¹H Assisted ¹³C/¹⁵N Heteronuclear Correlation Spectroscopy in Oriented Sample Solid-State NMR of Single Crystal and Magnetically Aligned Samples

Abstract

¹H-irradiation under mismatched Hartmann-Hahn conditions provides an alternative mechanism for carrying out ¹⁵N/¹³C transfers in triple-resonance heteronuclear correlation spectroscopy (HETCOR) on stationary samples of single crystals and aligned samples of biopolymers, which improve the efficiency especially when the direct ¹⁵N-¹³C dipolar couplings are small. In many cases, the sensitivity is improved by taking advantage of the ¹³C_α labeled sites in peptides and proteins with ¹³C detection. The similarities between experimental and simulated spectra demonstrate the validity of the recoupling mechanism and identify the potential for applying these experiments to virus particles or membrane proteins in phospholipid bilayers; however, further development is needed in order to derive quantitative distance and angular constraints from these measurements.

Introduction

Solid-state NMR spectroscopy is playing an increasingly important role in the determination of the structures of proteins in biological supramolecular assemblies, such as membrane proteins in phospholipid bilayers[1–4], coat proteins of virus particles[5], aggregates of amyloid proteins[6], etc. Oriented Sample (OS) solid-state NMR is particularly well suited to molecular assemblies that can be mechanically or magnetically[7] aligned in the field. This allows angular constraints to be derived from the measured frequencies relative to a single external axis, which fully defines the alignment tensor, unlike the situation for residual dipolar couplings (RDCs) in solution NMR where determination of the alignment tensor can be a major source of error [8]. Significantly, since each measurement is independent of the others on the molecule, any experimental errors in the measurements of frequencies or uncertainties in the magnitudes or molecular orientations of the spin-interaction tensors do not accumulate. This results in high-resolution and accurate structure determinations.

OS solid-state NMR has been successfully applied to DNA, membrane proteins, and viral coat proteins, primarily through the use of uniform and selective ¹⁵N labeling. ¹⁵N labeling of biopolymers has many advantages[9]. Uniform ¹⁵N labeling is easy and inexpensive to implement by expressing the protein or nucleic acid of interest in bacteria grown on chemically defined media where there is only a single source of nitrogen, typically a salt of ammonia. Since no nitrogens are directly bonded in biopolymers, and in the critical polypeptide backbone of proteins each amide nitrogen is separated from another by two carbon atoms and three bonds, homonuclear ¹⁵N/¹⁵N decoupling is not necessary at any stage of the experiments because of the combination of the low gyromagnetic ratio and spatial separation of nitrogen atoms. Many double-resonance ¹H/¹⁵N experiments have been developed to measure frequencies from the three spin-interactions available at a single ¹⁵N labeled site: ¹H chemical shift, ¹⁵N chemical shift, and ¹H/¹⁵N heteronuclear dipolar couplings[10–13]. A number of assignment schemes have been developed based on both through-space interactions and the regularity of structural and spectral features that accompany the mapping of the structure onto the spectra by the anisotropic spin-interactions in the secondary structures of the α-helix and β-sheet[14; 15]. Recently, through-space methods of identifying proximate nuclei have been improved by invoking assistance from a third spin[16; 17].

In order to further advance OS solid-state NMR methods, triple-resonance experiments on ¹³C and ¹⁵N double labeled samples are an essential next step. This would enable spectroscopic interrogation of essentially all sites in a biopolymer. Since all backbone sites of a protein would be labeled it offers the possibility of systematic assignment schemes, and with ¹³C detection higher sensitivity. However, in fully labeled biomolecules there is a significant problem. The ¹³C form a dense network of homonuclear dipole-dipole coupled nuclei that interfere with most multidimensional solid-state NMR experiments as well as ¹³C detection. We have taken two approaches to ameliorating these difficulties. One is to use ‘tailored’ ¹³C labeling through judicious choice of ¹³C-containing precursors in the growth media[18]. This enables either uniform dilution of the ¹³C nuclei or high levels of labeling at selected sites where nearby carbons are unlabeled. The strong homonuclear dipole-dipole couplings are attenuated by the effect of the dilution on their spatial proximity. The second approach is to develop NMR experiments that incorporate homonuclear decoupling on both the ¹H and ¹³C channels[19]. The implementation of triple-resonance experiments provides ¹³C chemical shift, and ¹H-¹³C angular constraints that complement those from ¹⁵N especially since some of them are out of the peptide plane, and as a result add unique information to the structure calculations. Moreover, ¹³C detection offers increased sensitivity compared to the corresponding ¹⁵N detection.

