Background-Free Proton NMR Spectroscopy with Radiofrequency Amplification by Stimulated Emission Radiation

Abstract

We report on the utility of Radiofrequency Amplification by Stimulated Emission Radiation (RASER) for background-free proton detection of hyperpolarized biomolecules. We performed hyperpolarization of ~0.3 M ethyl acetate via pairwise parahydrogen addition to vinyl acetate. Proton NMR signal with signal-to-noise ratio exceeding 100,000 was detected without radio-frequency excitation at the clinically relevant magnetic field of 1.4 T using a standard (non-cryogenic) inductive detector with quality factor of Q = 68. No proton background signal was observed from protonated solvent (methanol) or other added co-solvents such as ethanol, water or bovine serum. Moreover, we demonstrate RASER detection without radio-frequency excitation of a bolus of hyperpolarized contrast agent in biological fluid. Completely background-free proton detection of hyperpolarized contrast agents in biological media paves the way to new applications in the areas of high-resolution NMR spectroscopy and in vivo spectroscopy and imaging.

Graphical Abstract

Radiofrequency Amplification by Stimulated Emission Radiation (RASER) was employed for background-free detection of hyperpolarized NMR resonances even in the extreme case of protonated solvents and biological fluids. This approach paves the way to a wide range of high-resolution NMR- and in vivo MRI applications of proton detection of hyperpolarized molecules.

NMR hyperpolarization techniques allow for transient increase of nuclear spin polarization (P) by several orders of magnitude.^[1–3] The sensitivity boost enabled by NMR hyperpolarization is geared towards boosting high-resolution NMR spectroscopy as well as in vivo MR spectroscopic imaging (MRSI) to probe in vivo metabolism in real time.^{[4, 5] 13}С^{[6, 7]} and ¹⁵N nuclei^[8] have been the primary focus as the temporary carriers of hyperpolarization, because they have long T₁ and T₂, and a wide dynamic range of chemical shifts, which are sensitive to local environment and chemical structure.^{[9, 10]} Indeed, numerous clinical trials are pending with hyperpolarized (HP) [1-¹³C]pyruvate.^[11–13] Moreover, the greatly improved sensitivity enabled by hyperpolarization can be used for even larger relative gains at low magnetic fields.^[14]

Despite these substantial HP sensitivity gains, some MR sensitivity limitations remain, especially with regard to the large thermal background from molecules at much higher concentrations than the hyperpolarized agents.^{[2, 11, 12, 15]} Also, by definition, P cannot be greater than 1 or 100%, and many hyperpolarization techniques have already exhausted their potential by producing HP compounds with P near unity.^[16–19]

A number of additional strategies have been developed to boost the signal-to-noise ratio (SNR) of HP detection. For example, polarization transfer from HP ¹³C or ¹⁵N to spin-spin coupled protons is employed, and protons are utilized for HP readout.^{[20, 21]} This approach has merit because protons have substantially greater gyromagnetic ratio, γ, (by 4- and 10-fold compared to ¹³C and ¹⁵N respectively) – as a result, substantially increased SNR can be obtained – up to the square of the ratio of the gyromagnetic ratios, (γ_1H/γ_het)², in the best case.^[22] Moreover, the vast majority of MR instrumentation is optimized for proton detection. Despite this nominal benefit of proton detection, the practical utility is hampered by a substantial background signal of thermally polarized protons, e.g., water or fat protons of tissue, which are present at substantially higher concentration.^{[15, 23]}

Here we suggest a new approach to address the large thermal background problem, which is to create conditions, where HP nuclear spins spontaneously develop transverse magnetization without radio-frequency (RF) pulses. Such radio amplification by stimulated emission of radiation (RASER) can be obtained,^[24–26] when a) the hyperpolarization is negative, i.e. corresponds to a population inversion, and b) when the radiation damping rate 1/τ_RD satisfies the condition of 1/τ_RD > 1/T₂*, where T₂* is the effective transverse relaxation time, and the radiation damping rate is given by:

$$1/{\tau }_{\text{RD}}=-\left({\mu }_0/4\right)\cdot \eta \cdot Q\cdot {\gamma }^2\cdot \hslash \cdot n_{\text{S}}\cdot P$$ (1)

