The Impact of Gd³⁺ on DNP of [1-¹³C]Pyruvate Doped with Trityl OX063, BDPA, or 4-Oxo-TEMPO

Abstract

Hyperpolarized [1-¹³C]pyruvate has become an important diagnostic tracer of normal and aberrant cellular metabolism for in vitro and in vivo NMR spectroscopy (MRS) and imaging (MRI). In pursuit of achieving high NMR signal enhancements in dynamic nuclear polarization (DNP) experiments, we have performed an extensive investigation of the influence of Gd³⁺ doping, a parameter previously reported to improve hyperpolarized NMR signals, on the DNP of this compound. [1-¹³C]Pyruvate samples were doped with varying amounts of Gd³⁺ and fixed optimal concentrations of free radical polarizing agents commonly used in fast dissolution DNP: trityl OX063 (15 mM), 4-oxo-TEMPO (40 mM), and BDPA (40 mM). In general, we have observed three regions of interest, namely: (i) a monotonic increase in DNP-enhanced nuclear polarization P_dnp upon increasing the Gd³⁺ concentration until a certain threshold concentration c₁ (1–2 mM) is reached, (ii) a region of roughly constant maximum P from c₁ until a concentration threshold c₂ (4–5 mM), and (iii) a monotonic decrease in P_dnp at Gd³⁺ concentration c > c₂. Of the three free radical polarizing agents used, trityl OX063 gave the best response to Gd³⁺ doping with a 300 % increase in the solid-state nuclear polarization whereas addition of the optimum Gd³⁺ concentration on BDPA and 4-oxo-TEMPO-doped samples only yielded a relatively modest 5–20 % increase in the base DNP-enhanced polarization. The increase in P_dnp due to Gd³⁺ doping is ascribed to the decrease in the electronic spin-lattice relaxation T_1e of the free radical electrons which plays a role in achieving lower spin temperature T_s of the nuclear Zeeman system. These results are discussed qualitatively in terms of the spin temperature model of DNP.

1. INTRODUCTION

Since the pivotal invention of the fast dissolution technique in dynamic nuclear polarization (DNP) by Ardenkjaer-Larsen and co-workers in 2003,¹ DNP has attracted renewed and increasing interest especially in chemistry and biomedical NMR spectroscopy and imaging. DNP, a technology that has been used in the production of polarized targets for nuclear and particle physics experiments since the 1950s,^2–5 uses microwave irradiation to transfer high electronic polarization to the target nuclear spins (e.g. protons and deuterons) with optimal results at low temperature (1 K or less) and moderate magnetic field (> 1 T). While polarization transfer occurs at cryogenic temperatures, the incorporation of a fast dissolution device¹ that rapidly dissolves the frozen polarized sample allows the production of injectable liquids at physiological temperatures containing highly polarized nuclei of biological interest. Since the hyperpolarized state decays by spin-lattice T₁ relaxation, an important requirement for successful dissolution DNP experiments is that the nucleus of interest should have sufficiently long T₁ (typically at least 10 s) to preserve some or preferably, most of the nuclear spin polarization during the dissolution transfer from the polarizer to an NMR magnet/imaging system, a process that normally takes 5–10 s. Notable examples of biomedical applications of this technology include pH mapping in tumors using hyperpolarized ¹³C-bicarbonate,⁶ grading the aggressiveness of prostate cancer with hyperpolarized ¹³C-pyruvate,⁷ and in general, real-time monitoring of in vitro and in vivo biochemical/metabolic activities^8–30 via hyperpolarized ¹³C NMR spectroscopy (MRS) and imaging (MRI). Other long T₁ nuclei that have been polarized via the fast dissolution DNP method include ¹⁵N,^{31 6}Li,^{32 89}Y,^33–35 and ^107,109Ag,³⁶ among others.

Free radicals commonly used as polarizing agents in DNP include carbon centered free radicals such as trityl OX063 and BDPA and nitroxyls (Chart 1). The ESR properties of the free radical are crucial parameters in DNP. In this work, the results are qualitatively discussed in the context of the thermodynamic model of DNP involving the interaction of three thermal baths: electron Zeeman system, electron dipolar or spin-spin interaction system, and the nuclear Zeeman system.^2–5 Thermal mixing, which is the expected predominant DNP mechanism in our experiments, occurs when the ESR linewidth D of the free radical is greater than or comparable to the nuclear Larmor frequency.^2–5 In this regime, microwave irradiation dynamically cools the electron spin-spin interaction reservoir that has an energy matching the nuclear Zeeman reservoir, and there is thermal contact between the two reservoirs thereby resulting in equal spin temperature T_s for both systems.^2–5 The nuclear polarization levels achieved in DNP can be maximized by using appropriate glassing agents,^35,37 free radicals with narrow ESR linewidths,^35,38 and lower polarizer operating temperatures.³⁹ Another factor that can improve the maximum polarization (minimum T_s for nuclear spins) in DNP is the use of Gd³⁺ which has been described in earlier experiments on ¹³C samples doped with trityl OX063.^40,41 Consequently, several hyperpolarized ¹³C MRS and MRI experiments have reported the routine use of trace amounts (1–2 mM) of Gd³⁺ compounds and complexes such as GdCl₃, ProHance®, Magnevist®, Dotarem®, and 3-Gd® in trityl-doped ¹³C samples to dramatically improve the liquid-state NMR signal enhancements after dissolution.^{6,17–30,42–44} The impact of Gd³⁺ on trityl-based polarizing agents has only been investigated over a limited concentration range (0–2 mM). Here, we present our studies on the influence of Gd³⁺ on the DNP of [1-¹³C]pyruvate samples doped with three different free radical polarizing agents: trityl OX063, BDPA, and 4-oxo-TEMPO (Chart 1). In this work, the effect of Gd³⁺ (hereafter refers to the Gd-HP-DO3A complex) on the solid-state nuclear polarization P_dnp and spin-lattice relaxation T₁ of [1-¹³C]pyruvate was investigated over a wider concentration range (up to 8 mM) with each radical.

