¹⁵N₄-1,2,4,5-tetrazines as potential molecular tags: Integrating bioorthogonal chemistry with hyperpolarization and unearthing para-N₂

A bioorthogonal reaction–based HP-MR strategy affords novel hyperpolarized products from tetrazine, including para-¹⁵N₂ gas.

Abstract

Hyperpolarized magnetic resonance (HP-MR) is a powerful, sensitive, and noninvasive approach to visualize molecular structure, function, and dynamics in vitro and in vivo. Current applications of HP-MR mostly rely on hyperpolarization of target compounds in dedicated hyperpolarizers because biomolecules can typically not be hyperpolarized directly in vivo. The injected hyperpolarized probes often undergo multiple metabolic pathways in living systems, and it remains challenging to localize and identify specific targets with high chemical selectivity. To address these current limitations in HP-MR, we report a novel hyperpolarization tagging strategy that integrates bioorthogonal chemistry and hyperpolarization to achieve the specific hyperpolarization of targets. This strategy is demonstrated by studies of hyperpolarized ¹⁵N₄-1,2,4,5-tetrazines, which undergo rapid and selective cycloaddition with cyclooctyne to provide hyperpolarized ¹⁵N₂-containing cycloaddition products and hyperpolarized ¹⁵N₂ gas. This work not only suggests great potential of ¹⁵N₄-1,2,4,5-tetrazines as molecular tags in HP-MR imaging (HP-MRI) but also supports the production of hyperpolarized para-¹⁵N₂ gas, a biologically and medically innocuous gas with great potential for HP-MRI. This bioorthogonal reaction–based hyperpolarization tagging strategy enables a new class of in vitro and in vivo applications.

INTRODUCTION

Hyperpolarized magnetic resonance (HP-MR) has been developed to overcome the low sensitivity of conventional MR, a limitation that arises from poor nuclear magnetization at thermal equilibrium (1). For example, at 7 T and 310 K, equilibrium ¹H nuclear magnetization is just 0.0024% (2); other interesting nuclei, such as ¹³C and ¹⁵N, have even lower gyromagnetic ratios, and thus, their MR detection is even more challenging. Hyperpolarization techniques induce nonequilibrium magnetization of target nuclei and therefore raise detectable signals by multiple orders of magnitude (3, 4). Particularly attractive is HP-MR imaging (HP-MRI) using heteronuclei (for example, ¹³C or ¹⁵N), which offers more comprehensive structural information than ¹H nuclear MR (NMR) and allows signal detection for extended periods of time due to their longer relaxation time compared to ¹H (5). Examples geared toward tracing metabolism and biological functions in living organisms include endogenous molecular species and derivatives, such as pyruvate (6), glucose (7), and amino acids (6, 8, 9). Other molecular probes include ¹⁵N-pyridine derivatives (10) for measuring pH and ¹³C-labeled drugs for tracking pharmacokinetics (11).

Despite these exciting advances, typical hyperpolarized probes, when subjected to in vivo applications, may readily undergo multiple metabolic pathways and cannot be directed to specific targets with high chemical selectivity. Furthermore, current HP-MRI mostly relies on ex vivo hyperpolarization of molecular targets. Exploiting HP-MRI for endogenous macromolecules in living systems remains an important yet unsolved challenge.

Here, we develop a novel hyperpolarization tagging strategy that uses hyperpolarizable reactive precursors that can undergo bioorthogonal cycloaddition reactions to achieve specific identification and localization of target molecules (Fig. 1A). Bioorthogonal chemistry is a powerful approach for the study of biomolecules in real time in living systems. It relies on rapid chemical ligation reactions between two bioorthogonal functional groups that are added to a biological sample. These two bioorthogonal partners react with each other in a chemoselective manner, which means that they are inert to any other chemical entity present. Meanwhile, the bioorthogonal chemistry should occur fast, in quantitative yield, and should be compatible with living systems (12, 13). Thus, bioorthogonal reaction–based hyperpolarization tagging appears as an attractive strategy that can selectively highlight and localize the target-containing bioorthogonal partner. Ideally, this marriage of hyperpolarized MR with bioorthogonal chemistry would enable molecular tracking of any biomolecule with the high signal-to-noise ratio afforded by hyperpolarization, simply by tagging it with the hyperpolarized reaction partner. Therefore, the development of a hyperpolarized probe that could participate in rapid bioorthogonal ligation has immense potential as a generally applicable and chemically specific tag for HP-NMR and HP-MRI.

