Proton-detected 3D ¹⁵N/¹H/¹H isotropic/anisotropic/isotropic chemical shift correlation solid-state NMR at 70 kHz MAS

Abstract

Chemical shift anisotropy (CSA) tensors offer a wealth of information for structural and dynamics studies of a variety of chemical and biological systems. In particular, CSA of amide protons can provide piercing insights into hydrogen-bonding interactions that vary with the backbone conformation of a protein and dynamics. However, the narrow span of amide proton resonances makes it very difficult to measure ¹H CSAs of proteins even by using the recently proposed 2D ¹H/¹H anisotropic/isotropic chemical shift (CSA/CS) correlation technique. Such difficulties due to overlapping proton resonances can in general be overcome by utilizing the broad span of isotropic chemical shifts of low-gamma nuclei like ¹⁵N. In this context, we demonstrate a proton-detected 3D ¹⁵N/¹H/¹H CS/CSA/CS correlation experiment at fast MAS frequency (70 kHz) to measure ¹H CSA values of unresolved amide protons of N-acetyl-¹⁵N-L-valyl-¹⁵N-L-leucine (NAVL).

Graphical abstract

Introduction

Chemical shift anisotropy (CSA) tensors measured using solution and solid-state NMR methods have always played a pivotal role in getting deeper insights into inter and intra molecular H-bonding, electrostatic, and π-electron interactions.^1-12 All these interactions heavily rely on the relative position of protons with respect to each other and/or other atoms. Subsequently, measurement of ¹H CSA has been of considerable interest in various research groups.^13-20 In particular, amide protons (NH) are mostly involved in H-bonding interactions providing structural strength to enormous chemical and biological systems and, consequently, ¹H CSA which is extremely sensitive to the local electronic environment and dynamics can lead to a comprehensive understanding of these interactions. In general, ¹⁵N isotropic/anisotropic chemical shifts of amide groups^11,21,22 are used to study H-bonding interactions. Since ¹⁵N nuclei indirectly participate in H-bonding interactions, therefore measurement of ¹H isotropic/anisotropic chemical shifts of amide protons that are directly involved in H-bonding is of a greater significance. Nevertheless, the distribution of isotropic chemical shift of amide protons is rather limited and its relationship with the strength of H-bonding interaction is not so obvious. This brings a strong motivation to develop methods to measure ¹H CSA of H-bonded amide protons.

Usually, the determination of ¹H CSA has always been challenging due to its small size and the presence of strong homonuclear dipolar interactions between ¹Hs. In the past several methods using rotor-synchronous pulse sequences are reported that work well for its extraction either at slow magic angle spinning (MAS) or fast MAS. In a previous report, site-resolved ¹H CSA measurement of amide protons in proteins is already demonstrated through ¹³C-detected 3D ¹H/¹⁵N/¹³C CSA/CS/CS experiment at a moderate MAS rate of 14 kHz.¹⁵ Herein, the overlapped ¹H resonances were resolved in the ¹⁵N/¹³C dimensions, allowing a site specific determination of ¹H CSA of amide groups. While ¹H CSA can be measured at moderate MAS rates, the development of methods for its measurement at fast MAS (> 60 kHz)¹⁸ are essential in order to avoid complexities associated with the experimental setup at slow MAS. Recently we demonstrated a 2D ¹H/¹H CSA/CS correlation method utilizing fast MAS and composite-180° pulse-based rotor-synchronized γ-encoded CSA recoupling sequences that result in undistorted ¹H CSA lineshapes with excellent efficiency.²³ In continuation to this, we also presented a 3D ¹H/¹H/¹H CSA/CSA/CS correlation experiment mediated through ¹H/¹H radio frequency-driven recoupling (RFDR) to extract relative orientation between two interacting ¹H CSA tensors at fast MAS.²⁴ However, 2D ¹H/¹H CSA/CS experiment cannot be utilized for CSA measurements of amide protons due to the lack of resolution in the ¹H CS dimension. In this regard, we present a 3D ¹⁵N/¹H/¹H CS/CSA/CS correlation experiment on N-acetyl-¹⁵N-L-valyl-¹⁵N-L-leucine (NAVL) that results in site-resolved ¹H CSAs measured through ¹⁵N isotropic chemical shifts. Manifold benefits of the 3D ¹H CSA measurement at fast MAS include: 1) enhanced sensitivity by ¹H detection,^25-37 2) requirement of a tiny amount (0.3 – 1 mg) of sample due to small rotor volume (micro-coil probe design),^38-44 3) direct correlation between ¹H CSA and ¹H CS which otherwise cannot be obtained due to overlap of ¹H CS resonances, 4) efficient and robust ¹H CSA recoupling,^23,24,45 and 5) low-power sequences can be implemented to avoid sample heating.^46-50

