Nitrogenase Cofactor Maturase NifB Isolated from Transgenic Rice is Active in FeMo-co Synthesis.

The engineering of nitrogen fixation in plants requires assembly of an active prokaryotic nitrogenase complex, which is yet to be achieved. Nitrogenase biogenesis relies on NifB, which catalyzes the formation of the [8Fe-9S-C] metal cluster NifB-co. This is the first committed step in the biosynthesis of the iron-molybdenum cofactor (FeMo-co) found at the nitrogenase active site. The production of NifB in plants is challenging because this protein is often insoluble in eukaryotic cells, and its [Fe-S] clusters are extremely unstable and sensitive to O2. As a first step to address this challenge, we generated transgenic rice plants expressing NifB from the Archaea Methanocaldococcus infernus and Methanothermobacter thermautotrophicus. The recombinant proteins were targeted to the mitochondria to limit exposure to O2 and to have access to essential [4Fe-4S] clusters required for NifB-co biosynthesis. M. infernus and M. thermautotrophicus NifB accumulated as soluble proteins in planta, and the purified proteins were functional in the in vitro FeMo-co synthesis assay. We thus report NifB protein expression and purification from an engineered staple crop, representing a first step in the biosynthesis of a functional NifDK complex, as required for independent biological nitrogen fixation in cereals.

Introduction

Nitrogen (N) fertilizers are a major input to sustain high yields of cereal crops.[1] However, high use of N fertilizers has negative impact on human health and the environment.[2,3] Engineering biological nitrogen fixation (BNF) in cereals is necessary to reduce our dependency on N fertilizers.[4] BNF involves the conversion of inert atmospheric N2 into biologically useable ammonia catalyzed by the enzyme nitrogenase. However, nitrogenases are only found in some bacteria and archaea, so plants cannot fix their own nitrogen and must either form symbiotic relationships with nitrogen-fixing prokaryotes or obtain fixed nitrogen compounds from the soil.[5] One potential strategy to develop nitrogen-fixing cereal crops is by engineering the transfer of bacterial nitrogen fixation (nif) genes to the plant genome,[6,7] but this process requires deep understanding of nitrogenase activity requirements and performance in eukaryotic cells.

There are three types of nitrogenases: molybdenum- (Mo), vanadium- (V), and iron-only (Fe) nitrogenases.[8] The Mo-nitrogenase, the most widespread and studied one, consists of two metalloproteins: MoFe protein and Fe protein. The MoFe protein is a dinitrogenase encoded by nifDK that contains the catalytic site which binds and reduces N2. The Fe protein is a dinitrogenase reductase encoded by nifH, which provides electrons to the MoFe protein.[9] Mo-nitrogenase has three essential metal clusters. The first is the [4Fe–4S] cluster located in the Fe protein, and the others are the P-cluster [8Fe–7S] and FeMo-co [7Fe–9S–C–Mo-R-homocitrate] located in the MoFe protein.[10]

The [4Fe–4S] clusters are synthesized by NifU and NifS, where NifU provides the scaffold for the assembly of [Fe–S] clusters, and NifS mobilizes sulfur (S) from cysteine for [Fe–S] cluster synthesis on NifU.[11,12] The P-cluster is formed by the reductive coupling of two [4Fe–4S] cluster pairs at the MoFe protein in the presence of the Fe protein, MgATP, and a reductant.[13,14] FeMo-co is one of the most complex [Fe–S] clusters discovered in nature thus far and is synthesized in a regulated and coordinated process, depending on a multitude of proteins.