Chart 1.

The free radical polarizing agents and Gd³⁺ complex used in this work.

2. EXPERIMENTAL METHODS

Materials

The free radical polarizing agents used in this work were obtained from commercial sources: (i) tris{8-carboxyl-2,2,6,6-benzo(1,2-d:5-d)-bis(1,3)dithiole-4-yl}methyl sodium salt (trityl OX063) [Oxford Instruments Molecular Biotools], (ii) 1,3-bisdiphenylene-2-phenylallyl (BDPA) [Sigma-Aldrich], and (iii) 4-oxo-2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) [Sigma-Aldrich]. [1-¹³C]pyruvic acid and [1-¹³C]sodium pyruvate were purchased from Sigma-Aldrich. ProHance® (Bracco Diagnostics, New Jersey) was obtained as 0.5 M Gd-HP-DO3A solution in water. All chemicals and glassing solvents (glycerol, sulfolane) used in this study were purchased from commercial sources and used without further purification.

Trityl-doped samples

[1-¹³C]sodium pyruvate solution (1.4 M) in 1:1 (v/v) glycerol:water glassing matrix was prepared and doped with trityl OX063 (15 mM). Small aliquots of these samples were prepared with different concentrations of Gd³⁺ (0–8 mM ProHance).

BDPA-doped samples

BDPA was prepared in sulfolane via sonication (80 mM) and mixed with equal amount (volume) of [1-¹³C]pyruvic acid as previously described.⁴⁵ The final concentration of BDPA in the mixture was 40 mM. Small aliquots were prepared with different concentrations of Gd³⁺ (0–8 mM ProHance).

TEMPO-doped samples

[1-¹³C]sodium pyruvate (1.4 M) in 1:1 (v/v) glycerol:water glassing matrix was prepared and doped with 4-oxo-TEMPO (40 mM). Small aliquots were prepared with different concentrations of Gd³⁺ (0–8 mM ProHance).

Microwave frequency sweep

The samples (100 μL aliquots) were polarized in a HyperSense® DNP (Oxford Instruments, England) for 3 minutes at each different microwave frequency and the NMR signal intensity was recorded with the built-in solid state NMR spectrometer. A series of hard rf excitation pulses was applied to destroy the residual magnetization before starting the polarization of the sample at the next microwave frequency. In addition, the ¹³C microwave DNP spectra with longer microwave irradiation time (1 hour) of trityl-doped ¹³C pyruvate samples were also recorded (see the Supporting Information) for comparison.

Measurement of solid-state nuclear polarization

The DNP-enhanced nuclear polarization of the frozen sample P_dnp was calculated by multiplying the NMR signal enhancement ε (ratio of integrated NMR intensity of the hyperpolarized NMR signal over the thermal NMR signal of the sample) with the calculated thermal nuclear polarization P_thermal of the sample at cryogenic conditions. The calculated ¹³C thermal polarization at 3.35 T and 1.4 K is P_thermal=6.147×10⁻²% using the Boltzmann distribution equation for an ensemble of ¹³C nuclear spins. The NMR spectra were acquired using a Varian VNMRS spectrometer (Agilent Technologies, Santa Clara, CA).

Measurement of solid-state nuclear spin-lattice relaxation time T₁

100 μL aliquots were polarized at 3.35 T and 1.4 K in the HyperSense until they reached their corresponding maximum polarizations. The microwave source was then turned off and the decay of the hyperpolarized NMR signal of the sample inside the polarizer (3.35 T, 1.4 K) was monitored by applying a 1.5-degree rf excitation pulse (see pulsewidth calibration in the Supporting Information) every 300 s using the Varian VNMRS spectrometer. The decay curves were fitted to an equation accounting for the magnetization decay due to T₁ decay and rf excitation.⁴⁶ Values of ¹³C T₁ of the sample in the frozen state were extracted from these fits.

Data Analyses

The NMR data were acquired using a Varian VNMR spectrometer and the spectra were processed using ACDLABS ver. 12 software (Advanced Chemistry Development, Inc., Toronto, Canada). The graphs, fits, and data analyses were done using Igor Pro ver. 6 (Wavemetrics Inc., Portland, OR).

