Direct Dynamic Nuclear Polarization of ¹⁵N and ¹³C Spins at 14.1 T Using a Trityl Radical and Magic Angle Spinning

Abstract

We investigate solid-state dynamic nuclear polarization of ¹³C and ¹⁵N nuclei using monoradical trityl OX063 as a polarizing agent in a magnetic field of 14.1 T with magic angle spinning at ~100 K. We monitored the field dependence of direct ¹³C and ¹⁵N polarization for frozen [¹³C, ¹⁵N] urea and achieved maximum absolute enhancement factors of 240 and 470, respectively. The field profiles are consistent with polarization of ¹⁵N spins via either the solid effect or the cross effect, and polarization of ¹³C spins via a combination of cross effect and solid effect. For microcrystalline, ¹⁵N-enriched tryptophan synthase sample containing trityl radical, a 1500-fold increase in ¹⁵N signal was observed under microwave irradiation. These results show the promise of trityl radicals and their derivatives for direct polarization of low gamma, spin-1/2 nuclei at high magnetic fields and suggest a novel approach for selectively polarizing specific moieties or for polarizing systems which have low levels of protonation.

Graphical Abstract

Introduction

Solid-State NMR is a powerful method to obtain structural information at the atomic level but the breadth of its applications is restricted by its inherently low sensitivity. Studying chemical species with low concentration and/or isotopes at natural abundance levels is often impractical due to time consuming signal averaging. To circumvent these limitations, dynamic nuclear polarization (DNP), combined with magic angles spinning (MAS), appears to be a powerful approach. MAS-DNP has enabled multiples studies that would have been otherwise impossible, ranging from catalyst surfaces [1–6], to proteins under biologically relevant conditions [7–13], to samples with NMR active isotopes at low natural abundance [14–19].

DNP transfers the higher electron polarization from a polarizing agent (PA) to the surrounding nuclei by irradiating with an appropriate microwave (MW) frequency. The MAS-DNP experiment consists of spinning the sample at low temperature (typically ~100 K) and irradiating it with a high power MW source [20,21]. The most efficient PAs are biradicals [22], which enhance the proton polarization via the Cross Effect (CE) mechanism [23,24]. Proton polarization is subsequently transferred to the targeted nuclei via cross-polarization [25]. This method is particularly powerful and can, under certain circumstances, be applied to samples with a low proton content [26–28]. Extensive work has been carried out to improve the efficiency of PAs [29–33] or to adjust their affinity to the target [34–37]. Nonetheless the overall sensitivity can remain low. This becomes especially challenging at high magnetic field due to the experimental [32,38–41] and theoretical dependencies of the DNP mechanisms [23,39,42,43]

For many samples, direct polarization of the targeted nuclei is attractive vs. cross-polarization from ¹H. This is could be a powerful method for observing low gamma nuclei which are not protonated due to either the inherent chemistry of the moiety of interest or because the sample has been extensively deuterated to mitigate relaxation effects. This approach, commonly used for dissolution DNP, has found some applications for MAS-DNP [44–52]. Recently Corzilius et al. have probed the efficiency of Gd and bis-Gd chelates to directly polarize ¹³C and ¹⁵N under biological conditions.[46,53] Following their work, we report here the first high field (14.1 T) study of trityl OX063 to directly polarize ¹³C and ¹⁵N under biologically relevant conditions.

In this article, we present ¹³C and ¹⁵N enhancements as a function of the main magnetic field (the DNP enhancement profile) using a model sample to illustrate the mechanisms at stake. We measure the polarization gains corrected by bleaching losses and depolarization factor as well as the polarization build-up times and MW power dependence. Additionally, we demonstrate the use of trityl OX063 to directly enhance an ¹⁵N-enriched protein sample and show that high sensitivity can be achieved by directly polarizing ¹⁵N spins to achieve SNR on par with the standard DNP approach using AMUPol and CP [29], with unprotonated ¹⁵N moieties exhibiting a relatively larger enhancement by direct polarization.

