Simultaneous Cross Polarization to ¹³C and ¹⁵N with ¹H Detection at 60 kHz MAS Solid-state NMR

Abstract

We describe high resolution MAS solid-state NMR experiments that utilize ¹H detection with 60 kHz magic angle spinning; simultaneous cross-polarization from ¹H to ¹⁵N and ¹³C nuclei; bidirectional cross-polarization between ¹³C and ¹⁵N nuclei; detection of both amide nitrogen and aliphatic carbon ¹H; and measurement of both ¹³C and ¹⁵N chemical shifts through multi-dimensional correlation experiments. Three-dimensional experiments correlate amide ¹H and alpha ¹H selectively with ¹³C or ¹⁵N nuclei in a polypeptide chain. Two separate three-dimensional spectra correlating ¹Hα/¹³Cα/¹H^N and ¹H^N/¹⁵N/¹Hα are recorded simultaneously in a single experiment, demonstrating that a two-fold savings in experimental time is potentially achievable. Spectral editing using bidirectional coherence transfer pathways enables simultaneous magnetization transfers between ¹⁵N, ¹³Cα⁽ⁱ⁾ and ¹³C′⁽ⁱ⁻¹⁾, facilitating intra- and inter- residue correlations for sequential resonance assignment. Non-uniform sampling is integrated into the experiments, further reducing the length of experimental time.

Introduction

High-resolution magic angle spinning (MAS) solid-state NMR spectroscopy is in the process of becoming a powerful tool for studying crystalline peptides [1] and proteins [2] as well as proteins immobilized in biological supramolecular assemblies such as membranes [3], fibrils [4], and virus particles [5]. Recent advances include high speed spinning, ¹H-detection [6], and non-uniform sampling (NUS) [7–9]. Rapid spinning at the magic angle attenuates the severe line-broadening resulting from the dense ¹H/¹H homonuclear dipole-dipole coupling network present in peptides and proteins. Due to its high gyromagnetic ratio and natural abundance, ¹H-detection has been the optimal choice for solution NMR for quite some time, and now with the ability to narrow ¹H resonances with fast MAS, it is finding increasing applications in solid-state NMR. Moreover, NUS can be used to reduce the amount of time required for the experiments.

In addition to detecting and resolving individual resonances, assignment to specific sites is an essential aspect of protein NMR spectroscopy. A number of ¹H-detection experiments in solid state have been reported that yield sequence-specific resonance assignments [10, 11] similar to the solution NMR experiment [12]. Current assignment procedures utilize double- and triple- resonance experiments to correlate ¹³C and ¹⁵N chemical shift frequencies with those of amide hydrogens (¹H^N). In contrast to solution NMR experiments, where backbone and side chain ¹H, ¹³C and ¹⁵N resonances are assigned using established two- and three- dimensional triple-resonance experiments, in order to measure the chemical shift frequencies of non-amide hydrogens in solid state NMR, four-dimensional experiments or multiple three-dimensional experiments are typically required. For example a four-dimensional experiment correlating ¹HXX¹H, where X stands for ¹³C or ¹⁵N, has been shown to fully resolve the spectrum of a protein [7]. However, four-dimensional experiments can be extremely time consuming. It is possible for two separate three-dimensional experiments with either ¹³C or ¹⁵N editing to replace four-dimensional experiments, however, this approach can also require long measurement times.

The long times required for the higher dimensional NMR experiments can be addressed by incorporating simultaneous cross-polarization (SIM-CP) from ¹H to ¹³C and ¹⁵N sites. Multiple two- and three- dimensional spectra can be obtained from a single experiment in this way [13–16]. These experiments were originally developed for ¹³C-detection at moderate spinning frequencies. Notably, ¹³C/¹³C homonuclear correlation under proton assisted recoupling (PAR) and ¹³C/¹⁵N heteronuclear correlation under proton-assisted insensitive nuclei (PAIN) can be performed using simultaneous spin-lock pulses on three channels under second-order recoupling schemes. In addition, recently, it has been shown that bidirectional cross-polarization between ¹³C and ¹⁵N nuclei can be used to transfer coherence without significant losses in signal intensities [15, 17]. This has enabled triple-resonance experiments to be executed with multiple acquisitions using a single receiver within a single experiment. Further savings of experimental time can be achieved with application of NUS [7, 18–21].