3. RESULTS AND DISCUSSION

3.1. DNP of trityl-doped [1-¹³C] sodium pyruvate

A number of experiments^1,35,41,47 have shown that the DNP of low-γ nuclei (e.g. ¹³C, ⁸⁹Y) doped with the free radical trityl OX063 proceeds predominantly via thermal mixing. The optimum concentration of trityl OX063 in ¹³C samples that gives the maximum solid-state polarization in a reasonable amount of microwave irradiation time has been reported to be ~15 mM^41,48 (see also the Supporting Information). Trityl OX063 (Chart 1) has an ESR linewidth D = 2.24 mT,⁴¹ which is among the narrowest of the free radicals used for fast dissolution DNP-NMR. In the thermal mixing regime, trityl OX063 is an efficient DNP polarizing agent for low-γ nuclei such as ¹³C because its narrow D translates to lower electron heat capacity yielding a lower spin temperature for the electron spin-spin interaction reservoir which is in thermal contact with the nuclear Zeeman system.^2–5 This results in lower spin temperature of the nuclear Zeeman system/high nuclear polarization P_dnp.

Figure 1 shows the microwave DNP spectra of 1.4 M [1-¹³C]sodium pyruvate in 1:1 (v/v) glycerol:water glassing matrix doped with a) 15 mM trityl OX063, b) 15 mM trityl OX063 plus 5 mM Gd³⁺, and c) 5 mM Gd³⁺. The microwave DNP spectra shown here were plotted by recording the ¹³C NMR intensity every after three minutes of microwave irradiation of the sample at different microwave frequencies. This relatively short irradiation time would give the approximate locations of the optimum DNP irradiation frequencies, namely the positive polarization peak P(+) and negative polarization peak P(−). A comparison with the microwave DNP spectra of ¹³C samples taken at a longer microwave irradiation time (1 hour; see the Supporting Information) reveals almost the same locations of P(+) and P(−), with a slight offset of 5–10 MHz down in frequency relative to the data taken with a 3-minute irradiation time shown in Figure 1. Theoretically, the longer irradiation times would give values closer to the actual locations of the optimum polarization peaks. For practical purposes, however, the shorter irradiation times could show the approximate locations of P(+) and P(−). Nevertheless, this slight offset only has a relatively minor effect on the polarization buildup results considering the broadness of the polarization peaks.

Figure 1.

Microwave DNP spectra of 100 μL samples containing 1.4 M [1-¹³C]sodium pyruvate in 1:1 glycerol:water glassing matrix doped with a) 15 mM trityl, b) 15 mM trityl and 5 mM Gd³⁺, and c) 5 mM Gd³⁺. These data were taken in the HyperSense at 3.35 T and 1.4 K using a 100 mW microwave source. The up and down arrows indicate the approximate locations of the positive and negative polarization peaks, respectively.

The addition of Gd³⁺ to trityl-doped [1-¹³C]sodium pyruvate samples leads to two main effects: (1) a decrease in the separation distance between polarization peaks P(+) and P(−) as shown in Figure 1b and (2) an increase in the NMR intensity. Since it was shown elsewhere⁴⁹ that the polarization buildup time constant τ_bu tends to become longer further out the tails of the microwave DNP spectrum, a question arises on whether or not the first effect is due to polarization buildup kinetic effect attributed to a relatively short irradiation time (3 minutes). The ¹³C microwave spectrum of the ¹³C pyruvate sample doped with 4 mM Gd³⁺ measured at the longer irradiation time (1 hour; see the Supporting Information) shows the same narrowing of the polarization peak separation distance. Currently, the exact cause for the decrease in the separation between P(+) and P(−) is not clear. For the second effect, the improvement in the DNP-enhanced NMR intensity in the presence of Gd³⁺ has been observed previously^40,41 and is ascribed to the shortening of the electronic spin-lattice relaxation time T_1e.

Figure 2.

a) Representative polarization buildup curves of 1.4 M [1-¹³C]pyruvate in 1:1 (v/v) glycerol:water glassing matrix doped with 15 mM trityl OX063 and mixed with different concentrations of Gd³⁺. These curves were taken by irradiating the samples at 94.072 GHz with a 100 mW microwave source at 3.35 T and 1.4 K. The solid lines are fits to a mono-exponential buildup equation. b) The maximum ¹³C nuclear polarization of [1-¹³C]pyruvate sample as a function of Gd³⁺ concentration at 3.35 T and 1.4 K. c) Polarization buildup time constant τ_bu versus Gd³⁺ concentration derived by fitting the polarization buildup curves with a single-exponential buildup equation.

To eliminate the possibility that the paramagnetic Gd³⁺ itself acts as polarizing agent, we also performed a microwave sweep on [1-¹³C] sodium pyruvate samples doped with 5 mM Gd³⁺ only as displayed in Figure 1c and found no indication that Gd³⁺ acts as polarizing agent in this microwave frequency window. This is expected because the frequency range the HyperSense is capable of is centered on g≈2 organic free radicals such as trityl OX063 and BDPA. The ESR resonance of g≠2 metal ions such as Gd³⁺ is located further upfield (lower microwave frequency in a fixed field) that is outside the microwave frequency window of the HyperSense. It should be noted however, that Gd³⁺ compounds such as GdCl₃ and the Gd³⁺ complex of 1,4,7,10-teraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) have been used as DNP polarizing agents for protons at 70–80 K where, neglecting the broad overlapping resonances of the higher order transitions, the narrow ESR linewidth of the central transition ±½→∓½ of the S=7/2 Gd³⁺ can transfer polarization to the proton spins via the solid effect.^50,51 However, at low temperature close to 1 K, the electron polarization is higher and most of the electron spins reside in the Zeeman ground state (m_s=7/2) and thus the transitions from this level form a broad signal which dominates the ESR spectrum.⁵² In addition to the fact that the Gd³⁺ ESR resonance is located far from that of the trityl OX063, the very broad ESR line of Gd³⁺ due to the higher-order transitions would be ineffective for DNP via thermal mixing. Therefore, the Gd³⁺ in this system does not act as polarizing agent in this case but rather aids the free radical electrons in achieving more efficient transfer of polarization to the nuclear spins.

