Bacterial strains used in this work are compiled in Table 1. Escherichia coli cells were cultivated in Luria Bertani medium (Miller, 1972) at 37°C. When needed, antibiotics were used at the following concentrations (in μg/ml): ampicillin, 200; kanamycin, 30; spectinomycin, 25; streptomycin, 25; tetracycline, 10.

Bradyrhizobium diazoefficiens cells were cultured oxically and microoxically basically as described earlier (Bueno et al., 2017). While Peptone-Salts-Yeast extract (PSY) medium (Regensburger and Hennecke, 1983; Mesa et al., 2008) was employed in routine oxic cultures, Yeast Extract-Mannitol (YEM) medium (Daniel and Appleby, 1972) was used as standard medium in our experiments. After growth under oxic conditions in PSY medium, cells were collected by centrifugation (8.000 g for 10 min at 4°C), and washed twice with YEM medium. Next, washed cells were used to inoculate, at a 600 nm optical density (OD₆₀₀) of 0.2, 17 ml or 500 ml rubber stoppered tubes or Erlenmeyer flasks containing 3 ml or 150 ml of YEM medium amended or not with 10 mM KNO₃, respectively. Next, cells were incubated for 24 h under low oxygen conditions, either at initial 0.5% O₂ or at 2% O₂ (in this case the headspace was exchanged every 8–16 h). The latter conditions were chosen to study the specific control of the FixK₂ and NnrR regulatory proteins. To analyze the effect of the different NOx, microoxically incubated cells were subsequently exposed for 5 h to 10 mM KNO₃, 500 μM NaNO₂, 50 μM NO (from a saturated NO solution [1.91 mM at 20°C]), and 0.15% (30 mM) N₂O. 10 μM or 100 μM of the NO-scavenger cPTIO [2-(4-Carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide; carboxy-PTIO potassium salt; Sigma] was added from the beginning to the WT and ΔnnrR strain cultures grown microoxically (2% O₂) in the presence of 10 mM KNO₃ for 24 h, in order to analyze the effect of removing the excess of NO on the expression of nosR-lacZ or N₂OR activity, respectively. Antibiotics were added to the B. diazoefficiens cultures at the following concentrations (μg/ml); chloramphenicol, 20; streptomycin, 200; kanamycin, 200; tetracycline, 100 (solid cultures), 25 (liquid cultures); spectinomycin, 200.

Plasmids used in this study are listed in Table 1. Primer sequences in this work are compiled in Supplementary Table S1. For construction of transcriptional reporter fusion plasmids, 5′ DNA fragments for the nosR (558; 132; 128 and 75 bp), nosZ (1024 bp) and nosD (875 pb) promoter regions were amplified using primers’ pair a1/PnosR.r, PnosRfull.f/PnosR.r, PnosRhalf.f/PnosR.r, PnosRno.f/PnosR.r, PnosZ.f/PnosZ.r and c1/c2, respectively (Supplementary Table S1). The PCR products were then individually ligated into the pGEM^®-T vector (Promega), digested with EcoRI or EcoRI-PstI and cloned into the lacZ fusion suicide vector pSUP3535 (Mesa et al., 2003), to yield plasmids pBG0301, pBG0304, pBG0305, pBG0306, pBG0302, and pBG0303, respectively (see Table 1 for details). The correct orientation of the inserts was verified by sequencing. Plasmids pBG0301, pBG0302, pBG0303, pBG0304, pBG0305, and pBG0306 were integrated by homologous recombination into the chromosome of WT B. diazoefficiens 110spc4, yielding strains 110spc4-BG0301, 110spc4-BG0302, 110spc4-BG0303, 110spc4-BG0304, 110spc4-BG0305, 110spc4-BG0306. Plasmid pBG0301 was also integrated into the chromosome of napA (GRAP1), nirK (GRK308), fixK₂ (9043), and nnrR (8678) mutants, yielding strains GRPA1-BG0301, GRK308-BG0301, 9043-BG0301, and 8678-BG0301, respectively (Table 1). Correct recombination into the chromosome of the corresponding recipient strain was checked by PCR analyses.

