¹⁵N Hyperpolarization of Imidazole-¹⁵N₂ for Magnetic Resonance pH Sensing via SABRE-SHEATH

Abstract

¹⁵N nuclear spins of imidazole-¹⁵N₂ were hyperpolarized using NMR signal amplification by reversible exchange in shield enables alignment transfer to heteronuclei (SABRE-SHEATH). A ¹⁵N NMR signal enhancement of ∼2000-fold at 9.4 T is reported using parahydrogen gas (∼50% para-) and ∼0.1 M imidazole-¹⁵N₂ in methanol:aqueous buffer (∼1:1). Proton binding to a ¹⁵N site of imidazole occurs at physiological pH (pK_a ∼ 7.0), and the binding event changes the ¹⁵N isotropic chemical shift by ∼30 ppm. These properties are ideal for in vivo pH sensing. Additionally, imidazoles have low toxicity and are readily incorporated into a wide range of biomolecules. ¹⁵N-Imidazole SABRE-SHEATH hyperpolarization potentially enables pH sensing on scales ranging from peptide and protein molecules to living organisms.

Spectral sensing or imaging of local pH variances in vivo has been of long-standing interest for characterizing a host of pathological conditions, including various cancers.¹⁻⁶ For example, a variety of MR-based approaches using both exogenous and endogenous agents (e.g., refs (6−17)) have been investigated as less invasive alternatives to using microelectrode probes.⁶ However, sensitivity presents a significant challenge to otherwise powerful MR-based methods due to the typically low concentrations of probe molecules compared to water in vivo.

One way to combat such MR sensitivity limitations is hyperpolarization. NMR hyperpolarization techniques significantly enhance nuclear spin polarization (P), resulting in large gains in NMR signal.¹⁸⁻²⁰ One such approach is signal amplification by reversible exchange (SABRE), a technique that relies on exchange of parahydrogen (para-H₂) and to-be-hyperpolarized substrate molecules on a catalyst²¹⁻²³—in solutions or in “neat” liquids.²⁴ Polarization of target nuclear spins (e.g., ¹H,²¹¹⁵N,^25,26 or ³¹P²⁷) occurs spontaneously when the applied static magnetic field B_T is “matched” to the corresponding spin–spin couplings between the nascent para-H₂ hydride pair and the target nuclei (Figure 1a). Homonuclear (i.e., ¹H) SABRE was demonstrated first²¹ using B_T in the mT range; the approach was later extended to heteronuclei (e.g., ¹⁵N, ³¹P, etc.) via SABRE in shield enabling alignment transfer to heteronuclei (SABRE-SHEATH²⁵) utilizing B_T static fields in the μT range. Alternatives to spontaneous SABRE or SABRE-SHEATH include radiofrequency irradiation targeting level anti-crossings (LAC)²⁸ and low-irradiation generation of high tesla-SABRE (LIGHT-SABRE).²⁹ These RF-based approaches are attractive because they yield hyperpolarization directly in the magnet where detection takes place. However, the spontaneous/static-field approaches currently yield larger polarization levels, up to 10% P_15N (corresponding to >30 000-fold signal enhancement at 9.4 T). A key advantage of all SABRE hyperpolarization methods is their fast polarization buildup—achieving high P levels in only a few seconds. Moreover, spontaneous SABRE and SABRE-SHEATH are not instrumentally demanding and only require access to readily produced para-H₂ and a weak static magnetic field. Furthermore, SABRE-SHEATH addresses a critical challenge faced by all hyperpolarization techniques: Upon injection of hyperpolarized (HP) material into a system of interest, signals usually decay rapidly, with decay constants on the order of seconds up to a minute. However, with SABRE-SHEATH, long-lived ¹⁵N sites can be HP with relaxation time constants ranging from 1 min²⁶ to 10 min.³⁰ Furthermore, compared to ¹³C enrichment of leading ¹³C HP contrast agents (e.g., pyruvate-1-¹³C^31,32), spin labeling with ¹⁵N uses relatively straightforward chemistry replacing N-sites in N-heterocycles with ¹⁵N.^26,33

Figure 1.

(a) Generalized scheme of SABRE and SABRE-SHEATH hyperpolarization processes. (b) Chemical structure of the activated Ir-IMes hexacoordinate complex after activation with H₂. The complex undergoes fast exchange with para-H₂ and free imidazole-¹⁵N₂, which enables spontaneous polarization transfer from para-H₂ (in the form of Ir-hydrides) to ¹⁵N nuclei of imidazole-¹⁵N₂ in μT magnetic fields.^25,26