So far, the effect of Gd³⁺ has not been investigated systematically over a wide concentration range. Therefore, we have performed DNP experiments with trityl-doped (15 mM) [1-¹³C] sodium pyruvate samples in which the concentration of Gd³⁺ was varied from 0 to 8 mM. The samples were irradiated at 94.072 GHz near the positive polarization peaks of Gd-free and Gd-doped ¹³C samples. Figure 2a shows the representative ¹³C polarization buildup curves at 3.35 T and 1.4 K of samples containing different concentrations of Gd³⁺. Figure 2b provides a summary that highlights the three Gd³⁺ concentration regions of interest on the basis of its effect on solid-state nuclear polarization of [1-¹³C]pyruvate. The maximum ¹³C polarization in the absence of Gd³⁺ was 9.6 % after microwave irradiation of the sample for 3–4 hours. The ¹³C nuclear polarization increases nearly linearly as Gd³⁺ is increased from 0–2 mM then plateaus between 2–5 mM at a polarization level of ~40 % as shown in Figure 2b, and then declines at Gd³⁺ concentration c >5 mM. In comparison, a previous DNP study on neat [1-¹³C]pyruvic acid doped 15 mM trityl OX063 polarized at 3.35 T and 1.2 K yielded a ¹³C nuclear polarization improvement from ~20 % to ~40 % with the addition of 1.5 mM GdCl₃.⁴¹ In Figure 2c, the polarization buildup time constant τ_bu monotonically increases with Gd³⁺ doping over this same concentration range. In a simple scenario, a shorter electronic T_1e would mean that the free radical electron would polarize nuclear spins faster, thus τ_bu should decrease. This result however suggests that the polarization buildup kinetics in this case is more than just fast electronic relaxation and nuclear spin diffusion.

We interpret the increase in nuclear polarization with Gd³⁺ doping using the spin temperature model of DNP to explain the increase in the nuclear polarization with Gd³⁺ doping. Theoretically, the upper limit of DNP-enhanced nuclear polarization P_dnp,max for an ensemble of nuclei with spin I=1/2 is given by:^53,54

$$P_{\mathit{dnp},max}=tanh\left({\beta }_L\frac{{\omega }_e{\omega }_I}{4D}\frac{1}{\sqrt{\eta (1+f)}}\right)$$ (1)

where β_L= Ħ/k_BT_L (Ħ, k_B, T_L are the Planck’s constant divided by 2π, the Boltzmann constant, and lattice temperature, respectively), ω_e and ω_I are the electron and nuclear Larmor frequencies, respectively, D is the ESR linewidth, η=T_1Z/T_1D (T_1Z, T_1D are the electronic Zeeman and dipolar spin-lattice relaxation times, respectively), and f is the so-called “leakage factor” of nuclear relaxation. Equation 1 can be simplified to P_dnp,max=tanh(μB/k_BT_s,min) where the minimum spin temperature is expressed as $T_{s,min}=(2D\sqrt{\eta (1+f)}/{\omega }_e)T_L$.³⁵ These equations point out that the key to achieving high nuclear polarization levels under the thermal mixing DNP mechanism is to minimize the spin temperature T_s of the electron spin-spin interaction reservoir and in this model, lower T_s can be achieved by decreasing the values of these parameters: D, T_L, η, and f and/or increasing the magnetic field (higher ω_e). Since T_L and B are determined by the physical limits of the DNP hardware, the parameters that can be influenced by Gd³⁺ doping can be narrowed to D, η, and f. It is assumed that Gd³⁺ doping has a minimal effect on the ESR linewidth D over the concentration range examined here (0 to 8 mM). This leaves the factor $\sqrt{\eta (1+f)}$ as the composite parameter that effectively changes with Gd³⁺ doping.

Previous ESR studies⁴¹ on trityl-doped samples (15 mM trityl OX063 in neat [1-¹³C] pyruvic acid at 3.35 T and 1.2 K) have shown that the electronic Zeeman T_1Z of the free radical paramagnetic electron was reduced from approximately 1.2 s to 0.3 s with the addition of 1 mM GdCl₃ while the nuclear longitudinal relaxation T₁ seemingly remained unaffected at this GdCl₃ concentration. This result implies that the parameter η, in particular the electronic Zeeman relaxation time T_1Z, is reduced in the presence of Gd³⁺ thereby lowering the spin temperature of the electron spin-spin interaction reservoir. A thermodynamic view of this case is that the rate of cooling of the electron Zeeman system is faster than the rate of heating of the electron dipolar system.⁵⁵