Materials and Methods

Urea Samples

Three samples containing equal volumes of 2 M [¹³C, ¹⁵N] urea in D₈-glycerol/D₂O/H₂O (60/30/10; v/v/v) were prepared. The first contained 40 mM trityl OX063 (Oxford Instruments Molecular Biotools), the second contained 20 mM trityl OX063, and the third was radical-free. The same 3.2 mm sapphire MAS rotor was used when characterizing these samples.

Tryptophan Synthase Samples

U-¹⁵N tryptophan synthase (TS) was expressed and purified as previously described [54,55]. TS microcrystals were crystallized by 1:1 dilution of the soluble protein with 50 mM bicine in D₂O, pH adjusted to 7.8 with CsOH, containing 14% PEG-8000 (W/V) and 3 mM spermine. Crystals were harvested at 10,000 rpm for 10 minutes at 4° C and washed with bicine buffer in 90% D₂O, pH adjusted to 7.8 with CsOH, containing 1.5 mM spermine, 8% PEG-8000 (W/V), 10% DMSO (V/V) and an excess of the high-affinity α-site ligand N-(4’-trifluoro-methoxybenzenesulfonyl)-2-amino-ethyl phosphate (F9). Microcrystals were again harvested at 10,000 rpm for 10 minutes at 4 °C and loaded into 3.2 mm sapphire MAS rotors with a sample volume of ~25 μL; rotors were packed to identical levels using a single batch of microcrystals in an attempt to match the amount of protein in each rotor. Radical, either AMUPol or trityl OX063, was added directly to the sample in the rotor from concentrated stock solutions made using the same crystallization buffer to achieve a final radical concentration of 20 mM or 40 mM, respectively. A comparison sample containing 20 mM trityl OX063 was also made.

Simulation of trityl EPR line shape

The simulated EPR spectrum of trityl OX063 at 14.1 T was obtained using EasySpin [56]. We assumed a g-tensor with principle axis values [2.003193 2.003193 2.002583] [57] and linebroadening of 1 mT Gaussian and 1mT Lorentzian. These parameters were based on literature values for the g-tensor [57] and our prior experimental measurements of trityl OX063 EPR spectra at 240 GHz.

NMR/DNP measurements for urea samples

NMR and DNP measurements were carried out on a custom DNP NMR system operating at 14.1 T with a 395 GHz gyrotron described in detail elsewhere [58]. Reported MW powers were measured at the input to the MAS probe with an Ophir 3A-P-THz pyrometer calibrated by comparison with an absorbing water load. All [¹³C, ¹⁵N] urea spectra were collected at 7 kHz MAS. Direct (DPMAS) ¹⁵N and ¹³C polarization NMR spectra were obtained via a π/2 excitation pulse centered at the resonance of interest with ~100 kHz proton decoupling during signal acquisition. To measure absolute sensitivity, spectra were collected for both ¹³C and ¹⁵N using the radical-free sample with a recycle delay of 3T_1n to calculate the maximum achievable thermally polarized signal. Microwave power was optimized under conditions where efficient DNP was expected for the trityl radical, and DNP enhancement profiles for ¹³C and ¹⁵N were collected by varying the main magnetic field while holding the gyrotron frequency at 395.175 GHz. For each field measurement, the nuclear excitation and monitoring frequencies were set on resonance. Recycle delays were set to 300 s and 200 s for ¹³C and ¹⁵N measurements, respectively, and saturation pulse trains on both ¹H and ¹³C/¹⁵N channels destroyed any existing polarization prior to DNP buildup. For measurements of absolute maximum DNP enhancement, DNP buildup measurements, in which polarization is measured as a function of delay between saturation pulses and a π/2 excitation pulse, were performed at field positions for maximum ¹³C and ¹⁵N direct polarization. The ratios of maximum signal intensities for the MW-on experiments for the trityl-containing samples and the MW-off experiments for the radical-free sample, scaled by the number of scans for each spectrum, were used to estimate the absolute sensitivity enhancements. The absolute enhancement provides an assessment of the combined effects of depolarization under MAS and any trityl bleaching of nearby nuclear spins in order to provide an evaluation of the real gain by direction polarization from electrons to ¹³C and ¹⁵N using 40 mM trityl OX063 [40] (also called θ [59]). This measurement of absolute enhancement was used to scale the relative enhancements measured at other magnetic field settings.

