Rhodobacter capsulatus AnfA is essential for production of Fe‐nitrogenase proteins but dispensable for cofactor biosynthesis and electron supply

Abstract The photosynthetic α‐proteobacterium Rhodobacter capsulatus reduces and thereby fixes atmospheric dinitrogen (N2) by a molybdenum (Mo)‐nitrogenase and an iron‐only (Fe)‐nitrogenase. Differential expression of the structural genes of Mo‐nitrogenase (nifHDK) and Fe‐nitrogenase (anfHDGK) is strictly controlled and activated by NifA and AnfA, respectively. In contrast to NifA‐binding sites, AnfA‐binding sites are poorly defined. Here, we identified two highly similar AnfA‐binding sites in the R. capsulatus anfH promoter by studying the effects of promoter mutations on in vivo anfH expression and in vitro promoter binding by AnfA. Comparison of the experimentally determined R. capsulatus AnfA‐binding sites and presumed AnfA‐binding sites from other α‐proteobacteria revealed a consensus sequence of dyad symmetry, TAC–N6–GTA, suggesting that AnfA proteins bind their target promoters as dimers. Chromosomal replacement of the anfH promoter by the nifH promoter restored anfHDGK expression and Fe‐nitrogenase activity in an R. capsulatus strain lacking AnfA suggesting that AnfA is required for AnfHDGK production, but dispensable for biosynthesis of the iron‐only cofactor and electron delivery to Fe‐nitrogenase, pathways activated by NifA. These observations strengthen our model, in which the Fe‐nitrogenase system in R. capsulatus is largely integrated into the Mo‐nitrogenase system.

In this study, we show that R. capsulatus AnfA binds two highly similar palindromic sites in the anfH promoter. Based on conserved sequences in various α-proteobacteria, we define a general AnfAbinding site consensus, TAC-N 6 -GTA. Besides, we present evidence that the anfH promoter is the only Fe-nitrogenase-related promoter in R. capsulatus strictly depending on AnfA.

| Strains, plasmids, and growth conditions
Bacterial strains and plasmids used in this study are listed in Table A1 in Appendix 1. Rhodobacter capsulatus minimal medium V (RCV) was prepared as previously described (Demtröder, Pfänder, et al., 2019). In this medium, a fixed nitrogen source and molybdate (Mo) have been omitted. Traces of Mo arising from impurities of the chemicals used support residual Mo-nitrogenase activity but are low enough to permit the production of Fe-nitrogenase.
To examine diazotrophic growth, cultures were inoculated in 3 ml RCV medium in screw-capped 17-ml Hungate tubes before the exchange of headspace air for pure N 2 gas and incubation in the light. When required, 10 mM serine was added as a fixed nitrogen source, which (in contrast to ammonium) does not inhibit nitrogen fixation.

| Construction of Rhodobacter capsulatus anfH-lacZ reporter strains and β-galactosidase assays
The anfH promoter was narrowed down by nested deletions. For this, appropriate primer pairs were used to PCR-amplify promoter variants F1 to F6 (Figure 2a,b), thereby adding BamHI and HindIII sites. Corresponding BamHI-HindIII fragments were cloned into the broad-host-range vector pBBR1MCS (Kovach et al., 1995) before insertion of a lacTeT cassette (carrying a promoterless lacZ gene, a tetracycline resistance gene, and an oriT transfer origin) from plasmid pYP35 (Gisin et al., 2010) into the HindIII site. The resulting reporter plasmids carrying transcriptional lacZ fusions were designated pBBR_F1-lacZ to pBBR_F6-lacZ.
To generate site-directed substitution mutations in the anfH promoter (Figure 3a), plasmid pBBR_F1-lacZ served as a template.
The reporter plasmids were conjugationally transferred into the R. capsulatus wild-type strain B10S. Following phototrophic growth of the R. capsulatus reporter strains in RCV medium with 10 mM serine (no Mo added) until the late logarithmic phase, LacZ (β-galactosidase) activity was determined (Miller, 1972). In the absence of ammonium (-NH 4 + ), the superior regulator NtrC activates transcription of nifA and anfA in concert with the housekeeping sigma factor RpoD (Foster-Hartnett, Cullen, Monika, & Kranz, 1994;Kutsche et al., 1996). MopA and MopB independently repress anfA in the presence of molybdate (+MoO 4 2-; Wiethaus et al., 2006). NifA and AnfA activate their target genes by partnering with the alternative sigma factor RpoN. Noteworthy, NifA indirectly controls AnfA-mediated anfHDGK expression by controlling RpoN production (Demtröder, Pfänder, et al., 2019). Involvement of NifA-activated genes in biosynthesis of the iron-molybdenum cofactor (FeMoco) of Mo-nitrogenase and the iron-only cofactor (FeFeco) of Fe-nitrogenase and electron transfer to both nitrogenases is indicated. (b) Production of active Fe-nitrogenase in a strain lacking AnfA. In this study, we constructed strain YP515-BS85 containing mutations in the anfA and nifD genes (marked by red crosses) and a chromosomal substitution of the anfH promoter (P anfH ) by the nifH promoter (P nifH ) thereby putting anfHDGK expression under NifA control. This strain grew under N 2 -fixing conditions ( Figure 4b) suggesting that AnfA is dispensable for FeFeco biosynthesis and electron supply to Fe-nitrogenase. For further details, see text