Let us first consider the Gd³⁺ concentration range of 0 to 1 mM where P_dnp monotonically rises. As mentioned before, the nuclear T₁’s in this region are barely affected as the Gd³⁺ predominantly shortens only the electronic T_1Z (Figure 3). Figure 2b shows the dependence of the maximum solid-state ¹³C nuclear polarization on the concentration of Gd³⁺. The maximum ¹³C P_dnp for the Gd³⁺-free sample (1.4 M [1-¹³C] pyruvate and 15 mM trityl OX063 in 1:1 (v/v) glycerol:water glassing matrix) is 9.6 % which corresponds to a spin temperature T_s=8.96 mK. P_dnp increases almost linearly, which implies that the composite parameter $\sqrt{\eta (1+f)}$ monotonically decreases, on the gradual addition of Gd³⁺ from 0 to c₁=1 mM. Further addition of Gd³⁺ complex in the sample from c₁=1 mM to c₂=5 mM yielded a constant polarization of approximately 40 % which corresponds to T_s≈2 mK. This suggests that $\sqrt{\eta (1+f)}$ remains the same in this concentration range; in this case, a further decrease in electronic relaxation ratio (η) due to Gd³⁺ doping is probably offset by a slight increase in the leakage factor f as suggested by the decrease of solid-state nuclear T₁ with higher Gd³⁺ concentration shown in Figure 3. Finally for c>c₂, the polarization drops monotonically suggesting a monotonic increase in $\sqrt{\eta (1+f)}$ (higher spin temperature). A detailed ESR measurement of the electronic relaxation of the free radical at DNP conditions is required to identify the individual effects of Gd³⁺ doping on η and implicitly on f.

Figure 3.

Representative decay curves of the hyperpolarized solid-state NMR signal of 100 μL trityl-doped samples (1.4 M [1-¹³C]pyruvate in 1:1 (v/v) glycerol:water doped with 15 mM trityl OX063) mixed with different concentrations of Gd³⁺. Inset: ¹³C solid-state T₁ values extracted from the decay curves versus different Gd³⁺ concentrations.

The solid-state nuclear relaxation T₁ of [1-¹³C] pyruvate samples (1.4 M in 1:1 (v/v) glycerol:water doped with 15 mM trityl OX063) at 3.35 T and 1.4 K monotonically decreases with increasing Gd³⁺ concentration as shown in Figure 3. At this low temperature, the relative motions of the electron and nuclear spins are frozen out, and thus the nuclear spins can relax through the fixed paramagnetic impurities which, in this case, refer to the trityl OX063 free radical (with a fixed 15 mM concentration) and a varying Gd³⁺ concentration (0 to 8 mM) present in the ¹³C pyruvate samples. In general, the decrease in solid-state ¹³C T₁ with Gd³⁺ doping can thus be approximately explained by a relaxation equation of an ensemble of nuclear spins due to a low concentration of fixed paramagnetic impurities:^2,3

$$\frac{1}{T_1}\simeq \frac{8\pi }{5}\frac{S(S+1)}{3}\frac{N_s{\gamma }_e^2{\gamma }_n^2{\hslash }^2}{b^3}\frac{{\tau }_c}{1+{({\omega }_n{\tau }_c)}^2}\left(1-P_e^2\right)$$ (2)

where S is the spin number, N_s is the number of paramagnetic impurities per unit volume, b is the diffusion barrier, τ_c is the spin correlation time, and P_e is the electron thermal polarization. The correlation time τ_c is due to the fluctuating local field of the electrons. In the slow motion limit ω_nτ_c≫1, the term τ_c/(1+(ω_nτ_c)²) reduces to 1/ω_n²τ_c. The factor (1−P_e²) indicates that as electron thermal polarization approaches unity (100 %) by decreasing the temperature and/or increasing the field, the nuclear relaxation rate 1/T₁ approaches zero because the electron spins are all in the ground state; the fluctuations disappear and the source of nuclear relaxation is removed.^2,3 It can be easily seen from Equation 2 that increasing the concentration of paramagnetic impurities N_s can lead to faster nuclear relaxation rate (shorter T₁). Experimentally, although the decrease in the trityl OX063 free radical T_1e with Gd³⁺ doping was drastic (the T_1e at 3.35 T and 1.2 K decreased from ~1.2 s to 0.3 s with the addition of 1 mM GdCl₃),⁴¹ the decrease in nuclear T₁ is less dramatic (from T₁=10300 s for a Gd-free sample to T₁=8200 s for a ¹³C pyruvate sample doped with 8 mM Gd³⁺) as shown in Figure 3. These results suggest that the ¹³C nuclear relaxation in this case proceeds predominantly through the electron dipolar system and the small decrease in ¹³C T₁ with Gd³⁺ doping is attributed to the slight effect of the mechanism described in Equation 2. The solid-state T₁ decay curves of hyperpolarized trityl-doped ¹³C pyruvate samples with 0 mM and 4 mM Gd³⁺ (Figure 3) almost overlap in agreement with the results of a previous study⁴¹ where GdCl₃ doping barely changed the ¹³C T₁ in the concentration range 0–2 mM.

