NosP Detection of Heme Modulates Burkholderia thailandensis Biofilm Formation

Abstract

Aggregated bacteria embedded within self-secreted extracellular polymeric substances, or biofilms, are resistant to antibiotics and cause chronic infections. As such, they are a significant public health threat. Heme is an abundant iron source for pathogenic bacteria during infection; many bacteria have systems to detect heme assimilated from host cells, which is correlated with the transition between acute and chronic infection states. Here, we investigate the heme-sensing function of a newly discovered multifactorial sensory hemoprotein called NosP and its role in biofilm regulation in the soil-dwelling bacterium Burkholderia thailandensis, the close surrogate of Bio-Safety-Level-3 pathogen Burkholderia pseudomallei. The NosP family protein has previously been shown to exhibit both nitric oxide (NO)- and heme-sensing functions and to regulate biofilms through NosP-associated histidine kinases and two-component systems. Our in vitro studies suggest that BtNosP exhibits heme-binding kinetics and thermodynamics consistent with a labile heme-responsive protein and that the holo-form of BtNosP acts as an inhibitor of its associated histidine kinase BtNahK. Furthermore, our in vivo studies suggest that increasing the concentration of extracellular heme decreases B. thailandensis biofilm formation, and deletion of nosP and nahK abolishes this phenotype, consistent with a model that BtNosP detects heme and exerts an inhibitory effect on BtNahK to decrease the biofilm.

Graphical Abstract

INTRODUCTION

Iron is a vital nutrient that bacteria strictly rely upon for survival. Due to the low solubility of ferric iron, and the sequestration of iron by host proteins, pathogenic bacteria secrete high-affinity iron chelators called siderophores to compete for host iron sources.¹ Heme and hemoglobin, which account for as much as 70% of the host iron content, are accessible and preferred iron sources for pathogens.² Many pathogens are capable of infecting blood-rich organs and tissues where they produce hemolysin to rupture erythrocytes.³ The released heme is then captured by extracellular hemophores or membrane-bound heme receptors, transferred into the periplasm region in a TonB-dependent manner for Gram-negative bacteria, or translocalized through the cell wall with the assistance of cell-wall heme chaperon proteins for Gram-positive bacteria, and eventually delivered into the cytoplasmic region via the inner-membrane-bound ATP-binding cassette permease.¹

Upon internalization, heme can be degraded by heme oxygenases to release ferrous iron or bind to heme-dependent enzymes such as cytochromes, catalases, and nitric oxide reductases. In addition, heme also enters cell signaling circuits by acting as a cofactor of heme-based gas/redox sensors or binding reversely to labile heme-sensing proteins.⁴ Heme-based gas sensing and labile heme-sensing have been implicated in regulating many bacterial physiological processes. For example, in the symbiotic bacterium Rhizobium meliloti, O₂ sensing is mediated by the heme-bound sensory kinase FixL to fine-tune the transcription of nitrogen fixation genes nif and fix.^5,6 In many biofilm-producing bacteria such as Legionella pneumophila and Shewanella oneidensis, the heme-nitric oxide/oxygen (H-NOX) family protein has been shown to sense pico- to nanomolar concentrations of NO and modulate biofilm dynamics through the regulation of intracellular concentrations of an important secondary-messenger molecule, cyclic diguanylate monophosphate (c-di-GMP).⁷ Recently, more studies have revealed that labile heme itself can act as a signaling entity, and labile heme-responsive proteins play critical roles in maintaining bacterial heme/iron homeostasis, virulence factor secretion, etc. In Staphylococcus aureus, the membrane-bound sensory kinase HssS is activated through labile heme binding to phosphorylate its cognate response regulator HssR. Phosphorylated HssR then serves as a transcriptional activator for hrtAB heme efflux pump genes, allowing surplus heme to be removed from the cell to avoid toxicity.⁸ In Pseudomonas aeruginosa, genetic silencing of the labile heme shuttling and sensing protein PhuS in both the wild-type MPAO1 and the siderophore-deficient isogenic strains results in the overproduction of pyocyanin in the presence of heme during the transition to the stationary phase.⁹ Pyocyanin, a major virulence factor of P. aeruginosa and the terminal readout of Pseudomonas quinolone signal (PQS), has also been implicated to play roles in iron acquisition and iron/hydroxyl radical-driven cellular injury.^10,11 It has been postulated that the prematuration of pyocyanin is due to the imbalance of iron homeostasis caused by PhuS disruption.⁹

Our laboratory has recently identified a novel sensory hemoprotein called NosP. nosP genes are widely distributed among bacterial genomes where they are frequently found in the same operon as, or fused in the same polypeptide chain as, signal transduction genes encoding histidine kinases, c-di-GMP-metabolizing enzymes, and methyl-accepting chemotaxis proteins, suggesting a role for NosP in signaling.¹² Purified stand-alone NosP proteins in P. aeruginosa,¹² L. pneumophila,¹³ S. oneidensis,¹⁴ and Vibrio cholerae¹⁵ can bind heme and subsequently ligate NO. NO-ligated holo NosPs differentially regulate NosP-associated histidine kinases (NahKs) in comparison to the Fe^II-unligated form, underscoring the function of NosP as a heme-based NO sensor. This is further supported by the fact that P. aeruginosa disperses from the biofilm upon NO treatment but lacks the canonical NO-sensing H-NOX protein, and disruption of PaNahK function abolishes the NO-triggered dispersal phenotype.¹² Thus, we have concluded that NosP could serve as a heme-based NO sensor in bacteria to regulate biofilms. Interestingly, V. cholerae encodes a second NosP domain-containing protein, Vc_0130, which contains a NosP domain N-terminal to a c-di-GMP phosphodiesterase (PDE). Vc_0130 is an active PDE in its apo form; however, the enzymatic function is completely inhibited upon heme binding to the N-terminal NosP domain.¹⁶ Collectively, these observations led us to postulate that NosP may exhibit dual functions as a heme-based NO sensor and/or a labile heme-sensor, the exact function of which may vary among species.

In this work, we explore this hypothesis in Burkholderia thailandensis, a nonvirulent soil-dwelling bacterium. B. thailandensis is a model organism for the Gram-negative BSL3 pathogen Burkholderia pseudomallei, the causative agent of the neglected tropical disease Melioidosis.¹⁷ Melioidosis is extremely resistant to an array of antibiotic treatments,¹⁸ and relapse of this infection is quite common and associated with biofilm formation.¹⁹ A blast search using PA1975 (NosP from PAO1) as the inquiry sequence identified homologous NosPs from B. pseudomallei (BPSS1647) and B. thailandensis (Bth_ii0732), each sharing ~42% identity to the inquiry sequence. Like other NosPs, BpNosP and BtNosP are each in the same operon as a signal transduction two-component system consisting of a histidine kinase (BPSS1646 and Bth_ii0733 for B. pseudomallei and B. thailandensis, respectively) and a degenerate c-di-GMP phosphodiesterase response regulator (BPSS1648 and Bth_ii0731 for B. pseudomallei and B. thailandensis, respectively). In this work, we characterized BtNosP and its associated histidine kinase BtNahK. Our in vitro study suggests that BtNosP exhibits both NO sensing and labile heme-sensing functions; in vivo, BtNosP binds labile heme to inhibit BtNahK autophosphorylation, and the inhibited autokinase activity of BtNahK is associated with a biofilm formation defect in B. thailandensis.

MATERIALS AND METHODS

Cloning.

Bth_ii0731 (narR), Bth_ii0732 (nosP), and Bth_ii0733 $\left(nah\bar{K}\right)$ were polymerase chain reaction (PCR) amplified from B. thailandensis E264 genomic DNA. Sequences of narR and nosP were cloned into the pET-20b vector to generate constructs with C-terminal His₆ tags. Sequences of nosP and nahK were cloned into the pT7TEVHMBP vector (engineered from the pET-28a backbone) to produce constructs with cleavable N-terminal His₆-MBP (HMBP) tags. Plasmids were sequencing-confirmed and subsequently transformed into Escherichia coli DH5α competent cells for DNA amplification and extraction.

His₆-Tagged Protein Expression and Purification.

