Parahydrogen in Reversible Exchange Induces Long-Lived 15N Hyperpolarization of Anticancer Drugs Anastrozole and Letrozole

Abstract

Hyperpolarization modalities overcome the sensitivity limitations of NMR and unlock new applications. Signal Amplification By Reversible Exchange (SABRE) is a particularly cheap, quick and robust hyperpolarization modality. Here, we employ SABRE for simultaneous chemical exchange of parahydrogen and a nitrile containing anti-cancer drugs (letrozole or anastrozole) to enhance ¹⁵N polarization. Distinct substrates require unique optimal parameter sets, including temperature, magnetic field, and the magnetic field profile. The fine tuning of these parameters for individual substrates is demonstrated here to maximize ¹⁵N polarization. After optimization, including usage of pulsed μT fields, the ¹⁵N nuclei on common anti-cancer drugs, letrozole and anastrozole, can be polarized within 1–2 minutes. The hyperpolarization can exceed 10%, corresponding to ¹⁵N signal enhancement of over 280,000-fold at the clinically relevant magnetic field of 1 T. This sensitivity gain enables polarization studies at natural abundant ¹⁵N enrichment level (0.4%). Moreover, the nitrile ¹⁵N sites enable long-lasting polarization storage with [¹⁵N]T₁ over 9 minutes enabling signal detection from a single hyperpolarization cycle for over 30 minutes.

Graphical Abstract

Nuclear Magnetic Resonance (NMR) is a leading tool for qualifying, quantifying, and monitoring molecules and their dynamics. The advantage of this non-destructive technique stems from the inherent chemical shift resolution afforded by NMR, which offers functional and structural information. However, NMR has inherently low sensitivity compared to other spectroscopic methods. This is because of the minute thermal spin polarization (<10⁻⁵ at 3T for ¹⁵N), even at high magnetic fields. Hyperpolarization modalities increase the sensitivity of NMR by perturbing thermal spin polarization towards unity. Thus, the increase in signal afforded by hyperpolarization unlocks new doors for more sensitive applications using NMR.

Currently, the dominant hyperpolarization modality employed in tandem with NMR is Dynamic Nuclear Polarization (DNP). ^1–10 DNP generates high levels of hyperpolarization on a broad range of substrates, including water, and can directly hyperpolarize site specific locations on proteins. ^11,12 However, in addition to being a lengthy process (on the order tens of minutes), instrumentation for DNP is expensive (>1M), requiring cryogenic temperatures, a superconducting magnet, and high powered microwave sources.¹³ In contrast, Signal Amplification By Reversible Exchange (SABRE),^14,15 a Para-Hydrogen Induced Polarization (PHIP) technique,^16–18 can produce hyperpolarized (HP) substrates quickly (~1 minute or less) with little instrumentation cost (~$20k).

Nonetheless, SABRE often generates lower levels of hyperpolarization and has a more limited substrate scope compared to DNP. Thus, a significant effort has been made by the SABRE community to not only optimize the hyperpolarization process for distinct substrate groups but to expand the substrate scope to various biologically relevant molecules.^19,20 Previous work has paved the way for SABRE methods to be employed as a tool for fast, simple and efficient sensitivity boosters that can readily pair with current analytical methods. For instance, in addition to pioneering ligand-protein studies with DNP, recently Hilty and colleagues demonstrated the utility of using SABRE for the determination of ligand-protein binding.^21,22 In this work, ligands are first hyperpolarized in methanol. Then, the hyperpolarized solution is mixed with a protein solution at a Y-junction prior to entering an aqueous flow cell. Furthermore, work from Tessari and colleagues showed the feasibility of detecting amino acid and other metabolites in biological fluids (e.g., urine) using SABRE.^23–25

SABRE utilizes a transition metal catalyst that simultaneously and reversibly binds parahydrogen (p-H₂) and a target substrate. During the SABRE process, a temporary J-coupling network forms between the p-H₂ derived hydrides and target nuclei on the substrate. The active SABRE complex and the corresponding J-coupling network is illustrated in Figure 1A. Spin order can be transferred to target nuclei through the J-coupling network by RF pulses^26–30 or field cycling to low magnetic fields that approach level anti crossings (LACs).^31–33 SABRE efficiency is controlled by exchange rates, J-couplings, and Larmor frequencies of the spin system. Therefore, distinct spin systems (different substrates) will have a unique optimal parameter set. Fine tuning these parameters for individual substrates is instrumental for efficient SABRE hyperpolarization with the primary key benchmark of level of attainable polarization, P (¹⁵N polarization is denoted as [¹⁵N]P in this document).