Fig. 1. A novel bioorthogonal reaction–promoted hyperpolarization tagging strategy.

(A) Illustration of spin-hyperpolarized tagging via bioorthogonal chemistry. Bioorthogonal ligation between reactive partners X and Y incorporates the hyperpolarized chemical motif (red) within the target molecule (yellow). (B) ¹⁵N₄-1,2,4,5-tetrazine as a molecular tag and its dual roles in hyperpolarization and bioorthogonal ligation. First, the ¹⁵N nuclei in ¹⁵N₄-1,2,4,5-tetrazine are hyperpolarized by SABRE-SHEATH, and then, the hyperpolarized tetrazine undergoes rapid IEDDA with a strained azadienophile, leading to the hyperpolarized ¹⁵N₂-cycloaddition product and hyperpolarized ¹⁵N₂ gas.

We demonstrate the bioorthogonal reaction–promoted hyperpolarization of selected targets using 1,2,4,5-tetrazines as hyperpolarizable precursors (Fig. 1B). The use of 1,2,4,5-tetrazines is particularly advantageous because of its dual role in both hyperpolarization and bioorthogonal reactions. First, we demonstrate that the hyperpolarization of ¹⁵N-labeled tetrazines can be achieved by SABRE-SHEATH (Signal Amplification by Reversible Exchange in Shield Enables Alignment Transfer to Heteronuclei) (14–16). This hyperpolarization method takes advantage of recent developments using spin order transfer from para-hydrogen (para-H₂) (17–21) at very low fields (approximately 1% of the Earth’s field), using a comparatively simple setup (16, 22). Second, we expect that the hyperpolarized 1,2,4,5-tetrazines react selectively and rapidly with strained azadienophiles by inverse-demand Diels-Alder (IEDDA) reaction, one of the fastest bioorthogonal reactions reported (13, 23–28). 1,2,4,5-Tetrazines are well studied, and various tetrazine-tagged biomolecules have been successfully used in vivo and in vitro (24–27). Here, we show that ¹⁵N₄-1,2,4,5-tetrazine contributes to both hyperpolarization and bioorthogonal ligation by generating the hyperpolarized cycloaddition target product. Furthermore, the ¹⁵N₄-1,2,4,5-tetrazine–based IEDDA generates hyperpolarized ¹⁵N₂ gas, a typically ignored byproduct of the tetrazine ligation. Here, our hyperpolarization approach also allows for selective access to para- and ortho-¹⁵N₂, two fundamentally interesting spin isomers of ¹⁵N₂ (29). In particular, para-¹⁵N₂ gas is a biologically and medically innocuous gas with mathematical properties similar to para-H₂ (30). Although para-¹⁵N₂ has no net signal, even weak transient bindings to transition metal catalysts (including some biocatalysts) would be expected to unlock the spin order and create magnetization (31). However, hyperpolarized para-¹⁵N₂ has not been reported to the best of our knowledge. In summary, our study entails the hyperpolarization of ¹⁵N₄-1,2,4,5-tetrazines using SABRE-SHEATH, followed by cycloaddition of the hyperpolarized tetrazine with an azadienophile, which enables the generation of hyperpolarized ¹⁵N₂-labeled products and hyperpolarized ¹⁵N₂ gas (Fig. 1B).

RESULTS AND DISCUSSION

For this proof-of-concept study, our investigation focused on ¹⁵N-labeled 3-phenyl-1,2,4,5-tetrazine, namely, 3-phenyl-(6-H)-¹⁵N₄-1,2,4,5-tetrazine 1a and 3-phenyl-(6-D)-¹⁵N₄-1,2,4,5-tetrazine 1b (Fig. 2A), to evaluate their potential as a dual tag for hyperpolarization and bioorthogonal reaction. These tetrazines were synthesized from ortho-ester precursors with ¹⁵N₂-hydrazine hydrate, as fully described in the Supplementary Materials.