The proteins involved in FeMo-co biosynthesis can be functionally divided into four classes: molecular scaffolds (NifU, NifB, and NifEN), metallocluster carriers (NifX, NafY, and NifY), substrate providers (NifS, NifQ, and NifV), and NifH.[15] The process of FeMo-co biosynthesis is initiated by NifU and NifS with the assembly of [4Fe–4S] cluster units that are transferred to NifB.[16−18] NifB is an S-adenosyl-l-methionine (SAM) radical enzyme that carries a catalytic [4Fe–4S] cluster (called RS cluster) and two additional [4Fe–4S] clusters (called K1- and K2-clusters) used as substrates to generate the [8Fe–9S–C] product called NifB-co.[19−22] It has been shown that NifB-co production is 80% lower in Azotobacter vinelandiifdxN mutants, suggesting that FdxN is required for NifB-co biosynthesis.[23] Although the precise function of FdxN remains unclear, it is thought to provide NifB with electrons needed for NifB-co formation.[10] NifB-co synthesized by NifB is matured into FeMo-co on NifEN/NifH complexes,[24,25] which is then inserted into the active site of the apo-MoFe protein to reconstitute active Mo-nitrogenase.[26] As the function of NifB is to catalyze the first committed step in the biosynthesis of FeMo-co (the production of NifB-co), nifB mutants lack FeMo-co.[27,28] NifB-co is also the precursor cluster for the biosynthesis of FeV-co and FeFe-co found at the active sites of the alternative V-nitrogenase and Fe-only nitrogenase[17,29] and therefore required for all BNF.

The NifB proteins of A. vinelandii and Klebsiella oxytoca comprise an N-terminal SAM-radical domain containing the CxxxCxxC SAM-binding motif and a C-terminal NifX-like domain.[30,31] However, the simplest NifB architecture is a standalone SAM-radical domain because the NifX-like sequence is not essential for NifB activity. For example, the single-domain NifB proteins from Methanosarcina acetivorans, Methanobacterium thermoautotrophicum, Methanocaldococcus infernus, and Methanothrix thermoacetophila are functionally equivalent to A. vinelandii NifB.[19,30,32] This single-domain architecture facilitates the heterologous expression of stable NifB in Escherichia coli.[32,33] When a library of 28 NifB proteins was screened for expression in Saccharomyces cerevisiae, only six accumulated as predominantly soluble proteins targeted to mitochondria, and four of these had a single-domain architecture.[34] When the same NifB variants were expressed transiently in Nicotiana benthamiana and targeted to the mitochondria or chloroplasts, only three accumulated as soluble proteins. All three were single-domain proteins, corroborating their superior expression and performance in eukaryotic hosts.[35]

Here, we generated rice plants expressing stably M. infernus NifB or Methanothermobacter thermautotrophicus NifB targeted to the mitochondria, in each case, together with A. vinelandii FdxN. Both NifB proteins accumulated in a soluble form, and their functionality was confirmed by FeMo-co synthesis in vitro.M. thermautotrophicus NifB exhibited higher NifB-co conversion activity in vitro, compared withM. infernus NifB. The production of functional NifB in rice represents an important step toward the expression of active nitrogenase to achieve BNF in cereal crops.

Results

Genetic Elements and Rice Transformation

S. cerevisiae codon-optimized nifB from M. infernus (nifB) and M. thermautotrophicus (nifB) and fdxN from A. vinelandii (fdxN) were used for rice transformation because there are no rare codons in these three gene sequences in the context of rice codon usage.[36] The nifB gene (hereafter OsnifB to indicate expression in rice), nifB (hereafter OsnifB), and the fdxN gene (hereafter OsfdxN) were introduced into separate vectors for rice transformation. Expression was driven by the strong constitutive ZmUbi1 + 1st i promoter. An N-terminal mitochondrial leader sequence from the S. cerevisiae cytochrome c oxidase subunit IV (Cox4) was added to direct the proteins to mitochondria because this sequence was previously shown to target recombinant eGFP to the rice mitochondria effectively.[37] An N-terminal Twin-Strep (TS) tag was added between the Cox4 signal and the OsNifB or OsNifB proteins for detection and purification of OsNifB. A C-terminal hemagglutinin (HA) tag was added to OsFdxN to enable immunodetection. The OsnifB and OsnifB constructs were used separately, in each case, combined with OsfdxN and a third construct carrying the hygromycin phosphotransferase (hpt) gene for selection. Transgenic rice callus expressing nif transgenes were produced by direct DNA transfer, as described.[38,39] Plantlets were regenerated from the corresponding callus lines under hygromycin selection and grown to maturity, as described.[38,39]

