¹⁵N MRI of SLIC-SABRE hyperpolarized ¹⁵N-labelled pyridine and nicotinamide

Abstract

Magnetic Resonance Imaging (MRI) is a powerful non-invasive diagnostic method extensively used in biomedical studies. A significant limitation of MRI is its relatively low signal-to-noise ratio, which can be increased by hyperpolarizing nuclear spins. One promising method is Signal Amplification By Reversible Exchange (SABRE), which employs parahydrogen as a source of hyperpolarization. Recent studies demonstrated the feasibility to improve MRI sensitivity with this hyperpolarization technique. Hyperpolarized ¹⁵N nuclei in biomolecules can potentially retain their spin alignment for tens of minutes, providing an extended time window for the utilization of the hyperpolarized compounds. In this work, we demonstrate for the first time that radio-frequency-based SABRE hyperpolarization techniques can be used to obtain ¹⁵N MRI of biomolecule 1-¹⁵N-nicotinamide. Two image acquisition strategies were utilized and compared: Single Point Imaging (SPI) and Fast Low Angle SHot (FLASH). These methods demonstrated opportunities of high-field SABRE for biomedical applications.

Entry for the Table of Contents

¹⁵N Signal Amplification By Reversible Exchange hyperpolarization of biomolecules were carried out in situ in a MRI scanner. This allowed us to perform ¹⁵N MRI of ¹⁵N-pyridine and 1-¹⁵N-nicotinamide using two different pulse sequences: Single Point Imaging and Fast Low Angle SHot (FLASH). The demonstrated method is a promising approach for biomedical applications.

Signal Amplification By Reversible Exchange^[1] (SABRE) is a rapidly developing parahydrogen-based hyperpolarization approach.^[2] Although parahydrogen (pH₂) itself is silent for Nuclear magnetic Resonance (NMR)), simultaneous exchange of pH₂ and the substrate on an Ir-based organometallic complex provides polarization transfer from pH₂–derived hydrides to the substrate, and results in an enhancement of the nuclear spin polarization (P) by several orders of magnitude.^[3–7]

At first, SABRE was demonstrated using pyridine and nicotinamide as substrates.^[1] SABRE was performed at magnetic fields of ≈10 mT simply by shaking a sample under pH₂ atmosphere while the signal was acquired later at the high magnetic field of an NMR spectrometer (9.4 T). These simple “shake and run” experiments at low magnetic fields provided high enhancements of the ¹H signal by a factor of 550 and 345 for pyridine and nicotinamide, respectively. In addition, ¹H MRI of pyridine was obtained with balanced steady-state free precession^[8] (b-SSFP) pulse sequence.

In another experiment, SABRE was used to continuously hyperpolarize substrates^[9] at a low magnetic field (6.5 mT). This approach was used to enable MRI at Earth’s magnetic field. Continuous hyperpolarization^[10] resolves the problem of short hyperpolarization lifetimes (T₁-relaxation time is on the order of a few seconds for protons).

The hyperpolarization and imaging of substrates in situ, i.e. at the same, constant magnetic field, was shown as well.^[11] This approach allowed to accelerate the imaging by shortening the repetition time from 8 s to 0.2 s and to increase the spatial resolution in a way that additional dynamic processes (e.g., pH₂ flow motion) became observable.

Hyperpolarized protons have a relatively short T₁, meanwhile, heteronuclei offer a significantly longer relaxation time. ¹⁵N T₁ relaxation values of three to twenty minutes were reported.^[12–17] These comparatively long lifetimes significantly increase the time available for sample purification,^[18,19] in vivo administration, distribution, metabolism and MRI. The first reports of ¹⁵N polarization by dissolution DNP^[20] show a low efficiency of direct ¹⁵N polarization (P~ 4%) with super long build-up times about 2 hours.^[13] However, low temperature cross-polarization to nitrogen-15 during dissolution DNP conditions^[21] allows to achieve about 20% polarization for urea molecule with moderate build-up times (15 min).^[22] Despite the high polarization level, the amount of polarized substance is extremely low (100μL, 0.8M) that along with long build-up times and high cost of DNP instrument limit possible utilization of DNP polarized substances for ¹⁵N MRI.

