Amino Acid-Derived Sensors For Specific Zn²⁺ Detection Using Hyperpolarized ¹³C Magnetic Resonance Spectroscopy

Abstract

Alterations in Zn²⁺ concentration are seen in normal tissues and in disease states, and for this reason imaging of Zn²⁺ is an area of active investigation. Herein, enriched [1-¹³C]cysteine and [1-¹³C₂]iminodiacetic acid were developed as Zn²⁺-specific imaging probes using hyperpolarized ¹³C magnetic resonance spectroscopy. [1-¹³C]cysteine was used to accurately quantify Zn²⁺ in complex biological mixtures. These sensors can be employed to detect Zn²⁺ via imaging mechanisms including changes in ¹³C chemical shift, resonance linewidth, or T₁.

Graphical abstract

Enriched [1-¹³C]cysteine and [1-¹³C₂]iminodiacetic acid are developed as Zn²⁺-specific imaging probes using hyperpolarized ¹³C magnetic resonance spectroscopy. These sensors can be employed to detect Zn²⁺ via imaging mechanisms including changes in ¹³C chemical shift, resonance linewidth, or T₁. [1-¹³C]cysteine is able to accurately quantify Zn²⁺ in complex biological mixtures.

Zinc (Zn²⁺) is the second most abundant transition metal ion in the human body and plays various structural, regulatory and catalytic functions in physiology.^1,2 Multiple methods for the measurement of Zn²⁺ concentration have been investigated as biomarkers for human disease, including blood and tissue tests, and imaging methods.³ Existing imaging techniques include fluorescence and magnetic resonance-based approaches. Small-molecule fluorescent agents with varying Zn²⁺ affinities have been used to image live cells^4,5,6 Magnetic resonance imaging (MRI) using Zn²⁺-responsive contrast agents has been proposed as an alternative imaging method that would allow non-invasive high-resolution monitoring of Zn²⁺ metabolism. Although there has been some success using manganese-derived probes⁷ and paramagnetic chemical exchange saturation transfer agents,⁸ most reported Zn²⁺-responsive MRI agents are gadolinium-based.^9,10

Enabled by hyperpolarization techniques such as dynamic nuclear polarization (DNP) and parahydrogen induced polarization (PHIP),^11–14 hyperpolarized (HP) probes for in vivo detection of a variety of analytes using HP ¹³C MRS have been developed.^15,16 Hyperpolarized EDTA and EGTA were reported for multi-metal imaging.¹⁷ Here, we explored the use of HP ¹³C MRS for Zn²⁺-specific imaging. In particular, we identified L-cysteine-1-¹³C ([1-¹³C]Cys) and iminodiacetic acid-1-¹³C₂ ([1-¹³C₂]IDA) as candidate probes that demonstrate altered magnetic properties upon binding Zn²⁺, optimized their polarization parameters, and demonstrated that [1-¹³C]Cys can accurately measure Zn²⁺ using ¹³C HP MRS in phantoms and biological samples (Figure 1A).

Figure 1.

Approach to specific detection of Zn²⁺ via hyperpolarized magnetic resonance spectroscopy (A) Structures of [1-¹³C]Cys and [1-¹³C₂]IDA. (B) Specificity of 50 mM [1-¹³C]Cys and [1-¹³C₂]IDA for physiological cations (present at 1 equivalent). pH-dependence of chemical shift was measured at pH 7.4 and 6.5, the physiologically relevant extracellular range.

