Real-Time in Vivo Detection of H₂O₂ Using Hyperpolarized ¹³C-Thiourea

Abstract

Reactive oxygen species (ROS) are essential cellular metabolites widely implicated in many diseases including cancer, inflammation, and cardiovascular and neurodegenerative disorders. Yet, ROS signaling remains poorly understood, and their measurements are a challenge due to high reactivity and instability. Here, we report the development of ¹³C-thiourea as a probe to detect and measure H₂O₂ dynamics with high sensitivity and spatiotemporal resolution using hyperpolarized ¹³C magnetic resonance spectroscopic imaging. In particular, we show ¹³C-thiourea to be highly polarizable and to possess a long spin–lattice relaxation time (T₁), which enables real-time monitoring of ROS-mediated transformation. We also demonstrate that ¹³C-thiourea reacts readily with H₂O₂ to give chemically distinguishable products in vitro and validate their detection in vivo in a mouse liver. This study suggests that ¹³C-thiourea is a promising agent for noninvasive detection of H₂O₂ in vivo. More broadly, our findings outline a viable clinical application for H₂O₂ detection in patients with a range of diseases.

Graphical abstract

Oxidative stress is a hallmark of many pathophysiological diseases corresponding to the oxidative damage incurred by aberrant reactive oxygen species (ROS) production. These ROS are essential for cellular signaling.^1,2 They include highly reactive oxygen metabolites such as hydrogen peroxide (H₂O₂), superoxide (O₂^•–), and hydroxyl radical (⁻OH). Dysregulation of ROS homeostasis leads to oxidative stress and irreversible molecular damage, which is associated with various disease states such as cancer, inflammation, and cardiovascular and neurodegenerative disorders.^3–7

Despite their importance, the precise mechanisms of ROS signaling are poorly understood. The high reactivity and short lifetime of ROS have further complicated efforts for determining their spatiotemporal dynamics in vivo. ROS detection in biological systems therefore requires probes that rapidly react with ROS to produce stable quantifiable products. Indeed, several studies have reported the development of ROS-specific probes for live in vivo optical detection.^8,9 These probes capitalized on the enhanced contrast of a caged fluorescence emission that is activated upon ROS-induced transformation. The use of small molecule and nanoparticle probes in these studies has in turn, inspired us to exploit ¹³C magnetic resonance spectroscopic imaging (MRSI)¹⁰ to study ROS flux in vivo.

In contrast to optical techniques, MRSI enables noninvasive acquisition of deep-tissue images in vivo and the investigation of real-time cellular metabolism with high spatial and temporal resolution.^11,12 However, MRSI experiments have inherently low sensitivity and are therefore limited to the detection of abundant ¹H nuclei from water molecules and lipid chains in vivo. To gain information on intermediary cellular metabolites and their fluxes, we have utilized ¹³C MRSI by taking advantage of its wide chemical shift differences to encompass signals from both a substrate and its metabolic product(s).¹⁰ To overcome the naturally low abundance of a ¹³C signal, we hyperpolarized the ¹³C-labeled substrate by using dynamic nuclear polarization (DNP) to afford more than 10 000-fold signal enhancement of the ¹³C-probe and its subsequent metabolic products.^13,14 In combination with a rapid dissolution process, hyperpolarized (HP) ¹³C-substrates such as [1-¹³C]-pyruvate,¹⁵ [2-¹³C]-pyruvate,¹⁶ [¹³C]-glucose,¹⁷ [¹³C]-bicarbonate,¹⁸ [2-¹³C]-fruc-tose,¹⁹ and [2-¹³C]-dihydroxyacetone²⁰ have been used as metabolic probes to investigate cellular metabolism with high spectral resolution. Notably, [1-¹³C]-dehydroascorbate (DHA)^21,22 and [1-¹³C]-benzoylformic acid (BFA)²³ have been reported as HP ¹³C-substrates to probe oxidative stress specific to redox and H₂O₂ sensing, respectively.