The development of all hyperpolarization techniques has largely been driven by their use in biomedicine to image organ function and probe metabolic processes in vivo.^20,31,34,35 While several translational challenges of conventional SABRE have been addressed recently, i.e., demonstration of SABRE in aqueous media,³⁶⁻³⁸ and implementation of heterogeneous SABRE catalysts,^39,40 most SABRE-hyperpolarized compounds studied to date have limited biological relevance (although nicotinamide,²¹ pyrazinamide, and isoniazid⁴¹ have been demonstrated). Recently, ¹⁵N heterocycles have been shown to be potent for pH imaging.⁴² In this case, hyperpolarization was performed with the well-established yet expensive dissolution-DNP (dynamic nuclear polarization)⁴³ modality and pH sensing was achieved by detecting changes in ¹⁵N isotropic chemical shifts, which are >90 ppm for the protonated and deprotonated states of the ¹⁵N-heterocycles.⁴² As a result, ¹⁵N isotropic chemical shifts of ¹⁵N-hyperpolarized probes may be ideal reporters of in vivo pH. This approach has two key advantages compared to the current HP ¹³C-bicarbonate pH sensing approach.¹⁵ First, in vivo ¹⁵N T₁ is significantly longer than that for ¹³C (e.g., ∼10 s for ¹³C bicarbonate¹⁵). Second, pH sensing using bicarbonate requires measurement and detection of both ¹³C bicarbonate and its exchanging partner ¹³CO₂ via spectroscopic imaging (MRSI)—a demanding approach with respect to SNR, because the relative signal ratio of ¹³C bicarbonate and ¹³CO₂ peaks must be measured with good precision, whereas this approach only requires accurate measurement of ¹⁵N frequency, which can be performed with relatively low SNR.

A key challenge for in vivo pH sensing is a relatively narrow pH range for the extracellular compartments for most conditions of interest, requiring that a given pH probe provide a wide dynamic range of signal response over a relatively narrow range of pH values (i.e., ∼1.5 pH units). As a result, the pH sensor must have a pK_a close to physiological pH of ∼7. Initial studies of six-membered N-heterocycles (see Supporting Information Figure S1 and ref (42)) identified only one somewhat suitable candidate: 2,6-lutidine,⁴² with pK_a ∼ 6.6. However, 2,6-lutidine is not readily amenable to SABRE-SHEATH hyperpolarization.²⁴ The pK_a of imidazole is ∼7.0—a property that has already been exploited for in vivo tumor pH imaging via proton detection without hyperpolarization.^6,44 Therefore, imidazole nitrogen-15 sites are excellent candidates for ¹⁵N HP pH sensing. Indeed, proton binding induces easily measured ¹⁵N chemical shifts of ∼30 ppm (Figure 2).⁴⁵⁻⁴⁷ Note that both ¹⁵N sites have the same chemical shift in the deprotonated form because of fast proton hopping between these two sites in aqueous media.⁴⁵⁻⁴⁷ In the protonated form, both ¹⁵N sites are equivalent and have the same chemical shift. As a result, imidazole-¹⁵N₂ is an excellent delivery vehicle, because its two ¹⁵N sites carry twice the hyperpolarization payload of (single-site) pyridine derivatives.

Figure 2.

(a) Molecular diagram of imidazole-¹⁵N₂ protonation; note that the effective molecular symmetry in unprotonated (due to fast proton hopping between two ¹⁵N sites) and protonated states results in the same ¹⁵N chemical shift of both sites. (b) Determination of imidazole-¹⁵N₂ pK_a using isotropic ¹⁵N chemical shift in aqueous solutions. (c) Selected (thermally-polarized) ¹⁵N spectra of imidazole in water used for pK_a determination. (d) ¹⁵N NMR spectrum of HP imidazole-¹⁵N₂ (∼0.1 M) in methanol:water (∼1:1) produced via SABRE-SHEATH (B_T < 0.1 μT, [catalyst] ∼ 4 mM); note the inset spectrum showing the other HP enlarged resonances: the large changes (i.e., ≥10 ppm) of a ¹⁵N chemical shift of these species are caused by the imidazole position in the hexacoordinate complex (e.g., equatorial vs axial position, Figure 1b), binding state (e.g., free vs catalyst-bound states, Figure 1b),^25,26 and protonation states.⁴² (e) ¹⁵N spectrum of a ¹⁵N signal reference. (f,g) SABRE-SHEATH optimization of magnetic transfer field B_T and temperature, respectively. All NMR spectra are recorded using a 400 MHz Bruker NMR spectrometer.

Here, ¹⁵N-SABRE-SHEATH hyperpolarization of imidazole-¹⁵N₂ is demonstrated. Figure 2b shows the exchange process of imidazole-¹⁵N₂ and para-H₂ gas on the activated Ir-IMes hexacoordinate complex of the most potent SABRE hyperpolarization catalyst to date.²³ As shown in Figure 2, ¹⁵N signal enhancement ε_15N of ∼2000-fold is detected on each of the two ¹⁵N sites in a methanol:aqueous (pH ∼ 12) buffer (∼1:1) solution of ∼0.1 M substrate utilizing only 50% para-H₂ gas and the hyperpolarization setup described previously.²⁶ Note the broad appearance of the HP NMR line in pure methanol-d₄ (Figure S3) owing to intermediate proton chemical exchange between the two ¹⁵N sites described above; the ¹⁵N NMR line is no longer broadened in aqueous solution (Figure 2d). The additional ¹⁵N HP resonances (seen as narrow lines) are due to the presence of catalyst-bound ¹⁵N imidazoles (Figure 2d inset and Figure S3a)—which have different pK_a values, protonation states, and proton exchange rates. If 100% para-H₂ would have been utilized (vs ∼50% para-H₂ utilized here), the enhancement would be effectively tripled to ε_15N ∼ 6000-fold, corresponding to P_15N ∼ 2%. Temperature and B_T(48) of the SABRE-SHEATH procedure were optimized to achieve the largest enhancements under our conditions. We note that unusually (for SABRE) high temperature (>340 K, Figure 2g) was found optimal for 15N SABRE-SHEATH in the aqueous medium (Figure 2f,g).

