The Molecular Mechanism of Nitrate Chemotaxis via Direct Ligand Binding to the PilJ Domain of McpN

Nitrate is of central importance in bacterial physiology. Previous studies indicated that movements toward nitrate are due to energy taxis, which is based on the cytosolic sensing of consequences of nitrate metabolism. Here we present the first report on nitrate chemotaxis. This process is initiated by specific nitrate binding to the periplasmic ligand binding domain (LBD) of McpN. Nitrate chemotaxis is highly regulated and occurred only under nitrate starvation conditions, which is helpful information to explore nitrate chemotaxis in other bacteria. We present the three-dimensional structure of the McpN-LBD in complex with nitrate, which is the first structure of a chemoreceptor PilJ-type domain. This structure reveals striking similarities to that of the abundant 4-helix bundle domain but employs a different sensing mechanism. Since McpN homologues show a wide phylogenetic distribution, nitrate chemotaxis is likely a widespread phenomenon with importance for the life cycle of ecologically diverse bacteria.

the Pfam database revealed that PilJ domains are also employed by other bacterial sensor proteins such as sensor kinases, diguanylate cyclases, and transcriptional regulators. In this study, we aimed at identifying the function of the PilJ domain containing chemoreceptors in P. aeruginosa.

RESULTS
Nitrate is a specific ligand for PA2788-LBD. To identify ligands that bind to PA2788, we cloned the DNA sequence encoding the LBD of PA2788 into an expression vector. Protein was expressed in E. coli and purified by affinity chromatography. High-throughput ligand screening assays were then conducted using the thermal shift method (33). In this method, changes in the melting temperature (T m ) of a protein, representing the midpoint of thermal protein unfolding, are recorded. The binding of ligands typically increases the T m , and shifts of greater than 2°C are considered significant. We screened 480 ligands from Biolog compound arrays PM1, PM2A, PM3B, PM4A, and PM5 that served as bacterial carbon, nitrogen, phosphorus, and sulfur sources.
In the absence of ligand, the T m of PA2788-LBD was 43.5°C. Fig. 1A shows the Nitrate Chemotaxis in Bacteria ® changes in T m caused by each of the 95 compounds of array PM3B comprising nitrogen sources. In the presence of NaNO 3 , the T m was increased by 3.5°C, whereas NaNO 2 caused only a minor increase of 0.5°C. No significant T m shifts were obtained for the compounds in arrays PM1, PM2A, PM4A, and PM5. To verify ligand binding, isothermal titration calorimetry (ITC) experiments were conducted. In an initial control experiment, the heat changes derived from the injection of 2 mM NaNO 3 into buffer were recorded (Fig. 1B), with results showing that the levels of dilution heat were low and uniform. Titration of PA2788-LBD with the same ligand caused endothermic heat changes (ΔH ϭ 0.38 Ϯ 0.1 kcal/mol) indicative of an entropydriven binding process (TΔS ϭ 6.4 kcal/mol) characterized by a dissociation constant of 47 Ϯ 8 M. The same protein was also titrated with NaNO 2 , ammonia and uric acid (Fig. 1B), but an absence of binding was noted in all cases, confirming the thermal shift assay results. We were intrigued by the failure of PA2788-LBD to sense nitrite since other sensor proteins were found to sense nitrate as well as nitrite (34)(35)(36). ITC experiments performed with ligands provide information only on higher-affinity binding events. To confirm that PA2788-LBD does not bind nitrite with low affinity, we performed titrations of PA2788-LBD with nitrate in the absence and presence of 20 mM nitrite. In cases of nitrite binding, this would alter nitrate recognition. However, this was not the case (see Fig. S1A in the supplemental material), confirming that PA2788-LBD is a nitrate-specific receptor. Following the demonstration that PA2788 specifically binds nitrate, this chemoreceptor was named McpN. Nitrate is not recognized by the LBDs of the PilJ and PA4520 chemoreceptors. Inspection of the chemoreceptor repertoire of P. aeruginosa (9) suggests that two other chemoreceptors may also bind nitrate. First, the LBD of the PilJ (PA0411) receptor is composed of two consecutive PilJ domains (9). Second, chemoreceptor PA4520 was predicted to contain a NIT LBD (Pfam PF08376). This domain, representing approximately 3% of all extracellular prokaryotic LBDs (21), was predicted to recognize nitrate and nitrite (37).