Transgenic Rice Callus and Recovery of Plants Expressing OsnifB and OsfdxN

We recovered transgenic lines co-expressing NifB and FdxN at the mRNA level (Figure S1). Four lines each from MiB and MtB were selected for immunoblot analysis. We identified three lines, MiB32, MiB115, and MtB35 that accumulated the recombinant proteins at the highest levels. Accumulation of OsNifB, OsNifB, and OsFdxN in these callus lines and the corresponding regenerated plants was determined by immunoblot analysis using antibodies specific for NifB, and the TS and HA tags (Figures , S2, and S3). Based on their SDS-gel migration patterns, the recombinant proteins were correctly processed, resulting in the expected molecular weights of 38 kDa (OsNifB), 35 kDa (OsNifB), and 11 kDa (OsFdxN). As seen previously when expressed in S. cerevisiae,[34,40] the migration of the HA-tagged FdxN protein was less distinct, probably due to its smaller size. We thus confirmed that OsNifB, OsNifB, and OsFdxN proteins accumulated in the soluble form in rice callus (Figure a). OsNifB and OsFdxN were soluble in regenerated plants (Figure b). We were not able to detect accumulation of OsNifB in regenerated MtB35 plants.

Figure 1

Expression of OsNifB, OsNifB, and OsFdxN in callus and plants. Immunoblot analysis of cell-free extracts prepared from callus (a) and plants (b). OsNifB and OsNifB were detected with antibodies against NifB and the N-terminal TS tag. OsFdxN was detected with antibodies against the C-terminal HA tag. The red arrow indicates the signal from OsNifB detected with antibodies against NifB. Abbreviations: OsNifB: O. sativa-derived M. infernus NifB; OsNifB: O. sativa-derived M. thermautotrophicus NifB; OsFdxN: O. sativa-derived A. vinelandii FdxN; s.e.: short exposure during immunoblot detection; l.e.: long exposure during immunoblot detection; MiB32, MiB115, and MtB35 are three independent lines. N.B. OsNifB was not detectable in multiple regenerated siblings from line MtB35.

Purification of OsNifB and OsNifB

The OsNifB and OsNifB proteins were purified from rice callus lines (MiB32, MiB115, and MtB35) by strep-tag affinity chromatography (STAC), and the purification process was monitored by sampling the total extract, cell-free extract, flow-through, wash, and elution fractions for analysis by SDS-PAGE and immunoblotting (Figures a–c and S4). No significant amount of protein was lost during centrifugation and filteration of the cell extract, confirming that both OsNifB and OsNifB proteins were soluble in the mitochondria. The elution fractions featured a band matching the anticipated size of correctly processed OsNifB or OsNifB. Isolated NifB yields were 44 and 87 μg per 100 g fresh weight callus for OsNifB and OsNifB, respectively (Figure S5). Side by side comparison of NifB and NifB proteins isolated from yeast and rice suggested the correct targeting and the specific processing of Cox4 signals from both OsNifB proteins (Figure d,e).[34] The exact cleavage site of the Cox4 sequence was further investigated by N-terminal sequencing. Cleavage was specific after amino acid 26 in the Cox4 peptide (Figure f,g), which is only one amino acid away from where the endogenous Cox4 protein is processed in S. cerevisiae,[41] generating a single OsNifB protein moiety. Removal of the Cox4 signal, following the import of OsNifB and OsNifB to the mitochondria, therefore confirmed successful targeting.

Figure 2

STAC purification and N-terminal sequence of OsNifB and OsNifB. Purification of OsNifB from MiB32 (a), MiB115 (b), and MtB35 (c) callus. TE: total extract, CFE: soluble cell-free extract, FT: flow-through fraction, W: wash fraction, and E: elution fraction. Fractions were analyzed by SDS-PAGE, followed by Coomassie gel staining or immunoblot analysis using antibodies detecting the TS tag. Migration of STAC-purified OsNifB (MiB32) and ScNifB (d) and OsNifB (MtB35) and ScNifB (e). Cleavage sites of Cox4 in OsNifB (f) and OsNifB (g). The black arrow indicates the N-terminal processing site as determined by N-terminal sequencing. The underlined amino acids represent those detected by the Edman degradation procedure. The black stars indicate the cleavage site for endogenous Cox4 in S. cerevisiae.