The plasmid used as transcription template was based on the plasmid pRJ9519 which contains a B. diazoefficiens rrn transcriptional terminator (Beck et al., 1997). The nosRZDFYLX promoter was PCR-amplified with nosR_For_Transc and nosR_Rev_Transc primers, subsequently restricted with XbaI and EcoRI, and finally cloned as a 486-bp fragment into pRJ9519, yielding plasmid pDB4020. The correct nucleotide sequence was confirmed by sequencing.

End-point reverse transcription-polymerase chain reaction (RT-PCR) was performed to investigate the transcriptional architecture of nosRZDFYLX genes. First, B. diazoefficiens cells were grown under 0.5% initial O₂ concentration to an OD₆₀₀ of ∼0.4 in YEM medium supplemented with 10 mM KNO₃. Cell harvest and isolation of total RNA were done as described previously (Hauser et al., 2007; Lindemann et al., 2007; Mesa et al., 2008). First strand cDNA synthesis was performed with the SuperScript II reverse transcriptase (Invitrogen) according to the supplier’s guidelines, using 1 μg of total RNA and primers c2 and g2 that hybridize in the complementary sequence of nosD and nosX genes. The obtained cDNA was next used for amplification of putative intergenic regions between nosR and nosX (blr0314-blr0320) using primers’ pairs labeled as b1/b2-to-g1/g2 and flanking regions using primers’ pair labeled as a1/a2 and h1/h2 (Supplementary Table S1), essentially as described by Sambrook and Russell (2001). In negative controls, reverse transcriptase was omitted in the reaction. Positive control PCR reactions were performed with B. diazoefficiens genomic DNA as template.

The transcription start sites of nos genes were determined with the RACE (Rapid Amplification of cDNA Ends) method as described by Sambrook and Russell (2001). Cell cultivation and harvest as well as total RNA isolation were carried out as described above for the RT-PCR experiments. First strand cDNA synthesis was performed with the SuperScript II reverse transcriptase (Invitrogen) according to the supplier’s guidelines, using 0.8 μg of total RNA and primer SP1_nosR. After the reaction, dNTPs and primers were removed with the GeneJET PCR Purification Kit (Thermo Fisher Scientific) and products were eluted in 15 μl of 10 mM Tris-HCl, pH 8.5. Poli-A tails were added to 5′ end of cDNAs with the terminal deoxynucleotidyl transferase (Thermo Fisher Scientific) and final products were diluted with purified water to final volume of 1 ml. Amplification reactions were carried out with primers (dT)₁₇-adaptor-primer, adaptor-primer and SP2_ nosR primers using the following PCR program: 95°C for 5 min; (95°C for 30 s; 48°C for 30 s; 72°C for 45 s) × 5 cycles; (95°C for 30 s; 55°C for 30 s; 72°C for 45 s) × 30 cycles; 72°C for 10 min and hold at 4°C. DNA libraries were constructed by cloning the PCR products into pGEM-T easy vector (Promega). Plasmid DNA of individual clones was purified with QIAprep Spin Miniprep Kit (Qiagen) and Sanger sequenced using SP6 as primer. Transcription start sites were identified as the first nucleotide sequenced after the poly-A sequence.

Expression of nosR was also analyzed by qRT-PCR using an iQTM5 Optical System (Bio-Rad, Foster City, CA, United States). B. diazoefficiens WT and napA, nirK, fixK₂, and nnrR mutant strains were grown in YEM medium amended with 10 mM NO₃^- under initial 0.5% O₂ (WT, napA and nirK mutant strains) or 2% O₂ (WT, fixK₂ and nnrR mutant strains) for 24 h. Cell harvest, isolation of total RNA and cDNA synthesis were done as described previously (Hauser et al., 2007; Lindemann et al., 2007; Mesa et al., 2008). Primers for the PCR reactions (nosR_qRT_PCR_F/ nosR_qRT_PCR_R; Supplementary Table S1) were designed with the Clone Manager Suite 9 software to have melting temperatures between 57 and 62°C and generate PCR products of 50–100 bp. Each PCR reaction contained 9.5 μl of iQTM SYBR Green Supermix (Bio-Rad), 2 μM (final concentration) of individual primers and appropriate dilutions of different cDNA samples in a total volume of 19 μl. Reactions were run in triplicate. Melting curves were generated to verify the specificity of the amplification. Relative changes in gene expression were calculated as described by Pfaffl (2001). Expression of the 16S rrn gene was used as reference for normalization (primers 16S_qRT_For and 16S_qRT_Rev; Supplementary Table S1).