NMR/DNP measurements for Tryptophan Synthase (TS) samples

All TS spectra were collected at 10 kHz MAS. DPMAS ¹⁵N spectra for the trityl OX063 containing samples were obtained via a π/2 excitation pulse at the resonance of interest witĥ100 kHz proton decoupling during signal acquisition. Cross polarization (CPMAS) ¹⁵N spectra for the AMUPol containing sample were obtained via a π/2 excitation pulse on the proton channel followed by 2 msec of ¹⁵N cross polarization before ~100 kHz proton decoupling during signal acquisition. CPMAS spectra were not attempted for the trityl containing samples as proton measurements on [¹³C, ¹⁵N] urea samples showed little DNP enhancement. Similarly, DPMAS spectra were not collected for the AMUPol containing sample as direct polarization of ¹⁵N from AMUPol is not as efficient as DNP of ¹H followed by cross polarization. For measurements of relative DNP enhancement of ¹⁵N, DNP buildup measurements, in which polarization is measured as a function of delay between saturation pulses and the π/2 excitation pulse, were performed for the trityl-and AMUPol-containing TS samples. Trityl-containing samples were measured at a field position for maximum ¹⁵N polarization by the trityl monoradical using DPMAS and the AMUPol-containing sample was measured at a field position for maximum ¹H polarization by the AMUPol biradical using CPMAS. The ratios of maximum achievable signal intensities for the DPMAS and CPMAS experiments for the trityl-containing and AMUPol-containing samples, respectively, were scaled by the number of scans for each spectrum and the relative time needed for polarization buildup to calculate the sensitivity per unit time for the different radicals.

Results and Discussion

Field dependence for ¹⁵N and ¹³C polarization

Direct ¹⁵N and ¹³C polarization measurements of [¹³C, ¹⁵N] urea by 40 mM trityl OX063 as a function of external magnetic field are shown in Figure 1 along with a simulated EPR spectrum for the trityl radical. The FWHM of the simulated Trityl OX063 EPR lineshape is 4.37 mT. The difference in magnetic fields corresponding to the ¹⁵N enhancement maxima is 4.26 mT. The ¹⁵N Larmor frequency at this field corresponds to 2.17 mT. Thus, the solid effect (SE) mechanism for ¹⁵N enhancement would predict maximal enhancements separated by 4.34 mT while the CE mechanism would predict a separation of 2.17 mT. Given the inherent broadening due to the spin properties of the trityl radical, the observed polarization enhancement profile for ¹⁵N is consistent with either mechanism of polarization transfer since the positive and negative maximal enhancements occur within the EPR line shape at about twice the ¹⁵N Larmor frequency [47]. The enhancement profile for ¹³C polarization is more complex and consistent with polarization via a combination of the SE and the CE [47,60] since the maxima are separated by almost twice the ¹³C Larmor frequency, or 10.77 mT, and are well beyond the limits of EPR line shape. We also measured ¹H polarization, but found only a 14-fold polarization enhancement via the trityl radical, as would be expected for narrow line radical polarizing via the SE.

Figure 1:

¹⁵N and ¹³C absolute enhancement as a function of magnetic field for 2 M [¹³C,¹⁵N]urea in glycerol-d₈–D₂O–H₂O (60 : 30 : 10 volume ratio) containing 40 mM OX063 trityl radical at 98 K and a MAS spinning rate of 7 kHz. The gyrotron frequency was fixed at 395.175 GHz. For each magnetic field setting, ¹⁵N and ¹³C absolute enhancements were measured via direct polarization by comparing NMR signal with and without microwaves using a 300 s (¹⁵N) or 200 s (¹³C) buildup time and considering the signal reduction caused by the presence of trityl. For reference, a simulated EPR spectrum of trityl at 14.095 T is provided.

