Rational design of a genetically encoded NMR zinc sensor

Abstract

Elucidating the biochemical roles of the essential metal ion, Zn²⁺, motivates detection strategies that are sensitive, selective, quantitative, and minimally invasive in living systems. Fluorescent probes have identified Zn²⁺ in cells but complementary approaches employing nuclear magnetic resonance (NMR) are lacking. Recent studies of maltose binding protein (MBP) using ultrasensitive ¹²⁹Xe NMR spectroscopy identified a switchable salt bridge which causes slow xenon exchange and elicits strong hyperpolarized ¹²⁹Xe chemical exchange saturation transfer (hyper-CEST) NMR contrast. To engineer the first genetically encoded, NMR-active sensor for Zn²⁺, we converted the MBP salt bridge into a Zn²⁺ binding site, while preserving the specific xenon binding cavity. The zinc sensor (ZS) at only 1 μM achieved 'turn-on' detection of Zn²⁺ with pronounced hyper-CEST contrast. This made it possible to determine different Zn²⁺ levels in a biological fluid via hyper-CEST. ZS was responsive to low-micromolar Zn²⁺, only modestly responsive to Cu²⁺, and nonresponsive to other biologically important metal ions, according to hyper-CEST NMR spectroscopy and isothermal titration calorimetry (ITC). Protein X-ray crystallography confirmed the identity of the bound Zn²⁺ ion using anomalous scattering: Zn²⁺ was coordinated with two histidine side chains and three water molecules. Penta-coordinate Zn²⁺ forms a hydrogen-bond-mediated gate that controls the Xe exchange rate. Metal ion binding affinity, ¹²⁹Xe NMR chemical shift, and exchange rate are tunable parameters via protein engineering, which highlights the potential to develop proteins as selective metal ion sensors for NMR spectroscopy and imaging.

Fig. 1. Hyper-CEST z-spectra of 80 μM ZS with (blue) and without (grey) 400 μM ZnCl2 in 20 mM Tris (pH 7.4), 100 mM NaCl at 300 K. Red and green lines show Lorentzian fits to the Xe@aq and Xe@protein peaks. Chemical shift of Xe@aq is referenced as zero and Xe@protein peak is 50 ppm downfield-shifted from the Xe@aq peak.

Fig. 2. Time-dependent saturation transfer data for ZS-GFP in 20 mM Tris (pH 7.4), 100 mM NaCl at 300 K. Saturation pulses were positioned at +50 ppm and −50 ppm referenced to Xe@aq peak for on- and off-resonance. Left: 1 μM ZS-GFP with 1 mM EDTA. T_1on = 37.3 ± 1.6 s, T_1off = 36.2 ± 1.8 s, saturation contrast = −0.01 ± 0.01. Right: 1 μM ZS-GFP with 400 μM ZnCl₂. T_1on = 22.8 ± 0.7 s, T_1off = 38.2 ± 1.3 s, saturation contrast = 0.17 ± 0.02.

Fig. 3. Hyper-CEST z-spectra showing metal selectivity of ZS in 20 mM Tris (pH 7.4), 100 mM NaCl at 300 K. Chemical shifts of Xe@aq peaks by Lorentzian fitting are referenced as zero.

Fig. 4. X-ray crystal structures of apo-ZS and Zn²⁺-ZS. (a) Zn-edge anomalous map of Zn²⁺-ZS. Histidine residues (blue sticks and yellow mesh) and solvent molecules (red non-bonded × symbols and white mesh) within the Zn²⁺ binding site are contoured at 1.0 σ. The Zn²⁺ ion (gray sphere and green mesh) is contoured at 10.0 σ. (b) Superimposed cartoon representations of apo-ZS (white) and Zn²⁺-ZS (blue). (c) A detailed view of the Zn²⁺ binding site of apo-ZS (white) and Zn²⁺-ZS (blue). The side chain of H111 was refined and modeled in two alternate conformations. (d) Residues that form the Xe binding cavity are represented in white (apo-ZS), blue (Zn²⁺-ZS), magenta (unliganded wt-MBP) and green (maltose-bound wt-MBP).