A critical role of an oxygen-responsive gene for aerobic nitrogenase activity in Azotobacter vinelandii and its application to Escherichia coli

Since nitrogenase is irreversibly inactivated within a few minutes after exposure to oxygen, current studies on the heterologous expression of nitrogenase are limited to anaerobic conditions. This study comprehensively identified genes showing oxygen-concentration-dependent expression only under nitrogen-fixing conditions in Azotobacter vinelandii, an aerobic diazotroph. Among the identified genes, nafU, with an unknown function, was greatly upregulated under aerobic nitrogen-fixing conditions. Through replacement and overexpressing experiments, we suggested that nafU is involved in the maintenance of nitrogenase activity under aerobic nitrogenase activity. Furthermore, heterologous expression of nafU in nitrogenase-producing Escherichia coli increased nitrogenase activity under aerobic conditions by 9.7 times. Further analysis of NafU protein strongly suggested its localization in the inner membrane and raised the possibility that this protein may lower the oxygen concentration inside the cells. These findings provide new insights into the mechanisms for maintaining stable nitrogenase activity under aerobic conditions in A. vinelandii and provide a platform to advance the use of nitrogenase under aerobic conditions.

ity under aerobic conditions seems to be induced in an oxygen concentration-dependent manner only under nitrogen-fixing conditions. For example, cytochrome bd is considered to protect nitrogenase from oxygen, and its expression depends on the oxygen concentration under nitrogen-fixing conditions 35,36 . Also, FeSII and the production of alginate, suggested to protect nitrogenase from oxygen, are induced under nitrogen-fixing conditions 29 and higher oxygen concentrations 23,37 , respectively. These previous results led us to believe that the mechanisms to maintain nitrogenase activity under aerobic conditions are unnecessary and are rather suppressed under low oxygen conditions. Therefore, we assumed that the genes involved in the mechanism are expressed only under nitrogen-fixing conditions, and their expression strongly depends on the oxygen concentration. A comprehensive search for genes showing such an expression pattern could thus discover critical genes involved in protecting nitrogenase from oxygen.
We first examined the culture conditions for identifying these genes by measuring the expression levels of the oxygen-responsive cydA, encoding the subunit of cytochrome bd, and nifH, encoding the subunit of nitrogenase and being repressed in the presence of nitrogen 33 . The cells pre-cultured in the medium without a nitrogen source were inoculated into the medium with or without a nitrogen source in sealed vials, and the expression level of nifH was quantified 2 h and 4 h after the inoculation to determine the optimal incubation time for genes to be induced and suppressed in A. vinelandii. As a result, significant difference in expression levels was observed at 2 h compared to 4 h (Fig. S1) to determine the optimal incubation time of 2 h. Next, we examined the oxygen concentrations in the air layer of the vials required for inducing oxygen-responsive genes by quantifying the expression level of cydA after 2 h incubation in the medium with or without a nitrogen source under 5%, 10%, or 20% oxygen. As a result, cydA expression depended on the oxygen concentration only under nitrogen-fixing conditions (Fig. S2). These results suggest that the incubation of A. vinelandii cells under these conditions for 2 h is sufficient to induce and identify a group of genes showing oxygen-concentration-dependent expression only under nitrogen-fixing conditions.