BtNosP.

pET-20b-BtNosP and pT7TEVHMBP-BtNosP was transformed into BL21 (DE3) pLysS and Rosetta (DE3) pLysS cells, grown in yeast extract phosphate media (45 g of yeast extract, 0.017 M NaH₂PO₄, 0.072 M Na₂HPO₄, per liter) in the presence of 100 μg mL⁻¹ ampicillin and 34 μg mL⁻¹ chloramphenicol for BL21, or 10 μg mL⁻¹ kanamycin and 34 μg mL⁻¹ chloramphenicol for Rosetta. 1 L cultures were shaken at 250 rpm at 37 °C to reach an OD₆₀₀ between 0.5 and 0.7, after which 100 μM isopropyl β-d-1-thiogalactopyranoside (IPTC) was added to induce protein overexpression. Induced cultures were grown at 16 °C with agitation at 250 rmp for 16 h and then harvested via centrifugation at 7500 rpm for 15 min. Cell pellets were then lysed on ice via sonication in base buffer (50 mM Tris-HCl, 300 mM NaCl, 10% glycerol, pH = 7.5) supplemented with 1% Triton X-100, 5 mM 2-mercaptoethanol (BME), and 1 mM phenylmethylsulfonyl fluoride (PMSF), followed by ultracentrifugation at 40,000 rpm for 1 h to remove cell debris. The clear supernatant was loaded onto a Ni-NTA affinity column (QIAGEN) for His₆-tagged purification. Base buffer containing gradient concentrations of imidazole (10, 20, and 250 mM), 5 mM BME, and 1 mM PMSF was used to wash and elute proteins. Apo-BtNosP, eluted by 250 mM imidazole, was subsequently gelfiltered to desalt imidazole by the AKTA FPLC system fitted with a Superdex 75 HiLoad 16/60 column using FPLC buffer (50 mM Tris-HCl, 300 mM NaCl, 5 mM BME, pH = 7.5).

BtNahK and BtNarR.

pT7TEVHMBP-BtNahK and pET20b-BtNarR were transformed into Rosetta (DE3) pLysS and BL21 (DE3) pLysS competent cells for protein overproduction, respectively. Both cell types were cultured in 2xYT media (16 g of tryptone, 10 g of yeast extract, 5 g of NaCl per liter), supplemented with 10 μg mL⁻¹ kanamycin and 34 μg mL⁻¹ chloramphenicol for Rosetta or 100 μg mL⁻¹ ampicillin and 34 μg mL⁻¹ chloramphenicol for BL21. 500 and 10 μM IPTG were used to induce Rosetta and BL21 cells, respectively. The following cell harvest, lysis, and purification steps were the same as those mentioned above. Base buffer (50 mM Tris-HCl, 300 mM NaCl, pH = 7.5) containing 5 mM BME and gradient imidazole (10, 20, 50, 100, and 250 mM) were utilized to wash and elute HMBP-BtNahK and His₆-BtNarR, after which the eluted proteins were desalted on PD-10 desalting columns (Cytiva). All three types of proteins were stored in a final buffer condition of 50 mM Tris-HCl pH 7.5, 300 mM NaCl, and 10% glycerol. Proteins were kept frozen at −80 °C and freshly thawed before assays. Bradford assay was performed to determine protein concentrations.

Removal of HMBP Tag by TEV Digestion.

800 μg mL⁻¹ HMBP-tagged apo-BtNosP purified from Rosetta cells was treated with 20 units mL⁻¹ of TEV protease (New England Biolabs) in 1× TEV reaction buffer (50 mM Tris-HCl, 0.5 mM EDTA, 1 mM DTT, pH = 7.5) and incubated at 30 °C for 1 h. The reaction mixture was then loaded onto a Ni-NTA column preequilibrated with base buffer (50 mM Tris, 300 mM NaCl, pH = 7.5), and the nontag BtNosP was eluted using base buffer supplied with 20 mM imidazole. Imidazole was subsequently removed by passing nontag BtNosP solution through PD-10 desalting columns.

Holo-BtNosP Reconstitution.

For assays that study the spectral/kinetic/thermodynamic properties of BtNosP, the nontag version of BtNosP was used. Due to the low recovery rate and low concentrations of nontag BtNosP after TEV digestion, nontag BtNosP was incubated with 10× hemin at 4 °C for 2 h on a rotating platform. Excess hemin was then desalted through PD-10 columns. For assays that study the BtNosP/BtNahK interactions, which usually require high concentrations of proteins, His₆-BtNosP was incubated with 10× hemin at 4 °C for 16 h. Hemin-incubated samples were subsequently PD-10-desalted and injected into a AKTA FPLC system fitted with a Superdex 200 HiLoad 16/60 column to get rid of any hemin that is not protein-bound.

UV–Vis Spectroscopy.

All electronic spectra were obtained on a Cary-100 spectrophotometer against a background base buffer of 50 mM Tris, 300 mM NaCl. The Fe^III-BtNosP complex was generated by incubating holo-BtNosP with 10 mM potassium ferricyanide aerobically for 10 min, then desalted on a PD-10 column using base buffer at pH = 7.5. The Fe^II-BtNosP complex was made by reducing the Fe^III-BtNosP complex using base buffer containing 20 mM sodium dithionite at pH = 10.5 in a COY anaerobic chamber for 30 min. The Fe^II(NO)-BtNosP complex was prepared by incubating the Fe^II-BtNosP complex with 8 mM diethylammonium (Z)-1-(N,N-diethylamino)diazen-1-ium-1,2-diolate (DEA-NONOate, an NO donor with a half-life of around 16 min at room temperature) for 30 min anaerobically, after which DEA-NONOate was desalted by a PD-10 column using deoxygenated base buffer at pH = 7.5. The Fe^II(CO)-BtNosP complex was generated by bubbling CO gas for 10 min into Fe^II-BtNosP solution sealed in an air-tight cuvette. All treatments related to the ferrous state were performed carefully in an anaerobic chamber using deoxygenated buffer. Holo-BtNosP solutions of different oxidation and ligation states were adjusted to a proper concentration and placed into septum-sealed cuvettes to collect spectra on the spectrometer.

NO Dissociation Rate Constant (k_−NO) Measurement.

NO dissociation rate constant was measured via an established method using CO and dithionite trap.^12,14,15 Briefly, 8× CO-saturated sodium dithionite solution was prepared by dissolving dithionite powders in a base buffer (50 mM Tris pH 7.5, 300 mM NaCl) anaerobically, after which CO was bubbled in for 10 min. 100 μL of 8× CO-saturated dithionite solution was injected into 700 μL of Fe^II(NO)-BtNosP solution sealed in an air-tight cuvette to give a final dithionite concentration equal to 1× (1× = 0.3, 3, 15, and 30 mM). Immediately after injection, a time-dependent absorbance change ranging from 360 to 480 nm was monitored by the scanning kinetic program (Cary) at 23.3 °C during a total of 3 h time course, with a 23 s interval between two scans. The absorbance values of the first spectrum at time = 0 were subtracted from all of the subsequent spectra to generate a graph of difference absorbance versus wavelength. The difference absorbance values at 390 nm were then subtracted from those at 420 nm and plotted against time. A double-exponential model in GraphPad Prism

$$\mathrm{\Delta }A\left(t\right)=A_1\left(1-{\mathrm{e}}^{-k_{1}t}\right)+A_2\left(1-{\mathrm{e}}^{-k_{2}t}\right)$$ (1)

was utilized to fit the change in absorbance over time, where $k_{\text{slow}}$ was reported as the final NO dissociation rate constant.

Heme Dissociation Rate Constant $\left({\mathit{k}}_{\text{-heme}}\right)$ Measurement.

Heme dissociation rate constant was measured using a standard apo-myoglobin trap assay.²⁰ The Fe^III-BtNosP complex was mixed with apo-myoglobin in a septum-sealed cuvette to a molar ratio of at least 1:5 to ensure a pseudo-first-order reaction, immediately after which the absorbance between 360 and 480 nm was recorded every 23 s for 30 min by the scanning kinetic program (Cary) at 23.3 °C. The difference absorbance values at 409 nm were plotted against time and fitted with a double-exponential model $\mathrm{\Delta }A\left(t\right)=A_1(1\left.-{\mathrm{e}}^{-k_{1}t}\right)+A_2\left(1-{\mathrm{e}}^{-k_{2}t}\right)$ (eq 1). $k_{\text{slow}}$ was reported as the final heme transfer rate constant.

Determination of the Stoichiometry of Heme Binding to BtNosP.

Heme stock was prepared in 50 mM Tris, 300 mM NaCl with 5% v/v DMSO. The stock solution of heme was filtered through a 0.45 μm syringe filter to remove dimerized heme. The precise concentration of heme was determined by measuring the absorbance at 384 nm ($\epsilon =58.4{\mathrm{c}\mathrm{m}}^{-1}{\mathrm{m}\mathrm{M}}^{-1}$)²¹ using a Cary UV–vis spectrophotometer. 1 μL of heme solution was titrated each time into 1 mL of 5 μM nontag apoBtNosP in 50 mM Tris, 300 mM NaCl buffer (sample cuvette), and into 50 mM Tris, 300 mM NaCl buffer (reference cuvette). The UV–vis absorbance of the reference cuvette was subtracted from that of the sample cuvette under the double-beam mode of a Cary spectrometer.

Heme Dissociation Constant (K_D) Measurement.

