Design and characterization of hyperpolarized ¹⁵N-BBCP as a H₂O₂-sensing probe

Abstract

Hydrogen peroxide (H₂O₂) is a type of reactive oxygen species that regulates essential biological processes. Despite the central role of H₂O₂ in pathophysiological states, available molecular probes for assessing H₂O₂ in vivo are still limited. This work develops hyperpolarized ¹⁵N-boronobenzyl-4-cyanopyridinium (¹⁵N-BBCP) as a rationally designed molecular probe for detecting H₂O₂. The ¹⁵N-BBCP demonstrated favorable physicochemical and biochemical properties for H₂O₂ detection and dynamic nuclear polarization, allowing non-invasive detection of H₂O₂. In particular, ¹⁵N-BBCP and the products possessed long spin-lattice relaxation times and spectrally resolvable ¹⁵N chemical shift differences. The performance of hyperpolarized ¹⁵N-BBCP was demonstrated both in vitro and in vivo with time-resolved ¹⁵N-MRS. This study highlights a promising approach to designing a reaction-based ¹⁵N-labeled molecular imaging agent for detecting oxidative stress in vivo.

Graphical Abstract

Reactive oxygen species (ROS) play essential roles in regulating a wide range of biological and physiological processes.^1–4 Among various ROS metabolites, hydrogen peroxide (H₂O₂) exists in the highest concentration (10⁻⁷–10⁻⁸ M range) and is stable under physiological conditions.^{5, 6} Normally, the production of oxidative species is balanced with an innate antioxidant defense system. However, a disturbance in the redox balance often triggers the accumulation of H₂O₂, which can cause oxidative stress and oxidative damage to biomolecules.^7–9 Elevated level of H₂O₂ has been considered as a hallmark for various diseases, including cancer,^10–12 inflammation,^{13, 14} and neurodegenerative disorders.^15–17 The broad actions of H₂O₂ in pathophysiology highlight the importance of imaging the distribution of H₂O₂ at biological concentrations for diagnostic and therapeutic applications. Accordingly, efficient sensing probes of oxidative stress applicable for preclinical and clinical models are in high demand.

Elegant development of optical imaging probes selective for H₂O₂ has been disclosed.^18–21 The challenge yet remains regarding the investigation of dynamic H₂O₂ activities with high spatial and temporal resolution. Magnetic resonance spectroscopy and imaging (MRS/MRI) with hyperpolarized probes has been established to be a clinically relevant imaging platform for investigating in vivo metabolism.^22–30 Therefore, hyperpolarized probes that are sensitive to oxidative stress would present great potential for non-invasive detection of H₂O₂ in real-time and for translation to patients. Several ROS-sensing hyperpolarized probes have been developed. For instance, hyperpolarized ¹³C-thiourea³¹ and [1-¹³C]-benzoylformic acid³² were proposed as H₂O₂-sensing probes, yet both showed limited spin-lattice relaxation times (T₁). Compared to carbon-13-labeled probes, nitrogen-15-labeled agents are relatively underexplored as hyperpolarization probes. Among elegant examples reported, the ¹⁵N-labeled hyperpolarized probes have shown great potentials to deliver a long T₁ and a wide range of ¹⁵N chemical shift (900 ppm) for an extended scope of chemical complexity.^33–48 Notably, the T₁ of ¹⁵N-trimethylphenylammonium was reportedly over 400 s although the oxidative reaction afforded a small chemical shift (~1.5 ppm).

We envisioned that exogenous ¹⁵N-probes for reaction-based H₂O₂-sensing can be rationally designed and fine-tuned to achieve the desired molecular reactivity and imaging properties. Toward this goal, several physicochemical and biochemical properties were considered in the pursuit of practical ¹⁵N-probe candidates. First, the ¹⁵N-center in the molecular probe must possess T₁ values in the order of minutes to afford a sufficient imaging window. Second, the probe and its reaction product should have a distinguishable chemical shift difference for detection accuracy. Third, the probe must be biocompatible in terms of aqueous solubility, stability and cytotoxicity with high reaction selectivity and kinetics. In this work, we designed ¹⁵N-boronobenzyl-4-cyanopyridinium (¹⁵N-BBCP) as a novel H₂O₂-sensing probe and investigated hyperpolarized ¹⁵N-BBCP on its ability for sensitive and selective detection of H₂O₂ with time-resolved ¹⁵N-MRS both in vitro and in vivo.