We conducted transcriptome analysis under the above culture conditions and identified 213 genes whose expressions depended on oxygen concentration only under nitrogen-fixing conditions ( Fig. 1a; Table S1). Among these genes, we focused on the 130 genes whose expression was strongly induced under aerobic nitrogen-fixing conditions. Specifically, the focused genes showed increased expression over 1.5-fold in both comparisons: (i) in the medium without a nitrogen source under 20% oxygen compared to under 5% oxygen and (ii) in the medium without a nitrogen source compared to the medium with a nitrogen source under 20% oxygen ( Fig. 1b; Table S2). The focused genes included a group of genes with known functions, such as the genes related to aerobic conditions, nitrogen fixation, FeSII and cyd genes, suggested to contribute to the maintenance of nitrogenase activity under aerobic conditions. Meanwhile, many genes with unknown functions were included. We then focused on nafU (Fig. 1c), suggested to be related to ammonia production 34 but whose function was unknown 38 , because its expression was most induced under aerobic nitrogen-fixing conditions compared with other genes with unknown function (Fig. 1b). This result was validated by RT-qPCR, and we confirmed that nafU was strongly induced when A. vinelandii was cultured under 20% oxygen without nitrogen (Fig. S3). Identification of genes whose expression depends on oxygen concentration only under nitrogen-fixing conditions by transcriptome analysis. (a) Heat map generated from the results of transcriptome analysis after ANOVA analysis. The heat map was generated using JMP 16 (https:// www. jmp. com). The z-score was calculated for each gene, and genes were clustered using Ward's method. The cluster surrounded by a red square represents the cluster where genes are upregulated in response to oxygen concentration only without nitrogen source and hardly expressed with nitrogen source. N source " + ": with nitrogen source, N source "−": without nitrogen source, O 2 : O 2 concentration, n = 3 for under 20% O 2 with nitrogen source, n = 4 for all other conditions. b Fold changes of 213 genes from the cluster in the red square of the heat map (a). Scatter plots show the fold changes of log 2 (without nitrogen/with nitrogen) under 20% oxygen and log 2  and nafU-complemented strain (nafU comp) by genomic insertion (Fig. 2a). RT-qPCR analysis showed that the expression of nafU was not detected in the ΔnafU strain and was comparable in the nafU comp strain to that of WT (Fig. S4). As a result, the final OD 600 was lower in the ΔnafU strain than WT and nafU comp in the growth curve in the medium without a nitrogen source under aerobic conditions (Fig. 2b). In addition, we measured the nitrogenase activity of each strain at 5% and 20% oxygen concentrations. In the ΔnafU strain, nitrogenase activity was significantly reduced only under 20% oxygen concentration, suggesting that nafU contributes in maintaining nitrogenase activity under aerobic conditions (Fig. 2c). To further investigate the positive effect of nafU on nitrogenase activity under aerobic conditions, we constructed the nafU-overexpressing strain (nafU OE) by replacing the promoter of nafU with that of nifH, strongly induced under nitrogen-fixing conditions (Fig. 2d). Compared with WT, nafU OE strain showed about threefold higher transcription and expression of nafU (Fig. S5, S12). In addition, the use of the nifH promoter for nafU overexpression did not significantly reduce the expression of nifH in the nafU OE strain compared to WT (Fig. S6). Nitrogenase activity of nafU OE strain under 20% oxygen concentration was significantly higher than that of WT and showed the same level of nitrogenase activity as that under 5% oxygen (Fig. 2e). These results indicate that nafU contributes to the maintenance of nitrogenase activity under aerobic conditions.
Aerobic nitrogenase activity of nitrogenase-producing Escherichia coli is improved by nafU. nafU, identified by transcriptome analysis, was important in maintaining nitrogenase activity under aerobic conditions in A. vinelandii. Therefore, the positive impact of nafU may apply to the heterologous expression of nitrogenase under aerobic conditions. To test this hypothesis, we utilized the nitrogenase-producing E. coli constructed in our previous study 39 . This E. coli strain harbors two plasmids with four operons consisting of 17 nif-related genes, cloned from A. vinelandii genomic DNA and assigned to each operon based on its expression level. Codon-optimized nafU (see "Methods", Table S6) was overexpressed by introducing the expression vector in nitrogenase-producing E. coli (Fig. 3a). As a result, the overexpression of nafU greatly increased nitrogenase activity under 20% oxygen by approximately 9.7-fold ( Fig. 3b). In contrast, the nafU overexpression did not affect the growth of nitrogenase-producing E. coli (Fig. S7). These results indicate that nafU is also effective for maintaining nitrogenase activity under aerobic conditions in E. coli, suggesting that nafU is a useful gene for heterologous expression of nitrogenase under aerobic conditions in other microorganisms.

Functional estimation of NafU protein.
