Introduction

The B1 immunoglobulin-binding domain of staphylococcal protein G (GB1) is a small, 56-residues, stable globular protein. It consists of a four-stranded β-sheets spanned by an α-helix^[1]. It has been widely used as a research model for developing protein structural determination methods^[2-11] and studying protein stability, dynamics and folding^[12-18].

Taking GB1 as the research model, various paramagnetic metal-tags have been developed to help obtain significant distance and orientation restraints for in vitro and in vivo protein structural determination^[2-11]. For instance, paramagnetic EDTA-Mn²⁺, EDTA-Cu²⁺ and TETAC-Cu²⁺ were introduced to GB1 to help obtain longitudinal nuclear paramagnetic relaxation enhancement (PRE)^{[2-6, 8, 9]}, which can provide long-range distance restraints for structural determination. The paramagnetic lanthanide-tag DOTA-M7Py and 4PhSO₂-PyMTA were introduced to GB1 to help obtain pseudocontact chemical shifts (PCSs) and residual dipolar couplings (RDCs) data for in-cell protein structural determination based on PCS and RDC, which can provide nuclear distance and orientation restraints^{[10, 11]}.

Recently, it has been reported that there exist intrinsic weak Cu²⁺ binding sites in GB1, which resulted in anomalous increase of the longitudinal ¹⁵N PREs of residues involved^[4]. It implies that it is necessary to investigate the binding of GB1 with other ions, especially lanthanide ions, in order to ensure reliability of paramagnetic nuclear magnetic resonance (NMR) data such as PCSs, PREs and RDCs based on paramagnetic metal-tags. Herein, we characterized the interaction of GB1 with lanthanide as well as divalent metal ions by NMR spectroscopy.

1 Materials and methods 1.1 Protein expression and purification

Plasmids containing the coding sequence of GB1 were transformed into Escherichia coli strain BL21 (DE3) competent cells. Expression and purification of ¹⁵N labeled GB1 were performed as previously described^[19]. The purified protein was passed through a HiPrep 26/10 desalting column (GE Healthcare) equilibrated and eluted with Milli-Q water. The eluted protein was lyophilized and stored as aliquots at -20 ℃. The concentration of protein was determined by A280 nm (using a molar extinction coefficient of 9 970 L·mol^-1·cm^-1.

1.2 Chemical reagents

TmCl₃·6H₂O and LuCl₃·6H₂O from Sigma-Aldrich Co. or Alfa Aesar Co., Ltd. were used without further purification. CuSO₄, MnCl₂, MgSO₄, CoSO₄, CaCl₂ and ZnSO₄ salts of the highest purity available were purchased from Merck Co., Ltd. or Sigma-Aldrich Co. All the other chemicals of analytical grade were purchased from Sinopharm Chemical Reagent Co., Ltd.

1.3 NMR experiments

The lyophilized protein was dissolved in NMR buffer [containing 20 mmol/L MES, V(H₂O)/V(D₂O) = 9/1, pH 6.5)] with final concentration of 0.20 mmol/L. ¹H-¹⁵N HSQC spectra were acquired on Bruker 850 MHz or 600 MHz spectrometers equipped with a triple-resonance cryoprobe at 298 K. The HSQC spectra were acquired with sweep widths of 11 904.7 Hz (¹H) and 3 446.8 Hz (¹⁵N), sampling points of t₂ (¹H)×t₁ (¹⁵N) = 1 024×128, 4 scans per t₁ point, and recycle delay of 1.0 s.

1.4 Dissociation constant determination

Dissociation constant (K_D) was determined from the titration curve according to equation (1) as follows^[20], where Δδ is the measured chemical shift perturbation at each ion concentration, Δδ_max is the maximum chemical shift perturbation, [P]_t is the total protein concentration, and [L]_t is the total ion concentration.