Results and discussion, Experimental

Design and synthesis of ¹⁵N-BBCP as a H₂O₂ sensor

The structure of ¹⁵N-BBCP comprised aryl boronate as a H₂O₂-sensing unit and ¹⁵N-nitrile as a signaling unit, connected through a pyridinium linkage (Figure 1A). The synthesis of ¹⁵N-BBCP was achieved by conjugation of the readily prepared boronic ester 2 with ¹⁵N-4-cyanopyridine (¹⁵N-CP), which was prepared from the isotope-enriched ¹⁵N-hydroxylamine hydrochloride and 4-pyridinecarboxaldehyde (Figure 1B).

Figure 1. Design and synthesis of a novel ¹⁵N-labeled H₂O₂-sensing probe.

A) Proposed two-step H₂O₂-sensing mechanism involving boron oxidation followed by 1,6-elimination. B) Synthetic routes to ¹⁵N-CP and ¹⁵N-BBCP (R = H), with bromide ion (Br⁻) as the counterion.

In our design principle of ¹⁵N-BBCP as a reaction-based H₂O₂ sensor, the oxidation of aryl boronic ester/acid upon H₂O₂ was expected to unmask a phenol intermediate (hydroxybenzyl-4-cyanopyridinium, ¹⁵N-HBCP). Simultaneously, the 1,6-elimination of ¹⁵N-HBCP would release ¹⁵N-CP and quinone methide (QM) (Figure 1A). We envisioned that both ¹⁵N-HBCP and ¹⁵N-CP can be captured as signal readouts in the H₂O₂-sensing studies. For the signaling unit, ¹⁵N-labeled nitrile was selected as the hyperpolarizable imaging handle as it was expected to provide the long T₁ lifetimes desirable for a molecular probe, based on previous hyperpolarization studies of ¹⁵N-benzonitrile.⁴⁸ The H₂O₂ sensing will rely on the detection of distinct chemical shifts from the ¹⁵N-nitrile in ¹⁵N-BBCP, intermediate ¹⁵N-HBCP, and final product ¹⁵N-CP. We anticipated the chemical shift difference between ¹⁵N-BBCP and ¹⁵N-HBCP would be small due to the structural similarity. Thus, the pyridinium linker is a key component in our design, as it would lead to a larger chemical shift change when cleaved to pyridine. Indeed, thermal scans using ¹⁵N2-urea (0.0 ppm) as a reference revealed a significant chemical shift difference of 13.9 ppm between ¹⁵N-BBCP (194.8 ppm) and ¹⁵N-CP (180.9 ppm). Furthermore, the incorporation of the pyridinium linker that bears a positive charge is favorable in increasing the aqueous solubility of the probe.

Biochemical analysis of BBCP

We investigated the applicability of BBCP for in vivo studies and characterized its physicochemical and biochemical properties. BBCP demonstrated high aqueous solubility and stability in physiological buffer solution (PBS, pH 7.4) (Figure S1). The selectivity toward H₂O₂ over other biologically relevant ROS and reactive nitrogen species (RNS) was examined. Upon reaction of BBCP at various timepoints (5–30 min, room temperature), the reaction with H₂O₂ outperformed other ROS/RNS and showed a significant conversion to 4-cyanopyridine (CP) based on LC/MS analysis (Figure 2A). In addition, hypochlorite (OCl⁻) and peroxynitrite (ONOO⁻) showed moderate reactivity with BBCP, in which reactions of aryl boronates with OCl⁻ and ONOO⁻ have been reported.^{49, 50} However, the physiological concentration of peroxynitrite is minuscule (~1 nM for [ONOO⁻] vs ~100 nM for [H₂O₂])⁵¹, allowing the use of BBCP as a selective H₂O₂-sensing probe.

Figure 2. Evaluation of BBCP as a H₂O₂-sensing probe.