To understand the mechanism by which NafU protein maintains nitrogenase activity under aerobic conditions, we employed an analysis based on the functionally known proteins that are predicted to have a similar structure to NafU protein using a secondary structure prediction tool (JPred4) 40 . As a result, SlyB was predicted to have a secondary structure similar to NafU protein (Fig. S8). In some bacteria, SlyB is localized in the cell membrane and protects the cells from external stresses by increasing the membrane stability 41 . Another protein with a similar function, Hsp12 in yeast, is also known to localize in the membrane and protect the cells from stress by improving the membrane integrity 42 . Based on these analyses, we hypothesized that NafU protein may localize to the cell membrane and protect intracellular nitrogenase from oxygen by improving membrane integrity. We first examined the localization of NafU protein by cell fractionation and western blot analysis. The fractionation was validated by a quantitative proteome to examine whether each fraction included proteins with known localization. Proteome analysis showed that soluble cytoplasmic proteins such as NifH 43 were enriched in the cytoplasmic fraction but not in the membrane fractions. In addition, the proteins that have been shown to localize in the inner membrane, such as CydA 44 , were enriched in the inner membrane fraction, and other outer membrane proteins such as OprI 45 were enriched in the outer membrane fraction (Fig. S9). Then, each fraction validated by these results was subjected to western blot analysis against the His tag fused to NafU protein. The   Figure 2. The contribution of nafU gene to the maintenance of nitrogenase activity under aerobic conditions. (a) Construction of A. vinelandii ΔnafU and nafU comp strains by genomic insertion. The homologous regions (500-600 bp) upstream and downstream of the nafU gene were used for homologous recombination. P cat , Cm R , and Kan R represent the promoters of chloramphenicol acetyltransferase (cat), chloramphenicol resistant gene, and kanamycin resistant gene, respectively. (b) Growth curves of wild-type, ΔnafU, and nafU comp strains in MB liquid medium without nitrogen source under aerobic conditions. Data are presented as the mean ± SE of three independent experiments. P-values were calculated using the Tukey_HSD test. (c) Nitrogenase activities of wild-type, ΔnafU, and nafU comp strains under 5% and 20% oxygen. The cells were sampled at 14 h after the start of culture. The Data are shown as relative values when the average activity in "wild-type strain under 5% oxygen" is 100% (357.49 nmol ethylene h −1 mL −1 OD 600 -1 ). Data are presented as the mean ± SE of three independent experiments. P-values of the results under 5% oxygen and 20% oxygen were calculated using the Tukey_HSD test, respectively. (d) Construction of the nafU OE strain by genomic insertion. The homologous regions (approximately 1700 bp) upstream and downstream of the nafU gene were used for homologous recombination. P nifH represents the nifH promoter e Nitrogenase activity of wild-type and nafU OE strains under 5% and 20% oxygen. The cells were sampled at 14 h after the start of culture. Data are shown as relative values when the average activity in "wild-type strain under 5% oxygen" is 100% (479.21 nmol ethylene h −1 mL −1 OD 600 -1 ). Data are presented as the mean ± SE of five independent experiments. P-values were calculated using the Student's two-sided t-test. www.nature.com/scientificreports/ NafU-His band was detected in the inner membrane fraction (Fig. 4a,b; Fig. S11), indicating that NafU is an inner membrane protein.
Next, we performed transcriptome analysis of WT and ΔnafU strains to examine the possibility that NafU protein reduces the intracellular oxygen concentration. The increase or decrease in intracellular oxygen concentration upon nafU deletion was examined by the expression changes of oxygen-responsive genes (identified in Fig. 1). Since the transcriptional level of nafU was highest around 2 h after inoculation, similar to nifH (Fig. S10), WT and ΔnafU strains were incubated in the medium without a nitrogen source for 3 h after inoculation to induce the production of NafU protein. After the induction, and the cells were incubated for another 2 h with the vialed being sealed to reproduce the same culture conditions as in the transcriptome analysis in Fig. 1a, followed by transcriptome analysis. Based on the results (Table S3) www.nature.com/scientificreports/ (GSEA) [46][47][48] for the genes that were induced in response to oxygen (O 2 UP gene set) and suppressed in response to oxygen (O 2 DOWN gene set) identified by transcriptome analysis (Fig. 1a) in the medium without a nitrogen source. Specifically, in the transcriptome in the medium without a nitrogen source (Fig. 1a), the top 300 genes on log 2 (20% O 2 /5% O 2 ) were defined as the O 2 UP gene set and the bottom 300 genes as the O 2 DOWN gene set (Table S4). The GSEA of the ΔnafU strain compared to WT for those gene sets revealed a positive correlation between the genes whose transcription levels were increased in the ΔnafU strain and the O 2 UP gene set, as well as a negative correlation with the O 2 DOWN gene set (Fig. 4c,d). From these results, there is a possibility that NafU protein functions to decrease intracellular oxygen concentration.