OsNifB and OsNifB Catalyzes FeMo-co Synthesis In Vitro

The minimal protein components for FeMo-co synthesis in vitro are NifB, NifEN, and NifH.[26] The isolated OsNifB or OsNifB protein was mixed with [4Fe–4S] cluster-loaded A. vinelandii NifU purified from E. coli (EcNifU, as the source of [4Fe–4S] precursor clusters for NifB-co biosynthesis), A. vinelandii NifEN with the permanent [4Fe–4S] clusters but lacking FeMo-co precursor cluster (apo-NifEN), A. vinelandii NifH protein (NifH), and A. vinelandii NifDK with the P-cluster but lacking FeMo-co (apo-NifDK). Molybdate, R-homocitrate, and SAM were added as they are the required substrates for NifB-dependent in vitro FeMo-co synthesis.

The as-isolated OsNifB and OsNifB proteins were colorless and inactive in FeMo-co synthesis. However, when loaded with [4Fe–4S] clusters from NifU, the OsNifB and OsNifB proteins were functional in the FeMo-co synthesis assay, in which the OsNifB-dependent activation of apo-NifDK was measured by in vitro acetylene reduction activity of the reconstituted enzyme (Figure ). This shows that the NifB-co produced by OsNifB matured into FeMo-co at the NifEN/NifH complex, which was then transferred to apo-NifDK. The OsNifB-dependent activation of apo-NifDK resulted in nitrogenase activities of 35 ± 0.95 and 19 ± 0.94 nmol C2H4 min–1 mg–1 NifDK using OsNifB isolated from lines MiB32 and MiB115, respectively, while FeMo-co synthesis using OsNifB isolated from MtB35 resulted in fourfold higher nitrogenase activities (137 nmol C2H4 min–1 mg–1 NifDK) (Figure ).

Figure 3

In vitro FeMo-co synthesis and apo-NifDK reconstitution using the as-isolated OsNifB and OsNifB proteins supplemented with [4Fe–4S] cluster substrates. Activity is represented as nanomoles of ethylene produced per minute and milligram of NifDK. The activity of the positive control reaction for FeMo-co synthesis (containing pure NifB-co instead of NifB) was 305 ± 2 units, and the activity of the ATP-mix control reaction (containing holo-NifDK) was 1506 ± 95 units. MiB32, MiB115, and MtB35 denote OsNifB or OsNifB isolated from three independent lines. Data are means ± SD (n = 2).

To rule out that this higher OsNifB activity was affecting cell growth and development and precluding the generation of plants expressing OsNifB (Figure ), we generated more lines expressing OsNifB. The MtB15 line expressed OsNifB at high levels (in addition to OsFdxN) not only in callus (Figures a and S6a) but also in leaves of the corresponding regenerated plants (Figures b and S6b), which indicates that expression of NifB is likely not detrimental to the plants. Similarly, NifB expression was shown to be stable in T1 plants of the MiB115 line (Figures and S7).

Figure 4

Accumulation of OsNifB and OsFdxN in rice callus (a) and plants (b). OsNifB was detected with antibodies against the N-terminal TS tag. OsFdxN was detected with antibodies against the C-terminal HA tag. Abbreviations: OsNifB: O. sativa-derived M. thermautotrophicus NifB; OsFdxN: O. sativa-derived A. vinelandii FdxN. MtB15 is a line accumulating OsNifB in callus and leaves.

Figure 5

Expression of OsNifB in 10 different T1 plants from the MiB115 line. Immunoblot analysis was performed using leaf soluble protein extracts and antibodies detecting the TS tag. The control lane was loaded with leaf soluble protein extracts obtained from wild-type O. sativa plants.