Figure 1.

[A] Illustration of parahydrogen and a drug reversibly binding to a transition metal complex forming a temporarily J-coupling network, and [B] the anti-cancer drugs, letrozole and anastrozole, with target ¹⁵N nuclei (at natural isotopic abundance) highlighted in orange.

A number of FDA-approved drugs have been HP via SABRE to date including: metronidazole^34–36 ornidazole,³⁷ nimorazole,³⁸ pyrazinamide, isoniazid,³⁹ clotrimazole, fluconazole, voriconazole⁴⁰ and dalfampridine.^41,42 These drugs target hypoxia, tuberculosis, fungal and bacterial infections, and multiple sclerosis. Moreover, they are HP via SABRE using five- or six-membered N-heterocyles rather than nitrile moiety studied in this work. Here, we utilize SABRE in Shield Enables Alignment Transfer to Heteronuclei (SABRE-SHEATH),^43,44 and coherent SABRE-SEATH using pulsed μT fields, to extend the ¹⁵N heteronuclear scope to two bulky, nitrile⁴⁵ containing anti-cancer drugs, letrozole and anastrozole, shown in Figure 1B. Both drugs are current generation aromatase inhibitors employed to combat breast cancer⁴⁶ and are on the World Health Organizations Model List of Essential Medicine, making them highly attractive biological compounds. Heteronuclei, such as ¹⁵N, are attractive hyperpolarization targets because they have virtually no background signal and are associated with long T₁ relaxation times.^19,47 For a general overview mapping the progress of ¹⁵N hyperpolarization using SABRE, see the following review articles.^20,48–50 Therefore, we envision the utility of these HP compounds to be used for more sensitive drug screening or as potential exogenous HP contrast agents. Moreover, because aromatase inhibitors (including letrozole and anastrazole studied here) are excreted via urine, we envision that these drugs and the products of their metabolism can potentially be detected in urine samples, paving the way to new applications in the context of cancer management.

Specifically, we perform hyperpolarization build-up and lifetime studies and then optimize the hyperpolarization process with respect to solution temperature and polarization transfer field (PTF), including pulsed μT-fields. At the optimum PTF, the hydride-substrate J-coupling is most efficient at transferring spin order from hydrides to substrates. The optimum PTF is dictated by a (Level Anti Crossing) LAC between hydride and substrate spin states. The LAC is established when the hydride-hydride J-coupling and the frequency difference between hydride and substrate spins are very close to each other. In the case of hydride-¹⁵N systems, like the one studied here, the optimal PTF is typically found at 0.3 µT.

The presented series of experimental optimization studies reaches high (>10%) and long lived ([¹⁵N]T₁ over 9 mins) hyperpolarization on the anti-cancer drugs at natural isotopic abundance. This work not only broadens the scope of accessible substrates for SABRE-SHEATH, but also presents a methodical pathway for hyperpolarizing new SABRE substrates.

In this work, we focused on optimizing the hyperpolarization of chemically symmetric nitrile groups. Although ¹⁵N enhancements were observed on the triazole substituent (a triplet peak from the coordinating nitrogen) on both drugs, they were vastly overshadowed by the ¹⁵N enhancements observed on the nitrile substituents. In the present chemical system, the nitrile substituents hyperpolarize far more efficiently than the nitrogen containing heterocycles. First, we studied hyperpolarization build-up and the hyperpolarization lifetime (T₁ constant) for the nitrile substituents on letrozole and anastrozole. We then optimized the hyperpolarization process with respect to solution temperature and PTF. Lastly, we applied a dynamic PTF, referred to as coherent SABRE-SHEATH,⁵¹ to yield maximum hyperpolarization. As the drugs were not isotopically enriched, polarization values were determined using a ¹⁵N enriched pyridine sample thermalized at 9.4 T (see SI for more advanced and complete details on [¹⁵N]P calculations) following a previously reported method.⁴³

Hyperpolarization modalities are often associated with lengthy polarization build-up times that often bottleneck the hyperpolarization process.⁵² In SABRE, the polarization build-up time is relatively fast. Accordingly, we first conducted polarization build-up experiments on the nitrile substituents. Letrozole and anastrozole had similar build-up profiles, both reaching a steady state around 1 minute, with ¹⁵N T_b (build-up constant) of 21.7±1.4 seconds and 14.6±0.9 respectively, summarized in Table 1. The experimental build-up data was fit using a mono-exponential function and are shown together in Figure 2A. For experimental reproducibility, a bubbling time of 90 s was chosen for all succeeding experiments.