Discussion
Maintaining the activity of oxygen-sensitive nitrogenase, even under aerobic conditions, has been a challenge for the biological use of nitrogenase via heterologous expression. However, no previous studies have significantly increased nitrogenase activity under aerobic conditions, and the heterologous expression of nitrogenase has so far been conducted only under anaerobic conditions 32,49,50 . Our transcriptome analysis identified the genes that were upregulated in an oxygen concentration-dependent manner under nitrogen-fixing conditions. Among them, we focused on the most upregulated gene, nafU with a previously unknown function 38 , and revealed that this gene contributes to the maintenance of nitrogenase activity under aerobic conditions in A. vinelandii. We also demonstrated that NafU is an inner membrane protein and suggested the possibility that this protein may reduce intracellular oxygen concentration. Furthermore, by introducing nafU, we displayed the improvement of nitrogenase activity of nitrogenase-producing E. coli under aerobic conditions, which has not been achieved in previous studies. Therefore, our findings could provide a promising platform for the expression of functional nitrogenase under aerobic conditions in other microorganisms and pave the way for the realization of heterologous aerobic nitrogen fixation. Our transcriptome analysis also identified genes with known functions such as nif genes, involved in nitrogen fixation and expressed in an oxygen concentration-dependent manner under nitrogen-fixing conditions 51 . Therefore, it is conceivable that some genes with unknown functions that show similar expression patterns to Plasmids for producing nitrogenase and nafU in E. coli. Each nif gene was cloned from A. vinelandii genomic DNA, and nafU was codon-optimized for E. coli. P trc and P lac represent the trc and lac promoters, respectively. pTrc-nif001, pMW-nif002, and pSTV-nafU were constructed from pTrcHis2-TOPO, pMW219, and pSTV29, respectively. (b) Nitrogenase activity under 20% oxygen of nitrogenase-producing E. coli harboring vector expressing nafU or empty vector. Data are shown as relative values when the average activity in E. coli harboring "pSTV29 (empty)" is 100% (0.0100 nmol ethylene h −1 mL −1 OD 600 -1 ). Data are presented as the mean ± SE of three independent experiments. P-values were calculated using the Student's two-sided t-test. www.nature.com/scientificreports/ www.nature.com/scientificreports/ nif genes could be involved in nitrogen fixation. In addition, many genes involved in aerobic respiration, such as zwf 51 , ace 52 , and pyk 53 were also part of the identified genes. There are two possible reasons for this finding: (i) cells have to make a lot of energy through respiration for nitrogen fixation, an energy-consuming reaction that requires more than 16 ATP to fix one nitrogen molecule to two ammonia molecules 54 , (ii) A. vinelandii prevents oxygen molecules from entering the cells by activating respiration 24 . Azotobacter vinelandii has a much faster respiratory rate than other bacteria 55 . Genes involved in the biosynthesis of alginates, such as algA and algC 56 , have also been suggested to maintain nitrogenase activity under aerobic conditions. However, our transcriptome analysis demonstrated that the expression of these genes did not differ significantly between the presence and absence of a nitrogen source in the medium and were rather higher under low oxygen conditions (5%) in terms of RNA levels. This was presumably because A. vinelandii is an obligate aerobe and consumed all the oxygen of 5% in the vials by their high respiration, leading to cyst formation that requires alginate biosynthesis 56 . In this regard, the lack of a significant difference in the nifH expression level when incubated in sealed vials for 4 h may be due to the induction of cyst formation or cell death by the decrease in oxygen concentration.
A previous study examining gene fitness in A. vinelandii by Tn-seq suggested that the deletion of nafU does not affect growth under aerobic nitrogen fixing conditions at ~ 7 generations (about 14 h) 57 . This result is consistent with our result at 12 h (Fig. 2b), and the effect on growth can be seen only after increasing incubation time. On the other hand, we found that the deletion of nafU significantly reduces nitrogenase activity under aerobic conditions when measured at 14 h after the start of culture (Fig. 2c). Based on our speculation that NafU protein is present in the inner membrane to improve membrane stability, the accumulation of NafU protein in the membrane may be insufficient to be effective and mechanisms other than nafU may be dominant in the early stages of growth.
The nafU gene is conserved in the genus Azotobacter and some Pseudomonas, according to the results of a BLAST homology search, and does not form an operon with other genes 58 . However, the nafU gene is located near the nif clusters in the genome (Fig. 1c), suggesting that it has been maintained together as a gene that promotes nitrogen fixation. The upregulation of nafU in high-ammonium-excreting A. vinelandii strain also supports the importance of this gene in nitrogen fixation 34 . In addition, the secondary structure prediction of NafU protein by JPred4 40 suggested its similarity in secondary structure to SlyB (Fig. S8), present in some bacteria, localized to the membrane, and increases membrane stability to protect cells from stress 41 . In yeast, there is also a protein with a similar function, Hsp12 42 . In this study, we demonstrated that NafU protein localizes to the inner membrane and suggested that NafU protein may reduce intracellular oxygen concentrations www.nature.com/scientificreports/ based on the altered transcription of oxygen-responsive genes. These results suggest the possibility that NafU protein can prevent oxygen permeation by increasing inner membrane integrity. Based on this mechanism, it is quite reasonable that the expression of nafU was suppressed under low oxygen conditions and the medium with a nitrogen source, because the cells would actively have to take in oxygen and a nitrogen source under such conditions, where the expression of nafU is detrimental to survival. To validate this mechanism, further studies are needed on the biochemical aspects of NafU protein, such as its contribution to membrane stability and the identification of important sites for its function.