We considered the requirements for a potential hyperpolarized Zn²⁺ imaging probe to include 1) a large change in ¹³C chemical shift in the presence of Zn²⁺, 2) good selectivity to Zn²⁺ over biologically abundant cations such as Ca²⁺ and Mg²⁺, 3) insensitivity to changes in pH over the physiologic range, 4) a sufficient T₁ relaxation time for MRI, and 5) lack of known or predicted toxicity. We began our study by screening a variety of candidate Zn²⁺ ligands for their chemical shift response upon Zn²⁺ binding. Among the 34 tested compounds, 10 ligands demonstrated a chemical shift response to Zn²⁺ binding. We noted that ligands with logarithm of stability constants of Zn²⁺ (logK) greater than 6.9 all displayed a chemical shift response to Zn²⁺ binding, while ligands with logK lower than 5.2 did not have a chemical shift response to Zn²⁺ binding (SI Table S1). Ligands bearing a soft functional group such as pyridine tended to show a chemical shift response at lower Zn²⁺ stability constants while ligands with hard functional groups, such as hydroxyl showed a chemical shift response at higher stability constants. Cysteine and iminodiacetic acid were selected for further evaluation because of their large chemical shift response to Zn²⁺ binding (+4.6 and +7.3 ppm in the presence of equimolar Zn²⁺, respectively), excellent water solubility, low molecular weight, predicting a long ¹³C T₁, and convenience for ¹³C labeling. ¹³C-labeled cysteine was commercially available, and ¹³C-labeled IDA was synthesized from [1-¹³C]glycine and [1-¹³C]methyl bromoacetate in 3 steps with a 32% total yield (Scheme 1). Their specificity for Zn²⁺ versus other abundant physiologic cations and their sensitivity to pH changes were also tested. Both probes showed excellent specificity for Zn²⁺ over other main physiologic cations Na⁺, K⁺, Ca²⁺ and Mg²⁺, and limited chemical shift change over the physiologic pH range of 6.5 to 7.4 (Figure 1B). These preliminary investigations suggested that [1-¹³C]Cys and [1-¹³C₂]IDA would be specific and sensitive probes for detecting Zn²⁺ using HP ¹³C MRI.

Scheme 1.

Synthesis of [1-¹³C₂]IDA disodium salt.

A quantitative method for determining Zn²⁺ concentration was developed for these probes using NMR. A titration curve was plotted to show the chemical shift difference between [1-¹³C]Cys and internal standard urea as a function of the Zn²⁺ to [1-¹³C]Cys ratio. The peak moved linearly with increasing Zn²⁺ up to 0.25 equivalents of Zn²⁺ to [1-¹³C]Cys. The rate of chemical shift change decreased at higher Zn²⁺ concentrations (Figure 2A, 2B, SI Table S2, SI equation S1). This suggests that [1-¹³C]Cys may have multiple Zn²⁺ binding conformations, as previously reported.^18,19 The manifestation of the ¹³C signal as a single resonance indicates that the on-off rates for Zn²⁺ binding are rapid, compared with the absolute frequency difference between bound/unbound ¹³C resonances. A linewidth change with different Zn²⁺/[1-¹³C]Cys ratios was also observed. In particular, the linewidth increased significantly at lower Zn²⁺ concentrations with the widest linewidth being between 0.2 eq and 0.5 eq of [1-¹³C]Cys, thus demonstrating exchange broadening at different Zn²⁺/[1-¹³C]Cys ratios (SI Table S3). The chemical shift change of [1-¹³C]Cys with equimolar Zn²⁺ was tested at different temperatures ranging from 20 °C to 50 °C and at various [1-¹³C]Cys concentrations from 1 mM to 100 mM. These studies showed that the temperature and concentration of [1-¹³C]Cys (in the presence of equimolar zinc) had little or no influence on the chemical shift response (SI Tables S3 and S4).

Figure 2.

Response of [1-¹³C]Cys and [1-¹³C₂]IDA to Zn²⁺ can be detected by several magnetic resonance mechanisms. (A) ¹³C NMR Spectra for 50 mM [1-¹³C]Cys with varying Zn²⁺ concentrations (B) Standard titration curve of [1-¹³C]Cys in response to various Zn²⁺ concentrations. The insert shows the titration curve at low Zn²⁺ equivalents. (C) Spectra for 50 mM [1-¹³C₂]IDA with various Zn²⁺ concentrations highlighting formation of a Zn²⁺ bound species, and increased linewidth upon Zn²⁺ binding. * represents natural abundance citric acid peak (D) Standard titration curve of [1-¹³C₂]IDA.