Table 1.

Hyperpolarization build-up and lifetime constants for ¹⁵N on the nitrile substituents of letrozole and anastrozole with errors given from the standard deviation of the individual mono-exponential fits.

Anti-Cancer Drug	Build-up Constant, Tb [s] @ 0.3 μT	[15N]T1 [s] @ 9.4 T	[15N]T1 [s] @ 1 T	[15N]T1 [s] @ 0.3 μT
Letrozole	21.7 ± 1.4	9.3 ± 0.1	554.1 ± 12.7	34.7 ± 3.0
Anastrozole	14.6 ± 0.9	21.6 ± 0.2	420.8 ± 7.3	27.8 ± 2.9

Figure 2.

Optimization studies for the nitrile substituents nitrogen on letrozole and anastrozole. [A] Experimental polarization build-up data recorded at 40°C overlaid with mono-exponential fits. [B] Experimental [¹⁵N]T₁ data, recorded at 40°C and stored at 1 T, overlaid with mono-exponential fits. [C] Temperature sweep from 25°C to 50°C with 5°C increments. The polarization is listed on the left axis while the molar polarization (the product of polarization and spin density) is listed on the right. Experiments were done at natural isotopic abundance (¹⁵N is ~0.4% naturally abundant) using a substrate concentration of 30 mM (see the methods section for more details on sample composition). [D] A static field sweep from 0 to 3.6 μT with 0.3 μT increments with PTFs both parallel (+) and antiparallel (−) to the detection field.

Long lifetimes of HP states are important for monitoring biological (in vivo) and biochemical processes (in vitro: e.g., protein and drug-protein studies) that have slower or downstream interactions of interest. Previous studies demonstrated that [¹⁵N]T₁ constants can be extended by storage at an appropriate field.^19,40,53 Often, using a storage field at an intermediate strength (relative to traditional PTFs and detection fields), around 1 T, can significantly extended [¹⁵N]T₁ times, while noting that the relaxation properties and their field dependencies vary widely depending on the individual molecules and associated spin systems. Thus, in addition to hyperpolarization build-up times, we elucidate hyperpolarization lifetimes at three different fields. Hyperpolarization lifetime studies were investigated at high field (9.4 T), low field (0.3 μT) and an intermediate field (1 T).

Agreeing with trends characterized in literature, short lifetimes are observed at low and high fields, while longer lifetimes are observed at intermediate fields.^19,40,53 At high field, the [¹⁵N]T₁ for letrozole and anastrozole was measured to be 9.3±0.1 s and 21.6±0.2 s respectively, as reported in Table 1. These values are slightly lower than the [¹⁵N]T₁ values at low field for both letrozole and anastrozole, which were measured to be 34.7±3.0 s and 27.8±2.9 s, respectively. Figure 2B shows the remarkably long ¹⁵N HP lifetime data recorded at 1 T overlaid with mono-exponential fits yielding [¹⁵N]T₁ of 554.1±12.7 s (>9 minutes) and 420.8±7.3 s (>7 minutes) for letrozole and anastrozole, respectively.

Relaxation at high field is likely dominated chemical shift anisotropy (CSA), which scales quadratically with magnetic field. CSA is mitigated at low fields, which can explain why we observe longer [¹⁵N]T₁’s when transitioning to a low storage field. However, other sources of relaxation become dominant at low field strengths. Specifically, when the system is not replenished with fresh p-H₂ at 0.3 μT, spin mixing during binding events at low fields leads to relaxation (i.e. the hyperpolarization mechanism is reverted and the hydrides may act as a polarization sink). Evidence supporting this claim is found when comparing the [¹⁵N]T₁ of letrozole and anastrozole at low field. The suspected faster exchanging substrate (experiencing more binding events), anastrozole, has a shorter [¹⁵N]T₁ than the slower exchanging substrate (experiencing fewer binding events), letrozole.

At an intermediate field of 1 T, both CSA and spin mixing are mitigated. Instead, dipolar interactions between nuclei become the dominant sources of relaxation¹⁹, which are weak for the isolated ¹⁵N spin in nitriles. Nitrogen on a nitrile lacks nearby spin ½ nuclei that could induce strong dipolar relaxation. Thus, mitigating high and low field relaxation sources and selecting an isolated target nucleus gives rise to an exceptional environment for magnetization storage. Other methods used to extend T₁’s have been shown in literature, for example isotopic labelling where protons are replaced for deuterium,⁴⁷ or the usage of long-lived singlet states.⁵⁴

In addition to relaxation parameters, another critical parameter in SABRE are the exchange rates on the SABRE complex. The chemical exchange of p-H₂ and a target substrate need to be similar to the frequency of spin order transfer, governed by the temporary J-coupling network on the active SABRE catalyst, for efficient spin transfer.