Fig. 2. Tetrazine hyperpolarization.

(A) Structures of studied ¹⁵N₄-1,2,4,5-tetrazines 1a and 1b. (B) Single-shot hyperpolarized spectra of tetrazines 1a and 1b at magnetization or singlet modes, with peak identifications, observed enhancement (ε), and polarization level (p). Depending on the magnetic field at which hyperpolarization was induced, in-phase signal (magnetization) or antiphase signal (singlet) was observed. Enhancement values (ε) and polarization levels (p) were obtained by comparison of the hyperpolarized spectrum to a thermal reference spectrum of the respective ¹⁵N₄-1,2,4,5-tetrazine. a.u., arbitrary units; ppm, parts per million. (C) T₁ and T_s lifetime curves for 1a and 1b. Measurement at 0.3 mT. Sample as a solution of ¹⁵N₄-1,2,4,5-tetrazine (1.5 mM), pyridine (1.0 mM), and Ir(COD)(IMes)Cl [COD, 1,5-cyclooctadiene; IMes, 1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene; 0.15 mM] 0.15 mM] in methanol-d₄ (400 μl).

Hyperpolarization of ¹⁵N₄-1,2,4,5-tetrazines

Hyperpolarization of 1a and 1b was examined by standard SABRE-SHEATH procedure, as reported in our recent studies (22). Two different hyperpolarized states for ¹⁵N₄-1,2,4,5-tetrazine 1a were observed, depending on the chosen magnetic field at which the para-H₂ gas is applied to the sample (Fig. 2B). At very low magnetic fields (~0.4 μT), the ¹⁵N spin pairs of the tetrazine are hyperpolarized in the triplet states and display in-phase signal upon detection at 8.45 T (that is, magnetization is hyperpolarized). Conversely, at a relatively broad range of slightly elevated magnetic fields (~0.2 mT < B < ~50 mT), we observe antiphase signals after a 90° pulse. In this case, scalar order is hyperpolarized in the tetrazine spin pairs, associated with singlet states on ¹⁵N spin pairs; upon transfer to high magnetic field for detection, this scalar order is transformed into antialigned magnetization (I·S is adiabatically converted to I_z-S_z; see the Supplementary Materials for details) (32). Such a field-dependent selection of hyperpolarized states corroborates our previous work on the hyperpolarization of ¹⁵N₂-diazirines and ¹³C₂-pyridyl acetylenes (33).

For tetrazine 1a, the signal enhancement over 8.45-T thermal measurements is up to 3000-fold (0.9% polarization). At 0.3 mT, the magnetization has a relaxation constant T₁ of 1.4 ± 0.1 min, and at the same field, the relaxation constant of the scalar order of the ¹⁵N spin pairs (T_s) is 2.7 ± 0.3 min, indicating that the scalar order is protected from relaxation and has a lifetime about two times longer than magnetization. We also measured the enhancement level and lifetimes of the tetrazine 1b, expecting that the replacement of the tetrazine proton with deuterium could affect enhancement and lifetime (34). The hyperpolarized scalar order yielded 2900-fold enhancement, with T_s calculated to be 2.1 ± 0.7 min at 0.3 mT (that is, no significant change in lifetime within the error of the measurement). Note that it was not possible to create the hyperpolarized magnetization for deuterated compound 1b (or to measure its T₁) because the quadrupolar deuterium quenches hyperpolarization at microtesla fields (16).

Bioorthogonal reactions of hyperpolarized ¹⁵N₄-1,2,4,5-tetrazines and cyclooctyne

With hyperpolarization of tetrazines 1a and 1b established successfully, we examined whether the hyperpolarization could be retained in reaction products of IEDDA reactions. We chose strained cyclooctyne bicyclo[6.1.0]non-4-yn-9-ylmethanol 2 (35, 36) as the reaction partner of tetrazine in our studies and confirmed the sufficiently rapid rate of this cycloaddition reaction at room temperature (completed within seconds; see the Supplementary Materials for details). In particular, the formation of a single pyridazine product allows for a straightforward analysis, excluding potential complexity from multiple products, which arise when other known azadienophile partners, such as trans-cyclooctene, are used (13, 23). We obtained the thermal ¹⁵N reference spectra of both reactant tetrazine 1a (before reaction) and product 3a (after IEDDA reaction) from which we observe a clear distinction between the reactant and product by their ¹⁵N chemical shifts (Fig. 3A).