$ \begin{array}{l} \mathit{\Delta }\delta = \mathit{\Delta }{\delta _{\max }}\left\{ {{K_D} + {{[P]}_t} + {{[L]}_t} - [{{({K_D} + {{[P]}_t} + {{[L]}_t})}^2} - {{(4 \cdot {{[P]}_t} \cdot {{[L]}_t})}^{1/2\;}}]} \right\}\\ /(2\;{[P]_t}) \end{array} $

(1)

In the backbone HN chemical shift perturbation versus residue plots, the combined HN perturbation (Δδ_HN) is calculated as follows, where Δδ_H and Δδ_N in equation (2)^[21-23] are the chemical shift differences of proton and nitrogen nuclei, respectively.

$ \mathit{\Delta }{{\delta }_{\rm{HN}}}=\sqrt{\frac{1}{2}(\mathit{\Delta } \delta _{\rm{H}}^{2}+\frac{\mathit{\Delta } \delta _{\rm{N}}^{2}}{25})} $

(2)

2 Results and discussion 2.1 K_D values determination and binding sites of lanthanide metal ions with GB1

The ¹H-¹⁵N HSQC spectra were used to examine the binding sites of metal ions to GB1. Fig. 1(a) shows the HSQC spectra of GB1 titrated with diamagnetic Lu³⁺. Cross peaks shifting with Lu³⁺ addition was observed, suggesting that Lu³⁺ can bind to GB1. K_D values were obtained by individually fitting the titration curve from backbone amide resonances with significant chemical shifts perturbations (CSP) according to equation (1). Fig. 1(b) shows the fitting curves of representative 6 residues. Fig. 1(c) shows K_D values for 14 residues of which chemical shift changes in titration are more than 0.04. The mean K_D value is about 0.18 mmol/L.

Fig. 1(d) shows the CSP of GB1 titrated with 3.5 equivalents of Lu³⁺. The larger chemical shift perturbations mainly come from residues N8-K13, V39-W43 and V54-E56. We speculate the possible binding sites of Lu³⁺ in GB1 are negatively charged residues D40, E42 and E56.

We also acquired HSQC spectra of GB1 titrated with lanthanide paramagnetic Tm³⁺ (Fig. 2). Fig. 2(a) shows the HSQC spectra of GB1 in the presence of Tm³⁺ (red) and Lu³⁺ (blue), respectively. Tm³⁺ has anisotropic magnetic susceptibility tensor and thus can cause both PCSs and PRE^[24]. Therefore, in the presence of Tm³⁺, it was observed that the cross peaks of some residues shifted or disappeared in HSQC spectrum. Taking Lu³⁺ as a diamagnetic reference^[24], PCSs of backbone amide ¹H induced by Tm³⁺ were obtained [Fig. 2(b)]. The residues with larger PCSs are mainly located in the first (residues N8-L12), the third (residues N37-G41) and the fifth loop (residues T53-E56) of GB1. Magnetic susceptibility anisotropy values were obtained by fitting PCSs to the structure of GB1 (PDB 3GB1^[25]) by using Numbat software^[26]. Fig. 2(c) shows the correlations between experimental and back-calculated PCSs from which the consistency of PCSs can be observed, implying that the binding with lanthanide metal ions Lu³⁺ or Tm³⁺ don't greatly disturb the structure of GB1.

In all, the lanthanide metal ions Lu³⁺ and Tm³⁺ have weak binding affinity with GB1. The dissociation constant for GB1 and Lu³⁺ is about 0.18 mmol/L. The possible binding sites are residues D40, E42 and E56. Therefore, for GB1, the paramagnetic metal-tags with high binding affinity are needed to avoid possible interference to paramagnetic NMR data.

2.2 Divalent metal ions binding sites determination

Beside lanthanide metal ions, the binding of divalent metal ions with GB1 were also investigated. Fig. 3(a) shows the HSQC spectra of GB1 in the presence of Ca²⁺ (green), Mg²⁺ (blue) and Zn²⁺ (purple) at equal molar ratio, respectively. Compared to GB1, no chemical shifts changes were observed in the spectra after adding these ions, suggesting that GB1 probably can not bind with these three ions.