| Examination of AnfA binding to the anfH promoter
In vitro binding of the DNA-binding domain of AnfA (AnfA_DBD) to the anfH promoter was examined by electrophoretic mobility shift assays (EMSA) as previously described .
To overexpress AnfA_DBD, appropriate primers were used to PCRamplify a DNA fragment coding for the C-terminal 72 amino acid residues of AnfA ( Figure A1 in Appendix 2) thereby adding SacII and NcoI sites. The corresponding SacII-NcoI fragment was cloned into the expression vector pASK_IBA45+ (IBA GmbH Göttingen) resulting in hybrid plasmid pYP409. For purification of the Streptagged AnfA_DBD, Escherichia coli BL21 (DE) carrying pYP409 was cultivated with AHT induction before cell disruption and Strep-Tactin affinity chromatography as previously described (Hoffmann, Ali, et al., 2016).
The F1 fragment (carrying the wild-type anfH promoter) and its variants ( Figure 3a) were labeled with γ-32 P-ATP, and free γ-32 P-ATP was removed by gel filtration using Illustra ProbeQuant G-50 Micro Columns (GE-Healthcare). After 20 min incubation of labeled promoter variants with increasing amounts of the purified AnfA_DBD protein, bound and free DNAs were separated in 6% polyacrylamide gels. Radioactive bands were detected by phosphor screen exposure.

| Substitution of the Rhodobacter capsulatus anfH promoter by the nifH promoter
To replace the anfH promoter (P anfH ) by the nifH promoter (P nifH ), we exchanged the 266 bp anfA-anfH intergenic region by the 267 bp fdxD-nifH intergenic region. For this purpose, we constructed mutagenesis plasmid pYP516 containing the 3′ end of anfA (including the translation stop codon, TGA), a gentamicin (Gm) resistance cassette, P nifH , and the 5′ end of anfH (starting with the translation start codon, ATG). To replace the anfH promoter and to delete the anfA gene in a single step, we constructed mutagenesis plasmid pYP515 containing an anfA upstream fragment (but lacking the anfA coding region), a Gm cassette, P nifH , and the 5′ end of anfH. Plasmids pYP516 and pYP515 were conjugationally introduced into the R. capsulatus strain BS85 (ΔnifD), in which the nifD gene is disrupted by a spectinomycin (Sp) cassette. BS85 does not exhibit Mo-nitrogenase activity, and hence, any nitrogenase activity observed in this background can be assigned to Fe-nitrogenase (Demtröder, Pfänder, et al., 2019).
Promoter replacement mutants were identified by selection for Gm resistance and screening for loss of vector-encoded tetracycline resistance indicating marker rescue by double cross-over events.

| Localization of the Rhodobacter capsulatus anfH promoter by nested deletions
Rhodobacter capsulatus AnfA is essential for the expression of the anfHDGKOR3 operon (Demtröder, Pfänder, et al., 2019), but the anfH promoter has not been investigated. The coding regions of anfH and its upstream gene, anfA, are separated by 266 bp (Figure 2a). This intergenic region includes three conspicuous sequences, namely (a) a GC-rich inverted repeat sequence followed by a T-rich stretch likely acting as Rho-independent terminator of anfA transcription, (b) two 17 bp direct repeats each encompassing inverted repeat sequences, which are promising candidates as AnfA-binding sites, and (c) a highly conserved RpoN-binding site.
For clarity, the 17 bp sequences will from now on be called distal and proximal AnfA-binding sites.
To localize the anfH promoter (P anfH ), we analyzed the effects of nested promoter deletions on anfH expression. For this purpose, we generated transcriptional fusions between P anfH fragments,