Cross-polarization (CP) between abundant spins and dilute spins has been demonstrated successfully both in magic angle spinning (MAS) and stationary solid-state NMR experiments. The magnetization can be transferred from ¹H to either ¹⁵N or ¹³C easily because of the strong dipolar couplings, up to 11 kHz or 22 kHz in peptides, respectively. In the basic cross-polarization experiment introduced by Waugh and coworkers[20], the observed signals are enhanced by up to ten- or four-fold for ¹⁵N and ¹³C, respectively[21]. Here we are interested in using these large initial increases in magnetization to enable subsequent transfers to provide additional frequency dimensions for resolution, and to initiate the development of systematic assignment methods for OS solid-state NMR.

The first goal of the pulse sequence is to transfer magnetization from the initially polarized dilute spin to the second type of dilute spin. In general, this is referred to as double-cross-polarization (DCP)[22], although the details of the spectroscopy can vary significantly among the pulse sequences used to carry out this procedure. In the simplest example, the magnetization can be transferred from ¹H to ¹⁵N, and then from ¹⁵N to ¹³C; alternatively, the magnetization can be transferred from ¹H to ¹³C, and then from ¹³C to ¹⁵N. In general, the efficiency of the magnetization transfer between dilute spins (¹⁵N and ¹³C) is low due to the relatively small dipolar couplings between these two nuclei, which are generally less than 1 kHz because of their relatively low gyromagnetic ratios.

Here we demonstrate that it is possible to obtain two-dimensional ¹³C/¹⁵N HETCOR spectra for all directly bonded pairs of ¹⁵N and ¹³C_α in the backbone of a peptide or protein. Ideally, we would like all of the correlation resonances to have equal intensity, however, for now we have to settle for reliable detection of the correlation resonances for all pairs of ¹⁵N and ¹³C_α in the peptides or proteins. Since the input for the structure calculations is in the form of the orientationally-dependent frequencies, not intensities or line shapes, this is much less of a handicap than in other classes of experiments, such as spin-exchange where the intensities of the off-diagonal peaks are important.

Longer mixing times generally improve the extent of magnetization transfer, especially when the dipolar couplings are small. However, the longer mixing times can also result in non-selective magnetization transfer or spin diffusion, depending on the various laboratory and rotating relaxation times. For solid-state NMR of stationary samples, selective magnetization transfers between ¹H and ¹⁵N or ¹H and ¹³C are feasible when the homonuclear ¹H/¹H dipole-dipole couplings are strongly attenuated, allowing individual heteronuclear dipolar couplings to dominant during specified time intervals of multi-dimensional experiments. Polarization inversion spin exchange at the magic angle (PISEMA),[10] SAMMY,[11] and related pulse sequences are able to decouple the homonuclear ¹H dipolar networks and selectively transfer the magnetization between ¹H and ¹⁵N[23; 24]. Alternatively, dipolar-based INEPT[25], also is effective for transferring magnetization when the ¹H network of homonuclear couplings is suppressed. However, during many trials, we were unable to obtain efficient magnetization transfer between ¹⁵N and ¹³C using this family of pulse sequences. In contrast, spin-lock on both ¹⁵N and ¹³C with matched continuous wave irradiation does transfer magnetization between coupled ¹⁵N and ¹³C sites.