β-galactosidase activity was determined by using permeabilised cells from at least three independently grown cultures assayed in triplicate essentially as previously described (Cabrera et al., 2016). Specific activities were calculated in Miller units (Miller, 1972).

B. diazoefficiens cells were incubated microoxically (2% O₂) for 24 h in YEM medium supplemented or not with 10 mM NO₃^-. In the latter conditions, parallel replicates were also exposed to 100 μM of the NO-scavenger cPTIO. Next, cells were washed three times with YEM medium and 30 μl gaseous aliquots of 2% N₂O in 98% N₂ (0.15% N₂O final concentration in the headspace) were injected into the rubber stoppered Erlenmeyer flasks. After 5 h of incubation at 30°C at 185 rpm, gas-liquid phase equilibration was reached and 500-μl gaseous aliquots were taken from the headspace to analyze N₂O consumption by gas chromatography as described previously (Tortosa et al., 2015).

The protein concentration was estimated using the Bradford method (Bio-Rad Laboratories) with a standard curve constructed with varying bovine serum albumin (BSA) concentrations. N₂OR activity was determined by using cells from at least three independently biological grown cultures.

B. diazoefficiens cells incubated micooxically (2% O₂) in YEM medium in the presence or absence of 10 mM NO₃^- for 24 h, were harvested and the soluble fraction of the cells was obtained by following the protocol previously described by Delgado et al. (2003). The resulting membrane pellet was discarded and the supernatant, containing the soluble fraction, was concentrated to about 100 μl by using AmiconR Ultra-2 centrifugal filter devices (Millipore) and stored at -20°C until their use. Protein concentration was estimated as described above.

For immunodetection of NosZ, protein samples (10 μg of the soluble fraction) were separated by 12% SDS-polyacrylamide gel electrophoresis (PAGE) as described by Laemmli (1970). Then, proteins were transferred to nylon or PVDF membranes (Millipore). The membrane was then incubated in blocking buffer [5% non-fat dry milk in TTBS buffer containing 50 mM Tris-HCl pH 7.5, 0.15 mM NaCl and 0.1% Tween 20], with overnight shaking at 4°C. Afterward, the membrane was then washed with TTBS buffer (four times for 10 min each), before being incubated in 10 ml of blocking buffer containing 1/1000 (v/v) antibody dilution (anti-NosZ of Pa. denitrificans; Felgate et al., 2012). The membrane was subsequently incubated by shaking gently for 1 h at room temperature (RT). Further, the membrane was then washed with TTBS and incubated for 1 h at RT with a 1/3500 (v/v) dilution of the secondary antibody (sheep anti-IgG: peroxidase antibody produced in donkeys; A3415 Sigma–Aldrich) in blocking buffer. Next the membrane was washed four times with TTBS before adding 500 μl of ECL Select western-blotting detection reagent (GE Healthcare, Amersham) followed by Chemiluminescent signal detection in a Chemidoc XRS (Universal Hood II, Bio-Rad). The Quantity One software (Bio-Rad) was used for image analyses.