To verify whether these receptors also bind nitrate, we generated the purified individual LBDs of both receptors for thermal shift assay ligand screening using the PM3B array. In the absence of ligand, PilJ-LBD and PA4520-LBD unfolded with T m values of 41.8 and 34.4°C, respectively. In both cases, however, no significant ligand-induced T m shifts were noted ( Fig. S2A and B) and microcalorimetric titrations confirmed the absence of NaNO 3 and NaNO 2 binding (Fig. S2C). To exclude the possibility that endothermic and exothermic contributions to binding canceled out each other at a given analysis temperature, the experiments were repeated at 15°C; however, the same result was produced, confirming the absence of nitrate/nitrite binding to PilJ-LBD and PA4520-LBD.
McpN mediates nitrate chemotaxis under nitrate starvation conditions. To assess the function of McpN, we conducted quantitative capillary chemotaxis assays for NaNO 3 using the wild-type (wt) strain as well as a mutant deficient in the mcpN gene. These assays were conducted using the standard conditions that we routinely employ to study P. aeruginosa chemotaxis. This assay involves cell culture in MS minimal medium ( O) supplemented with glucose (note that this medium also contains 25 mM NH 4 NO 3 as a nitrogen source). However, only very minor, nonsignificant responses were detected over the entire nitrate concentration range (Fig. S3).
Previous studies have shown that chemotaxis to P i in PAO1 was not observed in rich medium containing significant amounts of P i but was induced by P i starvation (23). We hypothesized that this might also be the case for nitrate chemotaxis and followed an approach similar to that described previously by Wu et al. (23). Thus, cells were precultured in rich 2ϫ YT medium (10 g yeast extract liter Ϫ1 , 16 g Bacto tryptone liter Ϫ1 , 10 g NaCl liter Ϫ1 ) and then diluted 133-fold into N0 medium (which lacks nitrogen sources) and continued to grow for another 3 h. Under these conditions, strong chemotactic responses to NaNO 3 were obtained. Initial significant responses were obtained at a NaNO 3 concentration of 5 M, whereas maximal responses were observed at 500 M (Fig. 2). No nitrate chemotaxis was observed for the mcpN mutant, suggesting that it is the sole nitrate chemotaxis receptor. The complementation of this mutant with a plasmid harboring the mcpN gene restored nitrate chemotaxis (Fig. 2).
To identify the possible roles of the PilJ receptor and the NIT domain containing PA4520 chemoreceptor in nitrate chemotaxis, single mutants with mutations of the corresponding genes were also analyzed. As shown in Fig. 2A, their responses to nitrate were similar to those seen with the wt strain, confirming that the observed nitrate chemotaxis was mediated solely by McpN.
Nitrate reduces mcpN transcript levels. To explain the absence of taxis under conditions of nitrate abundance, we hypothesized that nitrate might repress expression of the mcpN gene. To verify this hypothesis, we quantified mcpN transcript levels by reverse transcription-quantitative PCR (RT-qPCR). These assays were carried out using RNA from cells grown using the same protocol used for the chemotaxis assays under conditions of nitrate abundance and limitation. As shown in Fig. 2B, mcpN transcript levels were approximately 16 times higher under nitrate-limiting conditions than under nitrate-abundant conditions. To verify that the absence or presence of nitrate was the cause for these differences, NaNO 3 was added to these cultures to reach a final concentration of 1 mM and samples were taken for RT-qPCR experiments after additional periods of growth of 20 and 40 min. The results showed that the addition of nitrate to cells grown under nitrate-limiting conditions reduced mcpN transcript levels to those seen under nitrate-abundant conditions, indicating that nitrate reduces mcpN expression (Fig. 2B).