Discussion

The engineering of staple crops to fix nitrogen has been an important goal of plant biotechnology for several decades. If successful, this approach offers the potential to reduce or even abolish our dependence on nitrogen fertilizers, while maintaining the nitrogen content of soils. Natural BNF occurs only in some prokaryotes and is catalyzed by a nitrogenase complex with assistance from various accessory proteins required to assemble and incorporate metal cofactors into nitrogenase.[10] Many of these components are extremely sensitive to O2, which is an additional challenge when transferring the trait to plants.[42] One solution is to express the nitrogenase and its accessory proteins in the plant mitochondria, a strategy that would reduce O2 exposure and supply energy for nitrogenase activity and a ready source of [Fe–S] clusters generated by proteins similar to the bacterial NifUS system.[7]

Although many nif genes are involved in the assembly and activity of nitrogenase and its metal cofactors in bacteria, not all are expected to be required to reconstitute nitrogenase activity in plants because some of the accessory functions can be fulfilled by endogenous proteins.[43] The minimal gene set that must be transferred to plants includes NifD, NifK, and NifH which form the nitrogenase enzyme and NifE, NifN, and NifB which catalyze essential reactions in the biosynthesis of FeMo-co, the active-site cofactor of nitrogenase.[10] These six genes have been expressed in the mitochondria of S. cerevisiae and in some cases, also transiently in N. benthamiana(34,40,44−48) but have yet to be expressed in any staple crop.

The solubility of NifB is a prerequisite for its activity. Previous work on S. cerevisiae showed that the NifB proteins from both M. thermautotrophicus and M. infernus were soluble in the mitochondria and accumulated at high levels.[34] We therefore generated rice plants co-expressing OsNifB or OsNifB with OsFdxN. In our initial experiments, we observed that the expression frequency and levels of OsNifB appeared to depend on the NifB variant. While we could detect OsNifB in three out of four callus lines we analyzed, OsNifB was only detectable in one out of four lines. At the plant level, protein accumulation could only be measured in OsNifB lines. We hypothesized that OsNifB might be detrimental to cell growth and development, thus limiting the number of plants able to regenerate, when expressing the protein. We therefore initiated new transformation experiments aiming to generate more lines expressing OsNifB. Indeed, it proved to be difficult to generate additional lines expressing this protein. It is possible that NifB activity might interfere with other essential developmental processes in the cells or compete for essential precursors in metabolic processes sharing common precursors. However, we were able to obtain one line that accumulated OsNifB in both callus and regenerated plants (Figure ). While moving forward with the nitrogenase engineering process, it would be desirable to circumvent this problem, for example, by using tissue-specific or regulated promoters.

Likely reasons to explain the different outcomes when expressing NifB and NifB in rice plants are not clear at present. An overlay of NifB and NifB structures (Figure S8) shows that secondary structure elements and relevant residues in both structures match, with only two differences: (1) the NifB H22 residue, which appears to stabilize the K-cluster,[21] has been modeled by AlphaFold in the rotated position compared to its equivalent H24 residue in NifB and (2) the C-terminal stretch of NifB proteins. This region was well resolved in the NifB structure containing the K-cluster, but it was shown to be disordered before K-cluster formation (in the absence of K2-cluster) in the crystal structure of another Archaeal NifB homolog.[19] It was proposed that the short C-terminal stretch acted as a strap closing the side of the NifB β-barrel structure and stabilizing the K2-cluster.[21] It should also be noted that the confidence of the AlphaFold model for this region of NifB was low.

Earlier studies involved the co-expression of NifB with NifU, NifS, and FdxN in S. cerevisiae.[34,40] Analysis of the ScNifB protein isolated from different S. cerevisiae strains showed that while NifUS was important for providing [4Fe–4S] clusters, FdxN was more important for NifB activity.[34] Earlier studies had shown that FdxN was required for efficient NifB-co biosynthesis,[23] but its exact role is still unknown. Several non-exclusive roles in NifB Fe–S cluster acquisition or maturation or as a participant of the NifB reaction have been proposed.[10] For example, NifB produced recombinantly in S. cerevisiae required FdxN to acquire the EPR signatures of its three clusters.[34] FdxN could also promote the reductive coupling of K1- and K2-clusters to form the [8Fe–8S] K-cluster, a reaction intermediate of NifB-co synthesis.[21] Finally, FdxN could serve as an electron donor to the NifB RS-cluster for the reductive cleavage of SAM and release of the dA• radical.