Intrinsic tryptophan fluorescence quenching assay was utilized to determine the binding affinity of BtNosP to heme using a Horiba Scientific Fluorometer. 1 mL of 1 μM nontag apo-BtNosP in 50 mM Tris 300 mM NaCl buffer was excited at 283 nm, and 1–10 μL of heme stock solution (80 μM) was titrated into the protein solution stepwise. The total added volume of heme solution is less than 40 μL. Each time after adding heme, the fluorescence signals were allowed to reach equilibrium by incubating for 1000 s. Then, the intrinsic fluorescence emission of tryptophan was recorded at 328 nm. The titration data were fitted into the following eq 2 using GraphPad Prism

$$F=\frac{F_0\cdot K_{\text{D}}+F_{\text{C}}\cdot \left[\text{Heme}\right]}{K_{\text{D}}+\left[\text{Heme}\right]}$$ (2)

where $F_0$ is the fluorescence signal of apoprotein, $F_C$ is the fluorescence signal when $\left[\mathrm{H}\mathrm{e}\mathrm{m}\mathrm{e}\right]$ is infinitely large, $\left[\mathrm{H}\mathrm{e}\mathrm{m}\mathrm{e}\right]$ is the total added concentration of heme, and $K_{\mathrm{D}}$ is the dissociation constant of heme.²²

[γ-³²P]-ATP Kinase Autophosphorylation Assay.

Unless otherwise stated, the buffer system, MgCl₂ and ATP concentrations, volumes of loaded samples, and percentage of gels will stay consistent for all of the assays mentioned. The time-dependent autokinase reaction was initiated by adding ATP-mix solution to HMBP-BtNahK solution to give a final component of 3 μM HMBP-BtNahK, 5 mM MgCl₂, 2 mM ATP, and 0.25 μCi μL⁻¹ γ-³²P ATP in the reaction buffer (50 mM Tris, 300 mM NaCl, pH = 7.5) at 25 °C. At various time points (0, 5, 10, 20, 30, 45, 60, 90 min), 20 μL of the reaction mix was removed and quenched with 5 μL of 5× SDS sample loading dye. 10 μL of quenched samples were then loaded onto a 12.5% Tris-Glycine polyacrylamide gel to resolve proteins, and the gel was stained with and destained of Coomassie blue stain, followed by an overnight (~16 h) drying process. The dried gel was exposed to a phosphor screen for 5 h before being scanned by a Typhoon Imager. The radioactive bands were then quantified via ImageJ software. To examine the influence of BtNosP on BtNahK phosphorylation, apo-BtNosP and holo-BtNosP in different oxidation and ligation states were made accordingly. HMBP-BtNahK and His₆-BtNosP were preequilibrated at room temperature for 15 min. Then, ATP-mix solution was added to start the reaction. The final concentrations of HMBP-BtNahK and His₆-BtNosP in a total of 20 μL of the reaction mixture were 5 and 40 μM, respectively. Reactions were allowed to proceed for 90 min and quenched by adding 5 μL of 5× SDS-sample loading dye, followed by electrophoresis, staining and destaining, overnight drying, exposure, and imaging. For the Fe^III-BtNosP and apo-BtNosP titration assay, 3 μM HMBP-BtNahK and 5–100 μM His₆-BtNosP were used in a 90 min reaction.

[γ-³²P]-ATP Phosphoryl-Profiling Assay.

To determine whether BtNahK can engage in phosphoryl transfer to BtNarR, 3 μM HMBP-BtNahK was first pre-autophosphorylated for 30 min at 25 °C. Then, phosphoryl transfer was initiated by adding 3 μM His₆-BtNarR. 20 μL of the reaction mix was removed and quenched by 5 μL of 5× SDS-loading dye at varying time points (0, 0.5, 2, 10, 15, 30, 60 min). Samples were then subjected to electrophoresis, staining and destaining, drying, exposure, and imaging. To confirm the phosphoryl transfer pathway, site-directed mutagenesis was performed to mutate the key residues of BtNahK (H109A) and BtNarR (D89A). The same phosphoryl-profiling experiment was assayed similarly with mutated proteins.

Pull-Down Assay.

Amylose resin was prewashed with milli-Q water three times, then equilibrated with a pull-down buffer (50 mM HEPES, 300 mM NaCl, 1% Triton-100, 1 mM PMSF, 2 mM DTT) three times freshly before the assay. 20 μL of amylose resin was incubated with 1 μM HMBP-BtNahK or 1 μM His₆-MBP and 3 μM His₆-BtNosP of different ligation and oxidation states in a volume of 500 μL pull-down buffer. The reaction mixtures were allowed to gently rotate in an anaerobic bag for 1 h at room temperature to prevent the oxidation of Fe^II-NosP. Then, the reaction mixtures were centrifuged at 4000 rpm to pellet the beads and remove supernatants. To remove nonspecific protein–protein interactions, beads were resuspended in 500 μL of fresh pull-down buffer and gently rocked for 5 min, after which the beads were centrifuged and pelleted again. The washing steps were performed three times. Then, 20 μL of 2× SDS-sample loading buffer was added to the beads, and samples were boiled for 3 min followed by centrifugation at 14,000 rpm to elute both the bait and prey proteins. 10 μL of the supernatant was loaded onto a 12.5% Tris-Glycine gel for electrophoresis. Proteins were subsequently transferred from the gel to a nitrocellulose membrane for western blot. HRP-conjugate anti-6xHis antibody (abcam) was utilized to detect the chemiluminescence signals of HMBP-BtNahK (bait), His₆-MBP (tag control), and His₆-BtNosP (prey) using an Azure Sapphire Imager. The band intensity of prey proteins was normalized to that of the bait proteins. The results are an average of three biological replicates.

B. thailandensis Genetics.

Wild-type B. thailandensis strain E264 and Bth_ii0733 (nahK) transposon mutant BT02981 strain were obtained from Prof. Collin Manoil’s Laboratory at the University of Washington.²³ To generate clean in-frame deletions of nosP and nahK, homologous recombination mediated by the natural transformation method was utilized.^24,25 Generally, ~1 kbps homologous regions flanking nosP were amplified using E264 genomic DNA as the template. The upstream and downstream flanking regions were then joined by overlapping extension PCR to yield a single DNA product, with an EcoRI cutting site incorporated in the middle of the PCR product. The joined homologous region was then cloned into a blunt-end cloning vector pJET1.2/blunt (Thermo Fisher), resulting in pJETΔnosP::EcoRI. Next, an FRT-nptII-FRT sequence that contains a kanamycin-resistant cassette flanked by two flippase-binding sites was amplified with forward and reverse primers containing 5′-EcoRI sites using pUC18R6K-mini-Tn7T-Km as a template. The FRT-nptII-FRT cassette was subsequently ligated into pJETΔnosP::EcoRI to yield pJETΔnosP::FRT-nptII-FRT. The final allelic exchange vector was linearized by HindIII digestion and introduced to naturally competent BtE264 cells cultured in M63 media. The successful double crossover colonies were selected from low-salt LB agar plates supplied with 500 μg mL⁻¹ kanamycin and verified by colony PCR. Next, the FRT-nptII-FRT cassette was flipped out by introducing a pFlpeTet vector into kanamycin-resistant BtΔnosP::FRT-nptII-FRT through biparental mating using RHO3 as a vector donor strain.^26,27 The pFlpeTet vector and the RHO3 strain were generously gifted by Prof. Peggy Cotter’s Laboratory at the University of North Carolina. Next, the pFlpeTet vector was cured at 42 °C, and final colonies showing sensitivity to both kanamycin and tetracycline were selected. The operon of those colonies was PCR-amplified and sequenced to confirm the identity of the BtΔnosP::FRT strain. The ultimate result of nosP in-frame deletion is that a 117 base-pair-long dummy sequence containing an FRT scar replaces the actual nosP gene. Therefore, the deletion strain was named BtΔnosP::FRT. BtΔnahK::FRT strain was generated the same way by incorporating a KpnI cutting site in the middle of the fused flanking regions instead. After deletion, an FRT scar replaced the nosP or nahK sequences. The beginning and the ending three amino acids (including start and stop codons) of nosP and nahK are maintained, and primer lengths were carefully examined to avoid frameshifting. The complement strains were made by the mini-Tn7 insertion method.²⁸ Sequences of nosP and nahK were fused with a constitutively expressed promoter of rpsL gene by overlapping extension PCR. The resulting fused DNA was cloned into pUC18R6K-mini-Tn7T-Km, and the resulting vectors were transformed into RHO3. The helper plasmid pTNS3, which carries the transposase gene, was transformed into RHO3 as well. The mini-Tn7 fragments carrying the complement sequences were subsequently inserted into the intergenic region between glmS-1 and Bth_i0287 through triparental mating. Successful genomic insertion was verified by colony PCR, and the identification of BtΔnosP::FRT/Comp and BtΔnahK::FRT/Comp was confirmed by sequencing.

Static Biofilm Assay.