The NarX/NarL two-component system (TCS) senses nitrate and regulates genes involved in nitrate metabolism (38). To identify a potential role of this TCS in mcpN expression, we quantified mcpN transcript levels in a mutant defective in the gene encoding the NarX sensor kinase. However, RT-qPCR data revealed no statistical differences in the transcript levels of mcpN (Fig. 2B).
McpN signals through the Che pathway. The 26 PAO1 chemoreceptors signal through four different chemosensory pathways, and McpN was predicted to signal through the Che pathway (11). To verify this prediction, we conducted chemotaxis assays for NaNO 3 using mutants defective in the genes encoding the CheA paralogues of the Che1 (CheA 1 ) and Che2 (CheA 2 ) pathways. As shown in Fig. S4, no nitrate chemotaxis was observed in the cheA1 mutant, whereas the responses of the cheA2 mutant were comparable to wt levels. These results thus confirm that McpN signals through the Che pathway (11).
The three-dimensional (3D) structure of McpN-LBD. McpN-LBD in complex with nitrate was crystallized in a buffer at pH 7.5, and its structure was resolved by X-ray crystallography to a resolution of 1.3 Å. According to the Matthews coefficient, the unit cell accommodates three chains. A structural alignment of these three chains resulted in root mean square deviation (RMSD) values below 0.5, indicating that these chains can be considered identical. Chains A and B of the unit cell form a dimer (Fig. 3A), whereas chain C forms another dimer with a symmetry-related chain. The McpN-LBD monomer is composed of 4 ␣-helices that pack into a 4-helix bundle. Dimerization is achieved through the interaction of 22 residues of chains A and B that establish 16 hydrogen bonds and occlude a surface of approximately 1,100 Å 2 in each monomer.
A single molecule of nitrate is bound to a site with a positive surface charge at the dimer interface. The binding site is situated on the dimer symmetry axis; consequently, the same amino acids from the two monomers establish interactions with bound nitrate. Whereas G58, A59, and M62 establish nonbonded contacts, R61 played a key role in recognition since it forms two hydrogen bonds with nitrate ( Fig. 3B). To verify the role of R61, we generated a McpN-LBD R61A mutant. The intrinsic tryptophan fluorescence emission spectrum (Fig. S5A) and the thermal unfolding properties of the mutant protein ( Fig. 1; see also Fig. S5B) were comparable to those seen with the native protein, indicating that this amino acid replacement did not cause major changes to the overall protein structure. Analysis of this protein by the thermal shift assay and ITC showed that this protein was unable to recognize nitrate ( Fig. S5B and C).
The McpN-LBD structure was aligned to all structures currently deposited in the protein data bank using the DALI algorithm, and the closest structural homologues are listed in Table 1. Surprisingly, the closest structure was a LBD of a histidine kinase that belonged to a different family, namely, CHASE3. The high level of structural similarity between this domain and McpN-LBD is illustrated in Fig. 4A. The only chemoreceptor LBD with significant structural similarity was the HBM domain of the McpS chemoreceptor, which is composed of two 4-helix bundles (39).    formed part of chemoreceptors. Most of the corresponding species were marine bacteria, and a significant proportion of them are able to oxidize elemental sulfur or sulfite (see Table S1 in the supplemental material). Furthermore, a number of human pathogens such as Enterobacter cloacae, Streptococcus pneumoniae, and Eggerthia catenaformis were among the species that harbor McpN homologues (Table S1). The sequence alignment of McpN-LBD homologues revealed only a very modest level of overall sequence identity of approximately 5%. However, the zone around the nitrate binding site, which we have termed the N-box, was highly conserved and the corresponding sequence logo is shown in Fig. 5A. We have shown above that not all PilJ domains bind nitrate, since no binding was observed for the PilJ LBD, which is composed of two PilJ domains (Fig. S2). As shown in Fig. S7, the N-box was not conserved in either of the PilJ domains of the PilJ chemoreceptor. We then scanned the TrEMBL database using PROSITE (41) and the following consensus pattern for the N-box: The random statistical probability of a match was well below 1 sequence.