Third-spin assisted polarization transfer (TSAR) has been proposed as a method to recouple dilute spins under MAS condition[26]. However, instead of spinning the sample, the offsets can be created under the mismatched Hartmann-Hahn conditions between the abundant and dilute spins[16; 17] that effect recoupling in stationary solid-state NMR experiments, as shown in Equation1:

$$H_{\pm }=-\frac{1}{8\mathrm{\Delta }{\omega }_I}{\displaystyle \sum _{n=1}^N{a}_{1n}{a}_{2n}{I}_z^{(n)}}(S_{+}^{(1)}{S}_{-}^{(2)}+S_{-}^{(1)}{S}_{+}^{(2)})$$ (1a)

$$H_{\mathit{\text{II}}}=\frac{1}{8\mathrm{\Delta }{\omega }_I}{\displaystyle \sum _{i<j}^N(a_{1i}{a}_{1j}{S}_z^{(1)}+a_{2i}{a}_{2j}{S}_z^{(2)})}(I_{+}^{(i)}{I}_{-}^{(j)}+I_{-}^{(i)}{I}_{+}^{(j)})$$ (1b)

$$H_{\mathrm{\Delta }}=\frac{1}{16\mathrm{\Delta }{\omega }_I}{\displaystyle \sum _{n=1}^N{a}_{1n}^2(S_z^{(1)}-I_z^{(n)})+a_{2n}^2(S_z^{(2)}-}{I}_z^{(n)})$$ (1c)

Where S⁽¹⁾ and S⁽²⁾ represent the dilute spins (¹⁵N and ¹³C), and I is ¹H. Δω_I is the mismatched Hartmann-Hahn condition that makes the Hamiltonians significant when the mismatch is small. The first term, the recoupling term, recouples two dilute spins by the ¹H, and therefore, the ¹H dipolar network is used to create couplings among dilute spins. It is this alternative pathway that provides an opportunity to transfer magnetization when the dipolar couplings between dilute spins are weak. The second term, the equilibrating bath term, enable the equilibration of the spin temperature among the spins. For spin diffusion driven by this mechanism[16], ¹⁵N is polarized to achieve higher spin temperatures, and then, a z-filter is applied to eliminate any residual ¹H magnetization that could result in non-selective transfer. Hence, the spin temperature is always higher for the ¹⁵N spins, and that drives the magnetization to redistribute to the proton bath and results in decreasing the transfer efficiencies. The third term can be neglected when the lattice sum of each dilute spin is equal; otherwise, the magnetization will be brought to the orthogonal frame according to the commutator. This term could decrease the transfer efficiency in the heteronuclear correlation experiments, because it is difficult to make the lattice sums of ¹³C and ¹⁵N equal to each other. Spin diffusion experiments among ¹⁵N in OS solid-state NMR with ¹H-irradiation under mismatched Hartmann-Hahn conditions show that even if several different spin dynamics are involved, the recouping term still can assist the magnetization transfers. The Hamiltonians also suggest the feasibility of heteronuclear correlations in triple-resonance experiments on stationary samples. Here we demonstrate ¹³C-detected HETCOR experiments where the ¹⁵N/¹³C transfer is assisted by ¹H-irradiation under mismatched Hartmann-Hahn conditions. In addition, we show that the enhancement of signal intensities and the selectivity depend upon experimental conditions.

Results and Discussion

Improvement in ¹³C/¹⁵N heteronuclear correlation spectra results from ¹H-irradiation under mismatched Hartmann-Hahn conditions. This is demonstrated with two samples under stationary conditions. One sample is a single crystal of ¹⁵N, ¹³C_α N-acetyl-leucine (NAL), which has four unique molecules in its unit cell. The other sample is selectively ¹⁵N, ¹³C_α alanine-labeled Pf1 bacteriophage coat protein in magnetically aligned virus particles. There are 8 alanine residues in the protein sequence.