Here we demonstrate ¹H-detection experiments that utilize SIM-CP and bidirectional coherence transfer to obtain two separate two- or three- dimensional spectra in a single experiment. These experiments utilize cross-polarization from ¹H to both ¹³C and ¹⁵N sites without a substantial loss in sensitivity compared to individual ¹H/¹³C or ¹H/¹⁵N cross-polarization. Experiments are optimized for bidirectional coherence transfer at 60 kHz MAS in order to transfer magnetization simultaneously and selectively from the amide ¹⁵N to the ¹³Cα of the same residue and the ¹³C′ of the preceding residue. The experiments also enable sequential homo- and hetero- nuclear correlations between amide ¹H and alpha ¹H sites to ¹³C and ¹⁵N nuclei in order to improve resolution and to facilitate resonance assignment.

Experimental

Sample preparation

Uniformly ¹³C and ¹⁵N labeled fd bacteriophage with its major coat protein containing the Y21M mutation was obtained using our previously published protocol [5]. Bacteriophage particles from a single batch were concentrated to ~100 mg/ml by ultracentrifugation at 60,000 rpm for 2 hours. The pH of the sample was adjusted to 8.0 using 5 mM sodium borate buffer containing 0.1 mM sodium azide prior to ultracentrifugation. ~30 μL of fully hydrated phage particles at pH 8.0 were transferred to a 1.3 mm zirconium rotor. The sample in the rotor contained ~1 mg of protein.

Uniformly ¹³C and ¹⁵N labeled Met-Leu-Phe tripeptide was purchased from Cortecnet (www.cortecnet.com). ~3.5 mg of microcrystalline peptide was packed in a 1.3 mm MAS rotor.

NMR Spectroscopy

The experiments were performed on a spectrometer with a ¹H resonance frequency of 900 MHz. The spectrometer was equipped with a Bruker Avance III HD console and a Bruker 1.3 mm ¹H/¹³C/¹⁵N triple-resonance MAS probe (www.bruker.com). The spinning rate was controlled at 60.0 kHz. The temperature of the nitrogen gas was maintained at −18° C throughout the experiments. The ¹H resonance frequency of water was used to monitor the temperature of the protein-containing sample, and served as an internal chemical shift reference frequency at 4.8 ppm at 20 °C. The chemical shift frequencies of the polycrystalline sample were referenced externally to solid samples with the methylene ¹³C resonance of adamantane at 38.48 ppm and the ¹⁵N resonance of ammonium sulfate at 26.8 ppm [22].

The experimental data were acquired using the pulse sequences diagrammed in Figure 1. Heteronuclear decoupling was accomplished using XiX for ¹H [23] and GARP [24] for ¹⁵N and ¹³C nuclei. 30% ramped radio frequency (RF) irradiation on the ¹H channel was used for initial cross-polarization (CP) from ¹H to ¹³C, ¹H to ¹⁵N, and back to ¹H during reverse CP. Spin-exchange between ¹⁵N and ¹³C was accomplished using spectrally induced filtering in combination with cross-polarization (SPECIFIC-CP) [25] with 10% ramped amplitude irradiation on the ¹³C channel. All RF irradiations on the ¹H, ¹³C and ¹⁵N channels were matched to the (n-1) Hartmann-Hahn resonance condition for optimal magnetization transfer. The amplitude of the ramped ¹H irradiation ranged from 120 kHz to 60 kHz. The ¹³C and ¹⁵N RF amplitudes were adjusted to be near 30 kHz to meet the Hartmann-Hahn condition. Mixing intervals of 1 ms and 0.4 ms were used for initial and final steps in all experiments. A 6 ms mixing time was used to exchange magnetization between ¹⁵N and ¹³C. Phase alternated RF pulses were used to suppress the water signals [26], with 15 kHz RF irradiation for 300 ms. For the fully hydrated biological sample water suppression was improved by increasing the amplitude of the ¹H irradiation to ~40 kHz and implementing supercycled phase alternated pulses. All spectra were processed with Bruker TOPSPIN and represented with SPARKY (University of California, San Francisco) programs.