However, 941 sequences containing PilJ domains which are likely to be nitrate binding domains were retrieved. The retrieved sequences formed part of all major families of signal transduction systems, namely, transcriptional regulators, sensor kinases, chemoreceptors, and diguanylate cyclases. There were 1,135 protein sequences with at least one PilJ domain in Pfam at the time of the search, and the N-box may be usable as a means to identify PilJ domains that are able to bind nitrate.
Nitrate chemotaxis in other bacterial species. Subsequent work was aimed at assessing nitrate chemotaxis in other species. To that end, we conducted quantitative capillary chemotaxis assays using different strains grown under conditions of nitrate abundance and limitation. Pseudomonas putida KT2440 and Pseudomonas fluorescens  (Fig. 6).
We then studied two bacterial species that are among the top 10 plant-pathogenic bacteria (42), namely, Pectobacterium atrosepticum and Xanthomonas campestris pv. campestris. Interestingly, the responses of P. atrosepticum were very similar to those of PAO1, since only very minor responses were observed under nitrate abundance conditions but strong responses were observed under nitrate starvation conditions (Fig. 6). There is no McpN homologue among the 36 chemoreceptors of this strain, but there is a single receptor with a NIT domain. X. campestris pv. campestris also showed significant chemotaxis under nitrate-limiting conditions and only minor responses under nitrate abundance conditions (Fig. 6). Altogether, these data suggest that the induction of nitrate chemotaxis by nitrate limitation is common to other bacteria.

DISCUSSION
Nitrate is a final electron acceptor for anaerobic respiration and also serves as a nitrogen source for aerobic growth. Taxis to nitrate has been observed for a significant number of bacteria such as E. coli, Salmonella enterica serovar Typhimurium (43,44), Pseudomonas spp. (45)(46)(47)(48), Shewanella spp. (49,50), Azospirillum brasilense (51), Rhodobacter sphaeroides, Agrobacterium tumefaciens (52), Thioploca spp. (53), and Synechococcus spp. (54). The three major bacterial pathways for nitrate metabolism include respiratory, assimilatory, and dissimilatory nitrate reduction (55), and any type of metabolism can lead to energy taxis. In some of the cases, it has been demonstrated that bacterial nitrate taxis is based on energy taxis (44,49,51,52). For example, deletion or inhibition of enzymes that participate in nitrate metabolism abolished nitrate taxis (49). In other reports, the molecular mechanism of nitrate taxis, such as in the case of, for example, lake water bacteria (46) or different denitrifying strains (45,47), is unclear. It is possible that those observations were based on chemotaxis. Here we identify the molecular mechanism of nitrate-specific chemotaxis that is initiated by the specific recognition of nitrate at a periplasmic chemoreceptor LBD. McpN homologues show a broad phylogenetic distribution, including those of archaea and bacteria belonging to the Firmicutes and Proteobacteria phyla (see Table S1 in the supplemental material), which indicates that nitrate chemotaxis may be a widespread mechanism.
Interestingly, among the species that harbor McpN homologues were a significant number of bacteria isolated from marine sediments that are able to oxidize sulfide or elemental sulfur (Table S1). There is evidence that the oxidation of reduced sulfur compounds in these bacteria is coupled to the reduction of electron acceptors such as nitrate (56). As a consequence, some sulfide oxidizers were found to store nitrate in vacuoles (57) at concentrations of up to 370 mM (58,59). This intracellular nitrate is used to oxidize sulfide in deeper anoxic zones of sediments. This process has been particularly well studied in Beggiatoa spp. (60), which are also among the species that contain an McpN homologue (Table S1). On the basis of experiments performed with nitrate-reducing/sulfide-oxidizing shelf sediment bacteria belonging to the Thioploca genus, a functional model was proposed (53). The authors showed that the nitrate concentration in the sediment was lower than that in the flume water and that nitrate chemotaxis directed bacteria to the sediment surface, where they filled their vacuoles with nitrate. They then migrated back into deeper sediment layers, where they oxidized sulfide to sulfate until the nitrate was depleted, which induced the upward movement. Taken together, the data thus suggest the particular importance of nitrate chemotaxis in marine sulfide/sulfur-oxidizing bacteria.