One-dimensional ¹³C and ¹⁵N NMR spectra obtained by conventional spin-lock cross-polarization are shown in Figure 2A and B; the four resolved resonances in each spectrum correspond to the four molecular orientations in the unit cell of the single crystal. The ¹⁵N-¹³C heteronuclear dipolar coupling doublets generated in the t1 dimension by the separated-local-field (SLF) experiment using the pulse sequence diagramed in Figure 1A are shown in Figure 2C and D. Both the chemical shift and heteronuclear dipolar coupling frequencies depend on the angles between ¹⁵N-¹³C_α bonds and the magnetic field. As a result, the ¹⁵N and ¹³C_α sites in a single peptide plane have the same ¹⁵N-¹³C dipolar coupling. By matching the ¹⁵N-¹³C dipolar couplings in the ¹⁵N- and ¹³C-detected ¹⁵N-¹³C dipolar coupling SLF spectra, all four resonances in the ¹⁵N and ¹³C one-dimensional spectra can be correlated. The ¹⁵N and ¹³C_α pairs are identified in Figure 2. The ¹⁵N-¹³C_α dipolar couplings are 200 Hz, 374 Hz, and 451 Hz for pair 2, pair 3 and pair 4, respectively. These doublets show the ¹⁵N-¹³C dipolar couplings are small, as predicted for the peptide system. The ¹⁵N-¹³C_α dipolar coupling of pair 1 is below the resolution limit of this particular experiment, and therefore we classify the heteronuclear dipolar coupling of this pair to be ~ 0 kHz.

Figure 2.

¹³C and ¹⁵N chemical shifts correlated by their ¹⁵N-¹³C heteronuclear dipolar couplings in SLF spectra of a ¹⁵N, ¹³C_α NAL single crystal. A. ¹⁵N decoupled ¹³C NMR spectrum. B. ¹³C decoupled ¹⁵N NMR spectrum. C. ¹³C-detected ¹⁵N-¹³C heteronuclear dipolar coupling SLF spectra. D. ¹⁵N-detected ¹⁵N-¹³C heteronuclear dipolar coupling SLF spectra. The observed ¹⁵N-¹³C_α heteronuclear dipolar couplings are: ~0 Hz (pair 1); 200 Hz (pair 2); 374 Hz (pair 3); and 451 Hz (pair 4).

Figure 1.

Timing diagrams for the pulse sequences. A. ¹⁵N-¹³C dipolar coupling / ¹³C chemical shift or ¹⁵N-¹³C dipolar coupling / ¹⁵N chemical shift separated-local-field (SLF) experiment. B. ¹⁵N/¹³C HETCOR experiment. C. ¹⁵N/¹³C HETCOR experiment with ¹H-irradiation. SLF experiments provide the local fields (dipolar couplings) affecting on the observed spins in the t1 dimension, and the pulse scheme in the t1 dimension can be modified for different demands (¹H-¹⁵N, ¹H-¹³C, or ¹⁵N-¹³C dipolar couplings). Therefore, the chemical shifts acquired in the t2 dimension can be resolved by these two-dimensional experiments[31; 32]. CW refers to continuous wave irradiation and SPINAL-16[33] refers to the modulation used for heteronuclear decoupling. Pre-saturation is accomplished with thirty 90° pulses separated by 200 µs delay. The z-filter is 5 ms. Mix time for ¹H/¹⁵N cross-polarization is 1ms, and for ¹⁵N/¹³C cross-polarization is 1ms or 3ms.

The conventional ¹⁵N/¹³C HETCOR experiments were performed using the pulse sequence in Fig. 1B. ¹⁵N and ¹³C are pre-saturated to ensure that the magnetization is transferred from other spins without interference from residual magnetization. The magnetization is first transferred from ¹H to ¹⁵N, and the ¹⁵N nuclei then evolve according to their chemical shift interactions by removing heteronuclear dipolar couplings. Subsequently, a 5ms z-filter is applied to remove any residual ¹H magnetization. The first 90° pulse brings the ¹⁵N magnetization to the z-axis, and is alternated between −Y and −X phase to achieve quadrature detection in the t1 dimension; the second 90° pulse brings the magnetization back, and is alternated between Y and −Y phase to suppress the effects of probe ringing. The magnetization is subsequently transferred with the continuous wave irradiation applied on the ¹⁵N and ¹³C channels simultaneously, and ¹³C chemical shifts are detected in the t2 dimension. Three resonances shown in Fig. 3A were obtained in the regular ¹⁵N/¹³C HETCOR experiment with 1ms ¹⁵N/¹³C mixing time. The missing resonance corresponds to resonance 1, which has ~ 0 kHz ¹⁵N-¹³C dipolar coupling. Notably, no magnetization transfer is observed for pair 1 even with the mixing time extended to 3 ms.

