Nitrate- and Nitrite-Sensing Histidine Kinases: Function, Structure, and Natural Diversity

Abstract

Under anaerobic conditions, bacteria may utilize nitrates and nitrites as electron acceptors. Sensitivity to nitrous compounds is achieved via several mechanisms, some of which rely on sensor histidine kinases (HKs). The best studied nitrate- and nitrite-sensing HKs (NSHKs) are NarQ and NarX from Escherichia coli. Here, we review the function of NSHKs, analyze their natural diversity, and describe the available structural information. In particular, we show that around 6000 different NSHK sequences forming several distinct clusters may now be found in genomic databases, comprising mostly the genes from Beta- and Gammaproteobacteria as well as from Bacteroidetes and Chloroflexi, including those from anaerobic ammonia oxidation (annamox) communities. We show that the architecture of NSHKs is mostly conserved, although proteins from Bacteroidetes lack the HAMP and GAF-like domains yet sometimes have PAS. We reconcile the variation of NSHK sequences with atomistic models and pinpoint the structural elements important for signal transduction from the sensor domain to the catalytic module over the transmembrane and cytoplasmic regions spanning more than 200 Å.

Keywords: allostery; cell signaling; histidine kinases; nitrate regulation; nitrate respiration; signal transduction; two-component systems.

Figure 1

Model for the NarX–NarL and NarQ–NarP cross-regulation network. Dashed arrows represent relatively slow reactions. The NarX and NarQ sensor populations are hypothesized to be in a two-state equilibrium determined by stimulus (ligand binding). Phospho-sensors catalyze response regulator phosphorylation, whereas dephospho-sensors catalyze regulator dephosphorylation. Phospho-regulators activate (+) or repress (−) transcription; representative target operons are shown. Reproduced with permission from the reference [88].

Figure 2

Phylogenetic tree of nitrate- and nitrite-sensing histidine kinases. Genes belonging to different bacterial orders are shown in different colors; genes with missing order information are shown in gray. Genomic neighborhoods of representative genes (labeled in blue) are shown in Figure 3. The tree was calculated for a set of 920 representative genes (centroids from clustering at the 80% sequence identity level using UCLUST [118]) using FastTree 2 [119] and drawn using FigTree [120]. Multiple sequence alignment, taxonomic annotation and phylogenetic tree for the analyzed NSHK sequences are available as Supplementary Datasets 1, 2 and 3.

Figure 3

Order of putative nitrate metabolism genes in representative genomes (GenBank IDs U00096.3, AE004091.2, AFOC01000011.1, CP000116.1, CP001561.1, CADIKA010000002.1, CH899769.1, NZ_BFCB01000003.1, L42023.1, CP000626.1, JH651379.1, ALWO02000023.1, FONY01000003.1). Bacteroidetes genes are labeled narHK and narRR, since they are notably different from narQ/narP and narX/narL, and the architectures of the sensor proteins are different (GAF-like domains are absent). No genes possibly involved in nitrate metabolism are observed in the vicinity of Escherichia coli narQ, Haemophilus influenzae narQ, and narP and Neisseria meningitidis narX and narL. narK₁ and narK₂ are sometimes annotated in the literature as narK and narT, respectively. nrtB/C are sometimes referenced to as ntrB/C; here, these genes are representatives of the ATP-Binding Cassette (ABC) transporters family that are involved in nitrate transport and are not the members of the ntrBC TCS, which controls expression of the nitrogen-regulated (ntr) genes in response to nitrogen limitation [122,123]. mobA encodes molybdopterin-guanine dinucleotide biosynthesis protein [124,125,126]; moaC encodes cyclic pyranopterin monophosphate synthase [127], which is a nitrate reductase and molybdopterin biosynthesis-associated protein; molybdopterin guanine dinucleotide is a cofactor for nitrate reductases [124,128,129]. The genes whose possible involvement into nitrate metabolism is not clear are colored gray; rub, rubredoxin; cyt, c-type cytochrome.

Figure 4

Singlet and doublet NSHK genes in the assembled dataset. The phylogenetic tree is the same as in Figure 2. Sequences belonging to genomes with a single NSHK are shown in green and those belonging to genomes with two NSHKs are shown in red. Metagenomic sequences and sequences without assigned strain are shown in gray.

Figure 5

Architecture of transmembrane nitrate sensors. (a) Architecture of nitrate-responsive TCS receptors: chemoreceptor McpN and different nitrate/nitrite sensor HKs: GAF-less proteins from Bacteroidetes and NarQ and NarX from Proteobacteria. Panel adapted from [37]. (b) Atomic model of NarQ in a realistic membrane. The protein is a homodimer. The numbers indicate the respective amino acid numbers at the domain junctions.