Nitrate serves PAO1 as the sole nitrogen source for growth, and the anaerobic growth of this strain is accomplished through the denitrification enzyme pathway that catalyzes the sequential reduction of nitrate to nitrogen gas (61). Nitrate chemotaxis was observed in pathogenic P. aeruginosa bacteria but not in the nonpathogenic P. putida and P. fluorescens, suggesting that it may be related to virulence. Previous studies have shown a link between virulence and nitrate metabolism for anaerobically grown PAO1, since a mutant with a mutation in the nitrate reductase gene was avirulent in Caenorhabditis elegans (61). PAO1 causes airway infections in cystic fibrosis patients, and the sputum nitrate/nitrite concentration was 774 M in cystic fibrosis patients, well above the concentration seen with the healthy control group (421 M) (62). Importantly, these concentrations are in the range of the optimal chemotaxis responses measured here (Fig. 2), indicating that nitrate chemotaxis may be related to pathogenicity, as in the case of S. Typhimurium, where taxis to host-derived nitrate is required for efficient host infection (44). PAO1 nitrate chemotaxis was observed only under nitrate starvation conditions ( Fig. 2A), whereas no taxis was observed in under nitrate abundance conditions (see Fig. S3 in the supplemental material), and similar results were seen for P. atrosepticum (Fig. 6). This is unusual, since chemotactic behaviors are typically either constitutive or inducible by the chemoeffector (63,64). However, striking similarities exist between P i and nitrate chemotaxis in PAO1. P i taxis was observed only under P i starvation conditions and not under P i abundance conditions (23,24). As in the case of mcpN, the presence of P i was shown to decrease the transcript levels of both P i chemoreceptor genes, i.e., ctpL and ctpH (65). P i was identified as a key signal molecule that controls the expression of many virulence genes and features in PAO1 (65,66). P i and nitrate are both inorganic anions, and it is tempting to speculate that chemotaxis repression mediated by chemoeffector abundance is a feature of this compound family.
Almost one-third of all chemoreceptor LBDs are recognized by the Pfam 4HB domain signature (9). Signaling of chemoreceptors with this domain has been extensively studied, and the 3D structure reveals a 4-helix antiparallel bundle (67,68). Although the McpN-LBD sequence is not recognized by the Pfam 4HB signature, its structure superimposes very well on that of the 4HB Tar-LBD (Fig. 4B). This, together with the fact that the closest structural McpN-LBD homologue is a CHASE3 domain ( Table 1), demonstrates that the 4-helix bundle is a conserved structural motif for ligand Nitrate Chemotaxis in Bacteria ® sensing formed by members of different LBD families. Although conserved in structure, the modes of ligand binding for McpN-LBD and Tar-LBD are different. The Tar-LBD dimer recognizes two signal molecules with high negative cooperativity that bind to the dimer interface at two sites that are not on the dimer symmetry axis (68,69). In contrast, a single molecule of nitrate binds to a single site located at the dimer symmetry axis of McpN-LBD (Fig. 3A). However, 4HB domains and McpN-LBD (Fig. 3D) have in common that the individual domains are present in a monomer-dimer equilibrium and that ligand binding shifts this equilibrium to the dimeric state (69,70).