Achievable maximum ¹⁵N and ¹³C polarization

Theoretical gains for ¹⁵N and ¹³C are 6500 and 2620, respectively, for ideal transfer of polarization when compared to equilibrium thermal polarization. Measurements of ¹⁵N and ¹³C polarization using 40 mM trityl OX063 at an optimal field setting and comparing microwaves on vs microwaves off spectra yield gains of 2510 and 530, respectively (Table 1). However, comparisons to a control sample, which does not contain radical, indicate that the absolute gains of 470 and 240, respectively, are significantly attenuated by bleaching and/or depolarization under MAS by the high concentration of trityl OX063 required to achieve significant DNP. It is unlikely that a further improvement could be achieved by decreasing the radical concentration to reduce bleaching/depolarization as measurements using a sample containing 20 mM trityl OX063 showed significantly less bleaching and/or depolarization but also substantially lower DNP enhancement. The optimal microwave power needed for maximum polarization enhancement using trityl OX063 was similar to the optimal power for AMUPol (Figure 2). We note that ¹⁵N DNP enhancement via the trityl radical builds more quickly at lower power levels, consistent with it having a narrower field profile. This suggests an alternative strategy for applications of DNP which are power limited (i.e. when using MW sources with lower power output, such as klystrons [61–63]). At the temperatures needed for DNP, the ¹⁵N and ¹³C T₁ relaxation times for frozen urea are quite long at 3026 s and 1869 s (Table 2), respectively, for the radical-free sample. They only decrease to 1676 s and 905 s, respectively, in the presence of 40 mM trityl OX063, leading to very long DNP buildup times. In an attempt to shorten the polarization buildup time, a [¹³C, ¹⁵N] urea sample containing both 40 mM trityl OX063 and 4 mM Gd-fullerene was also tested (data not shown). The addition of Gd-fullerene was observed to reduce the DNP buildup time as a result of shortening ¹³C and ¹⁵N T₁ relaxation rates without any observable further reduction in thermal equilibrium polarization for either ¹³C or ¹⁵N. However, DNP enhancements were also reduced due to shortening the T_1e relaxation rate of the trityl radical. A lower concentration (e.g. ≤ 1 mM) of Gd-fullerene or a similar compound may sufficiently shorten the nuclear spin relaxation rates without deleteriously affecting the trityl T_1e relaxation rate, but we have not extensively tested its concentration dependence.

Table 1:

¹⁵N and ¹³C maximum DNP enhancements for 2 M ¹³C-¹⁵N-urea in glycerol-d₈–D₂O–H₂O (60 : 30 : 10 volume ratio) containing 40 mM and 20 mM OX063 trityl radical at 98 K and a MAS spinning rate of 7 kHz. DNP enhancement values were calculated considering signal reduction caused by the presence of trityl.

trityl conc./mM	40			20
	εon/off	εDepol × εBleaching	εabs	εon/off	εDepol × εBleaching	εabs
15N	2510	0.19	470	220	0.43	94
13C	530	0.45	240	-	-	-

Figure 2.

Power dependence of DNP enhancement for ¹H-¹³C CPMAS (open diamonds), ¹³C DPMAS (solid squares), and ¹⁵N DPMAS (solid circles) for 2 M [¹³C-¹⁵N]urea samples containing either 20 mM AMUPol (CPMAS) or 40 mM trityl OX063 (DPMAS) with a glassing matrix of glycerol-d8–D₂O–H₂O (60:30:10 volume ratio) at 98 K and 7 KHz MAS rate. Power measurements were collected at the main magnetic fields corresponding to the maximal DNP enhancements for each approach.

Table 2:

Polarization buildup times for 2 M ¹³C-¹⁵N-urea in glycerol-d₈–D₂O–H₂O (60: 30: 10 volume ratio) containing 40 mM OX063 trityl radical and for an undoped sample at 98 K and a MAS spinning rate of 7 kHz

trityl conc./mM		40	0
15Na	mw on	1251	-
	mw off	1284	1869
13Cb	mw on	1078	-
	mw off	905	3026

^a.NMR field at 14.09326 T (negative maximum ¹⁵N enhancement).