There are several methods reported in literature that have been used to control the exchange of both p-H₂ and a target substrate. These methods consist of either increasing or decreasing the electronic density on the metal centre or by increasing or decreasing the steric crowding around the metal centre.^55,56 These strategies not only allow for more efficient spin order transfer, but also enable the hyperpolarization of new substrates. Work led by Duckett and colleagues demonstrated that SABRE can be used to hyperpolarize α-keto acids and nitrites by modulating exchange with the incorporation of a co-substrate.^57,58 In addition to the referenced strategies, direct heating or cooling can be used to modulate exchange rates.^40,59 Following these findings, a temperature sweep was performed on both target drugs. We chose to sweep from room temperature (25 °C) to 50 °C with 5 °C increments, shown in Figure 2C, to shed light on the exchange dynamics.

Prominently, we observed an increase in polarization as temperature was increased. Letrozole reaches maximum hyperpolarization at 50 °C (limited by experimental setup), while anastrozole reaches a maximum at 40 °C, but levels off at further elevated temperatures. The difference in observed maxima is likely due to the chemical environment around the nitrile substituents on both drugs. Anastrozole has additional steric bulk close to the nitrile substituent compared to letrozole, likely leading to faster exchange. Therefore, the faster exchanging substrate, anastrozole, will be expected to reach maximum hyperpolarization at lower temperatures relative to the faster exchanging substrate, letrozole. We note that the substrate exchange rate also controls the hydride exchange rate, as we have shown with ab initio calculations,⁶⁰ because hydride exchange does require substrate exchange events for monodentate ligands. These findings not only support that temperature modulation represents a key hyperpolarization parameter, but additionally suggests that relative exchange rates can be unveiled through HP temperature studies.

In traditional SABRE-SHEATH experiments, the sample is subjected to a very low (<1 μT) static PTF. At sub µT fields, LACs between coupled spin states arise. At a LAC, spin mixing is most efficient, aiding in polarization build-up.^31,43 For this reason, we investigated the μT regime to unveil the most efficient PTF. We swept from 0 to 3 μT with 0.3 μT increments with static PTFs aligned parallel (+) and antiparallel (−) to the detection field, as shown in Figure 2D. Hyperpolarization was most efficiently generated at ±0.3 μT for both drugs.

However, recent advances reported in literature demonstrate that a static field only generates a fraction of the hyperpolarization of a dynamic pulse (field). ^61–64 The referenced work has shown significant boosts in hyperpolarization by applying shaped PTF to the sample during polarization build-up in a variety of distinct patterns.^61–64 On this notion, we implemented a previously published pulse sequence referred to as coherent SABRE-SHEATH, originally used by Lindale et al. to increase hyperpolarization on acetonitrile (chemically similar to the nitrile substituents on letrozole and anastrozole).⁵¹

Coherent SABRE-SHEATH first evolves spin order using an efficient PTF, B_P, for a short time, τ_P, and then quickly switches to an elevated storage field, B_S, where little to no evolution is expected, for a lengthier duration, τ_S, that is pre-optimized (see SI for τ_S optimization) to allow fresh p-H₂ to associate with the catalyst. A graphical representation of the pulse sequence is shown in Figure 3A. A plot sweeping through discrete τ_P values, while B_P, B_S, and τ_P remain constant, is shown in Figure 3C for both letrozole and anastrozole.

Figure 3.

[A] Graphical representation of coherent SABRE-SHEATH pulse sequence, [B] NMR signal of anastrozole when τ_P = 51 ms, and [C] applying coherent SABRE-SHEATH pulse sequence sweeping from τp = 0 to τp = 250 ms while keeping B_P, B_S, and τ_P constant (B_P = 0.5 μT, B_S = 30 μT and τ_S = 200 ms).

Notably, the pulse sequence drives higher polarization than what can be generated using a static field. Both compounds reach maximum enhancements when τ_P = 51 ms. For anastrozole, the maximum enhancement corresponds to [¹⁵N]P over 10% (at natural isotopic abundance and mM concentration) and the corresponding spectrum is shown in Figure 3B. The boost in [¹⁵N]P is due to more efficient spin order transfer generated by turning the usually incoherent spin order transfer into a coherently driven, unidirectional mechanism. In addition to boosting the hyperpolarization level, the coherent SABRE-SHEATH data of Figure 3B also elucidates the hyperpolarization transfer dynamics.