EXPERIMENTAL DESIGN

SABRE showed great potential for heteronuclear MRI: ¹³C,^{[1,7,23] 15}N,^{[24] 19}F^[25,26] and ³¹P^[27,28] images of SABRE polarized substrates were obtained. Here, we focused on ¹⁵N nucleus because it can be used in biomedical studies, for example, in pH sensing^[24,29] or metabolic imaging.^[30–32]

Many SABRE studies were based on polarization transfer at microtesla magnetic fields. These fields are beneficial because polarization transfer to heteronuclei occurs spontaneously in strongly coupled spin systems. The approach based on the use of low magnetic fields was termed SABRE - Shield Enables Alignment Transfer to Heteronuclei (SABRE-SHEATH)^[33,34] and more than 30% ¹⁵N-polarization was demonstrated.^[18,34,35] However, this approach has two shortcomings:^{[24,33,35–37]} a need of fast magnetic field alteration and creation of polarizer with precise ultralow magnetic field.

In another approach, termed HF-SABRE, the spontaneous polarization transfer from pH₂-derived hydrides to ¹⁵N nuclei and MR observation are carryied out in situ at a constant, high magnetic field (~10 T).^[38] Here, pH₂ is continuously bubbled through the sample, which is placed in the magnet of NMR or MRI, and spontaneous polarization transfer occurs while high-frequency electro-magnetic fields are played out. Although some shortcomings of SABRE-SHEATH were mitigated by HF-SABRE, the ¹⁵N-polarization enhancement factors were significantly lower than those provided by SABRE-SHEATH.^[38,39]

A previous report on MRI of ¹⁵N HF-SABRE^[40] revealed that HF-SABRE generates (a relatively low) anti-phase polarization, which is difficult to image due to partial signal cancelation. Therefore, HF-SABRE will likely be limited to in vitro imaging and the calibration of hardware or radio-frequency (RF) pulse sequences. Moreover, continuous pH₂ bubbling was required to obtain 2D MRI with extensive signal averaging.

To improve the efficiency of SABRE polarization at high magnetic fields, several polarization transfer techniques employing RF irradiation were developed: Alternating Delays Achieve Polarization Transfer SABRE (ADAPT-SABRE),^[41] SABRE Insensitive Nuclei Enhanced by Polarization Transfer (SABRE-INEPT),^[42] Low Irradiation Generation of High-Tesla SABRE (LIGHT-SABRE),^[43] adiabatic RF-SABRE,^[44] Spin Lock Induced Crossing SABRE (SLIC-SABRE),^[45] and Quasi-resonance SABRE (QUASR-SABRE).^[46] Some of these approaches may deliver ¹⁵N HP comparable with the one achieved by SABRE-SHEATH,^[46] and also mitigate SABRE-SHEATH shortcomings.

In this work, we present a strategy for performing imaging of SLIC-SABRE hyperpolarized compounds inside an MRI scanner for potential biomedical applications. SLIC-SABRE sequence is an improved version of LIGHT-SABRE where fast singlet – triplet mixing on the Ir-complex is taken into account that makes that variant more robust than LIGHT-SABRE. It was achieved by introduction of a 90^{o 1}H-RF pulse before CW^[45] (see Fig. 2,C and Fig. 3,C). Note, that by repeating the SLIC-SABRE block several times polarization is accumulated in the form of free substrate see (SI,Fig. S2, B).

Figure 2.

¹⁵N single-point MRI of 0.1 M ¹⁵N-Py (A) and 0.1 M ¹⁵N-NA (B) in methanol-d₄ with 5 mM Ir(COD)(IMes)Cl hyperpolarized by means of SLIC-SABRE with a 7 sccm pH₂ flow rate at 9.4 T. 2D images were acquired using SLIC-SABRE hyperpolarization and SPI spatial encoding and acquisition elements (C). Total scan time was 5.5 min (A) and 27 min (B). The signal void (circle) corresponds to the pH₂ supply capillary. Experiments were carried out at room temperature. The number of SLIC-SABRE cycles, n, was 1 for ¹⁵N-Py and 5 for ¹⁵N-NA. In both cases, k-space was acquired once, acquisition spectral width (SW) was 12 kHz, spatial resolution was 565×565$\mathrm{ }$μm²/pixel. Field of view was 0.9×0.9 cm², acquisition matrix was 16×16, It was zero-filled to 128×128 before the image reconstruction.