The nafU gene also significantly affected nitrogenase activity positively under aerobic conditions in nitrogenase-producing E. coli. However, the nitrogenase activity of nitrogenase-producing E. coli with nafU was still low compared to that of A. vinelandii. Since genes involved in aerobic respiration were up-regulated under aerobic nitrogen-fixing conditions, these genes may also contribute to maintaining nitrogenase activity under aerobic conditions. Thus, the additional expression of these genes and nafU in the nitrogenase-producing E. coli may further increase the activity.
In conclusion, this study provides new insights into the mechanism by which A. vinelandii maintains a stable nitrogenase activity even under aerobic conditions. Furthermore, our findings can be applied to the heterologous expression of nitrogenase in E. coli to increase nitrogenase activity under aerobic conditions. This improvement has not been achieved in previous studies. In addition to nitrogenase, other oxygen-sensitive enzymes, such as hydrogenase, are useful but have not been used effectively because of their oxygen sensitivity. Therefore, our findings would greatly contribute to realizing the heterologous expression of active nitrogenase under aerobic conditions and the stable heterologous expression of other oxygen-sensitive enzymes under aerobic conditions.

Methods
Strains, media, and culture conditions. Azotobacter vinelandii Lipman (ATCC 9046) and its derivatives used in this study are listed in Table S5. The A. vinelandii strains were cultured on a modified Burk's medium plate 59  RNA isolation. For transcriptome analysis, A. vinelandii strains were cultured on MB + plates for 2 days, inoculated into MBliquid medium with an initial OD 600 of 0.1, and incubated for 24 h. Next, the cells were harvested, washed with ice-cold phosphate solution (0.1 g K 2 HPO 4 and (0.4 g KH 2 PO 4 in 450 mL distilled water), and inoculated into 25 mL of MB + or MBliquid medium with an initial OD 600 of 0.5. The culture vials were purged with N 2 for 1 min and sealed, and then O 2 was injected into the air layer to bring the O 2 concentration to 5%, 10%, or 20%. After 2 h of incubation, an equal volume of ice-cold phosphate solution was added to the culture medium, and the cells were collected at 3000×g for 2 min. After removing the supernatant, the cell pellets were flash-frozen in liquid nitrogen and stored at −80 °C until needed. When measuring the expression levels of nafU (Figs. S4 and S5), the cells were cultured on MB + plates for 2 days, in MB + liquid medium for 24 h, and in MBliquid medium for 2 h. In the transcriptome analysis of WT and ΔnafU strains (Fig. 4c,d), the cells were cultured on MB + plates for 2 days, in MB + liquid medium for 24 h, and in MBliquid medium for 3 h. Then, the vials were sealed, and the cells were collected after an additional 2 h of incubation.
Total RNA was extracted using TRIzol reagent (Invitrogen, NY, USA) according to the manufacturer's instructions. RNase-free DNase I (NIPPON GENE, Tokyo, Japan) were used with RNase Inhibitor (TOYOBO, Osaka, Japan) to remove the residual genomic DNA. The quantity and quality of the isolated RNA were measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, MA, USA) and BioAnalyzer 2100 (Agilent Technologies, CA, USA), respectively. For RNA-seq analysis, rRNAs were removed using the RiboMinus Transcriptome Isolation Kit (Thermo Fisher Scientific). The concentration of mRNA was measured by Bioanalyzer 2100 using an RNA 6000 Nano Kit (Agilent Technologies).