In contrast to NifB, NifH showed similar activity (400 nmol of C2H4 min–1 mg–1 NifDK) when co-expressed with either NifM alone or with NifM, NifS, and NifU in S. cerevisiae.[49] This may reflect the distinct mechanisms used to incorporate [4Fe–4S] clusters into the NifB and NifH proteins or different requirements for these clusters. While NifH contains a permanent [4Fe–4S] cluster only required for catalysis, NifB requires a [4Fe–4S] cluster for catalysis and two additional [4Fe–4S] clusters as substrates for NifB-co biosynthesis.

The host organism can also influence the activity of Nif proteins. For example, when NifH was co-expressed with NifM, NifU, and NifS, it was functional in S. cerevisiae but not inN. benthamiana, and in the latter case, reconstitution in vitro was necessary to restore activity.[47] Although NifU is the major provider of [4Fe–4S] clusters for nitrogenase in vivo, theKlebsiella pneumoniaenifUS double mutant still synthesizes NifB-co, albeit at lower levels compared to the wild-type stain.[18] This shows that [4Fe–4S] clusters for NifB-co biosynthesis can be provided by other sources, such as the iron–sulfur cluster assembly or sulfur mobilization systems.[18,50] The IscS protein purified from an A. vinelandii strain with deleted nifS catalyzes the same reaction as NifS.[51] We have been unable to obtain detectable protein accumulation of NifU and NifS in transformed rice, and this study was therefore limited to the expression of NifB and FdxN. The identification of NifU and NifS variants suitable for expression in rice should therefore be the focus of future studies.

In conclusion, we were able to express the M. infernus NifB, M. thermautotrophicus NifB, and A. vinelandii FdxN proteins in rice using the Cox4 peptide to ensure the efficient targeting of all three proteins to the mitochondria, where they were correctly processed. The enzymatic activity of purified OsNifB proteins was confirmed by using the in vitro FeMo-co synthesis assay. We therefore show that these two NifB proteins fulfil the requirements for functional NifB in the rice mitochondria, such as stability, solubility, and competence, to acquire [4Fe–4S] cluster substrates, but further research is required to demonstrate active NifB in planta. This may require the co-expression of additional nif genes. Nevertheless, the expression of functional NifB in this study is an important step toward the engineering of nitrogenase activity in cereals.

Materials and Methods

Construct Preparation

The sequences of M. infernusnifB (nifB), M. thermautotrophicusnifB (nifB), and A. vinelandiifdxN (fdxN), the mitochondrial targeting peptide Cox4, and the TS and HA tags were codon-optimized for S. cerevisiae using the GeneOptimizer tool (Thermo Fisher Scientific, Waltham, MA, USA) and synthesized by Thermo Fisher Scientific, as described.[34,40] The empty vector pUC57 (GenScript Biotech, Piscataway, NJ, USA) was digested with Acc65I and SalI, allowing the insertion of the ZmUbi1 + 1st i promoter. The Cox4-TS-nifB-nos cassette was generated by PCR using pN2XJ21[40] as the template and was introduced into the intermediate vector pUC57-ZmUbi1 + 1st i at the SalI and SphI sites to produce pMiNifB. The pMiNifB vector was digested with BamHI and BstEII, allowing the insertion of the nifB sequence generated by PCR using pN2SB103[34] as the template, to produce pMtNifB. The pMiNifB vector was digested with SalI and BstEII, allowing insertion of the synthetic Cox4-fdxN-HA cassette, to generate pAvFdxN. All restriction enzymes and T4 DNA ligase were obtained from Promega (Madison, WI, USA) or New England Biolabs (Ipswich, MA, USA). The plasmids were amplified in E. coli DH5α cells grown at 37 °C in lysogeny broth medium supplemented with 100 μg/mL ampicillin. The fidelity of DNA constructs was verified by Sanger sequencing (Stabvida, Caparica, Portugal). The sequences of cloning primers and DNA constructs for rice expression are listed in Tables S1 and S2.

