Introduction

The two-component system (TCS) plays an important role in bacteria signal transduction. It is mainly found in prokaryotes, and in a small number of eukaryotes, but it is rarely found in animals and plants and notably absent in human body^[1]. TCSs are involved in various signal transduction pathways such as environmental Pi concentration, pH, osmotic pressure, temperature, and so on^[2]. The typical TCS comprises a sensor histidine kinase (HK) and a response regulator (RR)^[3]. The signal transduction pathway is mediated by the phosphorylation and dephosphorylation between these proteins^{[4, 5]}. The Mg²⁺ is an essential component of the phosphorylation of numerous response regulators through coordination with the phosphoryl group and protein sidechains to stabilize the active conformation of RR, that is extremely conserved in many TCSs^{[6, 7]}.

Nuclear magnetic resonance (NMR) spectroscopy is powerful to explore the conformational dynamics of RRs in their apo state without Mg²⁺ and phosphoryl group. The ¹H-¹⁵N HSQC NMR spectrum provides a residue-specific fingerprint of the overall protein conformation. By monitoring changes in chemical shift and/or line broadening of previously assigned resonances in the protein spectrum, it would provide abundant information of conformation exchanges^[8]. Based on relaxation experiments, it has been characterized that NtrC has an exchange between inactive and active conformations, and both states are populated in unphosphorylated NtrC^[9]. However, the minor active conformation is hard to be visualized by traditional NMR experiments directly.

In order to visualize the minor active conformation preexisting in the apo state of RR directly, we use PhoB as a model. PhoB is a response regulator of the PhoR/PhoB signal transduction system from Escherichia coli and controls the expression of over 40 genes related to phosphate assimilation^{[10, 11]}. It consists of a N-terminal receiver domain (PhoB_N) that contains the conserved phosphorylation site and a C-terminal effector domain that performs the output function^[12-14]. The crystal structure of active and inactive PhoB_N have been reported, respectively^[13].

In previous studies, we have found that apo PhoB_N has monomer-to-dimer equilibrium^[13] and the dimer has multiple conformational changes in solution. The NMR spectrum of PhoB_N gives inhomogeneous signals and makes it hard to investigate. It has been reported that F20 stabilizes one of the dimer interfaces (α1-α5) through hydrophobic interactions with V21 and the aromatic rings of the Q24 side chain^[15]. We replaced F20 with an aspartate residue (PhoB_N^F20D) which disrupted the α1-α5 dimer interface, and demonstrated that PhoB_N^F20D becomes a monomer in the solution and has the same phosphorylation property as the wild type PhoB_N^[16]. The relaxation parameters of the backbone amide nitrogens of apo PhoB_N^F20D were measured and the overall rotation correlation time (τ_c) was calculated. It gives the value of 12.87 ns which corresponds to the molecular weight of the monomer. Therefore, the mutant PhoB_N^F20D can be used to explore the conformational exchange between active and inactive state simply at monomer state. Moreover, beryllium fluoride (BeF₃^-) is widely used as a phosphoryl analogue in macromolecular studies and it can be applied to stabilize the active conformation, meanwhile both bound and unbound BeF₃^- can be detected by ¹⁹F NMR directly^[17].

Here we analyze the interaction between PhoB_N^F20D and BeF₃^- with or without Mg²⁺ using NMR. We find that apo PhoB_N^F20D has a pre-active conformation in solution and BeF₃^- can stabilize this conformation without Mg²⁺. The minor pre-active conformation was visualized by ¹H-¹⁵N HSQC and ¹⁹F NMR spectra directly. Meanwhile, Mg²⁺ can shift the equilibrium between inactive and active conformation and makes PhoB_N^F20D completely become active conformation. The Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion experiments verify that there is an exchange between the inactive and active conformations of apo PhoB_N^F20D. Our results provide direct evidence of the existing of minor active conformation in the apo state of RR, and a better understanding of conformational exchange during the phosphorylation process of response regulators.

1 Experimental method 1.1 Expression of PhoB_N^F20D

The E.coli phoB gene was cloned into the pET-28a vector (Novagen) using the NdeI and XhoI restriction enzyme sites to express PhoB_N (1~125 residue) protein with thrombin-cleavable His₆ tag^[18] at the N-terminus. The PhoB_N^F20D mutant was produced by polymerase chain reaction (PCR). The constructed plasmid was transformed into E.coli BL21(DE3) strain for protein expression and purification. The expression cell stock was grown in LB medium containing 50 μg/mL kanamycin at 37 ℃ overnight. ¹⁵N-labeled PhoB_N^F20D was expressed in 500 mL M9 media supplemented with 0.5 g ¹⁵NH₄Cl (Cambridge Isotope Laboratories, Andover, MA, USA). When OD₆₀₀ of the culture reached to 0.7~0.8, the cells were induced by 1 mmol/L isopropyl thio β-D-1-galactopyranoside at 20 ℃ for 20~21 h before harvested.