Purification of the B. diazoefficiens holoenzyme was carried by using a modified protocol similar to the one described by Beck et al. (1997). 25 g (wet weight) of B. diazoefficiens 110spc4 cells grown oxically in PSY supplemented with 0.1% arabinose until late exponential phase were used for each purification batch. All purification steps were performed at 4°C. Cells were resuspended in 70 ml of TGED buffer (10 mM Tris-HCl [pH 8.0], 10% glycerol, 1 mM EDTA, 0.1 mM dithiothreitol [DTT]) containing 0.02 M NaCl and 1 mM ABSF and disrupted in a French pressure cell (three passes at 1000 psi). The crude extract was treated with polyethyleneimine to a final concentration of 0.3%. The pellet obtained after centrifugation (15 min; 27,000 × g) was washed with TGED buffer (0.2 M NaCl), and protein containing RNAP was washed three times in TGED buffer (0.8 M NaCl). In all recovery steps, the supernatant was collected and precipitated again by adding solid (NH₄)₂SO₄ to 65% final saturation (43 g per 100 ml). The precipitate was collect by centrifugation (30 min; 27,000 × g), dissolved in 30 ml of TGED buffer (0.02 M NaCl) and, dialyzed against 1 liter of TGED buffer (0.02 M NaCl). The dialyzed sample was loaded onto an HiTrap Q FF column (GE Healthcare), from which it was eluted by a linear 0.02–1.2 M NaCl gradient. Fractions containing RNAP (as judged by standard transcription assays) were pooled and loaded onto a heparin agarose column (HiTrap Heparin HP; GE Healthcare). Equilibration and elution buffers were similar to those used in the HiTrap Q FF chromatography. Peak fractions contained the RNAP (indicated by general IVT assays performed according Beck et al., 1997) were pooled, concentrated by ultrafiltration (YM30 membrane, Amicon), and dialyzed and stored in TGED buffer (0.02 M NaCl) containing 50% glycerol at -20° or -80°. The purity of the active fractions was tested by SDS-PAGE. Protein concentrations were determined with Bio-Rad assay solution, with BSA as the standard.

Multiple-round in vitro transcription (IVT) assays were carried out as described previously (Beck et al., 1997; Mesa et al., 2008). Plasmid pDB4020 was used as template to study the capacity of the FixK₂ protein to initiate transcription from the nosRZDFYLX promoter. Expression and purification of an oxidation-insensitive C-terminal Histidine-tagged C183S FixK₂ protein variant (C183S-FixK₂-His₆; Bonnet et al., 2013) were carried out as described in (Mesa et al., 2005). Purified FixK₂ protein was used at concentrations of 1.25 or 2.5 μM dimer.

Runoff transcripts of 286 and 180 nucleotides produced in vitro following the procedure used by Mesa et al. (2005) were used as RNA size markers. Transcripts were visualized with a PhosphorImager and signal intensities were determined with the Bio-Rad Quantity One software (Bio-Rad).

Analysis of the nosRZDFYLX sequence did not reveal any predicted transcriptional termination signals¹ which is an indication that they might be transcribed as an operon. Overlapping coding regions between nosR and nosZ, as well as between nosD, F, Y, and L stop and start codons, suggest translational couplings between nosRZ and nosDFYL. However, unlike these translational couplings, there is a short intergenic region of 14 nucleotides between nosZ and nosD and 11 nucleotides between nosL and nosX.

In order to investigate the transcriptional architecture of nosRZDFYLX genes, end-point RT-PCR was performed to detect intergenic regions between each pair of correlative genes. To ensure that the amplified RT-PCR product was from the template mRNA, each RT-PCR reaction had a negative control (without reverse transcriptase) and a positive control (genomic DNA). First, total RNA was isolated from B. diazoefficiens WT cells cultured with initial 0.5% O₂ concentration in the presence of NO₃^- and subsequently reverse transcribed to cDNA. As shown in Figure 1A, specific cDNA products were obtained for intergenic regions designed as b-to-g, but not from those labeled as “a” and “h” corresponding to flanking regions of the nosRZDFYLX genes. These findings reveal that B. diazoefficiens nosRZDFYLX genes constitute a transcriptional unit, although we cannot discard the presence of additional internal promoters.

To test any potential transcription from the DNA regions upstream of the nosR, nosZ, and nosD genes, we determined β-Galactosidase activity of chromosomally integrated transcriptional fusions between the DNA regions preceding the annotated nosR, nosZ, nosD genes and the reporter gene lacZ (Figure 1B). After growing B. diazoefficiens cells under an initial O₂ concentration of 0.5% O₂ in the presence of NO₃^-, the highest transcriptional expression was driven from the nosR-lacZ fusion compared to the nosZ-lacZ and nosD-lacZ fusions (Figure 1B). These results strongly suggest that transcription of nosRZDFYLX mainly depends on a promoter present in the DNA region upstream of nosR. However, although β-galactosidase activity from the nosZ-lacZ fusion was sixfold lower to that observed from the nosR-lacZ fusion, we cannot exclude the possibility that another internal promoter upstream of nosZ might exist.