Figure 3.

A. ¹³C-detected ¹⁵N/¹³C HETCOR spectrum of a ¹⁵N, ¹³C_α NAL single crystal. B. ¹³C-detected ¹⁵N/¹³C HETCOR with ¹H-irradiation at 40 kHz. The one-dimensional spectra aligned on the top and side are the same ¹³C and ¹⁵N CP spectra shown in Figure 2A and B, respectively. Note in Figure 3B that the signal from pair 1 is detectable only in the presence of ¹H-irradiation. The experiments were performed with 1ms (shown) and 3ms ¹⁵N/¹³C mixing time with similar results.

Four resonances are present in Fig. 3B with ¹H-irradiation at 40 kHz during the ¹⁵N/¹³C mixing period (Fig. 1B). Slices through all the resonances in the t1 dimension are shown in Fig. 4. The experimental results on the single crystal are fully consistent with the described model: ¹H-irradiation provides another pathway to transfer the magnetization from ¹⁵N to ¹³C_α via the proton bath. The weaker intensities of the pair 2 to pair 4 imply that the other two terms (Equation 1b and c) are involved as well and result in loss of signal intensity. The benefit of ¹H-irradiation is shown on pair 1. Although, there are other processes competing with recoupling, the experimental results show that recoupling can drive the magnetization transfer without direct dipolar couplings. Non-selective transfer from the proton bath when ¹H-irradiation is applied can be ruled out by inspecting the slices in Fig. 4, since the sharp peaks and flat baselines show only that a specific frequency is transferred.

Figure 4.

One-dimensional spectral slices along the ¹⁵N chemical shift dimension taken from the two-dimensional spectra shown in Figure 3. A. ¹⁵N/¹³C HETCOR. B. ¹⁵N/¹³C HETCOR with ¹H-irradiation at 40 kHz. From top to bottom, the slices correspond to pairs 1 to 4, respectively. In Figure 4A, the measured signal-to-noise ratios are 36 (pair 2), 83 (pair 3) and 111 (pairs 4). In Figure 4B, the measured signal-to-noise ratios are 17 (pair 1), 18 (pair 2), 18 (pair 3) and 74 (pair 4). Reductions in some signal intensities result from competing pathways. The narrow single-line peaks and the flat baselines indicate that the transfer from ¹⁵N to ¹³C is selective.

Recoupling has been shown to assist the ¹⁵N/¹³C transfers when the mismatches are lower than the Hartmann-Hahn condition by 10 kHz. According to the Hamiltonians, the same effect should occur when the mismatch is higher as well. The stronger ¹H-irradiation brings the ¹H magnetization into the rotating frame more efficiently, and should provide superior recoupling; once the proton bath recouples the dilute spins, all the correlations between ¹⁵N and ¹³C should be observed in the HETCOR spectra. However, we only observed the intra-molecular correlations in the ¹⁵N, ¹³C_α NAL single crystal sample in the presence of ¹H-irradiation at 40 kHz. When the field strength of ¹H-irradiation was increased to 60 kHz not only the intra- but also some inter-correlations were observed in the spectra (Fig. 5A). Comparing the spectrum in Fig. 3B, the only difference is the strength of ¹H-irradiation, and it shows that the recoupling efficiency is improved by increasing the strength of the B₁ field. All the correlations expected in the unit cell were observed when mixing time was extended to 3ms (Fig. 5B).