A modified microtiter dish assay was performed in 96-well plates to quantify surface-attached biofilms developed in the static condition.²⁹ To study the effect of NO, overnight cultures of wild-type E264 and nahK transposon disruption strains were grown in AB media (2 g L⁻¹ (NH₄)₂SO₄, 6 g L⁻¹ Na₂HPO₄, 3 g L⁻¹ KH₂PO₄, 3 g L⁻¹ NaCl, 100 μM CaCl₂, 1 mM MgCl₂, 3 μM FeCl₃, 0.2% glucose, 0.5% casamino acid). The next morning, overnight cultures were diluted into fresh AB media supplied with or without 50 μM DPTA-NONOate to an OD₆₀₀ of 0.05. Then, 180 μL/well diluted cultures were pipetted into the 96-well plate. To avoid the edge effect, peripheral wells were skipped from experimental groups, instead, filled with fresh AB media to prevent overevaporation during the incubation. The 96-well plates were incubated at 37 °C statically for 16 h. After 16 h, OD₆₀₀ of the planktonic cells was recorded using Victor X PerkinElmer Plate Reader. The rest of the planktonic cells were dumped out, and the plate was rinsed with milli-Q water. Then, the plate was dried and stained by 200 μL/well 0.1% crystal violet (CV) solution for 15 min. Excess crystal violet was rinsed off with milli-Q water. The plate was then allowed to dry for 1 h, after which 200 μL of 30% acetic acid was added per well to solubilize the crystal violet for 15 min with shaking. Then, CV₅₇₀ was recorded on a Victor X PerkinElmer Plate Reader. CV₅₇₀ was normalized to OD₆₀₀ to minimize the variation caused by the difference of planktonic growth, though OD₆₀₀ was usually consistent from group to group. To study the effect of heme, overnight cultures of wild-type E264, ΔnosP::FRT, ΔnahK::FRT, and complement strains were started in LB media. The next day, cells were pelleted, washed with iron-depleted AB media, and resuspended in the iron-depleted AB media supplied with 200 μM 2,2-dipyridyl (DIP) to an OD₆₀₀ of 0.5. Then, the resuspended cells were allowed to be shaken at 37 °C for 3 h to deplete their intracellular iron content. Then, the cultures were diluted into iron-depleted AB media supplied with 2.5 or 5 μM hemin to an OD₆₀₀ of 0.05, and 180 μL/well diluted cultures were pipetted into the 96-well plate. The plates were first kept at 37 °C for 36 h with a brief shaking every 1 h interval, then left at room temperature statically for another 36 h. The quantification of biofilms is the same as described above. The results are the average of four biological replicates. Students’ t-test was performed on GraphPad Prism.

RESULTS AND DISCUSSION

BtNosP is a Hemoprotein.

We characterized the electron absorption spectra of BtNosP in various oxidation and ligation states, which collectively demonstrate that BtNosP is a hemoprotein. Hemoproteins have a high extinction absorbance peak between 380 and 500 nm, called a Soret band, as well as absorbances in the range of 500–750 nm, generally called Q-bands.³⁰ These spectral properties are highly sensitive to the ligation and oxidation states of the heme cofactor. Purified apo-BtNosP does not have a Soret peak without the addition of exogenous hemin (Figure 1A, green solid line).

Figure 1.

(A) UV–vis spectroscopy characterization of tag-free BtNosP (~6 μM) in various oxidation and ligation states. Purified HMBP-tagged BtNosP was subjected to tag removal by TEV digestion and subsequently reconstituted by 10× hemin (hemin/protein ratio = 10) incubation. Excess hemin was removed by gel filtration. Each complex was made as described in the text. (B) Holo-Fe^III-BtNosP (~0.5 μM) spectrum obtained after 0.5× hemin (hemin/protein ratio = 0.5) incubation and gel filtration. A shoulder peak appears at 410 nm, suggesting that multiple heme states are present in holo-Fe^III-BtNosP.

To study the spectral properties of purified apo-BtNosP, we enzymatically removed the HMBP tag used for purification and reconstituted apo-BtNosP with 10× exogenous hemin, followed by gel filtration to remove excess hemin. This reconstituted holo-BtNosP was oxidized with potassium ferricyanide to yield holo-Fe^III-BtNosP. The ferric complex has a broad Soret peak at 383 nm, consistent with a high-spin 5-coordinate heme iron (Figure 1A, blue solid line). Proteins with 5-coordinate ferric heme iron have been extensively reported. For example, the myoglobin H93C mutant,³¹ and the human heme oxygenase-1 H25C mutant,³² which both utilize a cysteine ligand, exhibit broad Soret peaks around 383 nm. The secreted hemophore Rv0203 from Mycobacterium tuberculosis,³³ the outermembrane heme receptor HmbR from Neisseria meningitidis,³⁴ and the heme oxygenase-1 H25Y mutant,³² are oxygen-ligated pentacoordinate hemoproteins that have Soret peaks around 390–400 nm.

While the broad 383 nm complex formed by holo-Fe^III-BtNosP may be explained by a 5-coordinate high-spin species as described above, we also wondered if the broad peak could be explained by a mixture of complexes with different Soret maxima. We were able to rule out non-protein-bound heme as a second species, however, as excess heme is removed by gel filtration prior to characterization by UV–vis spectroscopy. We hypothesized that in the presence of excess heme, two heme molecules could stack in the BtNosP heme-binding site, resulting in a complex with an absorbance at ~383 nm. This type of complex has been observed in the Yersinia pestis heme-binding protein HmuT.³⁵ With HmuT, incubation with excess heme results in a 2:1 heme/protein complex with a Soret maximum at 374 nm, but incubation with substoichiometric heme results in a 1:1 complex with a Soret maximum at 404 nm. Therefore, we also performed an experiment where BtNosP was incubated with 0.5× hemin (Figure 1B). Here, we indeed see evidence of two species, one with a Soret maximum at 383 nm, like the complex formed with excess heme, and another with a Soret maximum at 410 nm. Furthermore, during the hemin titration experiments reported below, the complex formed during titration with substoichiometric heme absorbs at 410 nm. Thus, multiple species are present in the holo-Fe^III-BtNosP complex, which may indicate more than one heme-binding site or more than one heme-bound protein conformation; in the presence of substoichiometric heme, a complex is produced with a Soret absorbance of ~410 nm, and a second protein-bound ferric heme species (Soret absorbance ~383–385 nm) is formed over time or in the presence of excess heme. That we still observe a significant 383 nm complex with 0.5× hemin leads us to believe it may not be the result of stacked heme but rather an initial heme complex that shifts over time to a different species, perhaps due to a change in the ligand coordination sphere or a protein conformational change. This second possibility may indicate that the heme-binding properties of BtNosP are like the M. tuberculosis hemoproteins MmpL3 and MmpL11. A kinetics study demonstrated that these proteins bind heme in two phases, a fast phase with a transient peak between 400 and 410 nm, and a slower phase resulting in the dominant and final product peak at 374 nm.³⁶ There are still many remaining questions about the heme-binding mechanism of BtNosP, which will be the subject of future studies.

Treatment of the ferric complex with the reducing agent sodium dithionite, anaerobically at neutral pH, results in only partial reduction of the heme cofactor, as indicated by a slightly red-shifted shoulder on the ferric 383 nm peak. Given that sodium dithionite is a stronger reductant under alkaline conditions,³⁷ we increased the pH to 10.5 and performed the anaerobic reduction again. These conditions result in a completely reduced form of Fe^II-BtNosP with a sharp red-shifted Soret peak at 426 nm, along with split α/β bands at 530 and 560 nm, respectively (Figure 1A, red solid line). The split α/β peaks of the ferrous complex indicate the presence of a 6-coordinate heme iron center, similar to what we have observed for reduced NosP complexes from other organisms.^12–16

Generation of the Fe^II-NO complex results in the formation of a broad peak at 389 nm, and a poorly resolved peak in the visible region, suggesting NO ligation leads to the dissociation of both axial ligands of the ferrous complex (Figure 1A, pink dashed line), resulting in a 5-coordinate species. This is consistent with many hemoproteins, as NO ligation results in the delocalization of electron density from the Fe dz² orbital toward the axial ligand in trans, which weakens the axial ligand bond strength and can lead to dissociation.³⁸

The addition of CO gas into the ferrous complex yields the Fe^II-CO complex, displaying a sharp Soret band at 421 nm and split α/β bands at 540 and 568 nm, consistent with the formation of a 6-coordinate species (Figure 1A, black dot dash line). The spectrum of the Fe^II-CO complex is similar to many histidine-ligated hemoproteins, such as CooA,³⁹ and also similar to the CO-ligated myoglobin H93G mutant bearing a neutral thiol donor in trans of CO.⁴⁰ We are not able to determine whether the heme iron is coordinated by a histidine imidazole nitrogen, a cysteine thiol, an amino acid backbone element, or H₂O from the UV–vis spectra (Table 1). More characterization such as X-ray absorption spectra and magnetic circular dichroism are required to probe the coordination property of the heme iron.

Table 1.