In order to map transcription initiation within the nosR promoter region, we identified their Transcriptional Start Sites (TSS) by using 5′-RACE. As shown in Figure 2A, we identify two TSS (TSS₁ and TSS₂) that initiate at a G and T, 84 and 57 bp upstream of the putative translational start codon, respectively. Analysis of the 5′ region of nosR revealed the presence of a purine-rich Shine-Dalgarno-like sequence (GAGG) four bases in front of the nosR putative translational start codon. Exhaustive inspection of the nosR promoter region failed to identify any putative conserved -35/-10- or -24/-12-type elements associated to σ⁷⁰-dependent or σ⁵⁴-dependent promoters. However, we noticed the presence of an imperfect palindromic sequence (TTGATCCAGCGCAA) positioned at 40.5 and 67.5 bp from TSS₁ and TSS₂, respectively (Figure 2A). This sequence resembles reasonably well the consensus sequence of the binding site for FixK-type proteins, 5′-TTGA-N₆-TCAA-3′ (Fischer, 1994; Dufour et al., 2010) and specifically the consensus FixK₂ binding site [TTG(A/C)-N₆-(T/G)CAA] recently reported by Bonnet et al. (2013) based on the solved FixK₂-DNA complex structure.

In order to examine the importance of the FixK₂-like box identified within the nosR promoter region in its transcription, we studied the transcriptional expression derived from a battery of nosR-lacZ fusions harboring the full or half FixK₂-like box, or a deletion of this box (plasmids pBG0304, pBG0305, and pBG0306, respectively) (Figure 2B and Table 1). These plasmids were integrated into B. diazoefficiens WT and β-galactosidase activity was measured in cells cultured under initial 0.5% O₂ with NO₃^-. In contrast to the significant induction of the nosR-lacZ transcriptional fusion containing the full FixK₂-like site, expression of nosR-lacZ constructs carrying half or deleted FixK₂-like site was basal, which showed the importance of the presence of this FixK₂-like binding site in the induction of nosR (Figure 2B).

To address the effect of low oxygen and NOx in the expression of the nosRZDFYLX operon, we analyzed β-galactosidase activity of the nosR-lacZ transcriptional fusion in WT cells cultured oxically or under initial 0.5% O₂, both for 24 h, and later exposed to different NOx (NO₃^-, NO₂^-, NO, or N₂O) for additional 5 h-period. As shown in Figure 3A, β-galactosidase activity values were basal in cells incubated under oxic conditions. Similar basal levels were observed under oxic conditions in the presence of NO₃^- (data not shown). However, when cells were cultured under 0.5% O₂, expression of the nosR-lacZ fusion significantly increased (about fourfold) as compared to oxic conditions (Figure 3A). The presence of NO₃^-, but not of NO₂^-, NO, or N₂O, slightly increased nosR-lacZ expression (about 1.5-fold) compared to that observed in cells incubated microoxically in the absence of any NOx (Figure 3A).

Next, we were interested to confirm that the lack of NO₃^- reduction products does not affect nosR-lacZ expression. Therefore, β-galactosidase activity from the nosR-lacZ fusion was individually analyzed in napA or nirK mutant strains which are unable to reduce NO₃^- or NO₂^-, respectively (Velasco et al., 2001; Delgado et al., 2003). Again, a slight induction of the nosR-lacZ fusion in the WT cells cultured under 0.5% O₂ in the presence of NO₃^- was observed (Figure 3B), however, no change was detected in the napA mutant cultured under the same conditions, suggesting a requirement of NO₃^- reduction on nosR-lacZ expression. By contrary, induction by NO₃^- of the nosR-lacZ fusion was retained in the nirK mutant indicating that NO₂^- reduction products (NO or N₂O) are not required for activating the expression of nosRZDFYLX genes. These results were validated by qRT-PCR analyses (Figure 3C). Similarly as we observed by using the nosR-lacZ fusion, nosR expression was reduced in the napA mutant (3.18-fold) compared to WT cells, while it was not significantly affected in the nirK mutant (1.69-fold), all cultured in the presence of NO₃^-. However, we cannot conclude that the lack of NO₃^--mediated induction of nos genes observed in the napA mutant (Figures 3B,C) is due to the absence of NO₂^-, since the addition of NO₂^- to the medium did not increase nosR-lacZ expression (Figure 3A). Taken together, results from Figures 3A–C suggest that microoxia is the main signal that induces expression of B. diazoefficiens nosRZDFYLX genes.