Figure 5.

A. and B. ¹⁵N/¹³C HETCOR spectra obtained with ¹H-irradiation of 60 kHz with 1ms and 3ms ¹⁵N/¹³C mixing times, respectively, of a ¹⁵N, ¹³C_α NAL single crystal. One-dimensional spectra on the top and the side are the ¹³C and ¹⁵N CP spectra shown in Figure 2 A and B. Strong inter-molecular correlations in a single crystal are achieved only under the higher mismatched condition.

Simulations of the magnetization transfer in a ¹⁵N, ¹³C_α NAL peptide were carried out using SIMPSON 3.0.1[27]. To simplify the spin system, only ¹H nuclei attached to the ¹⁵N, C_α, C_β, and N-methyl groups were considered, and only the ¹⁵N/¹³C mixing period was simulated to evaluate the transfer efficiency. In the simulations, a 1kHz ¹⁵N-¹³C dipolar coupling was utilized. The regular ¹⁵N/¹³C transfer (Fig. 6A) is sensitive to ¹⁵N-¹³C dipolar couplings because they provide the only transfer pathway. The typical CP buildup curve[28] is observed in the simulations when the ¹⁵N-¹³C dipolar coupling is strong; the intensities increase proportionally to the magnitudes of the dipolar couplings, and the dipolar modulation also causes the oscillations in the magnetization buildup curve. The transfer efficiencies are poor when the ¹⁵N-¹³C dipolar couplings are small. However, in the presence of ¹H-irradiation (Fig. 6B), no matter the magnitude of the dipolar couplings, they tend to have similar transfer efficiencies. Both higher and lower mismatched conditions (59 kHz and 43 kHz according to the mismatched profile in Fig. 6D) were simulated. The higher field provides better transfer efficiencies, which is consistent with the experimental results shown in Fig. 3B and Fig. 5. This demonstrates that the transfer mechanism is dominated by recoupling with the proton bath. The data in Fig. 6C show that transfers for weak ¹⁵N-¹³C dipolar couplings benefit from ¹H-irradiation, but that it can decrease the efficiencies for strong ¹⁵N-¹³C dipolar couplings. The mismatched profile (Fig. 6D) was simulated under the small dipolar coupling condition. Two local maxima show up asymmetrically when mismatched conditions are about ±10 kHz, which is similar to the simulations of ¹H-irradiation assisted ¹⁵N spin diffusion, where the optimal mismatched conditions occur at ± 8%[16]. Both the experimental results and simulations suggest that the optimal conditions for ¹H-irradiation lie between ±5 kHz and ±15 kHz, mainly depending on the properties of the proton dipolar network. Better transfer efficiency can be observed at the higher mismatch condition because of the larger ¹H magnetization in the rotating frame. In practice, the higher mismatched condition drives the inter-peptide correlations that distribute the magnetizations to the proton bath and more distant dilute spins. As a result, the transfer efficiencies would be decreased as shown in Fig. 5.

Figure 6.

Simulations of the transfer efficiencies between ¹⁵N and ¹³C with various ¹⁵N-¹³C heteronuclear dipolar couplings. A. Conventional ¹⁵N/¹³C CP transfer. B. ¹⁵N/¹³C CP transfer with ¹H-irradiation at 59 kHz (solid line) and 43 kHz (dashed line). C. Logarithmic plot of enhancement by ¹H-irradiation. D. Mismatched profile of ¹H-irradiation. A. - C. the ¹⁵N-¹³C_α dipolar couplings were simulated for 0 (red), 100 (blue), 300 (green), 1000 (black) Hz with all other parameters kept the same. The buildup curves were sampled every 500 ms, and the lines connecting the data points show the trends. D. The simulation was done when ¹⁵N-¹³C_α dipolar coupling equal to zero, and the strength of ¹H-irradiation was varied from 0 kHz to 100 kHz.