Transformation of Rice Explants, Callus Recovery, and Regeneration of Transgenic Plants

Seven-day-old mature rice embryos (Oryza sativa cv. Nipponbare) were isolated as explants for particle bombardment. The embryos were transferred to Murashige & Skoog (MS) osmoticum (MSO) medium for 4 h in the dark before transformation with 10 mg gold particles coated with the transgene constructs (pMiNifB and pAvFdxN or pMtNifB and pAvFdxN) and the hygromycin phosphotransferase (hpt) selectable marker at a molar ratio of 3:3:1. The bombarded embryos were maintained on MSO medium for 16 h in the dark and then transferred to MS selection medium for 4 weeks in the dark, with one subculture after 2 weeks. Half of the resistant callus was kept under selection, and the other half was transferred to MS regeneration medium with a 12 h photoperiod for 3–4 weeks to regenerate transgenic plantlets. The transgenic plantlets were transferred to rooting medium (HMS) with a 12 h photoperiod for 2 weeks and planted to soil in the greenhouse with a 12 h photoperiod and 80% relative humidity. Media compositions are listed in Table S3.

Protein Extraction and Immunoblot Analysis

Soluble rice leaf protein extracts were prepared by grinding ca. 50 mg rice tissue (snap-frozen in liquid N2) in 2 mL Eppendorf tubes using 3 mm BeadBug steel balls and a microtube homogenizer (Benchmark Scientific, Edison, NJ, USA) operating at 400 rpm for 20 s. Leaf powder was resuspended in 7 volumes (v/w) of extraction buffer comprising 100 mM Tris-HCl (pH 8.6), 200 mM NaCl, 10% glycerol, 1 mM PMSF, 1 μg/mL leupeptin, and 5 mM EDTA and homogenized twice. Cell debris was removed by centrifugation (20,000g, 5 min, 4 °C), and the supernatant was collected and stored at −80 °C. Soluble rice callus extracts were prepared using a blender (see STAC purification section).

Rice proteins were separated by SDS-PAGE and then immunoblotted to Protran Premium 0.45 μm nitrocellulose membranes (GE Healthcare, Chicago, IL, USA) using a semi-dry transfer apparatus (Bio-Rad Laboratories, Hercules, CA, USA) at 20 V for 45 min. Loading equivalence was confirmed by staining polyacrylamide gels with Coomassie brilliant blue or nitrocellulose membranes with Ponceau S. The membranes were blocked with 5% non-fat milk in 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 0.02% Tween-20 (TBS-T) for 1 h at room temperature, before incubation with primary antibodies overnight at 4 °C. Primary polyclonal antibodies against NifB (generated in-house), and monoclonal antibodies against the strep-tag II (2-1507-001, IBA Lifesciences, Göttingen, Germany) and the HA tag (H6908, Sigma-Aldrich, St Louis, MO, USA) or Rubisco (AS03 037A, Agrisera, Vännäs, Sweden, used as loading control) were diluted at 1:2,000–1:5,000 in TBS-T supplemented with 5% bovine serum albumin (BSA). Secondary antibodies (Thermo Fisher Scientific) were diluted at 1:20,000 in TBS-T supplemented with 2% non-fat milk and incubated for 2 h at room temperature. Membranes were developed on medical X-ray films (AGFA, Mortsel, Belgium) using enhanced chemiluminescence.