To design new HP ¹³C-probes for ROS detection in vivo, we focused on small-molecule metabolites exhibiting several key properties. First, the substrate should be highly polarizable either by DNP or other methods.^24–26 Second, both the substrate and its metabolic product(s) should possess a long spin–lattice relaxation time (T₁), as needed for robust in vivo detection. Third, the HP probe should reveal a distinctive chemical shift for its ¹³C nuclei upon reaction with ROS in order to allow spectroscopic detection of both the probe and its products. Finally, the probe should be nontoxic (at micromolar to millimolar concentration) and be rapidly transported for ROS-mediated reaction (or react in the extracellular space).

To this end, we explored ¹³C-thiourea (¹³C-TU) as an HP probe for ROS detection. Thiourea is a water-soluble ROS scavenger that has been associated with antioxidant proper-ties.^27,28 In liver microsomes, flavin-containing monooxygenases (FMO) catalyze an NADPH-dependent S-oxygenation of thiourea in the presence of O₂,²⁹ leading to the formation of formamidine sulfenic acids (TUO) that in turn either react reversibly with glutathione (GSH) or undergo further oxidation to form thiourea dioxide (TUO₂). Notably, no thiourea transporters have been identified in mammals, even though studies have implicated urea transporters for a shared transport mechanism with thiourea in erythrocytes.³⁰ In addition, saturation kinetics in perfused rat inner medullary collecting duct showed saturation of the urea transporter with a K_M = 20 mM for thiourea.³¹ Moreover, studies on a single intra-peritoneal exposure of thiourea in different animal models have yielded a range of lethal dose (LD₅₀) concentrations of up to 1340 mg/kg body weight in rats.³² Together, these studies suggest that ¹³C-TU is an attractive HP probe candidate for ROS detection.

To test our hypothesis, we determined the selectivity of ¹³C-TU with different ROS that are responsible for many cellular signaling events. These ROS have been implicated in oxidative damage inflicted on fatty acid, DNA, and proteins as well as other cellular components (Figure 1A).³³ Briefly, thiourea was exposed to several biologically relevant ROS for 10 min, and the formation of oxidized products was monitored by ¹³C NMR. In the presence of ^•OOH (H₂O₂ or tBuOOH), we detected oxidation of TU (δ184 ppm) to form TUO₂ (δ 179.5 ppm), as identified by LC-MS (Figure 1B and S1). When reacted with other ROS such as superoxide (O₂^•−) and hydroxyl radical (^•OH), thiourea remained unchanged. Reaction with hypochlorite (⁻OCl) and peroxynitrite (ONOO⁻) yielded urea (δ 165 ppm) as the hydrolyzed byproduct (Figure 1B and C). While in vitro oxidation of thiourea is not limited to peroxide, we were especially interested in using ¹³C-TU to detect H₂O₂ in vivo due to its central role in cellular signaling.^34,35 Such a measurement is physiologically relevant because FMO-mediated oxygenation of thiourea in vivo is reported to proceed via reaction with H₂O₂.³⁶ We therefore investigated thiourea oxidation by reacting different concentrations of H₂O₂ (2-27 μM) with 10 μM ¹³C-TU and examining ¹³C-TUO₂ formation by ¹³C NMR (Figure 1D). By plotting the ratio of the integrated ¹³C-TUO₂ product peak to the total signal intensities (¹³C-TU + ¹³C-TUO₂) against peroxide, we showed a good linear correlation to increasing H₂O₂ concentration (Figure 1E) and confirmed its sensitivity for detecting low micromolar peroxide concentration in vitro.

Figure 1.