The NarX/NarL and NarQ/NarP two-component systems control transcriptional responses to nitrate and nitrite, which are the preferred anaerobic electron acceptors in E. coli (71). The LBDs of the NarX and NarQ sensor kinases are structural homologs of McpN-LBD (Table 1), and their 3D structures in complex with nitrate have been solved (72,73). Although McpN-LBD and NarX-LBD share only 21% sequence identity, their structures align very well and the nitrate binding site is conserved (Fig. 4C). McpN-LBD differs from NarX-LBD in several aspects. Our AUC studies showed that McpN-LBD has an intrinsic propensity to dimerize which is enhanced in the presence of nitrate. In contrast, NarX-LBD is monomeric even at a concentration of 10 mM and in the presence of nitrate (72). NarX and NarQ are characterized by a certain plasticity in ligand recognition, since they bind to nitrate, nitrite, and sulfite (34)(35)(36). In contrast, McpN-LBD recognizes nitrate exclusively and has no physiologically relevant affinity for nitrite ( Fig. 1; see also The NIT domain is present in different signal transduction protein families and was previously proposed to be a sensor domain for nitrate and nitrite (37). However, the recombinant NIT domain of PA4520 did not bind nitrate or nitrite (Fig. S2) and a mutant defective in this receptor was not affected in nitrate chemotaxis (Fig. 2). In addition, P. putida and P. fluorescens both possess a NIT domain containing a chemoreceptor which, however, did not mediate nitrate chemotaxis under the experimental conditions tested (Fig. 6). The NIT domain may thus represent a superfamily that contains subfamilies with different ligand binding properties and biological functions.
The demonstration of specific nitrate chemotaxis as reported here widens the range of known chemoeffectors and provides the basis for an assessment of this phenotype in other bacteria and for the elucidation of its physiological relevance.
Plasmid construction. The plasmids and oligonucleotides used are listed in Table 2 and in Table S2 in the supplemental material, respectively. Protein expression plasmids were constructed by amplification from genomic DNA of P. aeruginosa PAO1 for the DNA fragments encoding the LBDs of PilJ (amino acids 36 to 315), McpN (amino acids 44 to 179), and PA4520 (amino acids 38 to 321). The resulting PCR products were cloned into pET28(ϩ) to generate plasmids pPilJ-LBD, pMcpN-LBD, and pET4520-LBD. In all cases, plasmids were verified by sequencing. For the construction of the complementing plasmid pBBRMcpN, the mcpN gene was amplified using primers listed in Table S2. The resulting PCR fragment was cloned into the NdeI and BamHI sites of pBBR1MCS2_START, and the plasmid was transformed into P. aeruginosa PAO1-McpN by electroporation.
Protein overexpression and purification. E. coli BL21(DE3) was transformed with the expression plasmids, and the resulting strains were grown in 2-liter Erlenmeyer flasks containing 400 ml LB medium supplemented with kanamycin. Cultures were grown under conditions of continuous stirring (200 rpm) at 30°C. The growth temperature was lowered to 16°C when an optical density at 600 nm (OD 600 ) of 0.5 was reached, and protein expression was induced after 30 min by the addition of 0.1 mM isopropyl ␤-D-1-thiogalactopyranoside. Cultures were grown for another 14 h prior to harvesting of cells by centrifugation at 10,000 ϫ g and 4°C for 30 min. Cell pellets were resuspended in buffer A (20 mM Tris-HCl, 0.1 mM EDTA, 300 mM NaCl, 10 mM imidazole, 5% [vol/vol] glycerol, pH 7.6) and broken by French press treatment at a gauge pressure of 62.5 lb/in 2 . After centrifugation at 20,000 ϫ g for 1 h, the supernatant was loaded onto a 5-ml HisTrap column (Amersham Bioscience) previously equilibrated with buffer A. After washing with buffer A containing 35 mM imidazole was performed, protein was eluted by the use of a 35 to 500 mM imidazole gradient in buffer A. Proteins were dialyzed into the following buffers for analysis: for PA2788-LBD, 20 mM Tris-HCl (pH 7.4); for PA4520-LBD, 5 mM Tris-HCl, 5 mM MES (morpholineethanesulfonic acid), and 5 mM PIPES [piperazine-N,N=-bis(2-ethanesulfonic acid)] (pH 7.5); for PA0411-LBD, 50 mM HEPES (pH 7.5).