Fig. 3. IEDDA reaction and hyperpolarization transfer.

(A) Thermal spectra of tetrazine 1a compared to thermal spectra of cycloaddition product 3a. Upon the cycloaddition reaction, a noticeable difference in chemical shift is observed on both nitrogen atoms (that is, N1 and N2 in 1a versus N1′ and N2′ in 3a). (B) Spectra of hyperpolarized 1a and that obtained after the addition of 2. Hyperpolarized 3a and ¹⁵N₂ are observed. (C, E, and G) Representative T₁ or T_s decay measurements for product 3a magnetization, 3a singlet order, and 3b singlet order, respectively. Note the presence of observable ¹⁵N₂ in the postaddition spectra when hyperpolarizing magnetization (C) but the absence of this peak in hyperpolarized singlet (E and G), which strongly suggests that singlet ¹⁵N₂ has been generated in these experiments. (D, F, and H) Lifetime measurement data, exponential fit of the data, and calculated T₁ or T_s values for product 3a magnetization, 3a singlet order, and 3b singlet order, respectively. In the hyperpolarization-cycloaddition experiments, para-H₂ was bubbled into a solution of ¹⁵N₄-1,2,4,5-tetrazine 1a or 1b (1.5 mM), pyridine (1.0 mM), and Ir(COD)(IMes)Cl (0.15 mM) in methanol-d₄ (400 μl), and then, a solution of 2 (1.5 equiv.) in methanol-d₄ (200 μl) was added. The sample was held at 0.3 mT for a variable amount of time before transport to the magnet for detection.

To best monitor the hyperpolarized signals involved in the cycloaddition, we modified our conventional SABRE-SHEATH hyperpolarization setup and enabled direct injection of compound 2 into the solution of hyperpolarized tetrazine by adding a capillary tube into the pressurized NMR tube (see the Supplementary Materials for details). Note that all signals observed for 3 in the following experiments originate from tetrazine 1 and not from the SABRE-SHEATH of the product. Control experiments attempting SABRE-SHEATH hyperpolarization of compound 3a (the reaction product) provided no signal enhancements under the same conditions or even at higher temperatures with continuous bubbling of para-H₂.

We first examined the cycloaddition of 1a after hyperpolarizing magnetization (that is, triplet states were hyperpolarized by bubbling at 0.4 μT before cycloaddition). After addition of a solution of 2 to a sample of hyperpolarized 1a at 0.3 mT and subsequent transfer to high field, we observed sharp, in-phase peaks at 372 ppm that matched the position and pattern of the peaks observed in the thermal spectra of product 3a (Fig. 3B). An additional signal was detected at 310 ppm, which corresponds to hyperpolarized ¹⁵N₂ gas (thermal spectrum of ¹⁵N₂ in CD₃OD is provided in the Supplementary Materials). These data reinforce that the IEDDA reaction of hyperpolarized ¹⁵N₄-tetrazine 1a successfully generates hyperpolarized ¹⁵N-containing products, including both ¹⁵N₂-pyridazine 3a and ¹⁵N₂ gas. On the basis of the spectrum of hyperpolarized products, we calculated an enhancement of 540-fold over their thermal spectra. The magnetization lifetime T₁ for 3a was determined to be 13 ± 4 s, substantially shorter than that of tetrazine 1a (Fig. 3, C and D).

Next, we examined the reaction of 1a after hyperpolarizing scalar order by bubbling at 0.3 mT (Fig. 3, E and F). After the solution of 2 was injected to the solution of hyperpolarized tetrazine at the same field (0.3 mT) and transferred to high field for detection, we observed antiphase peaks at 372 ppm, with 140-fold signal enhancements over thermal spectra. The T_s for 3a was determined to be 13 ± 2 s, very similar in magnitude to T₁. This contrasts with the significant difference observed between T₁ and T_s for the tetrazine 1a.