A similar titration method was utilized for IDA for quantitative determination of Zn²⁺ concentration. When Zn²⁺ was added to [1-¹³C₂]IDA, a new resonance at a lower field appeared. The area under the peak (integration) increased linearly with increasing Zn²⁺ concentration from 0 to 0.5 equivalents of Zn²⁺ to [1-¹³C₂]IDA, which indicates that [1-¹³C₂]IDA binds to Zn²⁺ with a 2:1 probe:Zn stoichiometry, as previously reported (Figure 2C, 2D).²⁰ Additionally, the unbound [1-¹³C₂]IDA peak demonstrated increasing line broadening in response to Zn²⁺ (SI Table S5). It is worth noting that the [1-¹³C₂]IDA:Zn complex appears as two separate peaks while the [1-¹³C]Cys:Zn complex appears as a single peak, indicating that the on and off rates for [1-¹³C₂]IDA are slower than the frequency difference between bound and unbound resonances and the on and off rates for [1-¹³C]Cys are more rapid. These results are analogous to prior findings in ¹⁵N, ¹⁹F, and hyperpolarized ¹³C pH imaging probe development efforts.^21–27 Overall, these data demonstrate that the chemical shift of the 1-¹³C NMR signal of cysteine and IDA change in a predictable and quantitative manner in response to Zn²⁺ concentration.

Next, we developed and optimized a hyperpolarization method for ¹³C labeled [1-¹³C]Cys and [1-¹³C₂]IDA. For [1-¹³C]Cys, the optimized preparation was obtained by a mixture of 1.0 eq of [1-¹³C]Cys, 0.5 eq of 4N HCl and 2.65 eq of glycerol, with 20 mM OX063 radical.²⁸ Using this method, 13.4 ± 0.6% back-calculated polarization was obtained with a polarization time constant of 1227 ± 30 (n = 3). The T₁ relaxation time was 36.0 ± 1.8 seconds at a magnetic field of 3T. A Zn²⁺-dependent decrease in T₁ of [1-¹³C]Cys was observed when polarized [1-¹³C]Cys was mixed with various equivalents of Zn²⁺. A T₁ of 17.5 seconds was observed when 1 equivalent of Zn²⁺ was present (Table S6). For [1-¹³C₂]IDA, the optimized preparation included a mixture of 1.0 eq of [1-¹³C₂]IDA disodium salt, 14.0 eq of D₂O and 1.8 eq of DMSO, with 15 mM OX063.¹⁷ Using this method, 6.6 ± 0.9% (n = 3) back-calculated polarization was obtained with the polarization time constant of 1480 ± 38 (n = 3). The T₁ was 39.3 ± 1.0 seconds at 3T. The T₁ of both bound and unbound [1-¹³C₂]IDA resonances were also Zn²⁺ dependent, with the T₁ decreasing to 16.6 seconds when 1 eq of Zn²⁺ was present (Figure 3, SI Table S7). We hypothesize that the decrease in T₁ is due to an increase in molecular weight of the complex.²⁹ The high percent polarization, and relatively long T₁ of the Zn²⁺ ligands suggested feasibility for subsequent imaging experiments.

Figure 3.

Dynamic ¹³C NMR spectra following polarization of [1-¹³C]Cys (left) and [1-¹³C₂]IDA (right). Pulse conditions were TR = 3s, 5º hard pulses, 10 kHz spectral width, 30 timepoints, nominal spectral resolution = 0.5 Hz.

The ability of the probes to image Zn²⁺ concentration was tested using phantoms on a 3T MRI system. For cysteine, a copolarization method, as we have previously reported, was developed with [1-¹³C]urea as a chemical shift standard.²³ A four-compartment phantom containing various Zn²⁺ concentrations in each compartment was prepared (Figure 4A, B). As in the thermal equilibrium measurements, a chemical shift change was observed in response to Zn²⁺ in the ¹³C cysteine spectrum. By using the best-fit linear model of chemical shift asFigure 3 a function of Zn²⁺ concentration (Figure 2B inset), hyperpolarized [1-¹³C]Cys was able to accurately determine the concentration over the physiologically relevant range of 0.2 – 20 mM Zn²⁺. At the highest tested Zn²⁺ concentration of 40 mM, a slight deviation was observed, possibly due to signal loss and/or line broadening.

Figure 4.

HP [1-¹³C]Cys accurately quantifies Zn²⁺ concentration in phantom imaging experiments through use of chemical shift measurements. (A) Phantom imaging conducted using 40 mM [1-¹³C]Cys. (B) Phantom experiment conducted using 4 mM [1-¹³C]Cys, indicating high sensitivity for low concentrations of Zn²⁺.

