Crystal structure of the [2Fe–2S] protein I (Shethna protein I) from Azotobacter vinelandii

Several Azotobacter iron–sulfur proteins probably play roles in the complex redox chemistry that Azotobacter must maintain when fixing nitrogen. The 2.1 Å resolution crystal structure of the [2Fe–2S] protein I (Shethna protein I) from Azotobacter vinelandii reveals a homodimer similar to the structure of the thioredoxin-like [2Fe–2S] protein from Aquifex aeolicus, with the [2Fe–2S] cluster coordinated by the surrounding conserved cysteine residues.


Introduction
Azotobacter vinelandii is a model organism for nitrogen fixation (Peters et al., 1995). Although nitrogenase is inactivated by oxygen, A. vinelandii can grow in above-atmospheric concentrations of oxygen (Maier & Moshiri, 2000). This oxygen resistance arises via at least two mechanisms: firstly a high respiratory rate, which removes oxygen, called 'respiratory protection' (Jones et al., 1973), and secondly 'conformational protection', in which high oxygen levels pause nitrogenase activity, but on their removal nitrogen fixation resumes. Conformational protection is mediated by  protein II (FeSII) or Shethna protein, which forms a complex with nitrogenase that is catalytically inactive but resistant to oxygen (Robson, 1979;Moshiri et al., 1994;Schlesier et al., 2016). The FeSII protein was first purified by Shethna in the 1960s (Shethna et al., 1964(Shethna et al., , 1968) as a nonheme iron protein of interest, together with another protein, the so-called  protein I (FeSI).
Ferredoxins are iron-sulfur proteins that mediate electron transfer, and there are several families of them. 2Fe ferredoxins can be classified as plant-type, mitochondrial-type and bacterial ferredoxins. There is also another class comprising 3Fe, 4Fe, 7Fe and 8Fe ferredoxins with one or two FeS clusters of the cubane type (Zanetti & Pandini, 2013). FeSI is a member of the thioredoxin-like ferredoxin family (InterPro family IPR009737). Thioredoxins have no Fe-S clusters, but have pairs of cysteine residues to exchange disulfide-bond oxidation states (Saarinen et al., 1995). These thioredoxin-like ISSN 2053-230X ferredoxins are also present in some multimeric hydrogenases (Appel & Schulz, 1996;De Luca et al., 1998) and the NADHubiquinone oxidoreductase of respiratory chains (Yano et al., 1994).
In the A. vinelandii genome (Setubal et al., 2009), several genes encode proteins associated with nitrogen fixation. The dinitrogen reductase in the nitrogenase system is the MoFe protein. The major nif cluster of genes contains the nitrogenase structural genes, including those for MoFe and most of the nitrogenase-assembly factors. FeSI is in the middle of the major nif cluster, just after the nifN, nifE and nifX genes which encode nitrogenase-assembly factors. FeSI is predicted to be cotranscribed with a [4Fe-4S] ferredoxin-like protein and several other proteins of unknown function (Fig. 1a). Moreover, the transcript level of FeSI was reported to increase 2.3-fold under nitrogen-fixing conditions (Hamilton et al., 2011). Therefore, given the genomic location and transcript data, it is likely that FeSI plays some role in nitrogen fixation.

Macromolecule production
The fesI gene (Avin_01520) was amplified by PCR from A. vinelandii CA genomic DNA and cloned by Gibson assembly (Gibson, 2011) into a modified pRSET-A vector with a thrombin-cleavable 6ÂHis tag. The primers (Invitrogen) used for the pRSET-A vector were 5 0 -GGATCCACGCGG AACCAGACC-3 0 (forward) and 5 0 -GCCCGAAAGGAAG CTGAGTTGGCT-3 0 (reverse), and those for the genomic DNA were 5 0 -GGTCTGGTTCCGCGTGGATCCATGGCCA AACCCGAGTTCCATATC-3 0 (forward) and 5 0 -AGCCAA CTCAGCTTCCTTTCGGGCCTACCAGATCTCGGCAGG GGT-3 0 (reverse). The FesI-pRSET-A plasmid was transformed into Escherichia coli KRX cells. The cells were grown in 1 l Terrific Broth to an OD of 0.6-0.8 and were then induced with 0.1%(w/v) rhamnose. The cells were grown at 18 C for 18 h after induction, spun at 4000g for 15 min, resuspended in 50 mM Tris-HCl pH 7.9, 150 mM NaCl and disrupted by sonication. Cell debris was removed by centrifugation and filtration. The supernatant was loaded onto a nickel resin affinity column (Generon) and eluted with 500 mM imidazole in 50 mM Tris-HCl pH 7.9, 150 mM NaCl. The His tag was cleaved by adding 50 U thrombin (Sigma) and incubating at 4 C overnight. Finally, the protein was concentrated to $15 mg ml À1 for crystallization trials. Macromoleculeproduction information is summarized in Table 1.