Furthermore, we examined the effects of deuteration in the cycloaddition reaction and products. As explained above, only scalar order could be hyperpolarized on the deuterated tetrazine 1b. Therefore, we were restricted to measurements of T_s in the deuterated product 3b. Very similar to the observation in the reaction of tetrazine 1a, antiphase peaks at 372 ppm were detected with a similar enhancement level of 180-fold. Encouragingly, a significantly longer lifetime T_s of 24 ± 6 s was obtained from the deuterated product (Fig. 3, G and H). This is expected because the deuterium couples less (6.5-fold less) into the ¹⁵N₂ spin system than ¹H (all relevant J-coupling parameters of 1a, 1b, 3a, and 3b are provided in the Supplementary Materials).

One key observation is that the ¹⁵N₂ gas signal at 310 ppm (Fig. 3, B and C) is absent with hyperpolarized scalar order (Fig. 3, E and G) after the cycloaddition reaction. The absence of nitrogen gas signals in these data provides indirect evidence for the generation of an intriguing new hyperpolarized species, para-nitrogen (para-¹⁵N₂), that should have very similar spin properties to para-H₂, an extraordinary “quantum reagent” in a pure spin state. The H₂ molecule has an antisymmetric singlet state (para-H₂) and three symmetric “ortho” states. ¹H atoms are fermions; hence, they are antisymmetric with respect to exchange, so only para-H₂ can be in the (symmetric) ground rotational state J = 0 (37). The separation between J = 0 and J = 1 (in temperature units) is 174 K, so para-H₂ dominates at equilibrium at low temperatures. At room temperature in pure form, para-H₂ is extremely stable (100% para-H₂ drops to just 85% after 30 weeks) (30). With mathematical properties similar to para-H₂, para-¹⁵N₂ can therefore be expected to be exceptionally long-lived as a promising MRI agent of clinical safety. However, para-¹⁵N₂ cannot be prepared in the same way as para-H₂ because the rotational constant of N₂ is small and the nitrogen freezes before it achieves significant para excess under current conditions. Thus, the bioorthogonal reaction of hyperpolarized ¹⁵N₄-1,2,4,5-tetrazines represents a novel approach to permit characterization of this new quantum reagent.

To fully develop this bioorthogonal reaction–based hyperpolarization labeling strategy, we evaluated its feasibility under aqueous conditions toward in vivo biomedical applications. It should be noted that the SABRE-SHEATH hyperpolarization in water currently remains challenging, and its development is obstructed by a number of technical issues, including poor solubility of H₂ gas and iridium catalyst in water. Research to address these problems has been undertaken (16, 38–41), including a strategy we recently reported to improve SABRE-SHEATH hyperpolarization in water using a water-soluble iridium catalyst and increased temperatures (60° to 70°C) (42). However, this strategy was not successful in our current studies because of the instability of the tetrazine at elevated temperatures. Encouragingly, with 3:1 CD₃OD/D₂O as a cosolvent system, we were able to achieve hyperpolarization of tetrazine 1a with ~900-fold signal enhancement at 50°C and also detect the hyperpolarized signal from the cycloaddition product 3a under these conditions (see the Supplementary Materials for details). These preliminary results pave the way toward application of this strategy under aqueous conditions, although more optimal SABRE-SHEATH hyperpolarization in water remains to be achieved.

CONCLUSION

We report a novel MR strategy by integrating bioorthogonal reactions and hyperpolarization. This strategy is demonstrated on the hyperpolarized ¹⁵N₄-1,2,4,5-tetrazines, which undergo rapid cycloaddition with cyclooctyne to provide hyperpolarized cycloaddition products and hyperpolarized ¹⁵N₂ gas. This work suggests great potential of ¹⁵N₄-1,2,4,5-tetrazines as powerful molecular tags in NMR and MRI, with dual roles in hyperpolarization and bioorthogonal ligation. Excitingly, the observations in the current study support the production of hyperpolarized ¹⁵N₂ gas in both its ortho and para spin isomers. Future studies will be directed toward systematic optimizations on the ¹⁵N-tetrazine cycloaddition–based hyperpolarization tagging strategy and characterization of para-¹⁵N₂ gas.