Figure 3.

¹⁵N FLASH-MRI of 1 M ¹⁵N-Py with 50 mM of Ir(COD)(IMes)Cl (A) and 0.1 M ¹⁵N-NA with 5 mM Ir(COD)(IMes)Cl (B) in methanol-d₄ hyperpolarized by SLIC-SABRE with 35 sccm pH₂ flow rate at 9.4 T. 2D images were acquired using SLIC-SABRE hyperpolarization and FLASH spatial encoding and acquisition elements (C). Repetition time (TR) was 3.1 ms, echo time time (TE) was 1 ms. Experiments were carried out at room temperature. The number of SLIC-SABRE pumping cycles, n, was 10 and no signal averaging was employed. Acquisition spectral width (SW) was 10 kHz, spatial resolution was 0.15×2.4 mm²/pixel (A) and 0.3×4.8 mm²/pixel (B). Field of view was 1.9×1.9 cm² (A) and 3.8×3.8 cm² (B), Acquisition matrix was 128×8, it was were zero-filled to matrix size of 128×128 (A) and 256×256 (B) before image reconstruction.

Specifically, we hyperpolarized ¹⁵N-pyridine (¹⁵N-Py) and 1-¹⁵N-nicotinamide (¹⁵N-NA) via SLIC-SABRE at the high field of an MRI scanner. The parameters for the SLIC-SABRE of ¹⁵N-Py were similar to those optimized and reported before: amplitude of RF-pulse, ν₁ = 5 Hz, offset from the center of bound ¹⁵N-Py resonance, Δ_{rf $\cong$} 17 Hz and duration of continuous wave (CW) RF-pulse, t_cw = 1.17 s.^[47] For ¹⁵N-NA, we used the same parameters of SLIC-SABRE (ν₁, Δ_rf, t_cw) because when attached to active SABRE-complex it has similar J-coupling constants with hydride protons coming from pH₂. Additionally, we optimized the number of pumping cycles and pH₂ flow rate, as described previously^[47] (see Fig. S2, SI). All necessary parameters of the experiments are listed in the Tab. S4 (SI).

The resulting polarization enhancement factors for free substrate were: ε₁₅N ~ 1,374 for ¹⁵N-Py (P ≈ 0.3 %) and ε₁₅N ~ 834 (P ≈ 0.2 %) for ¹⁵N-NA at 7 T (Fig. 1). Optimization of the SLIC-SABRE pulse sequence parameters and the calculation of the signal enhancement are given in SI. In the Ref.^[45] the time period free from RF excitation to refresh system with pH₂ was used, but omitted in this paper. Introduction of such stage can potentially increase the level of hyperpolarization.

Figure 1.

¹⁵N NMR spectra of ¹⁵N-Py (A) and ¹⁵N-NA (B) hyperpolarized by SLIC-SABRE. Strongly enhanced signals of free and catalyst-bound substrate were observed, enhanced by ε₁₅N ~ 1,374 for ¹⁵N-Py and ε₁₅N ~ 834 for ¹⁵N-NA (see Fig. S1, SI). A pH₂ pressure of 3.4 bar and a flow rate of 80 sccm were used. The experiments were carried out at room temperature. Two different samples were used: 0.1 M ¹⁵N-Py and 5 mM Ir(IMes) methanol-d₄ solution (A) and 0.1 M ¹⁵N-NA and 5 mM Ir(IMes) methanol-d₄ solution (B).

The MRI sequence with SABRE polarization consists of the following steps: (i) preparation of hyperpolarization with SLIC-SABRE (this stage can be repeated n times to gain the maximal signal enhancement); (ii) excitation of ¹⁵N signal with hard ¹⁵N RF pulse; (iii) space encoding; (iv) acquisition of ¹⁵N signal. Two different space encoding MRI approaches were used: Single Point Imaging and Fast Low Angle SHot (compare Fig. 2,C and Fig. 3,C).