1.2 Protein purification and sample preparation

The cultured cells were resuspended in lysate buffer (containing 50 mmol/L Tris, 300 mmol/L NaCl, and 20 mmol/L imidazole), and disrupted by sonication on ice. The cell debris was removed by centrifugation (48 384 g, 40 min, 4 ℃). Liquid supernatant was filtrated and then loaded onto the Ni²⁺-NTA column which was equilibrated with buffer A (containing 50 mmol/L Tris, 300 mmol/L NaCl, and 20 mmol/L imidazole). PhoB_N^F20D was eluted by 70% (v/v) buffer B (containing 50 mmol/L Tris, 300 mmol/L NaCl, and 500 mmol/L imidazole). The His₆-fused protein was cleaved by bovine thrombin (2~3 h, room temperature). The redundant bovine thrombin was removed by benzamidine column. After that, the PhoB_N^F20D protein was loaded onto the Ni²⁺-NTA column again to remove the uncleaved protein and then further purified by Superdex 75 (buffer containing 50 mmol/L Tris, 100 mmol/L NaCl, and 5 mmol/L dithiothreitol) and desalting column.

1.3 NMR spectroscopy

All the NMR samples were prepared in the buffer containing 20 mmol/L HEPES (pH 7.0), 100 mmol/L NaCl, 5 mmol/L DTT and 10% (v/v) D₂O and 90% (v/v) H₂O. All protein NMR experiments were performed on 0.3 mmol/L or 0.8 mmol/L ¹⁵N-labeled PhoB_N^F20D. In addition, samples for ¹⁹F NMR and ¹H-¹⁵N HSQC titration experiments consisted of 20 mmol/L HEPES, 100 mmol/L NaCl, 5 mmol/L DTT, 1.5 mmol/L BeCl₂ and 15 mmol/L NaF, with or without 10 mmol/L Mg²⁺, pH 7.0.

¹H-¹⁵N HSQC experiments were performed on Bruker 600 MHz equipped with ¹H/¹⁹F(¹³C)/¹⁵N channels and triple-resonance cryo-probes with pulsed-field gradients at 298 K. ¹H-¹⁵N HSQC spectra were acquired with ¹H carrier set coincident with the water resonance and ¹⁵N frequency set to δ_N 119; spectra widths of F₂ (¹H) and F₁ (¹⁵N) dimensions were 7 183.908 Hz and 2 188.284 Hz, respectively. The bandwidth of 1D ¹⁹F spectrum was 40 760.871 Hz and the chemical shifts were calibrated according to our previous study^[17]. Mg²⁺ titration experiments in the presence of BeF₃^- were carried out with the concentrations of Mg²⁺ at 0.05, 0.10, 0.30 and 0.50 mmol/L, respectively.

A set of constant-time ¹⁵N CPMG relaxation dispersion experiments was performed on Bruker 850 MHz NMR spectrometer equipped with ¹H/¹⁹F(¹³C)/¹⁵N channels and triple-resonance cryo-probes with pulsed-field gradients at 298 K. The transverse relaxation rates (R_ex^eff) of PhoB_N^F20D were obtained with frequency of 40, 80, 120, 320, 480, 640, 800 and 960 Hz.

All NMR spectra were processed using NMRpipe^[19] and analyzed by NMRView^[20]. The data of CPMG relaxation dispersion data were fitting by NESSY^[21].

2 Results and discussion 2.1 Partially BeF₃^--bound PhoB_N^F20D observed in the absence of Mg²⁺

In recent years, BeF₃^--bound PhoB_N^F20D observed has facilitated numerous structural studies of protein in their phosphorylated forms^[22-24]. For most of the RRs, the binding of BeF₃^--bound PhoB_N^F20D observed is Mg²⁺-dependent. The crystal structure of BeF₃^--bound RR468, for instance, gives a clear picture of the interaction network [Fig. 1(a)]. Mg²⁺ plays an important role in the interaction between Asp53 and BeF₃^-. However, the BeF₃^- titration experiments of PhoB_N^F20D indicated that they can bind BeF₃^- with or without Mg²⁺ [Fig. 1(b)]. The three fluorine atoms in ¹⁹F NMR spectra [marked as F₁, F₂, and F₃ in Fig. 1] are the indicators of BeF₃^- binding and have been assigned as described in reference[17]. The intensities of these three ¹⁹F NMR signals are significantly weak in the spectrum without Mg²⁺ compared with those in the presence of Mg²⁺, suggesting that only a small amount of phosphorylated conformation can be formed in the absence of Mg²⁺. Another two redundant ¹⁹F NMR signals which are labeled with 1 and 2 have shown up in Fig. 1(b). They may be the signals of the nonspecific binding between PhoB_N^F20D and BeF₃^-. Analysis of the crystal structure revealed that there are many charged residues near the phosphorylation site of PhoB_N. Be²⁺ and F^- can form BeF⁺ or BeF₄^2- in solution, in addition to BeF₃^-. We speculate that these ions may interact with those charged residues. The intensity of peak 2 decreases when BeF₃^- increases. When Mg²⁺ was added, peak 1 and 2 disappeared. It indicates that Mg²⁺ participates in the coordination of BeF₃^- and PhoB_N^F20D, and helps BeF₃^- to adopt the coordination mode that is illustrated in Fig. 1(a).

