of the iron protein or anaerobic conditions to maintain activity during purification. The purified truncated ANFA exhibited some heterogeneity in native molecular weight, being at least a dimer in solution but with some higher molecular weight species present. This may represent an equilibrium between the dimeric and tetrameric forms of the protein. The purified protein activated the anfH promoter in in vitro transcription assays, transcription being stimulated by the presence of acetate ions in the buffer and by sequences upstream of the -12 to -24 promoter element. The DNA binding experiments indicate that there are at least two and possibly three binding sites for ANFA upstream of position -200. The space between these is approximately 40 base pairs, which should allow four helical turns between each to orient them on the same face of the DNAhelix. Although there does not appear to be any dyad symmetry at these sites, conserved sequences are identifiable in the regions protected from dimethyl sulfate. As there is only one known anf promoter, a n w , it is not possible to make sequence comparisons to aid identification of a consensus binding site for ANFA. However ANFA is known to activate the promoter of A. vinelandii n i p p r~m o t e r .~ This gene product is required for all M. Drummond, personal communication. layer chromatography as described by Austin and Dixon (1992). DNA 1-pl aliquots were removed for analysis of phosphate release by thin was the fuII-Iength anfH promoter construct, pSA8 (black bars), or the deleted promoter, pSA9 (gray bars). c, effect of AANFA concentration of ATP hydrolysis. Reactions containing the indicated concentrations of AANFA were incubated in acetate buffer in the absence of DNA. Reaction conditions and analysis of phosphate release were as for a and b. d, time course ofATP hydrolysis by AANFA. 300 nM AANFA was incubated in a reaction volume of 10 1. 11 a t 25 "C. Hydrolysis was initiated by the addition of 3 mM [q2P1ATP (95 cpdpmol ATP). 1-pl samples were removed at the times indicated and diluted with 9 1. 11 of 0.5% SDS, 2 mM EDTA. 1-111 aliquots were analyzed for phosphate release as described under "Methods." three nitrogenase systems, and two binding sites for NIFA have been identified upstream of the -12 to -24 region of this promoter (Joerger and Bishop, 1988). However, examination of the sequence upstream of the promoter does not reveal any homology with the conserved sequences found in the anfH promoter. Therefore, this region of the putative ANFA binding site is not conserved in the nifB promoter.
The region upstream of the anfH promoter also contains a n IHF consensus binding site located between the -12 to -24 downstream promoter element and the upstream binding sites for ANFA. Purified IHF protein stimulates transcription from the linear anfH promoter by AANFA, but the supercoiled promoter shows no IHF dependence except at very low concentrations of AANFA. This is in contrast to NIFA activation of the K. pneumoniae n i m promoter where IHF stimulates transcription of the supercoiled promoter (Hoover et al., 1990).5 The IHF site and potential ANFA binding sites are located considerably further upstream of the -12 to -24 region than their counterparts in the K. pneumoniae nifH promoter (Buck et al., 1986;Hoover et al., 1990). One possibility for the lack of IHF dependence of the supercoiled promoter is that there is an increased flexibility due to the length of the upstream region and so a greater likelihood of the bound activator contacting E d 4 at the promoter. Linearization of the DNAmay reduce this flexibility, and activation then becomes more dependent on IHF for loop formation. Alternatively, the ability of E d 4 to bind the promoter may influence the dependence on IHF for activation.
At AANFA concentrations above 200 nM, transcription from the anfH promoter is not stimulated by specific DNA binding, and activation occurs equally well from either the full-length or UAS-deleted promoter. At high activator concentrations, interactions between AANFA and E d 4 can occur either in solution or by nonspecific DNA binding, so there is much less dependence on the presence of the UAS. Transcriptional activation from the glnAp2 and nifZ promoters occurs in the absence of a UAS and requires a higher concentration of AANFA to activate transcription than the full-length anfH promoter. The glnAp2 promoter can form a stable closed complex with E d 4 in the absence of the activator (and at the anfH promoter protection of the -12 to -24 region by E d 4 from DNase1 is observed in the absence of AANFA).6 In contrast, stable closed complexes at nifZ are only detected in the presence of the activator (Minchin et al., 1989). In acetate buffer, which facilitates open complex formation, all three promoters are activated efficiently by AANFA. However, in chloride buffer, where requirements for open complex formation are more stringent, the nifZ promoter is not activated by AANFA, but the anfH and glnAp2 promoters are, although activation from the both the full-length and UAS-deleted anfH promoter is reduced. These differences may reflect, at least in part, the ability of the promoter to form a closed complex with E d 4 and the requirement for the activator t o stabilize this complex.
In common with other d4-dependent activator proteins AANFA requires a nucleoside triphosphate to form productive complexes with promoter DNA. ATP or GTP will support complex formation, this requirement being similar to that for the NTRC protein (Lee et al., 1993): NTRC possesses a phosphorylation and DNA-dependent ATPase activity that is required to catalyze open complex formation (Weiss et al., 1991;Austin and Dixon, 1992). The AANFA protein has a constitutive ATPase activity that, by analogy to NTRC, is presumably res. Austin quired to drive formation of open complexes, and it seems likely that there is a common mechanism by which ATP hydrolysis is employed to catalyze open complex formation by d4-dependent activator proteins . The levels of ATPase activity of AANFA are roughly similar to those achieved by phosphorylated NTRC and higher than those of S160F NTRC, a constitutively active mutant NTRC protein Dixon et al., 1991;Austin and Dixon, 1992). The ATPase activity of AANFA differs from that of NTRC in that it occurs equally well in both acetate and chloride buffer systems, whereas that of NTRC is greatly stimulated by acetate buffer. ATP hydrolysis by AANFA also increases linearly with increasing protein concentrations, in contrast to phosphorylated NTRC, which shows a nonlinear increase in both the presence and absence of DNA. We have interpreted this response as a mechanism whereby ATP hydrolysis is stimulated by cooperative interactions between phosphorylated NTRC dimers bound to multiple binding sites on DNA . The ATPase activity of the AANFA protein is not greatly stimulated by the presence of DNA and may not be so tightly coupled to DNA binding as in the case of NTRC. However, the ATPase activity of the full-length ANFA protein may be modulated by the presence of the N-terminal domain just as phosphorylation of the N-terminal domain regulates the ATPase activity of NTRC. In the case of ANFA, it may be the reduction-oxidation state of the potential metal binding site in the N-terminal domain that influences the ATPase activity of the catalytic domain of the protein. Purification of the full-length protein will facilitate understanding of the role of the N-terminal domain in the transcriptional activity of the protein.