¹⁵N-Single-point MRI enhanced by SLIC-SABRE.

High-resolution, 2D Single Point Imaging (SPI) was realized by using two spatial encoding gradients G_x and G_y without slice selection (Fig. 2,C). Each point of k-space was acquired consequently after the SLIC-SABRE pulse sequence, a 90° excitation pulse and two spatial encoding gradients. In total, 16×16=256 points were acquired, resulting in a high native spatial resolution of 565×565$\mathit{ }$μm²/pixel. It means that the experiments were repeated 256 times with constant hyperpolarization production. Here we took advantage of SABRE method that allows, unlike DNP, fast and continuous production of hyperpolarization. The gained resolution was sufficient to image the thin, 1/16 inches $\cong$ 1.6 mm capillary supplying pH₂ to the NMR tube (Fig. 2). Note that the fine details were resolved despite the fact that pH₂ was continuously supplied, which induces convection in the liquid sample. In the context of biomedical applications, however, the scan times of 5.5 minutes (Py) or 27 minutes (NA) are impractical. Moreover, ¹⁵N polarization had to be re-created for each point of k-space, which makes the method inherently slow.

¹⁵N-FLASH MRI enhanced by SLIC-SABRE.

Using a gradient echo pulse sequence such as Fast Low Angle Shot (FLASH) allows one to decrease the scan time by orders of magnitude. Here, an entire k-space was read out after one SLIC-SABRE block (Fig. 3,C). This is advantageous, because in biomedical conditions, it would be impossible to re-hyperpolarize an injected HP bolus for every excitation, as was done for SPI.

In this setting, the hyperpolarization by SLIC-SABRE took 13 s and the FLASH MRI less than 1 s (repetition time 3.1 ms, echo time 1 ms, 8 phase encoding steps, 128 readout points, Cartesian encoding). There was a delay of approximately 2–4 s between SLIC-SABRE and MRI due to hardware limitations while switching between two pulse sequences. We do not expect significant polarization losses during this delay (see HP decay kinetics in Fig. S3, SI). Note that the whole k-space was acquired after only one implementation of SLIC-SABRE.

As a result, ¹⁵N images (Fig. 3) were obtained using a 128×8 matrix that provide a spatial resolution of 0.15×2.4 mm²/pixel (¹⁵N-Py) and 0.3×4.8 mm²/pixel (¹⁵N-NA). For representation, the k-space data sets were zero-filled to 128×128 and 256×256 for ¹⁵N-Py and ¹⁵N-NA, respectively.

Conclusion.

The presented approach allowed us to hyperpolarize and image ¹⁵N biomolecules inside an MRI instrument, in situ, by SLIC-SABRE in less than a minute. The HP was continuously renewed inside the MRI scanner, which provides certain advantages in the context of biomedical applications.^[48] The HP was strong enough to allow subsecond ¹⁵N MRI of two ¹⁵N-labeled biomolecules: ¹⁵N-Py and ¹⁵N-NA (vitamin B₃). No external polarizer or sample transfer to an MRI scanner are required.

SLIC-SABRE was introduced as an efficient way to transfer polarization from pH₂ to a ¹⁵N substrate. Here, we obtained 1,374- and 834-fold signal enhancements for ¹⁵N-Py and ¹⁵N-NA, respectively. Although the polarizations are moderate the enhancements are quite significant and sufficient for fast MRI. We expect that future advances in SLIC-SABRE or related methods will increase the polarization further.

While in these studies organic solvents were used and the catalyst was not removed before imaging, recent advances in filtering,^[7,18] heterogeneous catalysis ^[49] and ¹⁵N SABRE-SHEATH in aqueous medium^[36] bode well for a future in vivo translation of the presented approach. The ¹⁵N sites can retain HP state for tens of minutes, but their detection sensitivity is approximately two orders of magnitude lower than that of protons. This shortcoming can be mitigated by transferring polarization from ¹⁵N to J-coupled ¹H nuclei and subsequent ¹H imaging. Various schemes have been designed for this indirect readout and have already been demonstrated in vivo.^[50–53]

This work is the first demonstration of hyperpolarized, heteronuclear MRI that was enhanced by SLIC-SABRE rather than by spontaneous polarization transfer. This method can be extended to other heteronuclei such as ¹³C, ¹⁹F, ³¹P, etc. for biomedical and other applications.