where μ₀ is the vacuum permeability, η is the filling factor, Q is the quality factor of the resonator, ℏ is Planck’s constant, γ is the nucleus gyromagnetic ratio and n_S is the spin number density.^[26] The RASER phenomenon was first demonstrated in low magnetic fields through the use of high-Q resonators.^[24–27] Besides the SNR boost achieved by high-Q RF coils, the NMR signal can in principle persist indefinitely (as long as P is maintained) resulting in line narrowing well below of the linewidth given by T₂*.^{[26, 28]} The latter effect leads to two key practical RASER benefits: improved spectral resolution and improved SNR (through indirect proton detection of ¹³C HP contrast agents). Therefore, RASER is a potential game-changing technology that can substantially boost the detection sensitivity of HP NMR and avoid all background signals because excitation pulses are not required.

Recently, RASER has been demonstrated using standard commercial NMR hardware at high magnetic fields of up to 14T.^[29–31] Moreover, RASER has been demonstrated at the clinically relevant field of 1.4 T using a bolus of HP compounds, i.e. under conditions of non-replenishable hyperpolarization.^{[28, 32–34]} It follows from eq. (1) that high-γ spins are more RASER potent, and therefore, can induce RASER at lower spin densities. Nevertheless, RASER has been already successfully demonstrated for low-γ ¹⁷O at 9.4 T via Dynamic Nuclear Polarization (DNP).^[35] As a result of these demonstrations, the RASER effect is poised to expand the reach of high-resolution HP NMR and in vivo MRSI by providing three clear benefits: I: substantially improved spectral resolution, II: higher NMR sensitivity and III: background free signals. These gains can be amplified further using high-Q resonators.^[24]

Here, we focus on the demonstration of background-free proton RASERs in the highly unfavourable case of fully protonated solvents. Moreover, we extend this concept to biological media, demonstrating biocompatible, background-free RASER.

We employ parahydrogen-induced polarization (PHIP)^{[36, 37]} to induce NMR hyperpolarization in ethyl acetate (EA). Our rationale for choosing EA is threefold. First, acetate can be hyperpolarized using both PHIP^[38] and DNP.^[39] Second, HP acetate has demonstrated in vivo utility.^[39] Finally, HP acetate was already imaged in vivo using proton detection.^[40] Pairwise parahydrogen (p-H₂) addition to vinyl acetate (Figure 1a) leads to production of proton-hyperpolarized EA (note that proton polarization must be transferred to [1-¹³C] nucleus and HP EA must be de-esterified for biological applications). The 300 mM vinyl acetate sample (~0.5 mL) was hydrogenated for 10 s using 4 mM catalyst in either methanol or methanol-d₄ at Earth’s magnetic field by p-H₂ (>98.5% para- fraction^[41]) bubbling via Teflon catheter inside standard 5-mm NMR tube as described previously.^[34] After cessation of p-H₂ bubbling, the catheter was removed, and the HP sample was inserted into the NMR spectrometer. As detailed in Figure 2, some samples were spiked with comparable volumes of neat ethanol, water or bovine serum. The acquisition of NMR signal was initiated approximately 2–3 seconds before sample insertion into the NMR spectrometer. The total acquisition time always was 16 s as indicated in Figure 1b and reflected in the time axis in Figure 2. RASER activity was observed in all studied samples, see Figures 2 and 3. As can be seen in panels (a) and (b) of both Figure 2 (FID’s) and Figure 3 (FT spectra), samples employing methanol-d₄ or protonated methanol exhibit similar signatures during the first burst and the remainder of the RASER activity. Despite the high proton concentration of methyl protons in methanol (ca. 74 M), their magnetization is too low to induce RASER activity. Furthermore, the thermal polarization is aligned with the magnetic field (not against the field, no population inversion) and thus, spontaneous emission of radiation is not possible, therefore no background signal from solvent protons is seen (see the corresponding insets). Only in exceptional cases, detailed below, the solvent may also attain negative polarization.

Figure 1.