MATERIALS AND METHODS

Hyperpolarization setup

A high-pressure gas delivery system was specially built for the SABRE-SHEATH process. Normal H₂ gas was converted to para-H₂ (~90.2% enrichment) using a commercial para-H₂ generator. The para-H₂ gas was delivered to the sample solution through a capillary at a pressure of about 0.680 MPa (100 psi). The magnetically shielded environment was prepared using a three-layer μ-metal magnetic shield. A solenoid placed inside the shield controlled the magnetic field via manual adjustment of the voltage using a dc voltage output and a resistor. A separate capillary for the injection of a secondary solution was added adjacent to the para-H₂ delivery line, with a valve placed at the site of injection to seal the pressure when bubbling gas (see the Supplementary Materials for the diagram).

Sample preparation

For typical hyperpolarization experiments, a solution of ¹⁵N-enriched tetrazine (1a or 1b; 1.5 mM), pyridine (1.0 mM), and Ir(COD)(IMes)Cl (0.15 mM) in methanol-d₄ (400 μl) was prepared (43).

Tetrazine hyperpolarization procedure

The Ir catalyst was preactivated by bubbling para-H₂ through a solution of tetrazine, pyridine, and Ir catalyst (sample preparation described above) for 30 min. Following preactivation, the tetrazine was hyperpolarized, either by magnetization or by singlet order.

To hyperpolarize magnetization, the solution was placed inside the magnetic shield, with the magnetic field adjusted to 0.4 μT (using a solenoid of 430 mm with 205 turns and a voltage of 7.5 V across 11.4 kilohms). After 3 min of bubbling of para-H₂, the gas flow was stopped, and the sample was manually transferred from the low field to an 8.5-T spectrometer for signal readout as quickly as possible. This manual transfer took ~8 s, and a 90° pulse-acquire sequence was used for readout.

To hyperpolarize the singlet, the sample was placed at a magnetic field of 0.3 mT, and para-H₂ was bubbled through the solution for 3 min. As described in the above procedure, the sample was then manually transferred to an 8.5-T spectrometer as quickly as possible and detected using a 90° pulse-acquire sequence.

Tetrazine hyperpolarization and cycloaddition reaction procedure

For the hyperpolarization of the cycloaddition products 3a and 3b, a solution of tetrazine (1a or 1b, respectively), pyridine, and Ir catalyst in methanol-d₄ was first hyperpolarized at 0.4 μT or 0.3 mT, depending on which spin order was studied (solution preparation and hyperpolarization procedure described above). After hyperpolarization, the para-H₂ gas flow was stopped, and the pressure was released through the exhaust outlet, after which the injection valve was quickly opened to inject a solution of cyclooctyne 2 (4.5 mM) in methanol-d₄ (200 μl) (1.5 equiv. of 2 with respect to tetrazine). Injection was completed in less than 1 s, and the sample was shaken for 3 s to reach complete reaction, visually evidenced by the color change from pink (that is, the color of the tetrazine) to transparent. The sample was then manually transferred to an 8.5-T spectrometer for product signal readout.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/3/eaar2978/DC1

Synthesis and cycloaddition reactions of 1,2,4,5-tetrazines

Hyperpolarization experiments

¹H, ¹³C, and ¹⁵N spectra

fig. S1. ¹H NMR comparison between tetrazine and cycloaddition product.

fig. S2. Experimental setup for hyperpolarization and hyperpolarized reaction experiments.

fig. S3. Hyperpolarized signal decay of magnetization and singlet at variable concentrations.

fig. S4. Hyperpolarization of magnetization and singlet as a function of magnetic field.

fig. S5. Comparison of the originally hyperpolarized singlet and diluted signal.

fig. S6. Small-tip-angle spectra of the hyperpolarized cycloaddition product 3a.

fig. S7. SABRE-SHEATH experiment using methanol-d₄/D₂O mixture as solvent.

table S1. Magnetization and singlet enhancements and lifetimes at variable concentrations.

References (44, 45)