In B. diazoefficiens, sensing and transduction of the decrease in O₂ concentration are mediated by two interlinked O₂-responsive regulatory cascades, the FixLJ-FixK₂-NnrR and the RegSR-NifA (Sciotti et al., 2003). A mild decrease in the O₂ concentration in the gas phase (≤5%) is sufficient to activate expression of FixLJ-FixK₂-dependent targets, however, a 10-fold lower O₂ concentration (≤0.5%) is necessary for NifA-mediated activation. In order to investigate how FixK₂ and NnrR control the microoxic expression of nosRZDFYLX genes, we analyzed β-Galactosidase activity from the nosR-lacZ fusion in the WT and ΔfixK₂ and ΔnnrR strains, incubated for 24 h oxically, and microoxically (2% O₂) in the absence or the presence of 10 mM of KNO₃. In these experiments, 2% O₂ concentration was chosen as a middle concentration between 5% (needed for FixLJ-FixK₂ cascade activation) and 0.5% (required for the activation of the low O₂-responsive NifA protein), in order to circumvent any possible influence by NifA regulation in our assays.

As observed in Figure 4A, microoxic induction of nosR-lacZ was completely abolished in the absence of a functional fixK₂ gene, however, it was retained in the ΔnnrR strain, suggesting that microoxic expression of nosRZDFYLX genes depends on FixK₂ but not on NnrR. When cells were cultured microoxically in the presence of NO₃^-, expression of the nosR-lacZ fusion was significantly reduced in the fixK₂ mutant (about threefold) compared to that observed in the WT cells (Figure 4A). However, β-galactosidase activity of the nosR-lacZ fusion was slightly reduced in the nnrR mutant (about 1.75-fold) compared to the WT (Figure 4A). This slight reduction of the expression of the nosR gene in the nnrR mutant is probably due to the toxic effect of NO that is accumulated in nnrR cells as previously reported by Bueno et al. (2017). To check this hypothesis, a NO scavenger (cPTIO) was added during growth of WT and ΔnnrR cells under microoxic conditions with NO₃^-. As shown in Figure 4A, while no effect of cPTIO was observed in WT cells, nosR-lacZ expression in ΔnnrR cells increased about 40% to that observed in the absence of cPTIO (right panel), which almost corresponds to the expression pattern of the WT. Thus, this indicates that nos expression could be partially recovered in the ΔnnrR mutant when NO was sequestered by cPTIO.

The different control of nosR expression by FixK₂ or NnrR was also confirmed by qRT-PCR analyses. When cells were cultured microoxically in the absence of NO₃^-, expression of nosR was reduced in the fixK₂ mutant (3.92-fold) compared to that observed in the WT cells (Figure 4B), however, it was almost not affected in the nnrR mutant (Figure 4B). When NO₃^- was added to medium, a significant reduction of nosR expression (10.38-fold) was observed in the fixK₂ mutant but only a slight decrease (2.4-fold) was detected in the nnrR mutant, both compared to the WT cultured in the same conditions (Figure 4B). Taken together, these results suggest FixK₂ as the transcriptional activator of nos genes in response to microoxic conditions.