A phantom imaging experiment to determine the sensitivity showed that HP [1-¹³C]Cys can detect 200 μM of Zn²⁺ when 4 mM of [1-¹³C]Cys was present (Figure 4B). When hyperpolarized [1-¹³C₂]IDA was tested in a phantom for Zn²⁺ imaging, the Zn²⁺ bound [1-¹³C₂]IDA peak appeared when more Zn²⁺ was present. The line broadening of both the Zn²⁺ bound and unbound peaks of [1-¹³C₂]IDA resulted in a loss of SNR between 0.2 eq Zn²⁺ and 0.3 eq Zn²⁺ (SI Figure S2). For this reason, accurate Zn²⁺ concentrations could not be obtained using this probe. Overall, these phantom experiments demonstrated that hyperpolarized [1-¹³C]Cys is a sensitive and accurate Zn²⁺ biosensor over the physiologic range, while issues with line broadening and signal loss limit the applicability of [1-¹³C₂]IDA.

Finally, we verified that HP [1-¹³C]Cys could accurately determine Zn²⁺ concentration in biological samples (Figure 5) by comparing imaging results against a commercially available fluorescence based Zn²⁺ quantification kit. Rat serum and prostate extracts were selected as biological samples with low and high Zn²⁺ concentration, respectively. In these samples, the Zn²⁺ concentration was accurately measured at 0.1 and 1.9 mM, respectively. The HP and kit Zn²⁺ measurements agreed within 0.1 mM. In order to verify that the signal change in these samples represented a specific response to Zn²⁺, additional tubes including serum spiked with exogenous Zn²⁺, and a prostate sample with additional tris(2-pyridylmethyl)amine (TPA), a high-affinity and specific chelator, were investigated.³⁰ We found that the Zn²⁺ concentration was accurately measured in the spiked serum sample, and that the addition of the TPA to the prostate extract blocked the chemical shift change. These data confirm that Zn²⁺ is both necessary and sufficient to induce the chemical shift change in cysteine, and moreover that this probe accurately and specifically determines Zn²⁺ concentration in complex biological samples.

Figure 5.

Hyperpolarized [1-¹³C]Cys phantom imaging experiment demonstrating accurate Zn²⁺ quantification in biological samples. Accurate detection of Zn²⁺ concentration in presence of exogenous Zn²⁺ and in the presence of Zn²⁺-chelating TPA demonstrates specificity of the probe for Zn²⁺ over other endogenous analytes.

For future in vivo applications, [1-¹³C]Cys in particular demonstrates both possible advantages and limitations compared against existing techniques. Owing to the short lifetime of the signal, it is unlikely that it would be internalized into cells in a time course reasonable to measure intracellular zinc concentration. Therefore, this method would likely be restricted to extracellular space Zn²⁺ imaging. However, zinc can be secreted from tissues including pancreas and prostate in response to glucose treatment. For example, Sherry et al. have developed imaging methods to detect the release of Zn²⁺ ions from β-cells in response to high glucose,³¹ as well as to differentiate between healthy and malignant prostate tissue in mouse models.³² Such a model could applied for initial in vivo testing of these probes. An additional limitation could be metabolic transformation, which could confound interpretation of spectroscopic data. Toxicity concerns are unlikely based on the known toxicology profile of cysteine.³³ The majority of Zn²⁺ imaging probes are gadolinium-based contrast agents, which face the concern of their deposition in brain and environment, a concern which would be avoided in this case.³⁴ A final potential benefit of this method is the broad range of zinc concentration in serum and other tissues, potentially enabling high image contrast. Overall, these preliminary data suggest both possibilities and limitations for future in vivo applications of these methods.

Taken together, these data demonstrate that [1-¹³C]Cys and [1-¹³C₂]IDA represent promising probes for imaging Zn²⁺ using hyperpolarized ¹³C MRI. The probes demonstrate large changes in signal and chemical shift in response to Zn²⁺, excellent selectivity over other biologically relevant cations and changes in pH, favorable T₁ and polarization parameters, and can be imaged in phantom experiments. [1-¹³C]Cys accurately quantified Zn²⁺ concentration in biological samples at physiologically relevant concentrations. For this reason, [1-¹³C]Cys is a particularly promising probe for future in vivo hyperpolarized magnetic resonance imaging of pathologies with alterations in Zn²⁺ homeostasis such as prostate cancer, neurodegenerative disease, and diabetes.