^b.NMR field at 14.09964 T (positive maximum ¹³C enhancement).

DNP buildup, depolarization and absolute enhancement for Tryptophan Synthase (TS)

To test the viability of direct ¹⁵N polarization using trityl OX063 in a protein sample, we measured the DNP polarization for a microcrystalline ¹⁵N-enriched TS sample. We note that for this protein we have found that cryoprotectants and crystallization conditions utilized for characterizing this protein by diffraction at low temperature (e.g. 10% DMSO and 8% PEG-8000 [64]) also work well for keeping the AMUPol biradical well distributed for the purposes of DNP. To evaluate trityl OX063 polarization relative to typical conditions for proteins, we also prepared a sample containing 40 mM trityl OX063 and a control sample containing 20 mM AMUPol (i.e. a sample with the same e⁻spin concentration but using a proven biradical PA). We first characterized the optimal DNP power needed for polarizing with 40 mM trityl OX063 and the DNP buildup time (Figure 3). The power needed for optimal polarization was similar to the power needed for optimal polarization using 20 mM AMUPol. DNP buildup curves observed for trityl containing samples were consistently biexponential. The buildup times for 20 mM trityl OX063 and 40 mM trityl OX063 were very similar with a fast component having a DNP buildup time of 38–41 s and a slow component exhibiting a buildup time of 636–681 s. However, the ratio of the fast and slow components altered from 1:6 at 20 mM trityl OX063 to 1:2 at 40 mM trityl OX063, consistent with an increase in radical concentration. We note that the bimodal buildup behavior can be explained via several possible mechanisms. The first is that there are two populations of trityl radicals—one which is evenly distributed and solvated with a second population existing in an aggregated form; this would agree with the differences we observe with radical concentration and is the most likely explanation. A second possibility is that polarization of the microcrystals proceeds via an inefficient spin diffusion since the ¹⁵N spin bath is very weakly coupled due to its substantially lower gyromagnetic ratio relative to ¹H [43,65–67]. A third possibility is that some of the microcrystalline protein is dissolving into the surrounding buffer. While we did not observe any loss of crystals and maintained the buffer reagents at a constant concentration to minimize crystal dissolution, some exchange of protein into the surrounding buffer is inevitable and could explain the observation of two polarization timescales. However, we consider this is a less likely explanation given the differences observed for 20 mM radical vs. 40 mM radical.

Figure 3.

¹⁵N direct DNP buildup curves and fittings for U-¹⁵N TS samples containing 40 mM (red circles) or 20 mM (blue squares) trityl OX063 at a magnetic field strength of 14.0931 T (field corresponding to maximal ¹⁵N enhancement for trityl OX063), 11 W of MW power at 395.175 GHz, 98 K and 10 kHz MAS. Solid lines are fits using double exponential functions. Dashed lines are fits of double exponential functions of polarization buildup against square root of buildup times to calculate the optimal repetition rate for maximum SNR per unit time. The optimized buildup times derived from fits are 67 s and 497 s for 40 mM and 20 mM trityl, respectively.

The optimal recycling delays calculated from the buildup curves were 67 s and 497 s for 40 mM and 20 mM trityl OX063, respectively. Overall, the 40 mM trityl OX063 provided better SNR per unit time. More importantly, the SNR per unit time for a DPMAS spectrum for the sample containing 40 mM trityl OX063 was 60% better than the optimal DNP CPMAS spectrum for the sample containing 20 mM AMUPol. We note that in other protein samples where the AMUPol concentration has been extensively optimized, the best results are typically observed at ~10 mM AMUPol for 600 MHz MAS-DNP. However, we chose 20 mM AMUPol for comparison in this study to match the radical concentration. Thus, it is possible that with a more optimal AMUPol concentration the SNR per unit time may be equivalent or even better than that observed for trityl OX063. Nonetheless, we were able to achieve significant polarization using a narrow line monoradical suggesting new directions for achieving optimal DNP at high magnetic fields (Figure 5). In comparing the MAS-DNP NMR spectra for the two radicals (Figure 6), we note that the DPMAS spectrum for the trityl-containing sample has a significantly higher relative intensity for the resonances at 240–250 ppm compared to the CPMAS spectrum for the AMUPol-containing sample. This region corresponds to histidine nitrogens in the TS enzyme which are not protonated. This highlights possible approaches for direct polarization of low gamma nuclei, particularly those which are not protonated and thus more difficult to enhance with a biradical and CPMAS approach.