Summary of UV–Vis Peak Positions

sample	wavelength (nm)
Apo	not detected
ferric	383
ferrous	426, 530, 560
ferrous-CO	421, 540, 568
ferrous-NO	389

Holo-BtNosP Has a Slow NO Dissociation Rate Constant.

As NO sensors that are predicted to sense pico-to nanomolar concentrations of NO, to date, all characterized NosPs from bacterial organisms have slow NO dissociation rate constants.^12,14,15,41 Thus, next, we measured the NO dissociation kinetics of BtNosP using the standard CO/dithionite trap method (Figure 2A,B).⁴² Here, we report the k_slow [(1.73 ± 0.6) × 10⁻⁴ s⁻¹] of the two-phase dissociation kinetics as the NO dissociation rate constant (Table 2), since this is the rate-determining step. However, the actual molecular steps of NO dissociation remain to be elucidated. This NO dissociation rate constant is independent of dithionite concentrations and CO (Figure S1). The measured k_fast value was (1.98 ± 0.49) × 10⁻³ s⁻¹; it contributed less than 35% of the total change in absorbance. There are several possible explanations for two observed kinetic rate constants. One possibility is that the 5-coordinate Fe^II-NO BtNosP is a mixture of distally and proximally coordinated complexes, each with a different NO dissociation rate constant (Figure S2A). Consistent with this possibility, excess DEA-NONOate was utilized to generate the NO complex in this experiment, which has been reported to lead to a mixture of geometries in the H-NOX family of NO-sensitive proteins.⁴³ It is also possible that k_fast can be attributed to the ligation of an unknown ligand (amino acid or water) after NO dissociation and before CO association (Figure S2B). Nevertheless, the NO dissociation rate constant of BtNosP is comparable to other heme-based NO sensors, including H-NOX proteins and other NosP family members (Table 2). We anticipate the NO association rate constant of BtNosP to be ~10⁸ M⁻¹ s⁻¹, the diffusion limit of NO, as is the case with other NosP proteins, which should yield a theoretical $K_{\mathrm{D}}$ for NO around picomolar, consistent with a NO-sensing function for BtNosP.

Figure 2.

NO dissociation kinetics of BtNosP using the standard dithionite/CO trap method⁴² at 23 °C. A CO-saturated sodium dithionite solution was injected into the Fe^II-NO complex of BtNosP (final sodium dithionite concentrations of 0.3, 3, 15, and 30 mM were tested individually; the NO dissociation rate constant was not affected by dithionite or CO concentration). The formation of the Fe^II-CO peak at 421 nm and the disappearance of the Fe^II-NO peak at 389 nm of BtNosP were monitored for 3 h. (A) Representative example of ΔAbsorbance versus wavelength during the 3 h experiment, the ΔAbsorbance spectra were obtained by subtracting the absolute absorbance spectrum at t = 0 from subsequent scans collected from the following data points. (B) Two-phase association exponential fit $\left[A\left(t\right)=A_1\left(1-{\mathrm{e}}^{-k_{1}t}\right)+A_2\left(1-{\mathrm{e}}^{-k_{2}t}\right)\right]$ (eq 1) of ΔΔAbsorbance versus time. ΔΔAbsorbance values were calculated by subtracting ΔAbsorbance values at 390 nm from those at 420 nm. The average of k_slow measured with CO-saturated 3 mM sodium dithionite trap is reported as the final NO dissociation rate constant. BtNosP displays a NO dissociation rate constant that is comparable to other well-established heme-based NO sensors, including other NosP family proteins, H-NOX family proteins, and the mammalian NO sensor soluble guanylate cyclase (sGC).

Table 2.

NO Dissociation Rate Constants of Selected Hemeprotein

protein	k−NO (10−4 s−1)	reference
sGC	3.6±0.9	44
VhH-NOX	4.6±0.9	45
SwH-NOX	15.2±3.5	46
SoNosP	2.25±0.5	14
PaNosP	1.8±0.5	12
VcNosP	4.6±0.1	15
BtNosP	1.73±0.60	this work

Holo-BtNosP Has a Fast Heme Dissociation Rate Constant.

BtNosP is purified predominately in the heme-free apo form. Thus, we investigated its heme-binding properties. Upon reconstitution of holo-BtNosP, the heme dissociation kinetics were investigated with the standard apo-myoglobin trap⁴⁷ (Figure 3A,B), an approach that has been utilized previously to estimate the apparent heme dissociation rate constants of heme oxygenase-2,⁴⁸ MhuD,²⁰ IsdA,⁴⁹ IsdC,⁴⁹ IsdG,⁵⁰ and IsdI.⁵⁰ Since apo-myoglobin exhibits extremely high affinity for heme and a rapid heme association rate constant (k_on ~ 1 × 10⁸ M⁻¹ s⁻¹),⁵¹ apo-myoglobin should rapidly sequester heme upon dissociation from BtNosP, such that the observed kinetic rate constant can be interpreted as the heme dissociation rate constant, k_−heme, of BtNosP. In these experiments, apo-myoglobin was used at ≥5× the concentration of holo-BtNosP to measure the pseudo-first-order heme dissociation rate constant. The resulting data was best fit with a two-phase exponential model, resulting in two measured rate constants (2.98 ± 0.60) × 10⁻³·s⁻¹ (~70%, k_slow) and (1.48 ± 0.19) × 10⁻²·s⁻¹ (~30%, k_fast). Two rate constants for heme dissociation were also reported for rGAPDH⁵² and MhuD²⁰ using the same apo-myoglobin assay. For rGAPDH, K_D and k_on for heme binding were also independently determined, and when the authors calculated k_−heme of rGAPDH using K_D*k_on, the theoretical k_−heme was closer to k_slow from the two-phase model.⁵² Therefore, we report k_slow as the k_−heme of BtNosP. This rate constant is around 10-fold faster than that of Vc_0130 (3.0 × 10⁻⁴·s⁻¹), a NosP domain we previously characterized from V. cholerae, which was proposed to have a heme-sensory function.¹⁶ This k_−heme for BtNosP heme dissociation is similar to that of the heme-regulated eukaryotic initiation factor 2α HRI (1.5 × 10⁻³·s⁻¹),⁵³ a predicted heme-sensor that regulates the translation of globin in response to heme sufficiency in erythrocytes.⁵⁴ In comparison to obligate heme proteins such as soluble guanylate cyclase (sGC)⁵⁵ and myoglobin,⁵¹ the heme dissociation rate constant of BtNosP is much faster, suggesting that BtNosP may not bind heme as tightly as obligate heme proteins (Table 3).

Figure 3.

Heme dissociation kinetics measurement using a standard apo-myoglobin trap assay⁴⁷ at 23 °C. Apo-myoglobin was injected into the Fe^III holo-BtNosP solution at a final apo-myoglobin/holo-BtNosP ratio of 5:1. The formation of holo-myoglobin peak at 409 nm was monitored for 30 min. (A) Representative example of ΔAbsorbance versus wavelength monitored for 30 min. ΔAbsorbance values were obtained by subtracting the first scan collected at t = 0 min from all of the subsequent spectra. (B) Two-phase exponential fit $\left[A\left(t\right)=A_1\left(1-{\mathrm{e}}^{-k_{1}t}\right)+A_2\left(1-{\mathrm{e}}^{-k_{2}t}\right)\right]$ (eq 1) of ΔAbsorbance at 409 nm over time. The slower rate constant from the two-phase fitting model (k_slow) was reported as the final k_−heme of BtNosP.

Table 3.

Kinetic Parameters for Binding of Heme to Selected Proteins

protein	kon (μMY−1 s−1)	k−heme (s−1)	KD (μM)	reference
BtNosP	5.6 × 10−3 − 2.8 × 10−2(k−heme/KD)	(2.98 ± 0.60) × 10−3	(5.3 ± 1.2) × 10−1	this work
		(1.48 ± 0.19) × 10−2
Vc_0130 (CdpA)	ND a	3.0 × 10−4	ND a	16
PaPhuS	1.8 × 10−1	3.6 × 10−2(KD*kon)	2.0 × 10−1	56
rGAPDH	1.78 × 10−2	3.3 × 10−4	1.9 × 10−2 − 3.9 × 10−1(k−heme/kon)	52
		7.0 × 10−3	2.4 × 10−2 (direct measurement)
HRI	11	1.5 × 10−3	1.4 × 10−4(k−heme/kon)	53
MhuD (first heme binding)	14–33(k−heme/KD)	2.53 × 10−1	7.6 × 10−3	20,57
		1.03 × 10−1
myoglobin	76	8.4 × 10−7	1.1 × 10−8(k−heme/kon)	51
sGC	ND a	8.3 × 10−6b	ND a	55

^aND, not determined.

^bFerric heme dissociation rate; ferrous heme does not dissociate upon apo-myoglobin challenge.

Thermodynamic Heme Dissociation Constant for BtNosP Is Similar to Labile Heme-Binding Proteins.