A) Response of BBCP (200 μM) with various ROS/RNS (2 mM) in PBS (pH 7.4). Concentration of CP measured by LC/MS after 5, 10, 15, and 30 min of reaction. B) LC/MS trace data of reactions of BBCP (200 μM) + H₂O₂ (1 mM, 5 equiv) in PBS (pH 7.4) after 5–65 min. Peak area in reference to BBCP at Abs = 280 nm (blue: BBCP; purple: HBCP; green: CP; yellow: quinone methide, QM). Data represented is qualitative, not quantitative. C) Effects of buffer pH on oxidation of BBCP to HBCP and fragmentation to CP. BBCP (100 mM) reaction with 1 equiv H₂O₂ in PBS + 10% D₂O monitored by ¹H NMR

The reaction mechanism of BBCP with H₂O₂ was also investigated. Treatment of BBCP with H₂O₂ generated CP and QM, with transient phenol species HBCP observed by the LC/MS measurements (Figure 2B), suggesting the proposed two-step oxidation and fragmentation as the sensing mechanism. To confirm the involvement of 1,6-elimination pathway, we prepared a negative control probe S3 with an ethylene linker group that is unable undergo the 1,6-elimination step (Scheme S2). The control reactions with S3 confirmed that H₂O₂ does not directly cleave the benzylic center, further supporting the subsequent two-step mechanism shown in Figure 1A. In addition, the UV/Vis kinetic measurements revealed that the reaction of BBCP with H₂O₂ proceeds rapidly with a second-order rate constant of k₂ = 12.6 ± 0.52 M⁻¹ min⁻¹ (25 °C, PBS, pH 7.4) (Figure S3). Interestingly, the oxidation and fragmentation rates were dependent on the pH of the reaction solution. Investigation of the effects of buffer pH revealed the oxidation rate is slowed at pH <6.0, and the fragmentation rate was impeded to a greater extent in acidic solutions. These data suggest the biological pH may have an effect on the H₂O₂ reactivity with ¹⁵N-BBCP and fragmentation to ¹⁵N-CP, therefore affecting the sensitivity of H₂O₂ imaging.

The cytotoxicity of BBCP was evaluated as hyperpolarized molecular probes are typically administered as a bolus in high concentrations (20–80 mM) to account for >10-fold dilution in blood. Cytotoxicity assay showed BBCP has minimal toxicity (>90% cell viability) at concentrations up to 10 mM after 6 h of treatment (Figure S4).³⁸ Altogether, BBCP fulfills essential properties as a molecular imaging agent for in vivo sensing of H₂O₂, including high aqueous solubility/stability, high ROS selectivity, fast reaction kinetics, and low cytotoxicity.

Dynamic nuclear polarization of ¹⁵N-BBCP

Polarization level and T₁ relaxation time of ¹⁵N-BBCP were estimated in vitro. ¹⁵N-BBCP samples (3.4 M, 4:1 DMSO:glycerol with 15-mM Ox063) were prepared and polarized using a SPINlab DNP polarizer (GE Healthcare) that operates at ~0.8 K and 5 T.⁵² After 3–4 h of polarization, the frozen sample was rapidly dissolved in hot (130 °C) dissolution media (0.1g/L EDTA, pH = 7.4 in DI water), producing 5.5–6.0 mL of 20-mM hyperpolarized ¹⁵N-BBCP solution. The in vitro T₁ of ¹⁵N-BBCP was estimated as 340.1 s at 1 T (Spinsolve, Magritek) and 122.8 s at 3 T (Achieva, Philips Healthcare) after correcting radiofrequency (RF) sampling losses (Figure 3A). The liquid-state polarization level at the time of dissolution was estimated as 17.3% from the 3-T data by comparing to thermal scans. Hyperpolarized signal was retained when 40-mM hyperpolarized ¹⁵N-BBCP was injected to healthy rats as a bolus (Figure 3B). No metabolic products were observed presumably due to low H₂O₂ concentrations regulated by the native antioxidative responses in well-balanced physiological conditions.

Figure 3. Hyperpolarization of ¹⁵N-BBCP and the two-step sensing mechanism of H₂O₂.

A) In vitro time series of hyperpolarized ¹⁵N-BBCP spectra. B) In vivo observation of hyperpolarized ¹⁵N-BBCP spectra acquired from a rat abdomen at 3 T (insert: temporal changes of ¹⁵N-BBCP peak at 194.8 ppm). C) In vitro time series of hyperpolarized ¹⁵N-CP at 3 T. D) In vitro time series and time courses of hyperpolarized ¹⁵N-BBCP and products, observed after mixing with H₂O₂ (blue: ¹⁵N-BBCP; purple: ¹⁵N-HBCP; green: ¹⁵N-CP). E) In vivo reaction-based detection of H₂O₂ via ¹⁵N-HBCP production from hyperpolarized ¹⁵N-BBCP.