These results represent the highest payload (defined as the product of ¹⁵N concentration and polarization) for any SABRE-hyperpolarized compounds with the exception of ¹⁵N-nicotinamide (50 mM and P_15N ∼ 11% at ∼100% para-H₂ limit), which was achieved in pure methanol-d₄ using preactivation with pyridine,³³ whereas here, ¹⁵N SABRE-SHEATH was performed in an aqueous medium, which is known to provide lower enhancements due to lower para-H₂ solubility.³⁷ A potential solution is a further significant increase of para-H₂ pressure (compared to ∼6.5 atm used here), which could potentially enable significantly larger polarization levels,^24,26 e.g., P_15N ∼ 10% or more. ¹⁵N T₁ of imidazole-¹⁵N₂ in methanol:aqueous (pH ∼ 12) buffer (∼1:1) solution in the presence of SABRE catalyst was 24 ± 1 s at 9.4 T, whereas further reduction of methanol fraction (to an estimated value of <10% by volume) resulted in a T₁ increase to 86 ± 2 s (Figure S2) indicating that the in vivo T₁ (with the absence of both alcohol and exchangeable catalyst) could potentially exceed 1 min.⁴⁹ The ¹⁵N hyperpolarization lifetime could also be further enhanced via long-lived spin states and the use of lower magnetic fields.³⁰

Motivated by potential biomedical translation, SABRE-SHEATH hyperpolarization of imidazole-¹⁵N₂ in aqueous media was performed at several different pH values (below and above the pK_a, Figure 3) demonstrating that (i) ¹⁵N chemical shift of the HP probe indeed changes by ∼30 ppm, and (ii) the ¹⁵N NMR resonances are sufficiently narrow to discriminate minute changes in pH in the physiologically relevant range. Therefore, this HP molecular probe can potentially enable in vivo pH sensing with an estimated ∼15 ppm range covering pH range 6.5 to 7.5, and it should provide resolution of 0.1 unit of pH per 1.5 ppm of ¹⁵N shift.

Figure 3.

¹⁵N NMR spectra of imidazole-¹⁵N₂ hyperpolarized via SABRE-SHEATH at various pH values (below and above pK_a) in aqueous solutions containing <50% methanol. Note a minor shift of ∼2 ppm between resonances shown in Figure 2c (not color matched) and here due to temperature difference of ∼40 °C.

Conventional ¹H-SABRE of methanol-d₄ solution yielded ε_H ∼ 50–100-fold (Figure S3e), i.e., values lower than the corresponding ¹⁵N enhancements (Figure S3a)—in agreement with previous ¹⁵N SABRE-SHEATH studies of ¹⁵N-pyridine.²⁶ Moreover, Figure S3d also shows in situ (or “high-field”) SABRE ¹H NMR spectroscopy of imidazole-¹⁵N₂ recorded inside a 9.4 T spectrometer (the spectrum was recorded approximately 2 s after para-H₂ bubbling (conducted at 9.4 T) was stopped—note (i) the partial SABRE signal enhancement of one of the imidazole protons, manifested as the signal with negative (emissive) phase—consistent with the previously described “high-field” SABRE effect;⁵⁰ and (ii) upfield ¹H signals from intermediate hydride species formed transiently during the catalyst activation process.³⁷ Taken together, the ¹⁵N SABRE-SHEATH and ¹H SABRE results indicate that imidazole-¹⁵N₂ reversible exchange (and SABRE in general) have the same key features as the most-studied SABRE substrate, pyridine.

While d-DNP could in principle be employed for hyperpolarization of imidazole-¹⁵N₂, it is an instrumentationally demanding and expensive hyperpolarization technique, and DNP hyperpolarization processes for this class of compound typically require ∼2 h of polarization build-up.⁴²¹⁵N SABRE-SHEATH allows preparation of HP imidazole-¹⁵N₂ (and potentially other imidazole-based biomolecules) in less than a minute using a very simple experimental setup, paving the way to pH sensing (imaging and localized spectroscopy) in vivo. Furthermore, in combination with recent demonstrations of SABRE in aqueous media³⁶⁻³⁸ and in “neat” liquids,²⁴ the presented work potentially enables the hyperpolarization of ¹⁵N-imidazole moieties for structural and functional studies of peptides and proteins.^51,52

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acssensors.6b00231.

• Table summarizing pK_a values; additional figures providing quantitative measurements of pK_a using ¹⁵N NMR spectroscopy; additional experimental details and other supporting figures (PDF)

The authors declare no competing financial interest.