3.2. DNP of BDPA-doped [1-¹³C]pyruvic acid

The carbon-centered free radical, 1,3-bisdiphenylene-2-phenylallyl (BDPA) has an ESR linewidth D comparable with that of trityl OX063 and, therefore, is expected to have similar DNP efficiency via the thermal mixing process. Although BDPA is insoluble in water, it is readily soluble in sulfolane and a few other solvents (e.g. methanol, diethylene glycol monobenzyl ether, DMSO). We have shown recently that comparable or slightly higher nuclear polarizations can indeed be achieved for [1-¹³C]pyruvic acid doped with BDPA (40 mM) than with trityl OX063 (15 mM) in the absence of Gd³⁺.⁴⁵ In addition, BDPA offers the advantage of easy removal in the dissolution liquid by simple filtration when water is used as the dissolution solvent.⁴⁵ The optimum concentration of BDPA for DNP was found to be 20–40 mM, but, for practical purposes, 40 mM BDPA concentration is used because maximum nuclear polarization is reached with less microwave irradiation time.

For DNP experiments with BDPA, we opted to use free [1-¹³C]pyruvic acid, which is miscible with sulfolane, instead of sodium [1-¹³C]pyruvate salt which is poorly soluble in the glassing solvents (DMSO or sulfolane) suitable for BDPA. The DNP samples consisted of 1:1 (v/v) sulfolane:[1-¹³C]pyruvic acid doped with 40 mM BDPA as described in a previous work.⁴⁵ The ¹³C microwave DNP spectra of BDPA-doped [1-¹³C] pyruvic acid in the presence and absence of Gd³⁺ (Figure 4) are similar to those obtained with trityl OX063. The narrowing in the separation distance of the positive and negative polarization peaks in the presence of Gd³⁺ is also present. The polarization buildup curves were taken at 94.055 GHz, close to the positive polarization peaks of the Gd-free and Gd-doped DNP samples.

Figure 4.

Microwave DNP spectra of 1:1 [1-¹³C]pyruvic acid:sulfolane samples doped with a) 40 mM BDPA and b) 40 mM BDPA plus 5 mM Gd³⁺. These data were taken in the HyperSense at 3.35 T and 1.4 K using a 100 mW microwave source. The up and down arrows indicate the approximate locations of the positive and negative polarization peaks, respectively.

Figure 5a shows the polarization buildup curves for [1-¹³C]pyruvic acid samples doped with different concentrations of Gd³⁺ and a summary of maximum P_dnp obtained with different Gd³⁺ concentrations is displayed in Figure 5b. Figure 5c shows that the polarization buildup time constant remains roughly constant (τ_bu≈800 s) over the Gd³⁺ concentration range of 0–8 mM. The relatively faster polarization buildup time τ_bu for these BDPA-doped samples is attributed to a faster spin diffusion,⁵⁶ a combined effect of the high density of ¹³C spins (7.7 M [1-¹³C]pyruvic acid after mixing with equal volume of sulfolane) as well as to the high BDPA free radical concentration (40 mM). The maximum ¹³C nuclear polarization attained for a Gd-free BDPA-doped (40 mM) pyruvic acid sample is close to 12 % corresponding to T_s=7.16 mK. Addition of Gd³⁺ in the sample of up to 1 mM led to a small increase in P_dnp (up to 14 % maximum polarization; T_s=6.12 mK) and this polarization level remained roughly constant in the Gd³⁺ concentration range from c₁=1 mM to c₂=4 mM. Further addition of Gd³⁺ at c>c₂ led to a monotonic decrease in the nuclear polarization.

Figure 5.

a) Polarization buildup curves of 1:1 (v/v) [1-¹³C]pyruvic acid:sulfolane doped with 40 mM BDPA and mixed with different concentrations of Gd³⁺ at 3.35 T and 1.4 K. The buildup was monitored by irradiating the samples at 94.055 GHz with a 100 mW microwave source. b) the maximum ¹³C nuclear polarization of [1-¹³C]pyruvate sample as a function of Gd³⁺ concentration derived from Figure 5ac) Polarization buildup time constant versus Gd³⁺ concentration derived by fitting single-exponential buildup equation of data from Figure 5a.

These trends are similar to those obtained for the trityl doped-samples, but the improvement in the base nuclear polarization (Gd-free P_dnp) for BDPA-doped samples is relatively small compared to the results obtained with trityl OX063. A question arises whether a lower concentration of BDPA would improve the maximum nuclear polarization currently achieved with Gd³⁺ doping. To this end, we have performed DNP measurements of the same samples doped with 20 mM BDPA. The base P_dnp was found to be close to 13 % and with the addition of 2.5 mM Gd³⁺, the polarization jumped to 17 % (T_s=5.03 mK) as shown in Figure 5b. With a lower BDPA concentration (20 mM) in the sample, the improved polarization level was achieved at the expense of a longer microwave irradiation time with a buildup time constant increased to τ_bu=2050 s (see the Supporting Information).

It should be noted that direct comparison of the effect of Gd³⁺ on the DNP with trityl OX063 and BDPA samples cannot be made because the substrates (sodium pyruvate and pyruvic acid) as well as the glassing matrices (water-glycerol and sulfolane) were different. The nature of glassing agents may have significant influence on DNP.^35,37 Preliminary data (Supporting Information) on the polarization of 1:1 sulfolane:[1-¹³C]pyruvic acid sample doped with 15 mM trityl OX063 yielded a value close to 12 % and with the addition of 2.5 mM Gd³⁺, the nuclear polarization only doubled to P_dnp≈25 %.