(a) Reaction scheme for pairwise p-H₂ addition onto vinyl acetate (VA) leading to HP ethyl acetate (EA). (b) Event sequence for the performed experiments. In the Earth’s magnetic field, p-H₂ was bubbled in either methanol or methanol-d₄ while ethanol, water, or serum was injected immediately after the reaction. In all experiments, the catheter employed for p-H₂ bubble through the solution was removed. Note that the samples were placed in the bore of the 1.4 T spectrometer ~5 s after the spectrometer’s receiver was open and left open for ~16 s (no RF pulse). Afterwards, a series of low flip angle (8°) pulse-acquisition (2 s pulse-acquisition + 8 s relaxation) was performed to monitor the decay of the HP states.

Figure 2.

¹H ALTADENA^[42] RASER signals of HP EA in methanol-d₄ (a), non-deuterated methanol (b), methanol with post-reaction injection of ethanol (c), methanol-d₄ with post-reaction injection of water (d) or serum (e). All signals were recorded without RF pulse. Magenta boxes highlight the RASER bursts occurring at the time the samples were placed in the NMR spectrometer.

Figure 3.

¹H (Fourier transformed) spectra of the ALTADENA RASER signals emitted by HP EA in various solvents (same as Figure 2): methanol-d₄ (a), non-deuterated methanol (b), methanol with post-reaction injection of ethanol (c), methanol-d₄ with post-reaction injection of water (d) or serum (e). In each subpanel, the magenta spectrum corresponds to the RASER burst observed at the time of sample insertion in the spectrometer while the black spectrum corresponds to the subsequent RASER activity. The respective spectra of thermally polarized samples yielding the signals from non-deuterated solvent (methanol, ethanol, water) or serum are superimposed in blue.

When the RASER acquisition was completed (Figure 1b), a series of 1D spectra was recorded using small tip angle (8°) pulses to monitor the decay of HP state as shown in Figure 4. The measured decay constants of the HP protons A and B allowed extrapolated calculation of original polarization P_H ~ −0.25.^[34] We also note that the first few pulse-acquire transients clearly exhibit line narrowing of H_B due to residual RASER activity. Note that methanol solvent polarization during RASER experiments also deviated from thermal equilibrium due to cross-relaxation from less concentrated HP H_B. This is evidenced by the negative amplitude (corresponding to a population inversion) of both solvent peaks during the first few frames in Figure 4. Despite this altered polarization of solvent, no solvent RASER activity was seen from the protonated solvent, Figure 3b.

Figure 4.

Fourier spectra of HP EA in non-deuterated methanol recorded every 10 s after ALTADENA RASER. The FWHM of the H_B signal is ~0.5 Hz for the first two timeframes (RASER still active) and then increases toward reaching a constant value of ~3 Hz (RASER non-active) after 40 s decay (bottom insert), i.e. more than 1 min after the hydrogenation reaction was terminated. The solvent (blue, top insert) is also hyperpolarized (note the negative amplitude of both solvent resonances) by PHIP. The negative amplitude of solvent resonances changes to the positive because the sample gains Boltzmann polarization with time in 1.4 T magnet.

Additional control experiments were performed to simulate injection of HP solution. The spectra shown in Figures 2c,d and Figures 3c,d were recorded after mixing the solution of HP EA in methanol with ethanol, and in methanol-d₄ with water respectively (both without deoxygenation). Despite the HP EA dilution and likely decrease in T₂*, RASER effects are clearly observable from HP protons, but not from the solvent. Specifically, the HP RASER SNR in Figure 3c exceeds 10⁵, yet no background proton signal is detected above the noise.

Finally, the HP EA in methanol-d₄ was injected with serum (blood derived biofluid free from white and red blood cells) resulting in more significant decrease of T₂* as evident from the thermal spectrum (Figure 3e) and a likely decrease of T₁. Nevertheless, the initial RASER signal “burst” was detected after injection of the biological fluid without any background signal, as seen on the magenta trace in Figures 2e and 3e. This observation is important, because it shows RASER detection in biological fluids at the clinically relevant magnetic field of 1.4 T without the use of any specialized RF hardware.

In summary, we demonstrated that RASER phenomenon can be employed for background-free detection of HP resonances even in the extreme cases of protonated solvents and biological fluids. This approach can potentially be employed for a wide range of high-resolution and in vivo applications of proton detection of HP molecules. In particular, we envision that in vivo use could benefit tremendously as water background suppression techniques are typically difficult and cause undesirable polarization losses and artefacts. Development of NMR imaging exploiting the RASER effect is the next logical step and subject of ongoing work in our laboratories.