Reverse transcription-quantitative PCR (RT-qPCR) analysis. cDNA was synthesized from the isolated RNA using a High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Quantitative real-time PCR (qPCR) was performed using Fast Start SYBR Green Master (Roche, Mannheim, Germany) or Power SYBR Green PCR Master Mix (Thermo Fisher Scientific), and the measurement was performed using a C1000 Touch Thermal Cycler (Bio-Rad, CA, USA). The sequences of the primers used for qPCR are listed in Table S6. The expression level of gyrA was used as an internal control 60 to normalize that of nifH and cydA (Figs. S1 and S2), and the expression level of rho was used for the other experiments (Figs. S3, S4, S5a, S6, and S10) because the transcriptome analysis results (Fig. 1a) showed that the expression level of rho had less variation among the culture conditions. The relative gene expression was quantified using the standard curve method.  (Fig. 1a) and HiSeq X Ten (150 bp nucleotide paired-end sequence) to estimate the function of nafU (Fig. 4b,c). All raw data files were deposited in the ArrayExpress. The RNA-seq reads were aligned to the A. vinelandii DJ (CP001157.1) genome sequence using EDGE-Pro 61 . The read counts were extracted and normalized using edgeR 62 . The heat map was created using JMP 16 (SAS Institute Inc., Cary, NC, USA) for genes that showed significant differences by ANOVA analysis.
GSEA analysis using GSEA 4.1.0 [46][47][48] was performed on the transcriptome data after alignment and normalization to determine the function of NafU protein. The transcriptomic results of WT and ΔnafU strains were subjected to enrichment analysis in the absence of nitrogen (Fig. 1a) to examine the effect of NafU protein on intracellular oxygen levels. Specifically, among the genes included in the cluster diagram shown in Fig. 1a, in the medium without a nitrogen source, the top 300 genes on log 2 (20% O 2 /5% O 2 ) were defined as the O 2 UP gene set and the bottom 300 genes as the O 2 DOWN gene set (Table S4). GSEA analysis was performed with default settings for these gene sets, except phenotype labels being set as ΔnafU vs. WT, Collapse/Remap to gene symbols as No_Collapse, and Permutation type as a gene set.

Construction of A. vinelandii strains.
All primers used for the construction of the A. vinelandii strains are listed in Table S6. DNA fragments were combined with linearized vectors by In-Fusion HD Cloning Kit (Takara Bio). For the construction of the strain with a disruption of nafU (ΔnafU), homologous sequences of 500-600 bp in the downstream and upstream of nafU gene were amplified by PCR from the A. vinelandii genomic DNA extracted with the DNeasy Blood & Tissue Kits (QIAGEN, Hilden, Germany) using primer sets of "nafU DIS A" and "nafU DIS B" , respectively. Then, these fragments were combined with linearized pUC19 (Thermo Fisher Scientific) by PCR using a primer set "pUC19, " flanking the chloramphenicol resistance gene amplified from pSTV29 (Takara Bio) using a primer set of "Chloramphenicol. ". The constructed plasmid was named pUC19-ΔnafU and introduced into A. vinelandii to delete nafU by genome insertion, according to previous reports 54,63 . The ΔnafU cells were selected on an MB + plate containing 170 mg/L chloramphenicol.
The nafU gene was complemented in the ΔnafU strain as below. The DNA fragment containing 647 bp downstream of nafU and the fragment containing 577 bp upstream of nafU, nafU ORF, and its terminator were amplified using "nafU comp A" and "nafU comp B, " respectively. The fragment of the kanamycin resistance gene amplified from pMW219 (NIPPON GENE) using a primer set of "Kanamycin" was connected downstream of Cat promoter amplified by a primer set of "Cat promoter. " These fragments were combined with linearized pUC19. The constructed plasmid was named pUC19-nafU comp and introduced into the ΔnafU strain by genome insertion. The nafU Comp strain was selected on an MB + plate containing 2 mg/L kanamycin.
The strain overexpressing nafU was constructed as follows. Homologous sequences of approximately 1700 bp downstream and upstream of the nafU gene were amplified from the genomic DNA of A. vinelandii using primer sets "nafU OE A" and "nafU OE B", respectively. These fragments were combined with linearized pUC19, the fragment consisting of the Kan resistant gene under Cat promoter, the terminator of nafU, and a part of His tag amplified from pUC19-nafU comp using a primer set of "Kan + P cat + terminator + His", the fragment containing nafU and a part of His tag amplified from the genomic DNA of A. vinelandii using a primer set of "His + nafU", and the fragment of nifH promoter amplified from the genomic DNA of A. vinelandii using a primer set of "nifH promoter". The constructed plasmid was named pUC19-nafU OE and introduced into the ΔnafU strain. The nafU OE strain was selected in the same way as the nafU comp strain.
The nafU-His strain under the native promoter was constructed using almost the same method as the nafU OE strain with an upstream homologous region designed to contain the nafU gene. The primer sets "nafU Native A, "nafU Native B", and "Kan + P cat + terminator + His" were used. The constructed plasmid was named pUC19-nafU-His under the native promoter.