Figure 6 shows the effect of Gd³⁺ doping on the solid-state nuclear T₁ of the BDPA-doped samples inside the HyperSense polarizer at 3.35 T and 1.4 K. The solid-state T₁ relaxation for a Gd-free [1-¹³C] pyruvic acid sample doped with 40 mM BDPA is close to 3000 s and it decreases to 2200 s with the addition of 8 mM Gd³⁺. On the other hand, a Gd-free ¹³C pyruvic acid sample doped with 20 mM BDPA has a T₁=6800 s and with 10 mM Gd³⁺ doping, T₁=3400 s (see the T₁ decay curves in the Supporting Information). The higher polarization obtained with Gd-doped [1-¹³C]pyruvic acid in the presence of 20 mM BDPA suggests that achieving optimal results requires a delicate balance of the electronic and nuclear relaxation parameters⁵³ that are embedded in Equation 1.

Figure 6.

Representative decay curves of the hyperpolarized ¹³C NMR signal of 1:1 (v/v) [1-¹³C]pyruvic acid:sulfolane doped with 40 mM BDPA mixed with different Gd³⁺ concentration. The solid lines are fits to an equation accounting for the decay of hyperpolarized NMR signal due to rf excitation and T₁ decay. Inset: a summary of the solid-state ¹³C nuclear T₁ relaxation values of BDPA-doped samples mixed with different Gd³⁺ concentration.

3.3. DNP of [1-¹³C] sodium pyruvate doped with 4-oxo-TEMPO

The nitroxide-based free radical 4-oxo-TEMPO has an ESR linewidth (D=5.25 mT at 2.5 T and 1 K)⁵³ that is much wider than that of the carbon-centered free radicals, trityl OX063 and BDPA. As a consequence, the DNP of 4-oxo-TEMPO-doped ¹³C substrates, which proceeds predominantly via thermal mixing, yields relatively lower polarization compared with trityl and BDPA. Nevertheless, 4-oxo-TEMPO and other nitroxyls are becoming routine polarizing agents in fast dissolution DNP-NMR because of their commercial availability and lower cost. The optimum concentration of 4-oxo-TEMPO for ¹³C DNP samples was found to be around 30-50 mM.^37,57

Figure 7 shows the microwave DNP spectra of 1.4 M [1-¹³C]pyruvate in 1:1 (v/v) glycerol:water glassing matrix doped with 40 mM 4-oxo-TEMPO in the presence and absence of Gd³⁺. Due to the limited frequency range of the microwave source in the HyperSense, only the positive polarization peak of the DNP spectrum is accessible for measurement; the negative polarization peak is located at higher microwave frequency. The microwave DNP spectrum of the Gd-free sample in Figure 7a only shows positive NMR intensity whereas the Gd-doped sample in Figure 7b starts to show negative NMR intensity in the frequency range 94.20–94.30 GHz, indicative of negative spin temperature. The latter hints that part of the negative polarization peak is shown, most likely due to a substantial shift of P(−) to lower frequency similar to the ¹³C microwave DNP spectra of trityl and BDPA-doped samples with Gd³⁺. It should be pointed out that the ESR linewidth of 4-oxo-TEMPO⁵³ is larger than the Larmor frequencies of both ¹H and ¹³C, thus both nuclear species could be polarized at the same microwave frequency via the thermal mixing process.³⁷ Unlike the trityl-doped ¹³C samples where it was shown in a previous study⁴⁸ that ¹H proceeds mainly via the solid effect at 3.35 T and 1.2 K, a proper way to prepare identical initial state when plotting the ¹³C microwave DNP spectrum of TEMPO-doped ¹³C samples may necessitate destroying the proton polarization in addition to depolarizing the remnant ¹³C magnetization from a previous microwave frequency. In our current instrumental setup, we could not verify this experimentally because the built-in NMR coil of the HyperSense polarizer cannot be tuned to ¹H but this point should be noted for future experiments of TEMPO-doped samples because of the thermal contact of ¹H and ¹³C nuclear Zeeman systems via the electron spin-spin interaction reservoir.

Figure 7.

Microwave DNP spectra of 100 μL samples of 1.4 M [1-¹³C]pyruvate in 1:1 (v/v) glycerol:water doped with a) 40 mM 4- oxo-TEMPO and b) 40 mM 4-oxo-TEMPO and 3.5 mM Gd³⁺. These data were taken in the HyperSense at 3.35 T and 1.4 K using a 100 mW microwave source. The up arrows indicate the approximate location of the positive polarization peak. Due to the limited microwave frequency range of the source, measurement at frequency ω_e>94.30 GHz, where the negative polarization peak is located, is not accessible.