Differential scanning fluorimetry (DSF). DSF assays were performed on a MyIQ2 Real-Time PCR instrument (Bio-Rad). Compounds from different arrays (Biolog, Hayward, CA, USA) were dissolved in 50 l water, which, according to the manufacturer, corresponds to a concentration of 10 to 20 mM. The composition of these arrays is provided in http://208.106.130.253/pdf/pm_lit/PM1-PM10.pdf. Screening was performed using 96-well plates. Each well contained 2.5 l of the dissolved compound, 20.5 l protein, and 2 l SYPRO Orange (Life Technologies). The control well contained protein without ligand. Samples were heated from 23°C to 85°C at a scan rate of 1°C/min, and fluorescence changes were monitored. T m values correspond to the minima of the first derivatives of the raw data.  Sedimentation velocity (SV) runs were carried out at a rotor speed of 48,000 rpm using 400-l samples with the dialysis buffer as the reference. A laser was used at a wavelength of 235 nm in the absorbance optics mode. Least-squares boundary modeling of the SV data was used to calculate sedimentation coefficient distributions with the size-distribution c(s) method (74) implemented in SEDFIT v14.1 software. Buffer density ( ϭ 1.003 g/ml [0.99989 g/ml in the presence of NaNO 3 ]) and viscosity [ ϭ 0.013137 poise [0.01313 poise in the presence of NaNO 3 ]) at 10°C were estimated using SEDNTERP software (75) for the buffer components. The partial specific volume used was 0.7192 ml/g as calculated from the amino acid sequence using SEDNTERP software.
Intrinsic tryptophan fluorescence spectroscopy. McpN-LBD and McpN-LBD R61A mutants were dialyzed into 20 mM Tris-HCl (pH 7.4), and the reaction mixtures were adjusted to a concentration of 5 M. Proteins were placed into a PTI QM-2003 fluorimeter (Photon Technology International, Lawrenceville, NJ), and emission spectra were recorded at wavelengths of 305 to 400 nm following excitation at 295 nm. Spectra were recorded at 20°C using a slit width of 4 nm with a scan speed of 1 nm/s. Spectra were corrected with the buffer emission spectrum.
Quantitative capillary chemotaxis assays. Assays were conducted using two different protocols that differed under the cell culture conditions. Under conditions of nitrate abundance, overnight cultures in MS minimal medium supplemented with 20 mM glucose as a carbon source (note that this medium contains 25 mM NH 4 NO 3 ) were used to inoculate fresh medium to reach an OD 600 of 0.05. Cells were cultured at 30°C or 37°C until an OD 600 of 0.4 to 0.5 was reached. Under conditions of nitrate limitation, 150 l of an overnight culture in rich 2ϫ YT medium was used to inoculate 20 ml of N0 medium (MS lacking a nitrogen source). Growth was continued for 3 h (pseudomonads) or 4.5 h (P. atrosepticum), at which point the cells had reached an OD 600 of 0.15 to 0.2. For X. campestris, M9 minimal medium supplemented with 20 mM D-glucose, 5 mM NaNO 3 , and 5% (vol/vol) LB was used for the conditions of nitrate abundance, whereas M8 minimal medium (M9 without nitrogen source) supplemented with 20 mM D-glucose and 5% (vol/vol) LB was used for the conditions of nitrate limitation. Cells were grown for 6 h until the OD 600 reached 0.25 to 0.3.
Under both conditions, cells were washed twice by centrifugation (1,667 ϫ g and 6 min at 4°C) and resuspension in chemotaxis buffer (50 mM potassium phosphate, 20 M EDTA, 0.05% [vol/vol] glycerol, pH 7.0) and then resuspended in the same buffer to reach an OD 600 of 0.1. Aliquots (230 l) of the resulting cell suspension were placed into the wells of a 96-well microtiter plate. Capillaries (Microcaps; Drummond Scientific [reference P1424]) (1 l) were heat-sealed at one end and filled with buffer (control) or chemoeffector solution prepared in chemotaxis buffer. The capillaries were immersed into the bacterial suspensions at its open end. After 30 min, capillaries were removed from the wells, rinsed with sterile water, and emptied into 1 ml of chemotaxis buffer. Serial dilutions were plated onto M9 minimal medium plates supplemented with 20 mM glucose and incubated overnight at 30 or 37°C. CFU counts were determined and corrected with the buffer control.