Crystallization
Thin plate-shaped crystals, which belonged to space group P2 1 , were obtained by sitting-drop vapour diffusion using a reservoir solution consisting of 0.1 M HEPES pH 7.5, 70%(v/v) MPD (2,4-methylpentanediol). Crystals were cryoprotected in the mother liquor with $30%(v/v) polyethylene research communications  Subunits are shown with the [2Fe-2S] clusters and ligating cysteine residues shown as sticks. The right subunit is coloured according to the secondary structure. Strands (S1-S4), helices (H1-H4) and loops are shown in magenta, cyan and salmon, respectively. The residues (His7, Phe9, Thr53, Gly55 and Tyr69) at the dimer interface are shown as sticks. Hydrogen bonds are shown as black dashed lines. glycol (PEG) 400 and then flash-cooled in liquid nitrogen. Crystallization information is summarized in Table 2.

Data collection and processing
X-ray diffraction data were collected from cryoprotected crystals at 100 K on beamline I03 at Diamond Light Source, UK. For phase determination, a data set was collected at the Fe K-edge at a wavelength of 1.734 Å . The collected data were processed and scaled with xia2 (Winter, 2010) using the 3dii (XDS) setting (Kabsch, 2010). Data-collection and processing statistics are given in Table 3.

Structure solution and refinement
Phases were calculated with the AUTOSHARP pipeline (Vonrhein et al., 2007) and an initial model was built with ARP/wARP (Langer et al., 2008). Cycles of model building and refinement with noncrystallographic symmetry restraints were performed in REFMAC5 (Murshudov et al., 2011) and Coot (Emsley et al., 2010). MolProbity (Williams et al., 2018) was used for validation. The final refinement statistics are presented in Table 4

Results and discussion
The A. vinelandii FeSI structure is a homodimer burying 1420 Å 2 as calculated with the PDBePISA service at EMBL-EBI (Krissinel & Henrick, 2007). It consists of a sheet with four -strands, two long -helices that pack against the sheet, two short -helices and several loops. The sheet from the monomer packs against its symmetry mate, forming the dimer interface. On one side of the dimer interface, two hydrogen bonds in parallel strands (S1 and S3) between the side chains of His7 and Tyr69 of two chains and, on the other side, hydrogen bonding between the main chains of Thr53 and Gly55 on strand S2 of two chains, along with the interaction of Phe9 with its symmetry mate through -stacking, favour the dimerization (Fig. 1b)

Figure 2
FeSI electron density (2F o À F c ) contoured at 1 around the [2Fe-2S] cluster in chain A, with residues of interest labelled. The Gly63-Ala64 bond is a cis-peptide.
r.m.s.d. of 0.1 Å over 104 C atoms. Both clusters are coordinated in the same way and were refined at 100% occupancy. The [2Fe-2S] cluster is positioned between two loops that are located between strand S1 and helix H1 and between strand S2 and helix H2. It is bound to four cysteine residues: Cys11, Cys24, Cys56 and Cys60. There is a nonproline cis-peptide, well resolved in the density, between Gly63 and Ala64 (Fig. 2), which are Gly-Pro in many homologous sequences (Supplementary Fig. S1). The most similar protein to FeSI with a known structure is the [2Fe-2S] ferredoxin from Aquifex aeolicus Yeh et al., 2000), with an r.m.s.d. of 2 Å over one chain and 38% sequence identity. A. aeolicus does not fix nitrogen, so its thioredoxin-like ferredoxin must have a function that is unrelated to nitrogen fixation.
FeSI has 31% sequence identity to the C. pasteurianum ferredoxin , which has no experimental structure. The structure of C. pasteurianum ferredoxin was predicted in trRosetta (Yang et al., 2020) and the model has an r.m.s.d. of 0.6 Å to FeSI over one chain. C. pasteurianum ferredoxin specifically and strongly interacts electrostatically with the nitrogenase protein MoFe via three negatively charged residues: Glu31, Glu34 and Glu38 (Golinelli et al., 1997). These residues are on an -helix on the opposite side of the protein to the [2Fe-2S] cluster. In FeSI, the equivalent residues are Gly31, Asn34 and Gln38, all of which are on the outside face of the first helix of the structure (Fig. 3). The change in charge from three negative glutamates to a neutral glutamine, asparagine and glycine suggests that it is unlikely that FeSI interacts directly with A. vinelandii MoFe by this mechanism and that its involvement in nitrogen fixa-tion is of a different nature to the C. pasteurianum protein. Moreover, previous cross-linking experiments did not find any interaction of FeSI with the C. pasteurianum or A. vinelandii MoFe proteins .
The FeSI structure completes the structural picture of the original two Azotobacter FeS proteins purified by Shethna in the 1960s. The structure of the FeSI protein from A. vinelandii appears to be a typical thioredoxin-like ferredoxin and will provide information for understanding its function in relation to nitrogen fixation and its evolutionary relationships to other ferredoxins.