Reaction-based H₂O₂-sensing using hyperpolarized ¹⁵N-BBCP

Hyperpolarized ¹⁵N-BBCP was further tested with H₂O₂ in vitro to validate its reaction-based H₂O₂-sensing performance (Figure 3D). Immediately upon the addition of H₂O₂, hyperpolarized signals from ¹⁵N-BBCP as well as its oxidation product, ¹⁵N-HBCP 0.6 ppm upfield, and the fragmentation product, ¹⁵N-CP 13.9 ppm upfield, appeared, demonstrating that both oxidation and fragmentation occur rapidly and the T₁ values of ¹⁵N-species are sufficiently long to monitor the H₂O₂-sensing reactions in real-time. A separate in vitro experiment with hyperpolarized ¹⁵N-CP confirmed the long T₁ of ¹⁵N-CP (184.3 s) at 3 T (Figure 3C).

Finally, the feasibility of ¹⁵N-BBCP to detect H₂O₂ in vivo was investigated in a rat model at the clinical 3 T MRI. With an intravenous bolus injection of 40-mM hyperpolarized ¹⁵N-BBCP, following an intraperitoneal injection of H₂O₂, both ¹⁵N-BBCP and ¹⁵N-HBCP peaks appeared from the imaging slab that included the liver and kidneys (Figure 3E). Despite the small chemical shift difference, the accumulation of ¹⁵N-HBCP production was clearly observed in the later time points. Considering the hyperpolarized ¹⁵N-BBCP presented in healthy rats did not provide detectable metabolic products (Figure 3B), the appearance of ¹⁵N-HBCP peak indicates in vivo reaction of ¹⁵N-BBCP with H₂O₂. However, ¹⁵N-CP was not observed, suggesting the 1,6-fragmentation step has a lower reaction conversion in vivo in the two-step sensing mechanism, as implied in Figure 2C.

Potentials, limitations and future works

With currently limited reports on in vivo studies of ¹⁵N-probes,^{47 15}N-BBCP presents exciting progress as an exogenous reaction-based ¹⁵N-probe extended to in vivo imaging. While ¹⁵N-BBCP was rationally designed to achieve the desired molecular reactivity, several factors need to be considered and improved for future studies. For instance, BBCP showed moderate toxicity at a high concentration (10 mM) (Figure S4), and it should be noted the reaction byproduct quinone methide (QM) is reported to be toxic.⁵³ Therefore, next-generation reaction-based ¹⁵N-probes should consider the toxicity of the probes and have innocuous byproducts.

While a larger chemical shift difference is highly desired due to the low gyromagnetic ratio of ¹⁵N, both in vitro and in vivo results suggest that ¹⁵N-probes can be utilized to detect 1 ppm or less of chemical shift difference using a conventional clinical 3-T system. MR spectroscopic imaging approach may improve spectral separation of ¹⁵N-BBCP and ¹⁵N-HBCP by reducing intra-voxel B0 inhomogeneity, but at the expense of signal-to-noise ratio.

Unlike the in vitro setting that contained sufficient H₂O₂, in vivo application of hyperpolarized ¹⁵N-BBCP to scavenge H₂O₂ did not detect ¹⁵N-CP while its conversion to ¹⁵N-HBCP was still minimally observed. This is probably due to the limited concentration of H₂O₂ or because the provided H₂O₂ was rapidly decomposed into water and oxygen, as well as a slow fragmentation rate to generate ¹⁵N-CP. Pathological conditions that persist in elevated levels of H₂O₂ or an oxidative burst may produce a detectable amount of ¹⁵N-CP. Thus, further verification with disease models or a chemical stimulation that induces high H₂O₂ production such as cancer treated with β-lapachone⁵⁴ will be needed to evaluate in vivo performance of hyperpolarized ¹⁵N-BBCP as a H₂O₂-sensing probe.

Conclusions

In summary, we designed and characterized ¹⁵N-BBCP as a reaction-based H₂O₂-sensing probe that is applicable to dynamic nuclear polarization. Currently, hyperpolarized ¹⁵N-probes have been largely unexplored for in vivo imaging. Our study demonstrates a novel exogenous reaction-based ¹⁵N-molecular probe that has been applied in vivo. The probe exhibited ideal properties, including high aqueous solubility, stability, low cytotoxicity, and rapid reaction kinetics. In addition, ¹⁵N-BBCP retained a long polarization lifetime that allowed monitoring of the reaction with H₂O₂. Notably, the presented work demonstrates the possibilities of using reaction-based ¹⁵N-probes to image and characterize oxidative stress for future diagnostic and therapeutic applications.