Figure 5:

Comparison ¹⁵N spectra for ¹⁵N-enriched Tryptophan synthase collected using CPMAS and 20 mM AMUPol (black spectrum) vs DPMAS and 40 mM trityl OX063 (red spectrum). CPMAS and DPMAS spectra were collected with acquisition times of 6372 s (2048 scans) and 6926 s (128 scans), respectively, and scaled for their relative acquisition times. Spectra acquired at 14.1 T with 10 kHz MAS at 98 K. Input MW power was set to 11 W at 395.175 GHz. For direct ¹⁵N MAS experiment, the recycle delay was set to 54 s for each transient. A larger relative enhancement is observed in the DPMAS spectrum at 240–250 ppm corresponding to unprotonated ¹⁵N moieties.

Conclusions

In this report we have studied for the first time the potential for direct polarization from a trityl radical to ¹³C and ¹⁵N nuclei at 14.1 T, corresponding to a magnetic field that is almost three times higher than previous studies [47,60]. The analyses of [¹³C, ¹⁵N] urea samples show a significant polarization gain for both nuclei despite the significantly higher magnetic field. We also estimated quenching and depolarization under MAS and recorded the DNP enhancement profiles as a function of magnetic field. The DNP enhancement profiles are consistent with a CE mechanism for polarizing ¹⁵N nuclei and a mixture of CE and SE mechanisms for polarizing ¹³C nuclei, although some polarization of ¹⁵N via the SE cannot be ruled out. These mechanistic observations are consistent with our estimates of depolarization and quenching effects, with more significant depolarization and bleaching observed for ¹⁵N nuclei than for ¹³C nuclei as would be expected with a predominately CE mechanism for polarizing ¹⁵N nuclei and a mixture CE and SE mechanisms for polarizing ¹³C nuclei.

Finally, we examined the viability of direct polarization for enhancing ¹⁵N sensitivity in a biologically relevant sample. Remarkably, the time unit sensitivity was on par and even slightly better when comparing 40 mM trityl OX063 and DPMAS spectra to 20 mM AMUPol and CPMAS spectra. This is an unexpected result considering the exceedingly long time required to directly polarize ¹⁵N spins. Thus, the direct polarization using trityl radicals can be considered even for high magnetic field applications and one may expect further considerable improvements with the development of trityl-based bir

Figure 4:

DPMAS ¹⁵N spectra with and without microwaves for ¹⁵N-enriched Tryptophan Synthase sample containing 40 mM trityl OX063. A recycle delay time of 176 s was used for both experiments. Spectra acquired at 14.1 T with 10 kHz MAS at 98 K. Input MW power was set to 11 W at 395.175 GHz for the MW-on experiment.

Table 3:

Signal-to-noise comparison for ¹⁵N spectra of ¹⁵N-enriched Tryptophan synthase collected using CPMAS and 20 mM AMUPol vs DPMAS and 40 mM trityl. Spectra acquired at 14.1 T with 10 kHz MAS at 98 K. Input mw power was set to 11 W at 395.175 GHz.

experiment/sample	SNR per time unit	DNP buildup times/s (ratios)
DPMAS/40 mM trityl	1.6	41,681 (1:2)
DPMAS/20 mM trityl	0.9	38,636(1:6)
CPMAS/20 mM AMUPol	1.0	-