Pulse sequence

The pulse sequence used for the proton-detected 3D ¹⁵N/¹H/¹H CS/CSA/CS correlation experiment is shown in Figure 1. ¹H magnetization is transferred to ¹⁵N nuclei using ramped-amplitude cross-polarization (RAMP-CP).⁵¹ After which ¹⁵N nuclei are allowed to evolve during t₁ under their isotropic chemical shifts in the presence of ¹⁵N-¹H heteronuclear low-power CW decoupling on the ¹H channel. ¹⁵N magnetization is then back transferred to ¹H by the second RAMP-CP. Residual ¹H transverse magnetization is removed by the use of the homonuclear rotary resonance (HORROR) sequence^52-54 on the ¹H channel just before the application of second CP.⁵⁵ While the ¹⁵N magnetization is stored along the z-axis by the first 90° pulse on ¹⁵N just before the HORROR irradiation to avoid time-evolution and transversal relaxation, the second 90° pulse on ¹⁵N prepares magnetization for the second CP step. ¹H magnetization is stored along the z-axis by the use of a 90° pulse just after the second CP step and is allowed to evolve under the recoupled ¹H CSA using the symmetry-based $R{18}_8^7(270°90°)$ sequence with the composite-180° pulse element during the t₂ period.²³ A 180° pulse in the middle of CSA recoupling period (t₂) is applied on the ¹⁵N channel to decouple ¹⁵N-¹H hetronuclear couplings in contrast to the previously reported pulse sequences wherein 180° pulses are applied in the middle of every cycle time of the CSA recoupling pulses.^15-17,56 Finally, ¹H magnetization is prepared for detection by the application of a 90° read pulse and data is acquired in the presence of ¹⁵N-¹H WALTZ decoupling on the ¹⁵N channel. While the States-TPPI (Time Proportional Phase Incrementation)⁵⁷ approach is applied to the t₁ period to obtain pure absorption line shape, the amplitude modulated signals are recorded in the t₂ period.

Figure 1.

Proton-detected 3D ¹⁵N/¹H/¹H CS/CSA/CS pulse sequence. ¹⁵N chemical shifts are expressed during t₁ under the low-power ¹H CW decoupling, while ¹H CSAs are recoupled during t₂. Simultaneous recoupling of ¹⁵N-¹H hetronuclear interactions are avoided by the application of 180° pulse on the ¹⁵N channel in the middle of CSA recoupling pulses. The phase cycling scheme used in the 3D pulse sequence is as follows: ϕ₁ = {4 (0), 4 (180)}, ϕ₂ = {90}, ϕ₃ = {0}, ϕ₄ = {0}, ϕ₅ = {90}, ϕ₆ = {2 (0), 2 (180)}, ϕ₇ = {90}, ϕ₈ = {270}, ϕ₉ = {{0, 180}, {90, 270}}, ϕ₁₀= {90}, ϕ₁₁ = {270}, ϕ₁₂ = {0}, ϕ₁₃ = {0}, ϕ₁₄ = {0}, ϕ_acq = {0, 180, 180, 0, 180, 0, 0, 180}.