We determined the stoichiometry of heme binding to BtNosP by titrating heme into 5 μM apo-BtNosP using double-beam absorption spectroscopy so that the absorbance of non-protein-bound heme in the reference cell is automatically subtracted from the sample cell that contains a mixture of protein and heme solution. The dominant peak is at ~410 nm; this peak increases with increasing heme concentration and reaches saturation when the concentration of hemin added is ~5 μM, suggesting hemin binds to apo-BtNosP in a 1:1 ratio (Figure 4A). The peak at 410 nm matches the shoulder peak we observe in the UV–vis data in the presence of 0.5× heme (Figure 1B), consistent with our hypothesis that the 410 nm species is favored by BtNosP under a low hemin/protein ratio. However, we did not observe the peak shift from 410 to 383 nm during this experiment, despite a heme/protein ratio of ~4.5:1 at the end of the titration. Failure to observe the peak shift from 410 to 383 nm could be due to not enough excess hemin, an insufficient time of incubation with excess heme, or it is possible that a small peak at 383 nm could be masked due to subtraction of the large free hemin absorbance in the double-beam experiment. Again, the mechanism of heme binding will be the subject of future studies in our laboratory.

Figure 4.

(A) Stoichiometry of heme binding to BtNosP was determined by titrating increasing concentrations of hemin into 5 μM apo-BtNosP (sample cell) and 50 mM Tris·HCl, 300 mM NaCl buffer (reference cell) in a double-beam mode absorbance experiment. Difference absorbance was automatically calculated by subtracting the absorbance of the reference cell from that of the sample cell. The titration reached a plateau at ~5 μM hemin, suggesting BtNosP binds heme in a 1:1 ratio. (B) Intrinsic tryptophan fluorescence signal of 1 μM apo-BtNosP was quenched by an increasing amount of hemin. The intrinsic tryptophan fluorescence of BtNosP was excited at 283 nm, and aliquots of hemin stock solution (1–10 μL) were titrated into the protein solution, incubating for 1000 s, and then emission signals at 328 nm were recorded at each titration point. The relative fluorescence intensity was calculated by normalizing the absolute fluorescence intensity at each titration point to the absolute fluorescence intensity obtained from the first titration point where [Hemin] = 0 nM. Relative fluorescence intensity as a function of hemin concentrations was fitted into eq 2. Final reported K_D is an average of four biological replicates.

Next, we used an intrinsic tryptophan fluorescence quenching assay to measure the heme dissociation constant (K_D) for apo-BtNosP, resulting in a K_D(heme) of (0.53 ± 0.12) μM (Figure 4B and Table 3). This heme-binding affinity for BtNosP is similar to PaPhuS, a cytoplasmic heme shuttling protein that has reported K_D values of 0.2⁵⁶ and 0.41 μM.⁵⁸ The K_D(heme) of BtNosP is around 20-fold larger than that of rGAPDH (0.024 μM),⁵² a labile heme chaperon that allocates heme to target proteins in cells.⁵⁹ Notably, in comparison to the K_D(heme) for obligate heme-binding proteins MhuD, a noncanonical heme oxygenase from M. tuberculosis,⁵⁷ and myoglobin,⁵¹ the K_D(heme) of BtNosP is much larger (Table 3), indicating BtNosP does not bind heme as tightly as typical obligate heme-binding proteins. Taken together, these kinetic and thermodynamic parameters suggest that BtNosP is capable of binding labile heme reversibly, underscoring its potential function of labile heme-sensing in vivo.

BtNahK Exhibits Autokinase Activity and Transfers a Phosphoryl Group to BtNarR.

NosP genes are frequently co-cistronic with genes encoding two-component systems and/or c-di-GMP-metabolizing enzymes.^12–16 In B. thailandensis, nosP is encoded in the same operon as a histidine kinase nahK and a degenerate “HD-GYP”-type c-di-GMP phosphodiesterase response regulator narR. We hypothesized that BtNosP-mediated signal transduction might proceed through this two-component system. Therefore, we characterized the autophosphorylation activity of the histidine kinase BtNahK. We first performed a time-dependent BtNahK autophosphorylation assay using trace [γ-³²P] ATP visualized by phospho-imagery. We were able to detect an increasing amount of [γ-³²P] labeled BtNahK as a function of time, indicating that BtNahK has autophosphorylation activity, as expected (Figure 5A,B). Once we confirmed that BtNahK exhibits autokinase activity, we proceeded to study whether the receiver domain of the response regulator BtNarR can engage in phosphoryl transfer with BtNahK. A similar [γ-³²P] ATP phosphoryl-profiling experiment was performed by mixing prephosphorylated BtNahK with BtNarR. Our results indicate that a ³²P-labeled phosphoryl group is transferred from BtNahK to BtNarR in a time-dependent manner (Figure 6A,B).

Figure 5.

(A) Radiography (top) and Coomassie blue stained gel (bottom) images of BtNahK autophosphorylation in the presence of trace [γ-³²P]-ATP monitored during a time course of 90 min. Lanes 1–8 represent autophosphorylation reaction samples that are quenched at 0, 5, 10, 20, 30, 45, 60, and 90 min. (B) Plot of relative intensity of BtNahK autophosphorylation as a function of time by normalizing the intensity of autoradiography bands to the intensity of Coomassie blue stained gel bands. Increased BtNahK autophosphorylation was observed as a function of time, suggesting BtNahK is an active histidine kinase. Experiments were performed in triplicate and representative data is shown.

Figure 6.

(A) Radiography (top) and Coomassie blue stained gel (bottom) images of phosphoryl profiling of BtNahK to BtNarR with trace [γ-³²P]-ATP during a time course of 60 min. Lane 1, BtNahK control; lanes 2–8, BtNahK/BtNarR phosphoryl transfer reaction samples quenched at 0, 2, 10, 15, 30, and 60 min; lane 9, BtNarR control. (B) Plot of relative intensities of BtNahK and BtNarR as a function of time by normalizing the intensity of autoradiography bands to the intensity of Coomassie blue stained gel bands. A ³²P-labeled phosphoryl group is transferred from BtNahK to BtNarR in a time-dependent manner. Experiments were performed in triplicate and representative data is shown.

To determine the amino acid residues involved in this phosphoryl transfer, site-directed mutagenesis was used to generate point mutants at the predicted sites of phosphorylation on BtNahK (H109A) and BtNarR (D89A). Our results indicate that BtNarR is not phosphorylated in the D89 mutant, confirming its role in receiving a phosphoryl group from BtNahK. Similarly, the H109A mutant of BtNahK does not autophosphorylate. Successful phosphoryl transfer from BtNahK to BtNarR only occurs in the presence of both wild-type proteins; mutation of BtNahK H109 and/or BtNarR D89 completely disrupts the phosphoryl transfer pathway (Figure 7).

Figure 7.

Radiography (top) and Coomassie blue stained gel (bottom) images of the phosphoryl transfer pathway. H109A mutant of BtNahK does not autophosphorylate (lanes 4 and 5); phosphoryl transfer from wild-type BtNahK to D89A mutant of BtNarR (lane 3) and from H109A mutant of BtNahK to wild-type BtNarR (lane 4) were disrupted; phosphoryl profiling was able to proceed only in the presence of both wild-type BtNahK and BtNarR (lane 2). Experiments were performed in triplicate and representative data is shown.

Holo-BtNosP Inhibits BtNahK Autophosphorylation.

Since BtNosP and BtNahK are in the same operon, we hypothesized that BtNosP and BtNahK interact, and that BtNahK activity might be modulated by BtNosP. Indeed, similar studies conducted in our laboratory have revealed that NosP regulates its co-cistronic histidine kinase or c-di-GMP-metabolizing enzymes in response to environmental cues such as NO and heme.^12–16 To evaluate the effect of BtNosP on BtNahK autophosphorylation, apo-BtNosP and holo-BtNosP in different oxidation and ligation states were incubated with BtNahK and trace [γ-³²P]ATP, after which the reaction was quenched and the extent of autophosphorylation was quantified. We found that BtNahK autophosphorylation is strongly inhibited by holo-BtNosP, with the Fe^III complex having the most significant inhibitory effect. Apo-BtNosP, however, does not seem to influence autophosphorylation (Figure 8). This is further supported by a titration experiment, where we observed a dose-dependent inhibition of BtNahK autophosphorylation by titrating Fe^III holo-BtNosP (Figure 9A) but not by titrating even a large excess of apo-BtNosP (Figure 9B). Heme-mediated inhibition of associated enzyme activity was also observed in Vc_0130, in which heme binding to the N-terminal NosP domain regulates the C-terminal PDE domain activity.¹⁶ Interestingly, the BtNosP Fe^II-NO complex restores BtNahK activity relative to the BtNosP Fe^II complex, suggesting a possible NO detection function for the ferrous NosP protein. The Fe^II-NO complex does, however, inhibit autokinase activity mildly in comparison to kinase alone or kinase with apo-BtNosP (Figure 8). Using these data alone, it is difficult to conclude if BtNosP is primarily a sensor for heme, NO, and/or both.