(A) Different kinds of ROS generated via multiple enzymatic steps. (B) Response of thiourea (TU) to various ROS at thermal equilibrium. All ROS (H₂O₂, tBuOOH, NaONOO, and NaOCl) were added to TU in D₂O (except for KO₂-treated sample in DMSO) and the spectra recorded with ¹³C NMR at 125 mHz. Reaction with H₂O₂ and tBuOOH yielded thiourea dioxide (TUO₂) as the oxidized product. (C) TU was oxidized to TUO₂ before getting hydrolyzed to urea. (D) ¹³C NMR spectra of 10 μM ¹³C-TU after 10 min reaction with various concentrations of H₂O₂ (2–27 μM). (E) Linear correlation of the ratio of integrated peak intensities of ¹³C-TUO₂ to the total signal intensities (¹³C-TU + ¹³C-TUO₂) against H₂O₂ concentration, R² = 0.994.

Next, we sought to measure the rate of thiourea oxidation by H₂O₂ using UV/vis spectroscopy. We first performed a UV sweep of thiourea solutions (12.5–200 μM) and determined a maximum absorbance at λ_max = 237 nm with an extinction coefficient ε = 0.0113 μM⁻¹ cm⁻¹ (Figure S2). We measured the kinetics of thiourea oxidation by detecting A₂₃₇ change (depletion of thiourea) when reacted with H₂O₂. Briefly, two thiourea solutions (0.0625 and 0.125 mM) were prepared in a buffered solution at pH 7.4, incubated with different H₂O₂ concentrations (0.125–8 mM), and monitored for change in A₂₃₇ over 50 min (Figure 2A). By plotting the initial rate against peroxide concentration, we found the results to be in close agreement with a previous report,³⁷ that the rate of thiourea oxidation is first-order with respect to each thiourea and H₂O₂ concentrations (Figure 2B). Kinetics measurements reveal that the reaction between thiourea and H₂O₂ proceeds rapidly to produce thiourea dioxide with a second-order rate constant (k₂) of 0.082 ± 0.002 M⁻¹ s⁻¹ (Figures 2B and S3).

Figure 2.

(A) UV/vis measurement of TU oxidation with H₂O₂. Representative measurement of thiourea consumption with different H₂O₂ concentrations (0.125–8 mM) as measured by A₂₃₇ change. (B) Oxidation follows a first order reaction kinetic with respect to TU and H₂O₂. Second-order rate constant, k₂ = 0.082 ± 0.002 M⁻¹ s⁻¹.

We then proceeded to test the ability of ¹³C-TU to undergo hyperpolarization using a preclinical HyperSense (Oxford Instruments) and a clinical SPINlab (GE Healthcare) polarizer. The polarization of a ¹³C-labeled substrate exploits the principle of the nuclear Overhauser effect, which transfers nuclear spin polarization from one spin bath to another (e.g., free radical species to the ¹³C nuclei of thiourea) using microwave irradiation at near absolute zero temperature (1.2 K for HyperSense, 0.8 K for SPINlab). The hyperpolarized sample was then rapidly dissolved, and the corresponding T₁ measured with a clinical 3T MRI scanner. Samples polarized with the HyperSense