Experimental

All NMR measurements were carried out at a 600 MHz solid-state NMR spectrometer (JEOL RESONANCE Inc., JNM-ECZ600R) equipped with a 0.75 mm double-resonance ultrafast MAS probe (JEOL RESONANCE Inc.) operating at ¹H and ¹⁵N Larmor frequencies of 599.67 MHz and 60.76 MHz, respectively. Approximately 0.3 mg of NAVL was packed into a 0.75 mm zirconia rotor and all data were collected at an ambient temperature under 70 kHz MAS. The ¹H and ¹⁵N 90° pulse lengths were set to 0.6 and 1.4 μs, respectively. A recycle delay of 5 s and an acquisition time of 8.19 ms were used in the 3D experiment. 8 transients were coadded for each 32 t₁ and t₂ increments. While a longer contact time of 4 ms was used for the first CP (¹H to ¹⁵N) to maximize the magnetization transfer efficiency from protons to ¹⁵N nuclei, a short contact time of 0.4 ms was used for the second CP (¹⁵N to ¹H) to select only those protons that are directly bonded to nitrogens in NAVL. The double quantum CP condition in both CP steps was achieved using 15 and 55 kHz RF field strengths for ¹H and ¹⁵N, respectively.^47,49,50,58 The use of double quantum CP condition becomes essential especially for heat-sensitive samples like proteins to avoid sample degradation due to rf heating during the experiment. However, such matching condition can be sensitive to chemical shift offsets due to the limited rf power. In the present study as the ¹H chemical shift range of amide groups is limited, we used a weak rf field on ¹H and, consequently, a relatively stronger rf field on ¹⁵N during CP was applied to cover the ¹⁵N chemical shift dispersion. A ¹⁵N-¹H heteronuclear continuous-wave (CW) decoupling of 9.5 kHz was applied on the ¹H channel during the t₁ evolution period, whereas ¹⁵N-¹H WALTZ decoupling with 13 μs pulse of RF field strength 18.9 kHz was applied on the ¹⁵N channel during t₃. The phase-alternated HORROR sequence was applied for 8 ms duration with 35 kHz ¹H RF field. The γ-encoded symmetry-based pulse sequence $R{18}_8^7(270°90°)$ was used for the ¹H CSA recoupling. The composite-180° pulse in each R element of $R{18}_8^7(270°90°)$ pulse sequence is a combination of 270° and 90° RF pulses with corresponding phases alternating between (70°, 250°) and (-70°, -250°). The amplitude modulated t₂ signal was obtained at every 8τ_r period. Delta NMR software (JEOL RESONANCE Inc.) was used to process all the NMR data. While a Fourier transformation after zero filling was applied in the t₁ and t₃ dimensions, a DC balance to remove DC offset effects (the average of the final 1/8^th FID points in t₂ is subtracted from total data points) followed by zero filling and real Fourier transformation were applied in the CSA recoupling dimension (t₂) to process the 3D ¹⁵N/¹H/¹H CS/CSA/CS correlation data. The CSA parameters were extracted through numerical fittings using SIMPSON program.⁵⁹ Powder averaging was achieved using 678 (α, β) crystal orientations and 26 γ angles.⁶⁰

Results and Discussion

Molecular structure and 1D ¹H spectra of NAVL collected at 70 and 100 kHz MAS are shown in Figure 2. While the resolution and sensitivity are seen to be significantly improved for protons H3, H4, H5, H6 and H7 (peaks in the 0 - 6 ppm range) with increasing the MAS rate due to better suppression of ¹H/¹H homonuclear dipolar interactions, proton resonances from amide groups of Val and Leu remain unresolved even with a MAS rate as fast as 100 kHz. In other words, spectral resolution provided by 100 kHz MAS is not sufficient enough to resolve amide proton resonances with similar isotropic chemical shift values. This problem is crucial especially for the CSA measurement of structurally relevant protons from the 2D ¹H/¹H CSA/CS correlation experiments that require well-resolved ¹H resonances for the precise determination of CSA. To demonstrate this limitation further, we carried out 2D ¹H/¹H CSA/CS correlation measurement on NAVL at 70 kHz MAS.²³ Since CSA and heteronuclear dipolar coupling have the same spatial and spin symmetry, the previously reported pulse sequence simultaneously recouples both interactions. Subsequently, in order to accomplish ¹⁵N-¹H heteronuclear decoupling such that any recoupling of heteronuclear dipolar interactions is avoided, we applied a 180° pulse on the ¹⁵N channel in the middle of the CSA recoupling pulses. It is to be noted that NH proton CSA can also be measured in natural abundant samples wherein ¹⁴N-¹H heteronuclear interactions are decoupled through on-resonance ¹⁴N CW irradiation at ultrafast MAS.⁴⁵ The fast MAS spectrum of NAVL (Figure 3A) shows well-resolved resonances barring NH peaks, as a result CSA of amide protons cannot be determined. While the 2D ¹H/¹H CSA/CS correlation spectrum fails to provide any site-specific information about CSA of amide protons of NAVL, it can still be used to extract CSA of OH proton of –COOH group in NAVL. A spectral slice taken parallel to CSA dimension at the ¹H isotropic chemical shift of OH resonance is simulated using SIMPSON⁵⁹ (Figure 3B) and ¹H CSA parameters obtained from the line shape fitting are: δaniso = δ_zz-δ_iso = 19.1 ppm and η = (δ_yy-δ_xx)/δaniso = 0.3. Here, δ_iso is the isotropic chemical shift calculated using the relation (δ_xx+δ_yy+δ_zz)/3, and the principal components of the chemical shift tensor (δ_xx, δ_yy and δ_zz) are defined as |δ_zz-δ_iso| ≥ |δ_xx-δ_iso| ≥ |δ_yy-δ_iso|.

Figure 2.

Molecular structure and ¹H NMR spectra of a powder sample of NAVL recorded at 600 MHz under 70 and 100 kHz MAS frequencies.

Figure 3.