In conclusion, this work demonstrates a systematic optimization study for nitrile moieties of two bulky anti-cancer drugs achieving high degrees of [¹⁵N]P with long HP state lifetimes.

First, the polarization build-up was investigated for both letrozole and anastrozole. Both anti-cancer drugs had similar build-up constants, reaching a [¹⁵N]P steady state in ~1 minute. For comparison, DNP has a build-up time on the order of tens of minutes to an hour.¹³ Additionally, by exploring a clinically relevant field of 1 T, we were able to achieve [¹⁵N]T₁ constants of ~9 minutes and ~7 minutes on letrozole and anastrozole, respectively. Next, we explored exchange dynamics of the SABRE system by modulating the solution temperature and swept through the μT regime to elucidate the optimal PTF. Maximum enhancements were measured at 50 °C for letrozole and 40 °C for anastrozole, corresponding to [¹⁵N]P of 8.3% and 7.6%, respectively, at the experimental optimal PTF of 0.3 μT. We also showed that controlling the solution temperature, in a small window (25 °C), can more than double the net [¹⁵N]P. Finally, we pushed the hyperpolarization to over 10% on anastrozole (at natural isotopic abundance and mM concentration), by applying coherent SABRE-SHEATH, showing the importance of exploring non-static fields.

The feasibility of studying SABRE hyperpolarization at natural abundance of ¹⁵N substantially streamlines the experimental workflow and potentially enables screening of many other nitrile-containing drugs, biomolecules, etc. This work lays a foundation for more sensitive drug screening and trace analysis using high-resolution NMR spectroscopy.

Materials and Methods

Chemicals:

Letrozole and anastrozole were purchased from Tokyo Chemical Industry Co., LTD, and Sigma Aldrich, respectively, and used as delivered. Deuterated methanol was purchased through Cambridge Isotope Laboratories and degassed prior to experimental use. The pre catalyst, [Ir(IMes)(COD)Cl] (IMes = 1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene, COD = 1,5-cyclooctadiene), was synthesized in lab with commercially available starting materials following a previously published procedure.⁶⁵

Sample Preparation:

All samples are prepared using standard Schlenk line conditions, maintaining an oxygen free environment. Samples are prepared with 30 mM substrate (letrozole or anastrozole) and 3 mM pre-catalyst in 500 μL deuterated methanol. Samples are then transferred into a 7” medium wall NMR tube (Wilmad 524-PV-7) and connected to our in-house, fully automated, pneumatic shuttling system.⁶⁶ Samples are then subjected to 100 psi of parahydrogen and allowed to activate, by bubbling, for 5 minutes prior to experimentation.

Hyperpolarization Build-up:

Samples are first shuttled above the detection field to a polarization transfer field (PTF = 0.3 μT) and are then subjected to bubbling for a distinct time before shuttling back down to high field (9.4 T) for detection.

Hyperpolarization Lifetime:

Samples are first shuttled above the detection field to a polarization transfer field (PTF = 0.3 μT) and were then subjected to 90 s of bubbling. Samples are then stored at a desired field for various times before detection. The PTF and detection field are used as low and high storage fields, respectively, while the fringe field from the detection field is used as an intermediate storage field (~1 T).

Temperature Dependence:

Samples are first heated to a desired temperature in the detection field using the Bruker VT interface. Samples are then shuttled to a PTF (0.3 μT) and are subjected to 90 s of bubbling. Samples are not actively heated while bubbling, thus the sample experiences light cooling towards room temperature during bubbling. Samples are then shuttled down to high field for detection.

Static Field Dependence:

Samples are shuttled to a desired PTF controlled by a DC power supply and house-built solenoid coil inside mu-metal shields. Samples are then subjected to 90 s of bubbling within the PTF and shuttled down to high field for detection.

Coherent SHEATH:

Samples are shuttled to a dynamic PTF controlled by a DC power supply and home-built solenoid coil inside mu-metal shields. The dynamic PTF alternates between two distinct fields, an “evolution field”, denoted Bp, and a “storage field”, denoted Bs, for different durations, τp and τs, respectively, n times for 90 seconds. The dynamic PTF is generated using a simple circuit board, TTL lines, and solid-state relays. In this method, a pulse sequence can be developed using Bruker software to automate the process. Largest signal enhancements were observed at Bp = 0.5 μT while maintaining a Bs of ~30 μT, which was the highest achievable field in our experimental design.