Figure 8a shows the representative polarization buildup curves of 100 μL aliquots of 4-oxo-TEMPO-doped ¹³C pyruvate samples mixed with different concentrations of Gd³⁺ at 3.35 T and 1.4 K. These buildup curves were taken at 94.07 GHz, the approximate location of P(+) for the TEMPO-doped ¹³C samples. In the absence of Gd³⁺, the base DNP-enhanced nuclear polarization of the TEMPO-doped sample was found to be P_dnp≈5 % which is approximately half the base P_dnp yielded on the same sample doped with 15 mM trityl OX063 at 3.35 T and 1.4 K shown in Figure 2a. As mentioned before, this is expected since the broad 4-oxo-TEMPO D corresponds to a higher electronic heat capacity which leads to relatively higher spin temperature achieved for both the electron dipolar system and nuclear Zeeman reservoir.^35,53 A summary of the maximum nuclear polarization achieved as a function of Gd³⁺ concentration is displayed in Figure 8b. Additionally, the polarization buildup time constants τ_bu displayed in Figure 8c remained roughly the same in the 0–5 mM Gd³⁺ doping range. The relatively short τ_bu’s here is attributed to the high concentration of free radical present in these samples.

Figure 8.

a) Representative polarization buildup curves of 1.4 M [1-¹³C]pyruvate samples doped 40 mM 4-oxo-TEMPO and mixed with different concentrations of Gd³⁺ at 3.35 T and 1.4 K. The buildup was monitored by irradiating the samples at 94.07 GHz with a 100 mW microwave source. b) the maximum ¹³C nuclear polarization of [1-¹³C]pyruvate sample as a function of Gd³⁺ concentration derived from Figure 8a. c) Polarization buildup time constant vs. Gd³⁺ concentration derived by fitting the polarization buildup curves with a single-exponential buildup equation.

Similar to the pattern observed in trityl OX063 and BDPA-doped samples, the nuclear polarization monotonically increases as Gd³⁺ is added until a concentration c₁≈1 mM, although the increase was relatively modest (only 20 % improvement of base P_dnp) compared to the results obtained with trityl OX063 (approximately 300 % increase in base P_dnp). Also, the solid-state nuclear polarization was roughly constant in the Gd³⁺ concentration c₁<c<c₂, where in this particular case c₂≈3.5 mM and at c>c₂, a monotonic decrease in P_dnp is observed. These results point out that the influence of Gd³⁺ on P_dnp are similar for all three radicals, however the effect is most pronounced in trityl-doped samples where the free radical concentration used is relatively lower (15 mM). Similar measurements performed on samples doped with reduced 4-oxo-TEMPO concentration (20 mM; see the Supporting Information) yielded roughly the same improvement in P_dnp with Gd³⁺ doping at a longer microwave irradiation time. Concomitant with these changes in P_dnp with Gd³⁺ doping is a monotonic decrease in the solid-state nuclear T₁ as the sample is doped with Gd³⁺ from 0 to 8 mM as shown in Figure 9. The ¹³C T₁ drop from 2400 s with 0 mM Gd³⁺ to 1400 s with 8 mM Gd³⁺. Finally, it is currently unclear why the optimum concentration of trityl OX063 in DNP samples is generally lower than the optimum concentration of 4-oxo-TEMPO and even BDPA. It should be noted, however, that the chemical structure of trityl OX063 as shown in Chart 1 is special compared to 4-oxo-TEMPO and BDPA: its paramagnetic electron is a) enclosed in a very symmetric structure to minimize the g-anisotropy and b) surrounded mainly by non-NMR active nuclei which lowers the hyperfine interaction.⁵³

Figure 9.

Representative decay curves of the hyperpolarized solid-state NMR signal of 100 μL TEMPO-doped samples (1.4 M [1-¹³C]pyruvate in 1:1 (v/v) glycerol:water doped with 40 mM 4-oxo-TEMPO) mixed with different concentrations of Gd³⁺. The experiments were performed after the samples achieved maximum polarization level and the microwave source was turned off. Inset: Summary of the ¹³C solid-state T₁ values extracted from the decay curves versus different Gd³⁺ concentrations.

4. CONCLUSION

In summary, we have extensively investigated the effects of Gd³⁺ over a wide concentration range on the DNP of a biologically important substrate [1-¹³C]pyruvate polarized with different free radicals used in fast dissolution DNP, trityl OX063, 4-oxo-TEMPO, and BDPA. We have found three regions of interest for all three radicals in the plot of DNP-enhanced polarization versus Gd³⁺ concentration: (i) a monotonic increase in DNP-enhanced nuclear polarization P_dnp upon increasing the Gd³⁺ concentration until a certain threshold concentration c₁ (1–2 mM), (ii) a region of constant maximum P_dnp from c₁ until a concentration threshold c₂ (4–5 mM), and (iii) a monotonic decrease in P_dnp at Gd³⁺ concentration c > c₂. Of the three free radical polarizing agents examined here, the symmetric, carbon-centered trityl OX063 (15 mM) gave the best response to Gd³⁺ doping with 300 % increase in the base solid-state DNP-enhanced polarization whereas addition of Gd³⁺ at the optimum concentration to BDPA (40 mM) and 4-oxo-TEMPO-doped (40 mM) samples yielded only a modest 5–20 % increase in the base polarization. The improvement in P_dnp with Gd³⁺ doping is ascribed to the decrease in the electronic relaxation parameter η which results in a lower spin temperature achieved by the nuclear Zeeman system. Extensive ESR studies are needed to elucidate the details of the exceptional improvement in the DNP-enhanced polarization of trityl-doped samples with Gd³⁺ doping compared to samples doped with BDPA or 4-oxo-TEMPO.