2.2 Pre-active conformation of PhoB_N^F20D stabilized by BeF₃^-

The ¹H-¹⁵N HSQC spectrum of PhoB_N^F20D in the absence of Mg²⁺ shows that the protein can partially bind BeF₃^-, which is consistent with the ¹⁹F NMR spectrum [Fig. 1(b) and Fig. 2(b)]. Another two sets of signals appear in the Mg²⁺-absent spectra [green signals in Fig. 2(b) and 2(d)]. They are the spectra of apo PhoB_N^F20D [red signals in Fig. 2(b) and 2(c)] and BeF₃^--PhoB_N^F20D in the presence of Mg²⁺ [blue signals in Fig. 2(c) and 2(d)]. The enlarged spectra in Fig. 2(a) show the examples of Gln24 and Asp53 at different conditions. We also verified that the orders of addition of Mg²⁺ or BeF₃^- have no effects on the results of the spectra. The crystal structures of PhoB_N in inactive and active states have been compared in Fig. 2(e). The conformational differences between two conformations mainly locate in two helical regions, helix α3 and helix α4. The helix α3 is slanted outward at the C-terminus of helix in the active conformation with respect to that in the inactive conformation. The helix α4 is longer in the active conformation than that in the inactive conformation.

Comparison of ¹H-¹⁵N HSQC spectra indicates that PhoB_N^F20D has pre-active conformation which is stabilized by BeF₃^- (Fig. 2). We choose some residues in the Mg²⁺-absent spectrum (green signals in Fig. 2) and calculate their intensities. The ratio of the two sets of signals in Fig. 2(b) indicates that there are approximately 15~20% pre-active conformation of apo PhoB_N^F20D. This pre-active conformation could completely transfer to active conformation when Mg²⁺ is added. The slices of Gln24 and Asp53 from HSQC titration spectra were taken as examples. As shown in Fig. 3(a) and 3(b), the signals of inactive conformation are gradually decreased, and the signals of active conformation are gradually enhanced with the increase of Mg²⁺ concentration. Accordingly, the signals of BeF₃^--PhoB_N^F20D in ¹⁹F NMR spectra increase while the Mg²⁺ concentration increases [Fig. 3(c)]. These phenomena indicate that Mg²⁺ plays an essential role in conformational exchanges of PhoB_N^F20D.

2.3 Conformational exchange of apo PhoB_N^F20D

¹⁵N NMR CPMG relaxation dispersion experiments were originally described by Loria et al.^[25] and Vallurupalli et al.^[26] in combination with an independently determined exchange-free transverse relaxation rate^[27]. For a resonance undergoing exchange, the transverse relaxation rate (R₂^eff) has been extensively used to quantify conformational exchange processes in the millisecond to high microsecond time regime in CPMG experiment. The conformational exchanges of apo PhoB_N^F20D were determined by CPMG experiments. The CPMG relaxation dispersion profiles of PhoB_N^F20D are shown in Fig. 4. We discovered that R₂^eff values of a number of residues have exchanged, such as Leu82, Val92, Arg93 and Leu112. These residues are mainly located in helix α4, and it is known that the conformation of helix α4 differs in the inactive and the active conformation of PhoB_N from its crystal structures (1B00 and 1ZES). The CPMG data of these residues are globally fitted and they give similar exchange constants of R_ex^eff (Fig. 4). We only determined the exchange rates of residues in helix α4. The exchange rates of residues in helix α3 can not be calculated due to their poor signal to noise ratio.

It is well established that conformational dynamics plays a critical role in protein function, particularly in signal transduction and regulation which typically utilizes an inactive to active state conformational transition as a means of signaling^[28-30]. Molecular dynamics (MD) simulations have been used to investigate the process of RR conformational transition with atomic details which is difficult to be obtained by experimental methods^{[31, 32]}. We directly visualized the pre-active conformation of apo PhoB_N^F20D by ¹⁹F NMR. These results reveal that pre-active conformation has a low population in the absence of phosphorylation and Mg²⁺ binding for PhoB_N^F20D which supports a "population shift" model for PhoB_N^F20D activation. Moreover, our results also indicate that Mg²⁺ binding can shift conformational transition towards active form.

3 Conclusion

In summary, apo PhoB_N^F20D has pre-active conformation in solution which can be stabilized by BeF₃^-, and Mg²⁺ accelerates this process and makes PhoB_N^F20D become active conformation. Our results provide basic information for better understanding of the phosphorylation process of response regulator.