exhibited a polarization level of 3.2 ± 0.1% and T₁ = 30.2 ± 0.8 s, while the SPINlab produced a polarization level of 10.4 ± 1.1% and T₁ = 53.8 ± 3.7 s (Figure 3a and Figure S4). Notably, higher polarization of the sample from the SPINlab might be attributed to the use of a hand-held electromagnetic carrier, which preserves polarization during sample transfer,³⁸ and the longer T₁ measurement was due to the exclusion of gadolinium. Together, polarization by the HyperSense and SPINlab represented a 12 200- and 39 500-fold enhancement, respectively, of the ¹³C-TU signals relative to thermal equilibrium. We then evaluated ¹³C-TU for spectroscopic detection of H₂O₂ in vitro by characterizing its oxidation product(s) with ¹³C-MRS. Briefly, 4 M solutions of ¹³C-TU and 15 mM of OX063 (Oxford Instruments) radicals in 60% (v/v) glycerol were hyperpolarized at 1.2 K and irradiated with microwaves (139.88 GHz) using a HyperSense polarizer. After hyperpolarization, the sample was rapidly dissolved in 100 mM Trizma-HCl buffer (pH ∼ 7.6) to a final thiourea concentration of 50 mM and reacted with 50 μM H₂O₂ for ∼30 s before spectral acquisition with a 3T scanner. Representative spectra from the peroxide-treated ¹³C-TU showed several distinct peaks. In particular, we identified ¹³C-TUO₂ (179.6 ppm) as the oxidation product of ¹³C-TU (184 ppm) within 50 s of H₂O₂ reaction (Figure 3B and C). Additional byproducts of TU oxidation were also detected at 164.5 and 168.7 ppm, which were assigned to ¹³C-urea and ¹³C-TU₂, respectively.³⁹ Resonance frequencies of TU and the oxidized products are assigned based on the resonance of a 1 M ¹³C-bicarbonate phantom. Indeed, peroxide-mediated oxidation of TU proceeds via a mono-oxygenated intermediate (TUO), which further oxidizes to TUO₂. Additionaly, the TUO intermediates may undergo reversible disulfide (TU₂) formation in the presence of excess TU.⁴⁰ Any resulting TUO₂ can then be readily hydrolyzed to form urea under mildly basic conditions.⁴⁰ Finally, we observed a previously unidentified peak at 188.1 ppm, which likely corresponds to the ¹³C-TUO intermediate. Indeed, time course analysis of the reaction revealed a nonlinear decay in ¹³C-TUO signal at t = 35–40 s, which was followed by an increase in the ¹³C-TUO₂ signal at t = 50 s (Figure 3C, Figure S5). In contrast, a linear decay was observed in ¹³C-TU as a result of the spin–lattice relaxation time of the hyperpolarized signal (Figure S5). These findings suggest that despite the spin-lattice relaxation time, we were able to detect ¹³C-TUO oxidation to ¹³C-TUO₂ in real time. Together, these results demonstrate that the identification of a transient ¹³C-TUO intermediate is required for ¹³C-TUO₂ production and that TU can successfully detect H₂O₂ by ¹³C-MRS.