Fig. 3(b) shows the HSQC spectra of GB1 in the presence of paramagnetic Mn²⁺ at 0.2/1, 0.6/1 and 1/1 molar ratio of Mn²⁺/GB1, respectively. Compared to GB1, in the presence of Mn²⁺, the cross peaks of some residues, such as K10, G38, V39, D40, G41, E42, W43 and E56, severely broadened and even disappeared due to PRE effects. The resonance from D40 disappears at 0.2/1 molar ratio of Mn²⁺/GB1, suggesting that Mn²⁺ binding site in GB1 may be in the region near D40 residue. Chemical shifts induced by Mn²⁺ binding have not been observed, suggesting that the binding of Mn²⁺ with GB1 is weak and have little impact on GB1 structure.

We also performed NMR titration of Mg²⁺ and Zn²⁺ in the presence of Mn²⁺, respectively [Fig. 3(c) and 3(d)]. No changes were observed in the spectra with Mg²⁺ and Zn²⁺ addition, suggesting that neither Mg²⁺ nor Zn²⁺ can substitute for Mn²⁺ to bind with GB1, which is consistent with the observation on Mg²⁺ and Zn²⁺ titration.

Fig. 3(e) shows the HSQC spectra of GB1 titrated with paramagnetic Co²⁺. Compared to GB1, the cross peaks of residues G38, V39, D40, G41 and E56 shifted a little due to PCSs and binding. Only the cross peak of residue V39 exhibited a decrease in peak intensity, probably due to the PRE effect. The observations suggest that Co²⁺ has a very low binding affinity with GB1.

Fig. 3(f) shows the HSQC spectra of GB1 titrated with paramagnetic Cu²⁺. Compared to GB1, the cross peaks of some residues, such as Y3, E19, A20, V21, D22, A26, D40, G41, E42 and E56, severely broadened or disappeared due to PRE effects with Cu²⁺ addition. The cross peaks of some residues, such as K4, K10, T18 and E56, shifted a little due to PCS effects. By analyzing the distribution of these residues in GB1 structure, we speculate that Cu²⁺ has two binding sites in GB1, among which one is D40, E42 and E56, and the other is E19 and D22 (Fig. 4). Furthermore, the binding affinities of these two binding sites are different. At 0.6/1 molar ratio of Cu²⁺/GB1, the cross peaks of D40 and E42 have disappeared while E19 and D22 residues still exist, suggesting that Cu²⁺ has a higher binding affinity with the former.

In all, for divalent metal ions, GB1 can bind with paramagnetic ions such as Cu²⁺, Mn²⁺ and Co²⁺, but not with diamagnetic ions such as Ca²⁺, Mg²⁺ and Zn²⁺. The possible binding sites of Mn²⁺, Co²⁺ and Cu²⁺ in GB1 were shown in Fig. 4(b) and 4(c), respectively. For Mn²⁺ and Co²⁺, the possible binding sites are residues D40/E42/E56. For Cu²⁺, there exist two binding sites, one is residues D40/E42/E56, and the other one is E19/D22. The former has relative higher binding affinity than the latter. According to the Irving-Williams series^[27] which ranks the relative stability of complexes formed by divalent metal ions, the stability of the complex with Cu²⁺ is higher than that with Mn²+ and Co²⁺, which is the possible reason that the E19/D22 sites weakly bind with Cu²⁺, but not with Mn²⁺ and Co²⁺.

3 Conclusion

In this research, we determined the binding of GB1 with various divalent cations and lanthanide ions, respectively, using 2D ¹H-¹⁵N HSQC spectroscopy. GB1 weakly binds with paramagnetic divalent cations Cu²⁺, Mn²⁺ and Co²⁺, and lanthanide ions Lu³⁺ and Tm³⁺, but not with diamagnetic divalent cations Ca²⁺, Mg²⁺ and Zn²⁺. Our studies demonstrate that NMR spectroscopy is a powerful tool to study weak binding between protein and metal ions, and also hint that care must be taken to avoid possible interference to paramagnetic NMR data when using GB1 as a model protein.