| Effects of AnfA-binding site mutations on anfH expression
To dissect the function of the distal and proximal AnfA-binding sites in P anfH , we generated pBBR_F1-lacZ variants carrying site-directed

| Effects of AnfA-binding site mutations on promoter binding by AnfA
The AnfA protein encompasses three domains, namely a GAF, an AAA+, and a HTH domain, involved in environmental sensing, activation of RNA polymerase, and promoter binding, respectively.
To test the direct binding of AnfA to P anfH , we performed electrophoretic mobility shift assays (EMSA). For this, the radiolabeled F1 fragment (carrying the wild-type anfH promoter) and its variants

| NifA-driven anfHDGKOR expression restores Fe-nitrogenase activity in a strain lacking AnfA
Productive nitrogen fixation by the Fe-nitrogenase requires more than the expression of the anfHDGKOR3 operon. Current knowledge suggests that AnfA is required only for anfHDGKOR3 expression and has little impact on the expression of NifA-activated genes like nifB and rnfA, which are essential for FeFeco biosynthesis and electron transfer to Fe-nitrogenase, respectively (Demtröder, Pfänder, et al., 2019;Schüddekopf et al., 1993). We, therefore, speculated that AnfA might be dispensable for the production of active (N 2 -fixing)

F I G U R E 2
Effect of nested deletions in the R. capsulatus anfH promoter on anfH-lacZ expression. (a) Cis-regulatory elements in the anfA-anfH intergenic region. The DNA sequence encompasses the AnfA translation stop codon (TGA), the Rho-independent anfA transcription terminator, two AnfA-binding sites (AnfA_BS), the RpoN-binding site (RpoN_BS), and the AnfH translation start codon (ATG). Arrowheads mark inverted repeat sequences. The start sites of anfH promoter deletion variants F1 to F6 are indicated. (b) Reporter fusions between anfH promoter deletion variants and lacZ. Promoter variants F1 to F6 were cloned into a broad-host-range vector, before insertion of a lacZ cassette (designed for transcriptional fusions) immediately downstream of the anfH start codon resulting in reporter plasmids pBBR_F1-lacZ to pBBR_F6-lacZ (Materials and Methods). (c) Expression of anfH-lacZ fusions. R. capsulatus reporter strains carrying pBBR_F1-lacZ to pBBR_ F6-lacZ were phototrophically grown in RCV minimal medium with 10 mM serine but without Mo addition, conditions allowing anfHDGKOR3 expression. LacZ (β-galactosidase) activity is given in Miller units (Miller, 1972). The results represent the means and standard deviations of five independent experiments R. capsulatus reporter strains carrying pBBR_F1-lacZ (WT) and its variants (Mut1 to Mut7, Mut2/6, and Mut2/7) were phototrophically grown in RCV minimal medium (no Mo added) with 10 mM serine. LacZ (β-galactosidase) activity is given in Miller units (Miller, 1972). Data for WT control are the same as in Figure 2c. operon is the only member of the AnfA regulon, nitrogen fixation by Fe-nitrogenase should become entirely NifA-dependent in the P anfH → P nifH strain (Figure 1b).
To determine the effect of P anfH → P nifH substitution on anfH-DGKOR3 expression, we generated two R. capsulatus strains, which carry the same P anfH → P nifH promoter substitution, but either in the wild-type (anfA + ) or ΔanfA background. Subsequently, we introduced transcriptional anfH-lacZ reporter fusions in these strains and, as a control, in the wild-type strain B10S by chromosomal integration of plasmid pMH187 as previously described (Demtröder, Pfänder, et al., 2019). The resulting reporter strains were grown with either serine or ammonium as a nitrogen source before LacZ activity was determined.
To test whether nitrogen fixation by Fe-nitrogenase was entirely NifA-dependent in the P anfH → P nifH strains, we introduced a polar nifD mutation (ΔnifD) in these strains, before examination of diazotrophic growth. Regardless of the nifD mutation, these strains were expected to express all the other NifA-dependent nitrogen fixation genes involved in cofactor biosynthesis and electron transport ( Figure 1b). Since ΔnifD strains lack Mo-nitrogenase, diazotrophic growth of these strains depends entirely on Fe-nitrogenase (Demtröder, Pfänder, et al., 2019). As controls, we included the wild type, the parental ΔnifD strain, and a ΔanfA-ΔnifD strain lacking both nitrogenases.
The wild-type and ΔnifD strains grew well with N 2 as the sole nitrogen source, while the ΔanfA-ΔnifD strain failed to grow diazotrophically ( Figure 4b) consistent with earlier studies (Demtröder, Pfänder, et al., 2019;Kutsche et al., 1996). Both P anfH → P nifH strains grew almost as well as the parental ΔnifD strain suggesting that P anfH → P nifH substitution indeed decouples production of functional Fe-nitrogenase from AnfA.
Taken together, our findings show that substitution of the AnfAdependent anfH promoter by the NifA-activated nifH promoter restores anfHDGKOR expression and Fe-nitrogenase activity in a strain lacking AnfA. In other words, AnfA appears to be dispensable for FeFeco biosynthesis and electron delivery to Fe-nitrogenase.
vinelandii, C-N-GG-N 3 -GGTA (Austin & Lambert, 1994), and our consensus share the strictly conserved GTA motif (Figure 5a). The  (Figure 3b). Together these findings suggest that AnfA proteins in different bacteria require the GTA motif, but differ in their dependence on the TAC motif to activate their target promoters.