The construction of the above A. vinelandii strains was confirmed by Sanger sequencing of the inserted DNA region.
Cell fractionation. Cell fractionation was performed using a previously described protocol 64 with some modifications. A. vinelandii WT and nafU OE strains were cultured on MB + plates for 2 days and inoculated into MB + liquid medium with an initial OD 600 of 0.1. After cultivation for 24 h, the cells were collected, washed with ice-cold phosphate solution, and incubated in MBliquid medium for 4 h with an initial OD 600 of 0.2. Cell pellets were washed three times with PBS (pH 7.4) containing 15 mM EDTA to reduce cell aggregation and were frozen in liquid nitrogen. The frozen pellets were suspended in ice-cold 10 mM Tris-HCl (pH 8.0) and sonicated on ice using a sonicator (Q125, QSONICA, CT, USA) at 20 kHz and 50% amplitude for 10 s pulse with 10 s intervals until the cumulative J reached 300 J. After sonication, unbroken cells and debris were pelleted at 18,000×g for 5 min at 4 °C. The supernatant was collected, and the pellet was resuspended in ice-cold 10 mM Tris-HCl (pH 8.0), followed by sonication under the same conditions as above. This procedure was repeated four times to completely disrupt the cells. The lysate was then ultracentrifuged at 100,000×g for 90 min at 4 °C, and the supernatant was collected as the cytoplasmic fraction. The pellet was suspended in 12 mL of ice-cold 10 mM Tris-HCl (pH 8.0) containing 1.7% sarkosyl on ice and incubated for 20 min on ice. The supernatant after ultracentrifugation at 100,000×g for 90 min at 4 °C was collected as the inner membrane fraction. The pellet was suspended in 12 mL of PBS containing 2% (w/v) SDS and incubated at 95 °C for 30 min. After centrifugation at 5700×g to remove cellular debris, the supernatant was collected as the outer membrane fraction. Each fraction was concentrated by ultrafiltration using an Amicon Ultra-15 centrifugal filter unit (Merck, Darmstadt, Germany) according to the manufacturer's instructions. www.nature.com/scientificreports/ To confirm the overexpression of nafU in the constructed strain nafU OE (Fig. S5, S12), only the membrane protein was extracted as follows. The nafU-His strain that produces nafU-His under the control of the native promoter and nafU OE strain were cultured under the same conditions as above. The harvested cell pellets were frozen with liquid nitrogen, suspended in 100 μL of PBS containing 15 mM EDTA per 0.1 g of wet pellet weight, and sonicated as above. The lysate was then ultracentrifuged at 100,000×g for 90 min at 4 °C, and the pellet was suspended in PBS containing 2% (w/v) SDS and boiled at 95 °C for 30 min. The solution was then centrifuged at 5700×g for 5 min at RT, and the supernatant was collected as the membrane fraction. Each fraction was concentrated by ultrafiltration using an Amicon Ultra-15 centrifugal filter unit (Merck, Darmstadt, Germany) according to the manufacturer's instructions.
Western blot analysis. Each protein sample was separated by 12% (w/v) SDS-PAGE and transferred onto a polyvinylidene fluoride (PVDF) membrane. The membrane was blocked with 5% skim milk in PBS, probed with anti-His-tag mAb-HRP-DirecT (MBL, Tokyo, Japan) at a 1:5000 dilution, and visualized using Chemi-Lumi One Super (Nacalai Tesque, Kyoto, Japan).
The protein-transferred PVDF membrane was stained with 0.025% (w/v) CBB in 40% (v/v) MeOH for total protein normalization, destained with 50% (v/v) MeOH containing 7% (v/v) acetic acid, and scanned for ImageJ quantification to confirm the overexpression of nafU at the protein level. After scanning, complete decolorization was carried out using 50% MeOH containing 25% (v/v) acetic acid, followed by western blot analysis using an anti-His-tag mAb-HRP-DirecT.Blocking was performed with 0.3% (w/v) skim milk in TBS containing 0.05% (v/v) Tween 20, and the antibody was diluted with IMMUNO SHOT (COSMO BIO, Tokyo, Japan) to increase the band intensity.