¹⁵N, ¹³C_α alanine-labeled Pf1 bacteriophage was used to demonstrate the feasibility for the applications to aligned protein samples. The experiments were performed at a relatively low temperature, to ensure that a single conformation of the protein was present. [29]. All the resonances were assigned by comparisons to previous results. Signals from seven out of the eight alanine residues were observed with the regular HETCOR experiment (Fig. 7A). A7, A11, and A21 are strong and well resolved. A29 is merged with A21, and A34 and A36 are relatively weak comparing to the other peaks. The signal from A46 is not observed. With ¹H-irradiation at a lower mismatched condition of 33.8 kHz, two changes are apparent in the spectrum in Fig. 7B: the signal-to-noise ratio from A34 is increased by 50%, and the signal from A46 can be observed. For the higher mismatched condition at 67.7 kHz, A34 still gains 50% enhancement, while A21 is significantly reduced, which distinguishes between A21 and A29 in Fig. 7C. Other peaks were reduced in intensity in the presence of ¹H-irradiation, which serves to reinforce the agreement between the single crystal spectra and simulations. The superimposed spectrum shown in Fig. 7D contains all resonances that are expected from ¹⁵N, ¹³C_α alanine-labeled Pf1 bacteriophage. Thus be performing multiple experiments with varying strengths of ¹H-irradiation, it is possible to detect all ¹⁵N/¹³C correlation peaks from a protein, despite their widely varying magnitudes of heteronuclear dipolar couplings.

Figure 7.

Two-dimensional ¹⁵N/¹³C HETCOR spectra of selectively ¹⁵N, ¹³C_α alanine-labeled Pf1 bacteriophage that is magnetically aligned in the field. The protein has eight alanine residues. A. Two-dimensional ¹⁵N/¹³C HETCOR without ¹H-irradiation. B. Two-dimensional ¹⁵N/¹³C HETCOR with ¹H-irradiation at 33.8 kHz. C. Two-dimensional ¹⁵N/¹³C HETCOR with ¹H-irradiation at 67.7 kHz. The slices shown in panels A., B., and C. were taken at 53.7 ppm in the ¹³C chemical shift dimension, corresponding to the resonance from A29. D. This spectrum is a superposition of the spectra in panels A.(red), B.(blue), and C.(green); it displays all of the ¹⁵N/¹³C correlations in the selectively ¹⁵N- alanine labeled Pf1 bacteriophage, and demonstrates the possibility of observing all ¹⁵N/¹³C correlations regardless of the strength of their heteronuclear dipolar couplings by obtaining spectra under several different transfer conditions.

Conclusions

¹³C-detected triple-resonance experiments can detect all ¹⁵N/¹³C correlation resonances with high sensitivity, as long as they are performed with ¹³C-detection under several different conditions of ¹H radiofrequency irradiations. The frequencies of the ¹⁵N/¹³C correlation resonances provide valuable constraints for the calculation of protein structures in OS solid-state NMR. Moreover, ¹⁵N/¹³C transfers can be used as a filter to eliminate interference from natural abundance signals from the lipids [18]. ¹H-irradiation under mismatched Hartmann-Hahn conditions assists magnetization transfer from ¹⁵N to ¹³C. The experimental results obtained on single crystal and aligned protein samples, and simulations demonstrate the validity of this recoupling mechanism. The transfer mechanism relies primarily on the ¹H dipolar network, which is well suited for OS solid-state NMR experiments. The spin dynamics are complex once the spins are recoupled by ¹H-irradiation, and therefore, some of the transfer efficiencies may be lost, especially when the ¹⁵N-¹³C dipolar couplings are large. However, this method is complementary to other procedures especially because it provides good transfer efficiencies for small ¹⁵N-¹³C dipolar couplings. Both mismatched conditions are used in order to enhance the intensities of signals with different dipolar couplings. Combined with conventional spin-lock CP transfer, these experiments have the potential to simultaneous measure all of the ¹³C and ¹⁵N chemical shift frequencies and correlate proximate residues in a protein, which provides the basis for systematic assignment schemes.