(A) Two-dimensional ¹H/¹H CSA/CS correlation spectrum of NAVL at 600 MHz under 70 kHz MAS obtained using a previously reported pulse sequence²³ with a 180° pulse in the middle of the CSA recoupling period on the ¹⁵N channel to decouple ¹⁵N-¹H heteronuclear interactions. In the 2D ¹H/¹H CSA/CS correlation experiment 6 scans were collected every 32 t₁ increments and a recycle delay of 5 s was used (total experimental time = 16 minutes). Dotted circle indicates the two overlapped amide proton resonances. (B) The experimental CSA lineshape extracted by taking a slice (green line) parallel to the ¹H CSA dimension of the 2D spectrum at the ¹H isotropic chemical shift frequency of OH resonance of NAVL. Best fitting simulated lineshape (brown) was obtained using SIMPSON simulations.

The straightforward solution to overcome the issue of overlapping ¹H resonances of amide groups of NAVL is to carry out CSA measurement in a 3D manner by adding a high-resolution ¹⁵N dimension so as to get well-resolved ¹⁵N/¹H correlations. To this end, we implemented 3D ¹⁵N/¹H/¹H CS/CSA/CS pulse sequence shown in Figure 1 to obtain amide proton CSA of Val and Leu residues of NAVL. As discussed earlier, ¹⁵N chemical shifts are allowed to evolve during the incrementable t₁ period, and ¹⁵N filtered ¹H signal after the two CP steps then evolves under recoupled ¹H CSA during the t₂ period in the absence of ¹⁵N-¹H heteronuclear couplings. The 2D F1/F3 (¹⁵N/¹H CS/CS) projection (Figure 4B) extracted from the 3D ¹⁵N/¹H/¹H CS/CSA/CS spectrum of NAVL (Figure 4A) shows well-separated ¹⁵N/¹H chemical shift correlations for Val and Leu residues of NAVL. Additionally, ¹H isotropic chemical shifts of amide protons are more precisely determined and can be used for the structural refinement using the quantum chemical calculations. To determine CSA of amide protons, the 2D F1/F2 (¹⁵N/¹H CS/CSA) projection (Figure 5A) was extracted as the first step from the 3D ¹⁵N/¹H/¹H CS/CSA/CS spectrum of NAVL (Figure 4A). In the subsequent step, spectral slices were taken parallel to the ¹H CSA dimension of the 2D F1/F2 (¹⁵N/¹H CS/CSA) projection at the ¹⁵N isotropic chemical shifts of Val and Leu residues of NAVL (Figure 5A). Finally, numerical simulations were performed to get the best fit of experimental ¹H CSA lineshapes (Figure 5B). The ¹H chemical shift parameters obtained are as follows: Val (δ_iso = 9.0 ppm, δaniso = 9.1 ppm, η = 0.7) and Leu (δ_iso = 9.2 ppm, δaniso = 9.1 ppm, η = 0.7). While we observed similar CSA parameters for the amide protons of NAVL, 3D ¹⁵N/¹H/¹H CS/CSA/CS correlation experiment presented in this study is necessary for the cases where CSA of amide protons with unresolved peaks in the proton CS dimension are different and in principle leads to a direct correlation between ¹H CSA and ¹H CS. Subsequently, a set of CS and CSA of each amide proton can be extracted simultaneously.

Figure 4.

(A) 3D ¹⁵N/¹H/¹H CS/CSA/CS correlation spectrum of NAVL at 70 kHz MAS frequency. (B) 2D F1/F3 (¹⁵N/¹H CS/CS) projection extracted from the 3D ¹⁵N/¹H/¹H CS/CSA/CS spectrum of NAVL. Total experimental time = 22.5 hours.

Figure 5.

(A) Two-dimensional F1/F2 (¹⁵N/¹H CS/CSA) projection extracted from the 3D ¹⁵N/¹H/¹H CS/CSA/CS spectrum of NAVL. (B) Spectral slices (green) taken parallel to the ¹H CSA dimension at the ¹⁵N chemical shifts of Val and Leu residues of NAVL, and corresponding simulated lineshapes (brown).

Conclusion

In summary, we have presented a proton-detected 3D ¹⁵N/¹H/¹H CS/CSA/CS correlation experiment at fast MAS (70 kHz) on NAVL. The main objective of this experiment is to determine site-resolved ¹H CSA of overlapped amide proton resonances through ¹⁵N isotropic chemical shifts. The measured CSA parameters of amide protons of Val and Leu residues of NAVL are found to be identical. In addition, we have also reported ¹H CSA of well-resolved OH resonance of NAVL obtained from 2D ¹H/¹H CSA/CS correlation experiment. We believe that the proton-detected 3D pulse sequence presented in this study can be implemented for studies relying on ¹H CSAs in more complex chemical and biological systems wherein a limited or no resolution is a major problem to overcome. Additionally, a combination of partially-deuterated samples, higher magnetic field, and faster spinning speed (>100 kHz) can further reduce the difficulties due to spectral resolution from macromolecular systems.^{26,31,34,61,62}