Figure 8.

(A) Radiography (top) and Coomassie blue stained gel (bottom) images of BtNahK autophosphorylation inhibited by BtNosP. 40 μM His₆-BtNosP was incubated with 5 μM HMBP-BtNahK in the presence of trace [γ-³²P]-ATP for 90 min. (B) Relative extent of autophosphorylation of BtNahK in the presence of BtNosP was calculated by first normalizing the radiography band intensity of BtNahK to the Coomassie blue band intensity, followed by normalizing all groups to the relative autophosphorylation intensity of BtNahK alone control. The BtNahK alone control was set up as a standard to be compared with other groups. Statistical analysis was performed by one-way analysis of variance (ANOVA), using an average of four biological replicates.

Figure 9.

Radiography images and relative quantification of autophosphorylation extent of 3 μM BtNahK titrated by (A) 0–80 μM Fe^III-BtNosP and (B) 0–100 μM apo-BtNosP. Dose-dependent autophosphorylation inhibition effect was observed by incubating BtNahK with Fe^III-BtNosP but not apo-BtNosP. The plots shown represent an average and the standard deviation of three biological replicates.

Interestingly, dual regulation by heme and NO has been observed in heme-regulated eukaryotic initiation factor 2α (HRI).⁶⁰ Under conditions of heme deficiency, HRI is an active kinase that phosphorylates eukaryotic initiation factor 2 (eIF2). Phosphorylated eIF2 subsequently inhibits the translation of globin to prevent globin accumulation in red blood cells. When heme pools are sufficient, HRI becomes inhibited, allowing the translation of globin and consequently the assembly of hemoglobin.⁵⁴ NO ligation to HRI’s N-terminus heme-binding domain leads to the activation of both HRI autokinase and eIF2 kinase activities, suggesting NO and heme deficiency serve as activators of HRI.⁶¹ Our biochemical data seem to suggest a model of BtNosP/BtNahK that is homologous to HRI, with possible dual regulation by heme and NO.

Holo-BtNosP Binds BtNahK More Tightly Than apo-BtNosP.

The observation that holo-BtNosP and apo-BtNosP affect BtNahK autophosphorylation differently led us to speculate that heme binding may impact the binding affinity between BtNosP and BtNahK. To test this postulation, we designed a pull-down assay using amylose resin-immobilized HMBP-tagged BtNahK as a bait protein and His₆-tagged BtNosP as a prey protein. The amount of prey protein pulled down by the bait protein should be proportional to the relative affinity of each bait/prey pair. The trend of relative affinity between BtNosP and BtNahK (Figure 10B) was opposite to the extent of autophosphorylation inhibition in each pair (Figure 8B), with the holo-Fe^III complex displaying the highest affinity for BtNahK and the apoprotein exhibiting the lowest relative affinity for BtNahK (Figure 10A,B). To confirm that His₆-BtNosP was pulled down due to its interaction with BtNahK and not the MBP tag, a control experiment was performed using His₆-MBP as the bait protein. As expected, no forms of BtNosP were pulled down by the His₆-MBP tag alone (Figure S3). The pull-down data is in strong agreement with the autophosphorylation inhibition data, demonstrating that BtNosP inhibition of BtNahK activity is a function of binding affinity. The pull-down data, together with the autophosphorylation inhibition data, suggest that BtNahK activity is regulated directly by BtNosP, likely in response to labile heme and/or intracellular NO concentration.

Figure 10.

(A) Representative example of a Western blot image of the pull-down assay. His₆-BtNosP was pulled down by HMBP-BtNahK immobilized on amylose resin. Images were developed by anti-His₆ antibody, HRP conjugate, and luminol enhancer to display chemiluminescence. The absolute intensity of prey His₆-BtNosP (background-subtracted) and bait HMBP-BtNahK was quantified by ImageJ. Then, BtNosP band intensity was normalized to that of BtNahK. (B) Relative affinity of BtNosP to BtNahK. The highest relative affinity (Fe^III complex) was set up as a standard to be compared with other groups. Statistical analysis was performed using three biological replicates, one-way ANOVA GraphPad Prism.

BtNosP/BtNahK Pathway is Heme-Sensitive and Essential for B. thailandensis Biofilm Formation.

Since NosP/NahK in P. aeruginosa, S. oneidensis, and L. pneumophila all contribute to biofilm regulation in response to NO, and the S. oneidensis ΔnosPΔnahK double deletion strain fails to form mature biofilms,^12–14 we hypothesized that the BtNosP/BtNahK pathway is also involved in biofilm regulation in B. thailandensis. To first answer the question of whether NO modulates B. thailandensis biofilms through the BtNosP/BtNahK pathway, we performed a static biofilm assay with and without the addition of 50 μM DPTA-NONOate using wild-type and nahK transposon disruption strains of B. thailandensis.²³ To our surprise, B. thailandensis biofilms are unaffected by NO treatment. However, the nahK transposon mutation strain consistently produces less biofilm than the wild-type strain (Figure 11). This is similar to what we have seen in S. oneidensis, where a mutant strain lacking nosP/nahK cannot form mature biofilms.¹⁴ These observations suggest that in our experimental conditions, NO is not modulating B. thailandensis static biofilms through the BtNosP/BtNahK pathway, and that nahK is important for biofilm formation.

Figure 11.

Wild-type and nahK transposon disruption (nahK::Tn23) strains were grown in AB media for 16 h at 37 °C in the presence and absence of 50 μM DPTA-NONOate. Surface-attached biofilms were stained using 0.1% crystal violet solution and then solubilized in 30% acetic acid. The amount of attached biomass was quantified by measuring the absorption of the crystal violet (CV) stain at 570 nm and normalized to planktonic cell optical density (OD) measured at 600 nm.

NO typically has a concentration-dependent effect on cells. In bacteria, at low concentrations (ñM), NO acts as a signaling molecule to eradicate bacterial biofilms without affecting the planktonic growth.⁶² At high concentrations (>μM), NO kills bacteria by generating reactive nitrogen species and causing DNA and other cellular damages. B. thailandensis and B. pseudomallei generate NO endogenously through denitrification, a cellular respiration process under low oxygen tension where nitrate and nitrite are utilized as alternative terminal electron acceptors.⁶³ A previous study has shown that the nosP/nahK/narR operon of B. pseudomallei is upregulated upon nitrate and nitrite exposure, and the upregulation by nitrate is dependent on the NarX/NarL nitrate-sensing two-component system.⁶⁴ This suggests that nosP/nahK/narR operon of B. pseudomallei is NO-sensitive, which is not what we found in B. thailandensis, at least with respect to NO regulation of biofilm formation.

Several theories could explain this possible discrepancy. First, B. thailandensis may be intrinsically more resistant to NO and NO-generating species than B. pseudomallei. Indeed, B. pseudomallei is sensitive to NO-dependent killing, whereas B. thailandensis is less sensitive under the same conditions.¹⁸ It has been shown that NaNO₃ inhibits B. pseudomallei pellicle biofilms completely at 10 mM;⁶⁵ however, we have seen that B. thailandensis is still capable of forming pellicle biofilms at 20 mM NaNO₃ (data not shown). Second, it is possible that NO might not change the total biomass of the biofilm attached to the surface in our static assay, or that the total biomass changes are too small to be detected in the static biofilm assay, but instead, it may affect the structure of the mature biofilm. For instance, we have observed that the flow-cell biofilms of nosP and nahK mutants in S. oneidensis fail to form mature mushroom-like structures without significantly changing the total biomass.¹⁴

Next, we moved on to study whether the BtNosP/BtNahK pathway is heme-sensitive. To rule out any potential polar effects caused by transposon insertion, we generated in-frame deletion strains of BtΔnosP::FRT and BtΔnahK::FRT (Figure S7), as well as two complement strains BtΔnosP::FRT/Comp and BtΔnahK::FRT/Comp. The wild-type, deletion strains, and complement strains were predepleted with intracellular iron by DIP treatment, then 2.5 and 5 μM hemin were supplied as the sole iron source in a static biofilm assay. A 0 μM hemin supplied condition was also included as a control. However, complete lack of an iron source results in a severe growth inhibition effect for all tested strains, thus making it difficult to determine differences between strains (Figure S4). However, comparing increasing hemin concentration (from 2.5 to 5 μM), we observed that the wild-type strain formed about half as much biofilm in 5 μM hemin in comparison to the 2.5 μM hemin condition, despite almost identical growth (Figure 12).

Figure 12.

Wild-type, two in-frame deletion strains BtΔnosP::FRT and BtΔnahK::FRT, two complement strains BtΔnosP::FRT/Comp and BtΔnahK::FRT/Comp were grown in iron-depleted AB media supplied with 2.5 or 5 μM hemin for 36 h at 37 °C plus 36 h at room temperature. Surface-attached biofilms were stained using 0.1% crystal violet solution, then solubilized in 30% acetic acid. The amount of biofilm was quantified by measuring the absorption of crystal violet (CV) stain at 570 nm and normalized to planktonic cell optical density (OD) measured at 600 nm.