Methods

¹⁵N, ¹³C_α NAL single crystal spectra

The NMR experiments were performed on a Varian Inova spectrometer with ¹H, ¹³C, and ¹⁵N frequencies of 500.125 MHz, 125.76 MHz, and 50.68 MHz, respectively. A home-built modified Alderman Grant coil (MAGC) triple-resonance probe[19], which was designed for lossy biological samples, was used for these experiments. All the spectra were obtained at room temperature. The ¹H carrier frequency was set at 4.7 ppm; the ¹³C carrier frequency was set at 54.94 ppm; and the ¹⁵N carrier frequency was set at 79.26 ppm. ¹³C- and ¹⁵N-detected spectra were acquired with 512 and 1024 complex points in the direct dimensions, respectively. CP spectra were acquired with 64 scans and 40 µs dwell time. ¹⁵N-¹³C dipolar coupling SLF spectra were acquired with 4 scans, 100 µs dwell time and 96 points in the indirect dimension, and 40 µs dwell time in the direct dimension. HETCOR spectra were acquired with 4 scans, 100 µs dwell time and 192 complex points in the indirect dimension, and 40 µs dwell time and 512 complex points in the direct dimension. The mixing time for ¹H/¹⁵N and ¹H/¹³C CP was 1ms, and ¹⁵N/¹³C CP was 1ms or 3ms. For all the experiments the recycle delay was 6 s. Pre-saturation consisted of thirty 90° pulses separated by 200 µs, and the z-filter delay was 5 ms. The strength of the continuous wave decoupling on ¹⁵N and ¹³C channels was 40 kHz, and all other irradiations were performed at 50 kHz, except for the variable ¹H-irradiation during the ¹⁵N/¹³C mixing period. The experimental data were zero filled to 512 and 1024 data points in the indirect and direct dimensions, respectively, and were multiplied by a sine bell window function prior to Fourier transformation.

¹⁵N, ¹³C_α alanine-labeled Pf1 bacteriophage spectra

The Pf1 filamentous bacteriophage sample was prepared as described previously[30]. The final bacteriophage sample was diluted to 40 mg/ml with sodium borate buffer (pH=8), and the final volume was 150 µl. The spectra were obtained at −2°C to ensure that the coat protein was in its ‘low temperature’ conformation. The ¹H carrier frequency was set at 4.7 ppm; the ¹³C carrier frequency was set at 100 ppm, and the ¹⁵N carrier frequency was set at 213 ppm. HETCOR spectra were acquired with 400 scans, 100 µs dwell time, 32 complex points in the indirect dimension, and 25 µs dwell time and 256 complex points in the direct dimension. Mixing time of ¹H/¹⁵N CP was 1ms and ¹⁵N/¹³C CP was 3ms. The recycle delay was 4 s. All other conditions were the same as described above for the single crystal.

Simulations

Only the ¹⁵N/¹³C mixing period was simulated to evaluate the transfer efficiency. The simulations were carried out with SIMPSON 3.0.1. The model was based on the single crystal structure[16]; however, the spin system was simplified to ¹⁵N, ¹³C_α, the ¹H attached to the N, C_α, C_β, and N-methyl group. Euler angles “β” in the input file were manually adjusted to 90° for ¹H-¹⁵N, ¹H-¹³C_α and ¹⁵N-¹³C_α dipolar couplings, and therefore, the dipolar couplings can be varied without angular considerations. For all the simulations, ¹H-¹⁵N and ¹H-¹³C_α dipolar couplings were 12.18 and 22.80 kHz according to the bond lengths. ¹⁵N-¹³C_α dipolar couplings were adjusted manually according to each case. ¹⁵N/¹³C mixing period was 3 ms for the mismatched profile simulation.

Highlights.

• Heteronuclear dipolar couplings differ by two in perpendicular and parallel bilayers

• Dipolar couplings correlate resonances in perpendicular and parallel bilayers

• Assignments result from the dipolar correlation and an assigned isotropic spectrum