Figure 3.

In vitro ¹³C MR spectra of hyperpolarized TU. (A) ¹³C-TU was hyperpolarized using a HyperSense (preclinical) or a SPINlab (clinical) polarizer. Polarization and T₁ of the hyperpolarized TU samples were measured using a clinical 3T MR scanner and a nonselective pulse-and-acquire MRS sequence (flip-angle = 5.625°, temporal resolution = 3 s) after a 30 s delay from the dissolution due to mixing of ¹³C-TU with H₂O₂ and transfer to the magnet. (B) Time-averaged spectra of hyperpolarized ¹³C-TU showed multiple oxidation products (TUO, TUO₂, TU₂, and urea) when reacted with 50 μM H₂O₂. (C) Temporal changes of hyperpolarized TU and products due to T₁ decay and the reaction to H₂O₂ were detected from time courses of hyperpolarized ¹³C TU and the oxidation products. See SI for experimental details.

To evaluate TU for in vivo H₂O₂ sensing, we applied HP ¹³C-TU to detect the endogenous H₂O₂ level in healthy male rats. Briefly, 4 M solutions of ¹³C-TU and 15 mM AH111501 (GE Healthcare) radicals were hyperpolarized using a SPINlab polarizer, rapidly dissolved into 50 mM final thiourea concentration and intravenously injected into an anesthetized rat for imaging with a 3T MR scanner. A slice-selective ¹³C spectroscopic image over an oblique plane encompassing several major organs was acquired with free induction decay chemical shift imaging sequence using a ¹³C/¹H dual-tuned volume coil (FID CSI, field of view = 80 × 80 mm², matrix size = 16 × 16, flip angle = 10°, acquisition time = 19 s, repetition time = 75 ms), and the corresponding ¹H MRI confirmed significant ¹³C-TU resonance in well-perfused tissues of a rat, particularly in the heart, liver, and kidney (Figure 4A). However, no metabolic products from thiourea oxidation were detected above the noise level, presumably due to native antioxidant responses that kept endogenous H₂O₂ concentration low (∼10^-8 M)⁴¹ and thus undetectable by our HP probe. To assess whether thiourea oxidation could be detected in vivo, we coinjected HP ¹³C-TU with H₂O₂ into the intraperitoneal cavity of an anesthetized mouse to simulate elevated ROS level under oxidative stress. A small transmit/ receive surface radiofrequency coil was then used for localized data acquisition from the mouse liver, and a ¹³C-MRS scan of the injection site revealed several oxidized species in the liver (Figure 4B). One major oxidative product detected downfield of ¹³C-TU (184 ppm) was identified as the mono-oxygenated ¹³C-TUO intermediate (188.1 ppm). Moreover, several smaller but detectable oxidative products were assigned to ¹³C-TUO₂ (179.5 ppm), ¹³C-urea (164.5 ppm), and ¹³C-TU₂ (168.1 ppm) based on their in vitro chemical shifts. Both ¹³C-TUO and ¹³C-TUO₂ exhibited a linear rate of decay in vivo with apparent rate constants (1/T₁) of k_TUO = 0.0342 s⁻¹ and k_TUO2 = 0.0249 s⁻¹, respectively, which were similar to ¹³C-TU decay (k_TU = 0.0308 s⁻¹; Figure 4c). In addition, we examined the extent of thiourea oxidation by calculating the relative TUO₂/TUO ratios in vitro and in vivo and found that the relative TUO₂ signal is 7.9-fold lower in vivo. These results suggest that ¹³C-TUO was not actively oxidized to ¹³C-TUO₂ in vivo despite the exogenous peroxide addition and that the lack of in vivo oxidation of ¹³C-TUO might be attributed to the rapid antioxidant response to clear any excess peroxide.

Figure 4.

In vivo detection of hyperpolarized ¹³C-TU. (A) A slice-selective 2D ¹³C chemical shift imaging (CSI) and a ¹H MRI over an oblique plane that contains heart, kidney, and liver of a healthy rat, acquired using a ¹³C/¹H dual-tuned volume coil. The CSI scan started 20 s after a bolus injection (i.v.) of hyperpolarized ¹³C-TU, estimating the spatial distribution of the TU and measuring its perfusion into each organ (field of view = 80 × 80 mm², matrix size = 16 × 16, slice thickness = 15 mm). No ¹³C-oxidized product was observed. (B) ¹³C-MR spectra of a mouse liver after an injection (i.p.) of hyperpolarized ¹³C-TU with and without H₂O₂. A nonselective pulse-and-acquire MRS sequence was used (flip-angle = 10°, temporal resolution = 3 s). The localization was achieved by positioning a small ¹³C transmit/receive surface coil (diameter = 28 mm) on top of the mouse liver and the intraperitoneal cavity. Oxidized products (TUO and TUO₂) and byproducts (TU₂ and urea) were detected in vivo immediately when TU and H₂O₂ were coinjected. No oxidized product was observed with the TU-only injection. (C) All the hyperpolarized signals of ¹³C-metabolites exhibited monoexponential decays as a result of the spin–lattice relaxations (T₁). See SI for experimental details.

In summary, we have introduced a potential new ¹³C-MRSI probe to detect ROS level in real-time. Indeed, the present studies represent an ongoing effort to measure oxidative stress noninvasively. Our data demonstrate that ¹³C-TU may be a viable ¹³C-MRSI agent with important applications for real-time H₂O₂ detection. Notably, we were successful in detecting thiourea oxidation products in vivo using hyperpolarized ¹³C-TU and coinjected H₂O₂. In addition, we found the highest ¹³C-TU resonance in well-perfused organs, implicating its potential use for extracellular ROS measurements in blood and/or plasma. Future studies with ¹³C-TU in diseased animal models may provide important insights into the role of oxidative stress in pathophysiology.

Abbreviations

HP

hyperpolarized

TU

thiourea

TUO

formamidine sulfenic acid

TUO₂

thiourea dioxide

MRS

magnetic resonance spectroscopy

MRSI

magnetic resonance spectroscopic imaging