Purification of OsNifB and OsNifB by Strep-Tag Affinity Chromatography

OsNifB and OsNifB were prepared for STAC purification at O2 levels below 1 ppm in an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI, USA or MBraun, Garching, Germany). Callus was disrupted in lysis buffer comprising 100 mM Tris-HCl (pH 8.5), 300 mM NaCl, 10% glycerol, 3 mM sodium dithionite (DTH), 5 mM 2-mercaptoethanol, 1 mM PMSF, 1 μg/mL leupeptin, 10 μg/mL DNAse I, and 1:200 (v/v) BioLock solution (IBA Lifesciences Göttingen, Germany) at a ratio of 1:3 (w/v). Total extracts were prepared by lysing the cell suspensions under anaerobic conditions using the Oster 4655 blender (Newell Brands, Atlanta, GA, USA) modified with a water-cooling system operating at full speed in 4 cycles of 2 min at 4 °C. Extracts were transferred to centrifuge tubes equipped with sealing closures (Beckman Coulter, Brea, CA, USA) and centrifuged (50,000g, 1.5 h, 4 °C) using the Beckman Coulter Avanti J-26 XP device. The supernatant was passed through Nalgene 0.2 μm filter cups (Thermo Fisher Scientific) to yield a cell-free extract of soluble proteins. This was loaded at 2.5 mL/min onto a 5 mL Strep-Tactin XP column (IBA Lifesciences) attached to an ÄKTA FPLC system (GE Healthcare). The column was washed with 150 mL of 100 mM Tris-HCl (pH 8.0), 300 mM NaCl, 10% glycerol, 2 mM DTH, and 5 mM 2-mercaptoethanol at 16 °C, and bound proteins were eluted with 15–20 mL of the same wash buffer supplemented with 50 mM biotin (IBA LifeSciences). The elution fraction was concentrated using the Amicon Ultra centrifugal filter (Millipore Sigma, Burlington, MA, USA) with a cut-off size of 10 kDa. Biotin was removed by passing the protein through PD-10 desalting columns (GE Healthcare) equilibrated with wash buffer. The desalted eluate was concentrated and snap-frozen in Nalgene cryovials and stored in liquid nitrogen.

Quantification of Purified OsNifB and OsNifB Proteins and N-Terminal Sequencing

The yield of purified OsNifB and OsNifB was determined by Coomassie gel titration against standards of the purified S. cerevisiae NifB (ScNifB and ScNifB) protein, as shown in Figure S5. Amino terminal amino acid sequencing was performed by Edman degradation (Centro de Investigaciones Biológicas, Madrid, Spain). 25 pmol OsNifB protein was separated by SDS-PAGE, transferred to 0.2 μm Sequi-Blot PVDF membranes (Thermo Fisher Scientific) in 50 mM borate buffer (pH 9.0), stained with freshly prepared 0.1% Coomassie R-250 (Sigma-Aldrich) in 40% methanol and 10% acetic acid, and then destained using 50% methanol.

FeMo-Co Synthesis and Apo-NifDK Reconstitution In Vitro

FeMo-co synthesis and apo-NifDK reconstitution assays were carried out in vitro in an anaerobic chamber, as previously described.[34] For the in vitro synthesis of FeMo-co, each 100 μL reaction contained 3 μM NifH, 1 μM OsNifB, 1.5 μM apo-NifEN, 0.6 μM apo-NifDK, 17.5 μM Na2MoO4, 175 μM R-homocitrate, 9 μM [Fe4–S4]cluster-loaded NifU (holo-NifU), 125 μM SAM, 1 mg/mL BSA, and the ATP-regenerating mixture (1.23 mM ATP, 18 mM phosphocreatine disodium salt, 2.2 mM MgCl2, 3 mM DTH, 46 μg/mL creatine phosphokinase). For the positive control FeMo-co synthesis assay, holo-NifU was omitted, and OsNifB was replaced with 2.5 μM NifB-co. The reactants were incubated for 60 min at 30 °C. For the acetylene reduction assays, 500 μL of the ATP-regenerating mixture and 2.0 μM NifH were added to the reaction tube. The reaction mixture was then transferred to 9 mL serum vials under an argon/acetylene (94%/6%) atmosphere. The reaction was incubated for 20 min at 30 °C. To measure ethylene formation, 50 μL of the gas phase was taken from the reaction vials and injected in the Shimadzu GC-2014 gas chromatographer equipped with the Porapak N 80/100 column (Shimadzu, Kyoto, Japan).

Statistical Analysis

Standard deviation (SD) of in vitro activity data was calculated based on two biological replicates (each one with two technical replicates).