Experimental Section

Iridium N-heterocyclic carbene complexes, Ir(COD)(IMes)Cl (IMes = 1,3-bis(2,4,6-trimethylphenyl)imidazole-2-ylidene, COD = cyclooctadiene) was used as a pre-catalyst to provide SABRE. This complex was prepared from commercially available precursors ([It(COD)Cl]₂, abcr GmbH, CAS: 12112-67-3; [IMes]⁺[BF₄]⁻, Sigma Aldrich, CAS: 245679-18-9) according to the previously described method.^[54,55] Pre-catalyst was activated by bubbling pH₂ through the reagent solution in the presence of a substrate during 20 minutes with pH₂ flow rate of 20 sccm and 1.7 bar overpressure. Activation was required to convert the pre-catalyst to the active Ir(IMes)(sub)₃Cl SABRE-complex.

¹⁵N-pyridine and 1-¹⁵N-nicotinamide used as SABRE substrates were synthesized according to the procedures described in the literature (for details of synthesis of ¹⁵N-pyridine see SI).^[56,57] 600 μl methanol-d₄ mixtures: 1 M solution of ¹⁵N-Py and 50 mM of Ir(COD)(IMes)Cl, 0.1 M solution of ¹⁵N-Py and 5 mM of Ir(COD)(IMes)Cl or 0.1 M solution of ¹⁵N-NA and 5 mM Ir(COD)(IMes)Cl, were used for MRI experiments. All experiments were carried out using standard 5 mm NMR tubes. pH₂ was supplied to the solution through the capillary system equipped with a mass flow controller and back pressure valves (Fig. 4). pH₂ was prepared using a commercially available pH₂ generator with a conversion temperature of 38 K that provides ~90% enrichment of the para fraction (Bruker, BPHG90).

Figure 4.

Scheme of the HF-SABRE MR setup. pH₂ was supplied to the sample solution placed inside the spectrometer trough the capillary system. The gas system contained mass flow controller and back pressure valves

MR images were acquired with a 400 MHz (B₀ = 9.4 T) microimaging instrument (Bruker, Avance III) equipped with a two-channel probe (¹H, ¹⁵N) and maximal gradient strength of 150 G/cm. The SLIC pulse sequence^[58] was used to transfer nuclear spin order from pH₂ to ¹⁵N in the strong magnetic field.^[45,58] NMR spectra were obtained on a 300 MHz (B₀ = 7 T) NMR spectrometer (Bruker, Avance).

¹⁵N-Single-point MRI enhanced by SLIC-SABRE.

Parahydrogen was delivered to the solution at 1.7 bar and 7 sccm flow rate during SLIC-SABRE and MRI. For SPI, two phase encoding gradients (Fig. 2, C) were applied after non-selective 90° excitation, and each point of k-space was acquired separately after SLIC-SABRE (a total of 16×16). Phase encoding gradients had 45% strength of maximal value and 300 μs duration. 1 or 5 SLIC-SABRE cycles of ca. 1.2 s each were applied for HP of ¹⁵N-Py and ¹⁵N-NA, respectively.

¹⁵N-FLASH MRI enhanced by SLIC-SABRE.

For both substrates, SLIC-SABRE (ca. 1.2 s each) was repeated 10 times, resulting in a total duration of 13 s. At the beginning, pH₂ was flushed through the solution with a flow rate of 35 sccm and 3.5 bar pressure. Importantly, the flow of pH₂ was stopped 4–6 s before the onset of the MRI (11 s after the onset of SLIC) to reduce magnetic field inhomogeneities and fast convection caused by the bubbles. There was a delay of approximately 2–4 s between SLIC-SABRE and MRI due to hardware limitations. Duration of the FLASH was less than a second, TR was 3.1 ms, TE was 1 ms. Phase encoding gradient had 8% strength of maximal value and 400 μs duration, readout gradient had 4% strength and 2.1 ms duration. The flip angle was 30^°. No k-space filter was applied.