A. vinelandii
Rhodobacter capsulatus AnfA is essential for anfHDGKOR3 expression and consequently, for nitrogen fixation by Fe-nitrogenase (Demtröder, Pfänder, et al., 2019). Remarkably, a strain lacking AnfA but driving anfHDGKOR3 expression by NifA regained the capacity to grow diazotrophically via Fe-nitrogenase (Figure 4b). This means that AnfA in the wild type is required for AnfHDGKOR3 production, but dispensable for FeFeco biosynthesis and electron delivery to Fenitrogenase. For a regulatory model, see Figure 1b. Our findings do not necessarily exclude other AnfA targets than the anfH promoter.
Indeed, AnfA affects the expression of different nitrogen fixation genes including iscN, nifE, fprA, and nifB (Demtröder, Pfänder, et al., 2019). None of these genes, however, is preceded by an obvious AnfA-binding site (or just a GTA motif) suggesting that AnfA control of these genes is indirect.
In contrast to the R. capsulatus nifB promoter, we found potential AnfA-binding sites in the nifB promoters of Rhodospirillum rubrum (TAC-N 6 -GTA) and Rhodomicrobium vannielii (CAC-N 6 -GTA) suggesting direct nifB activation by AnfA in these strains. In line with the requirement of NifB for the activity of all three nitrogenases in A.
vinelandii, the nifB promoter can be activated by NifA, VnfA, or AnfA, which bind to overlapping sites in the nifB promoter (Drummond et al., 1996).
Our finding that only a single Fe-nitrogenase-related target, the anfH promoter, strictly requires activation by AnfA in R. capsulatus, raises the question of why this diazotroph needs AnfA. One explanation is that AnfA contributes (indirectly) to fine regulation of NifA-dependent genes (Demtröder, Pfänder, et al., 2019). Also, Mo repression of anfA introduces a regulatory level to prevent the production of Fe-nitrogenase under Mo-replete conditions. This guarantees the exclusive activity of Mo-nitrogenase, which exhibits higher N 2 -reducing activity than Fe-nitrogenase (Hoffmann, Wagner, et al., 2016;Wiethaus et al., 2006).

ACK N OWLED G EM ENTS
We thank F. Narberhaus (Bochum, Germany) for helpful discussions and continuous support. We thank J. Eisfeld (Bochum, Germany)

F I G U R E 5
AnfA-binding sites in proteobacterial anfH promoters. (a) Comparison of AnfA-binding sites. Binding of AnfA to distal and proximal sites has been experimentally shown for R. capsulatus (this study) and A. vinelandii (Austin & Lambert, 1994). Affiliation of bacterial strains to the α-and γ-proteobacteria, and the numbers of nucleotides (N) between cis-regulatory elements are indicated. Known and presumed AnfA-binding sites encompass strictly conserved GTA and partially conserved TAC motifs (highlighted in red). Lower and upper case lettering in the consensus sequences indicates conservation in at least four or five of the respective sequences, respectively. (b) AnfAbinding site logo. The AnfA-binding site consensus based on all distal and proximal sites shown in (a) was generated using the weblo go.berke ley.edu program (a)

(b)
Bits for help with electrophoretic mobility shift assays. This work was financed by a grant from the Deutsche Forschungsgemeinschaft (DFG) (Ma 1814/4-2) to BM.

CO N FLI C T O F I NTE R E S T S
None declared.

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are provided in full in the Section 3.

O RCI D
Bernd Masepohl https://orcid.org/0000-0002-1747-7177 A PPE N D I X 2 F I G U R E A 1 Comparison of AnfA proteins