Preparation of protein digests. The protein sample of each fraction was subjected to chloroform/methanol precipitation 65 and the pellet was dissolved in solubilization buffer (4.97 mg sodium deoxycholate and 3.74 mg sodium N-lauroylsarcosinate in 1 mL of 50 mM TEAB solution). The protein concentration was determined by BCA assay. The samples were reduced and alkylated by incubation with 50 mM DTT at 37 °C for 15 min and 50 mM IAA at RT for 30 min. Next, the samples were mixed with Trypsin/Lys-C Mix (Promega, WI, USA) at a total protein: enzyme volume ratio of 50:1, and digested by shaking at 37 °C for 24 h and the reaction was stopped by 0.5% (v/v) trifluoroacetic acid. The digestion was confirmed by 12% SDS-PAGE with silver staining (Sil-Best Stain One, Nacalai Tesque). The digested peptides were desalted using a MonoSpin C18 column (GL Sciences, Tokyo, Japan). The desalted samples were labeled with TMTsixplex™ Isobaric Label Reagent (Thermo Fisher Scientific) according to the manufacturer's instructions, and 60 μL of each labeled sample was mixed. The mixture was dried by vacuum centrifugation for the proteome analysis. Proteome analysis. Proteome analysis was performed in the previously optimized method using an LC-MS system (LC, UltiMate 3000 RSLCnano System, and MS, LTQ Orbitrap Velos mass spectrometer; Thermo Fisher Scientific) equipped with a long monolithic silica capillary column (490 cm in length, 0.075 mm ID; Kyoto Monotech, Kyoto, Japan) 66,67 . All raw data files were deposited in the jPOST database 68 . We performed the data analysis according to the previously described method 69 with an exception of using the NCBI protein database of A. vinelandii DJ (CP001157.1) for alignment.

Measurement of acetylene reduction by A. vinelandii.
After cultivating each strain on MB + plates for 2 days, the cells were collected, washed with phosphate solution, and inoculated into MB + liquid medium with an initial OD 600 of 0.1. After 24 h of incubation, an equal volume of ice-cold phosphate solution was added to the culture, and the cells were collected at 5700×g for 5 min at 4 °C and washed three times with phosphate solution. The cells were then incubated in MBliquid medium with an initial OD 600 of 0.1, at 30 °C and 250 rpm for 14 h under aerobic conditions. After incubation, 2.5 mL of the cell culture and 2.5 mL of MBliquid medium were placed into a 17 mL vial and flashed with Ar for 1 min, followed by sealing with a butyl rubber (GL Science). In the vials, 5% or 20% of the headspace gas was replaced with O 2 , and 10% of the headspace gas was replaced with acetylene generated from CaC 2 . After incubation for 60 min with shaking at 30 °C and 250 rpm, the reaction was stopped by soaking the vials in ice water. Next, the headspace gas (100 μL) was injected into a GCMS-QP2010 (Shimazu, Kyoto, Japan) connected to an HP-PLOT Q column (30 m × 0.32 mm diameter with 20 μm film), and the amount of ethylene was quantified based on ethylene of known concentration. The injector, detector, and oven temperatures were 50 °C, 200 °C, and 35 °C, respectively. The amount of produced ethylene was normalized by OD 600 value of each sample.

Measurement of acetylene reduction by nitrogenase-producing E. coli.
Nitrogenase activity of recombinant E. coli was performed as previously described 19 . Briefly, each strain was cultured in LB medium at 37 °C and 200 rpm overnight, and the cells were collected by centrifugation at 3000×g for 5 min at 25 °C. The cell pellet was inoculated in 5 mL of KPM medium containing 0.1 mM IPTG and antibiotics with an initial OD 600 of 0.5, in a 21 mL headspace vial. The headspace was replaced with Ar multiple times and sealed with a butyl rubber septum. Then, 1 mL of a gas mixture of 1:200 ethane gas and acetylene gas obtained from CaC 2 was added to the vial using a syringe. After incubation at 30 °C for 16 h, the reaction was stopped by injecting 300 µL of 4 M NaOH into each vial. The produced ethylene (m/z: 26) and the internal standard ethane (m/z: 30) were detected using GCMS QP2020Ultra (Shimadzu) connected to a TurboMatrix HS110 headspace autosampler (PerkinElmer, MA, USA) and GS GasPro column (0.32 mm, 30 m) (Agilent Technologies), and the concentration was determined by the external standard method. The injector, detector, and oven temperatures were 40 °C, 250 °C, and 40 °C, respectively. The average of the results in samples without E. coli inoculation was used as the negative control and was subtracted from the average values in samples with E. coli.

Data availability
All data and materials that support the findings of this study are available from the corresponding author upon reasonable request. The raw data of the transcriptome analysis were deposited in ArrayExpress: E-MTAB-10710 for MiSeq data and E-MTAB-10712 for HiSeq X. The raw data of the proteome analysis were deposited in the jPOST database: JPST001241 and PXD027133.