The differential dependency of nosRZDFYLX expression on FixK₂ and NnrR was also confirmed at protein level by immunoblot analyses using antibodies raised against purified Pa. denitrificans NosZ (Felgate et al., 2012). Firstly, we were able to identify NosZ protein in the soluble fraction of B. diazoefficiens cells cultured under microoxic conditions (2% O₂) with NO₃^-, since a prominent band of about 70 kDa found in the WT was readily undetectable in the nosZ mutant (Figure 4C, lanes 1 and 2). The size of this band corresponds to the predicted molecular mass of B. diazoefficiens NosZ subunit (71.6 kDa; ProtParam tool²). NosZ was already detected in the WT cells cultured microoxically (Figure 4C, lane 3) but the presence of NO₃^- slightly increased NosZ steady-state levels (Figure 4C, lane 6). This is in line with the observed NO₃^--mediated induction of the nosR-lacZ fusion (Figures 3A,B, 4A). Similarly as the expression pattern observed for the nosR-lacZ fusion, NosZ was present in the soluble fraction of ΔnnrR cells cultured microoxically either in the absence or in the presence of nitrate (Figure 4C, lanes 5 and 8), although at a slightly lower concentration than in the WT cells. As expected, the band of about 70 kDa corresponding to NosZ was absent in the soluble fractions of the ΔfixK₂, independently of the presence or absence of NO₃^- in the incubation medium (Figure 4C, lanes 4 and 7).

Finally, we determined N₂O reductase (N₂OR) activity in B. diazoefficiens WT and fixK₂ and nnrR mutant strains as the capacity to reduce a defined initial N₂O concentration. As shown in Figure 4D, values of N₂OR activity in WT cells correlated with NosZ steady-state levels in B. diazoefficiens cells (Figure 4C), where a slight induction (about 1.6-fold) of activity was observed in the WT cells in the presence of NO₃^- (Figure 4D) compared to that observed in exclusively microoxic conditions. In line with the expression pattern of the nosR-lacZ fusion (Figure 4A), nosR expression (Figure 4B) and NosZ detection (Figure 4C, lanes 4 and 7), N₂OR activity was severely impaired in the ΔfixK₂ strain cultivated microoxically independently of the presence of NO₃^- (Figure 4D). Under microoxic conditions, cells of the ΔnnrR strain showed a milder decrease of N₂OR activity (about 1.75-fold) compared to that observed in WT cells (Figure 4D), which was significantly diminished further (about 10-fold) in the presence of NO₃^- (Figure 4D). As we have mentioned above, this strong decrease is probably due to the higher NO accumulation capacity of ΔnnrR cells grown microoxically with nitrate compared to WT cells grown under the same conditions (Bueno et al., 2017). In fact, when cPTIO was added during growth, ΔnnrR cells restored its ability to reduce N₂O reaching WT N₂OR activity values (Figure 4D). These data discard the involvement of NnrR as direct regulator of nos expression and suggest that the incapacity of ΔnnrR to reduce N₂O under microoxic conditions with NO₃^- is probably due to the accumulation of NO. Taken together, these results pointed out that FixK₂ is the key transcriptional regulator involved in nosRZDFYLX expression.

In order to investigate whether FixK₂ could have a direct role on nosRZDFYLX activation, we monitored RNA synthesis by multiple-round IVT. The nosR promoter region was cloned into the template plasmid pRJ9519 (Beck et al., 1997), which carries an rrn terminator, yielding plasmid pDB4020. In these experiments, purified C183S-FixK₂-His₆ (Bonnet et al., 2013), hereafter referred as FixK₂, and RNA polymerase (RNAP) holoenzyme from B. diazoefficiens that was purified in this work (see Material and Methods) were used. In the absence of FixK₂, B. diazoefficiens RNAP was unable to transcribe the nosR promoter efficiently (Figure 5, lane 3), whereas it produced a vector-encoded transcript that served as an internal reference. In the presence of FixK₂ (1.25 and 2.5 μM dimer), B. diazoefficiens RNAP transcribed the nosRZDFYLX promoter producing a single specific transcript larger than 286 nucleotides (Figure 5, lanes 4 and 5, respectively), which probably initiate at TSS₁. This suggested that the nosR promoter is directly activated by FixK₂ and that transcription from TSS₁ depends on FixK₂, at least, in in vitro conditions.