Figure 6

Conservation of NSHK domains relative to conservation of the whole protein. Shown are pairwise sequence identity values for all NSHKs in the dataset smoothed with Gaussian kernel density estimation and transformed to logarithmic scale. Red diagonals are used as guides: the values above the diagonals correspond to better domain conservation compared to the whole protein, whereas the values below the diagonals correspond to lower relative domain conservation.

Figure 7

Signaling-associated conformational changes in the NarQ sensor and TM domains. Superposition of the symmetric ligand-free structure (gray) and symmetric ligand-bound structure (colored) is shown. The binding of nitrate causes downward displacement of TM1 and upward displacement of TM2, alongside with rearrangements in the membrane plane. Left: Changes in the conformation of helix TM1. Center: Changes in the conformation of helix TM2. Right: Changes in the arrangement of the TM helices. Serine and threonine side chains are shown explicitly, glycine C_α atom positions are marked with the spheres. Positions of water molecules in the TM region are shown with gray (apo) or red (holo) spheres. The structures are aligned by the sensor domains. Adapted from [140].

Figure 8

Structure of the nitrate-binding pocket in NarQ [140]. The nitrate is coordinated by Arg50 side chains of the two protomers and stacks with Gly47. The interaction may also be stabilized by interaction between the partially negatively charged carbonyl oxygen atoms of Gly47 and the partially positively charged nitrogen atom of the nitrate ion. G47 and R50 form the G-x-x-R motif conserved among many nitrate-responsive sensor proteins.

Figure 9

Amino acid frequencies at different TM1 (a) and TM2 (b) positions in multiple sequence alignment of NSHK genes. Amino acids with similar properties are grouped together for clarity. Patterned occurrences of glycine, serine, and threonine are evident.

Figure 10

Abundance of glycines, serines, and threonines in transmembrane α-helices in NSHKs. Each TM helix (TM1 and TM2) on average contains one to two glycines and three to four serines or threonines.

Figure 11

Details of the signal transduction from the TM domain to and through the HAMP domain. (a) Ligand-free state. (b) Ligand-bound state. (c) Superposition of the ligand-free (gray) and ligand-bound (colored) states. A piston-like displacement of the cytoplasmic end of the helix TM1 relative to TM2 and the TM2-AS1 proline hinge is transmitted to the membrane-proximal end of AS2 and results in lever-like rotations of the HAMP domain protomers around the hinges. Since the HAMP domain protomers move in opposite directions, the positions of membrane-distal ends of helices AS2 also change relative to each other. Positions of the Leu225 Cα atom are marked with the spheres. The gray bar shows the position of TM1 ends in the apo state structure. The domains are aligned by the residues 175–177. Reproduced with permission from [140].

Figure 12

Coiled coil models of NarQ signaling helix domain. Two conformations are possible. Conformation 1 is shown in red and conformation 2 is shown in blue. The packing in layers 1–3 is different, while the packing in layers 4–6 is similar. The phase stutter is possible in the region of the residues 232–236. Residues Glu235, Lys236, and Thr237 belong to the characteristic conserved signaling helix motif (32). Reproduced with permission from [140].

Figure 13

Structural models of different GAF domains: homology models of NarQ and NarX GAF-like domains, which are based on the structure of the Acinetobacter baylyi phosphoenolpyruvate-protein phosphotransferase (ptsP) GAF domain (PDB ID 3CI6); E. coli potassium sensor HK KdpD GAF domain (PDB ID 4QPR [202]); phycocyanobilin-bound Leptolyngbya sp. JSC-1 phosphorylation-responsive photosensitive histidine kinase (PPHK) GAF domain (PDB ID 6OAP, [203]); cyclic AMP-bound Anabaena adenylyl cyclase cyaB2 GAF domain (PDB ID 1YKD [204]). NarQ and NarX GAF-like domains are reduced in size compared to well-characterized GAF domains.

Figure 14

Structural models of NarQ (left, based on PDB IDs 3SL2 and 4GT8 [131,223]) and SrrB (right, PDB ID 6PAJ [218]) catalytic domains. The active site binds ATP and Mg²⁺ ion. In SrrB, Cys464 and Cys501 form an intramolecular disulfide bond, which responds to the cellular redox environment and affects autophosphorylation kinetics [218]. In NarQ, there is a conserved pair of similar cysteine residues, Cys455 and Cys494, that may also form a disulfide bond and react to the redox environment.