RT-qPCR gene expression analysis. Total RNA was extracted using a High Pure RNA isolation kit (Roche Diagnostics) and treated with Turbo DNase (Invitrogen). RNA quality was verified by agarose gel electrophoresis and quantified spectrophotometrically. Subsequently, cDNA was synthesized from 500 ng RNA using SuperScript II reverse transcriptase (Invitrogen) and 200 ng of random hexamer primers (Roche) following the instructions of the manufacturers. Quantitative PCR was performed using iQ SYBR green supermix (Bio-Rad) in a MyiQ2 thermal cycler (Bio-Rad). The following protocol was used: 95°C (5 min), 35 cycles of 95°C (10 s) and 61°C (30 s), and melting curve analysis from 55 to 95°C, with an increment of 0.5°C/10 s. Gene expression data were normalized to expression of the rpoD reference gene. The primers used are listed in Table S2.
McpN-LBD crystallization and structure resolution. Crystallization conditions were screened using the capillary counter-diffusion technique and commercially available crystallization kits GCB-CSK, PEG448-49, and AS-49 (Triana Science & Technology, Granada, Spain). The protein, maintained at 1.5 mg/ml in 20 mM Tris-HCl-200 mM NaCl (pH 7.5), was incubated at 4°C with 1.7 mM NaNO 3 , and the excess of NaNO 3 was removed by centrifugation using Amicon concentrators (3-kDa cutoff). The protein-ligand complex was loaded into 0.2-mm-inner-diameter capillaries, and crystals of sufficient size appeared in 0.82 M K/phosphate-0.82 M Na/phosphate (0.1 M Na/HEPES, pH 7.5). Crystals were extracted from the capillary, flash-cooled in liquid nitrogen, and stored until data collection. Crystals were diffracted at beam line ID23-1 of the European Synchrotron Radiation Facility (ESRF). Data were indexed and integrated with XDS (76) and scaled with SCALA (77). Attempts at molecular replacement using homology models generated using the NarX sensor domain (PDB identifier [ID] 3EZI) and the NarQ sensor domain (PDB ID 5IJI) were unsuccessful. Phases were obtained using Arcimboldo (78) and searching for two helices that were 30 amino acids in length. Refinement was initiated with Refmac (79) and finalized with phenix.refine (80), tracking the quality with MolProbity (81). Refinement statistics and quality indicators are summarized in Table S3.
Site-directed mutagenesis. An overlapping PCR mutagenesis approach was used to construct the alanine substitution mutant McpN-LBD R61A. First, a NdeI/XhoI DNA fragment of pMcpN-LBD was cloned into the same sites of pCR2.1-TOPO and transformed into E. coli DH5␣ (dam positive [dam ϩ ]). Next, the resulting pCR-McpN-LBD plasmid was fully amplified by PCR using a complementary primer pair carrying the mutation. The parental plasmid was cleaved using DpnI, and plasmids with the desired mutation were recirculated with T4 DNA ligase (Roche). The presence of the mutation in the resulting plasmid, pCR-McpN-R61A, was confirmed by sequencing prior to cloning into the NdeI/XhoI sites of pET28(ϩ) to generate pMcpN-R61A (Table 2). Data availability. Coordinates and structure factors of McpN-LBD were deposited at the PDB with accession code 6GCV.

ACKNOWLEDGMENTS
We acknowledge the European Synchrotron Radiation Facility (ID23-1 and ID30A-1,3) and thank the beam line staff members for their invaluable support.
This work was supported by FEDER funds and Fondo Social Europeo through grants from the Junta de Andalucía (grant CVI-7335) and the Spanish Ministry for Economy and Competitiveness (grants BIO2013-42297, BIO2016-76779-P, and BIO2016-74875-P). We furthermore acknowledge NIH grant P30 DK089507, which financed the generation of bacterial mutants.