The heme-responsive phenotype is not significant in ΔnosP::FRT and ΔnahK::FRT strains. Consistent with the nahK transposon mutation strain, the ΔnahK::FRT also forms less biofilm than wild-type under all conditions (Figure 12). Heme as an iron source could also be broken down by cellular heme oxygenases to release free iron. It has been reported that iron also regulates bacterial biofilm formation.⁶⁶ To confirm that B. thailandensis is sensing heme rather than iron through the NosP/NahK signaling pathway, we compared the biofilm response of wild-type and knock-out strains grown with gradient hemin and FeCl₃ concentrations (Figure S5). Wild-Type biofilm is inhibited by hemin, but not FeCl₃, in a dose-dependent manner, highly suggesting assimilated heme rather than iron is the signaling molecule (Figure S5A,B). The heme-regulatory effect is abolished in ΔnosP::FRT and ΔnahK::FRT strains (Figure S5C,D) but restored in the complement strains (Figure S5E,F).

Decreased biofilm in strains lacking nahK is consistent with our biochemical data reported above. BtNosP acts as an inhibitor of BtNahK in the presence of heme (Figures 8 and 9). When the cellular labile heme concentration increases, the abundance of cellular holo-NosP likely also increases, thus leading to a greater inhibitory effect on BtNahK (Figure 9A). This heme-driven kinase activity inhibition is associated with a decreased production of biofilm in the wild-type, consistent with a biofilm formation defect in the nahK null mutation (Figure 12).

Interestingly, however, we have noticed the biofilm phenotype in the ΔnosP::FRT strain was not consistent in every replicate. When setting up the 96-well plate biofilm assay, technical replicates of each strain were pipetted into the plate by subculturing from the same parental liquid culture. Although the ΔnosP::FRT strain always produced less biofilms than wild-type, biofilm-positive ΔnosP::FRT technical replicates stochastically appeared, which may explain why this strain has a relatively large standard deviation in this biofilm assay (Figures 12 and S6). Regardless, complementing nosP and nahK back into the in-frame deletion strains has restored both the biofilm formation phenotype and the heme-driven biofilm inhibition phenotype (Figure 12).

The observation that lacking nosP alone can result in either biofilm-on or biofilm-off led us to hypothesize that other factors may also contribute to regulating BtNahK activity. One of the possibilities is that the BtNosP/BtNahK pathway might belong to a more complicated multikinase network (MKN). The NosP/NahK networks in both S. oneidensis and P. aeruginosa are members of MKNs. In S. oneidensis, SoNosP/SoNahK and SoH-NOX/SoHnoK are integrated into a NO-responsive MKN by phosphorylating the same response regulators HnoB (EAL phosphodiesterase), HnoC (transcriptional regulator), and HnoD (HD-GYP degenerate phosphodiesterase).^14,67,68 Furthermore, SoHnoK autophosphorylation is regulated by both SoH-NOX and SoNosP, both of which act as NO sensors.^14,67 It is therefore also possible that sensors other than BtNosP could regulate BtNahK function when the master regulator BtNosP is absent. In P. aeruginosa, PaNosP/PaNahK is just one member of a MKN, including four hybrid sensor kinases (PaNahK, PA1611, SagS, and RetS) that all feed phosphoryl groups into a histidine-containing phosphotransfer protein HptB.⁶⁹ Within this MKN, it has been shown that the various kinases are able to regulate one another by forming heterodimers. For example, PA1611 inhibits RetS activity through the formation of a PA1611-RetS kinase–kinase heterodimer.^70,71

In B. thailandensis, BtNarR serves as a phosphoryl receiver of BtNahK, but it is likely that other response regulators may also participate downstream of BtNahK to modulate its phosphorylation state. Indeed, biochemistry assays performed with homologous proteins purified from B. pseudomallei have revealed that NosP-associated kinase (BPSS1646) is integrated into a multikinase network with two other sensory kinases BPSS0813 and BPSL0703. It has been proposed that BPSS0813 and BPSL0703 are activated by unknown stimuli, then the activated BPSL0813 stimulates BPSS1646 autophosphorylation, possibly by heterodimer formation.⁷² Due to the high genome similarity between B. thailandensis and B. pseudomallei, a similar MKN likely operates in B. thailandensis to regulate BtNahK activity,^73–76 and the absence of NosP stochastically leads to regulation of NahK activity by other network interactions.

The model that has arisen from these studies for the role of NosP in B. thailandensis is that BtNosP is a labile heme-sensor. In vivo, it equilibrates between apo- and holo-forms depending on intracellular heme availability. Holo-BtNosP inhibits BtNahK autophosphorylation and results in decreased biofilm. Heme is a preferred iron source of many pathogenic bacteria.² Although B. thailandensis is not a virulent strain, its close surrogate B. pseudomallei can infect humans and invade blood-rich tissues. It has been shown that the inactivation of B. pseudomallei siderophore transport and heme uptake systems does not reduce virulence in the murine melioidosis model. However, mutant strains defective in siderophore/heme utilization demonstrate significantly lower organ burdens, especially in the lung and spleen, suggesting the spread of the pathogen among organs becomes attenuated under conditions of iron/heme deficiency.⁷⁷ This agrees with our observation that increasing extracellular heme concentrations drive B. thailandensis to switch from a more sessile growth phase to a more planktonic phase, which could lead to spread through the body and infection of more organs.

An intriguing part of this research is to understand the relationship between NO and heme in regulating the NosP signaling pathway. As indicated by our data, the BtNosP/BtNahK pathway is heme-sensitive, and kinetic parameters of BtNosP resemble both NO and heme-sensors. Despite no changes observed in the static biofilm assay in response to NO, we cannot rule out the possibility that BtNosP is regulated by both heme and NO in vivo. Our group has shown that in Vc_0130, a NosP family member protein, the C-terminal PDE activity is inhibited through heme binding to the N-terminal NosP domain.¹⁶ Follow-up research on Vc_0130 has revealed that this protein is essential for regulating V. cholerae cell detachment in response to NO, and prevention of heme binding by residue mutations also abolishes the NO-triggered detachment phenotype.⁷⁸ Recently, several studies have demonstrated that NO synthesis, heme-sensing, and protein insertion are intertwined. NO has been shown to block heme insertion into a variety of mammalian heme enzymes, including iNOS, eNOS, nNOS, CYP, and catalase.⁷⁹ NO also drives the maturation of its own eukaryotic receptor, sGC, with the assistance of a heme chaperon GAPDH. Heme-bound GAPDH rapidly detects and ligates NO generated by NOS, and GAPDH reallocates heme-NO to the apo-sGCβ-hsp90 complex. The incorporation of the heme-NO complex into sGCβ promotes the dissociation of hsp90 from, and the association of sGCα to, holo-sGCβ, generating a functional sGC heterodimer.⁸⁰ Here, we have demonstrated that NosP displays both NO and heme-sensing features. Our future experiments in this system include determining if NO is linked to heme binding by NosP.

CONCLUSIONS

In this study, we have presented data suggesting the BtNosP/BtNahK pathway is heme-sensitive and is essential for B. thailandensis biofilm formation. As a labile heme-sensor, increasing intracellular heme concentration switches BtNosP from the apo- to holo-form, consequently leading to inhibition on BtNahK autophosphorylation and decreasing biofilm. Biofilms, in many cases, confer antibiotic resistance of bacteria and cause issues such as latent infection and relapse of disease. Understanding the bacterial heme-regulatory machinery at the molecular level may provide a new insight into combating disease-causing bacteria by targeting their heme-sensing pathways. This study also underscores the complexity of NosP-mediated signaling systems, especially as dual-functional sensors for both NO and heme. In future work, we will seek to understand how NosP-mediated NO sensing affects heme homeostasis and to elucidate the molecular mechanism of holo-NosP maturation in vivo from multiple bacterial systems.

ABBREVIATIONS

NO

nitric oxide

CO

carbon monoxide

c-di-GMP

cyclic diguanylate monophosphate

PDE

phosphodiesterase

NahK

Nosp-associated histidine kinase

NarR

NosP-associated response regulator

IPTG

isopropyl β-d-1-thiogalactopyranoside

PMSF

phenylmethylsulfonyl fluoride

BSA

bovine serum albumin

DTT

dithiothreitol

BME

β-mercaptoethanol

TEV protease

tobacco etch virus protease

OD

optical density

CV

crystal violet

MBP

maltose-binding protein

Tris

tris-(hydroxymethyl)aminomethane

DEA-NONOate

diethylamine NONOate

DPTA-NONOate

dipropylenetriamine NONOate

FRT

flippase recombination site

SDS

sodium dodecyl sulfate

DIP

2,2-dipyridyl

PQS

pseudomonas quinolone signal