Figure 1.

Three channel pulse sequence diagrams for ¹H detection in solid state NMR with 60 kHz magic angle spinning. A. and B. Three-dimensional experiments for simultaneous ¹³C and ¹⁵N edited heteronuclear correlation. A. Experiment for inter and intra residue ¹³C/¹⁵N/¹H^N three-dimensional correlation. B. Experiment for three-dimensional correlation spectroscopy for simultaneous observations of ¹Hn/N(Cα)/¹Hα and ¹Hα/¹³Cα (N)/¹H^N resonances. Phase cycling for A ; ¹³C CP (0 0 0 0 2 2 2 2), ¹³C and ¹⁵N DCP (0), ¹⁵N reverse CP (0 0 2 2), ¹H reverse CP (0 0 0 0 1 1 1 1 2 2 2 2 3 3 3 3), receiver (1 3 3 1 3 1 1 3 2 0 0 2 0 2 2 0). Phase cycling for B; ¹³C and ¹⁵N CP (0), ¹³C DCP (0 0 0 0), ¹⁵N DCP (0 0 2 2), ¹H, ¹³C and ¹⁵N reverse CP (0), receiver (1 3 3 1).

Results

The pulse sequences utilized in this study are diagrammed in Figure 1. All of the experiments were performed with 60 kHz MAS. The sequence for a three-dimensional ¹H detection experiment is shown in Figure 1A. One and two-dimensional experiments with ¹H detection and simultaneous correlation to ¹³C and ¹⁵N nuclei were adapted from previous publications using simultaneous cross polarization (CP) [14, 15, 17] and concurrent cross polarization pulse schemes [27]. ¹H magnetization is transferred to ¹³C and ¹⁵N simultaneously using Hartmann-Hahn CP. Following CP and chemical shift evolution with ¹H decoupling, the ¹³C and ¹⁵N magnetizations are stored as Zeeman order using 90° pulses. Water signals and other unwanted ¹H signals are suppressed using RF irradiation applied in the XY plane [26]. Then the ¹³C and ¹⁵N magnetizations are flipped back to the transverse direction using 90° pulses. By applying matched RF irradiation on the ¹H, ¹³C, and ¹⁵N channels, polarization is transferred back to the ¹H nuclei, and subsequently detected under ¹³C and ¹⁵N heteronuclear decoupling.

For heteronuclear correlation, ¹⁵N and ¹³C chemical shift frequencies evolve in the indirect (t₁) dimension under ¹H irradiation for decoupling. The chemical shift frequencies are recorded in the phase sensitive mode by incrementing the RF phase by 90° during the initial CP. In one experiment, both ¹³C and ¹⁵N signals are acquired with the same phase. In a second experiment, out of phase signals are acquired by alternating the phase of either the ¹³C or ¹⁵N irradiation by 180°. We have also observed relatively inverted signals (i.e., amide ¹H and alpha ¹H) by altering the ¹³C and ¹⁵N RF phases during the reverse CP transfer to the ¹H nuclei.

Figure 1A illustrates a pulse sequence for a ¹³C to amide ¹H and ¹⁵N correlation experiment. The experiment transfers magnetization from ¹H to ¹³C followed by chemical shift evolution in the indirect (t₂) dimension under ¹H decoupling. The magnetization from ¹³Cα and ¹³C′ are transferred selectively to ¹⁵N using Hartmann-Hahn CP followed by ¹⁵N chemical shift evolution in the indirect (t₁) dimension under ¹Hdecoupling. Water resonance suppression pulses are applied before the reverse CP from ¹⁵N for ¹H-detection.

Three-dimensional correlation experiments with simultaneous ¹³C and ¹⁵N editing were carried out using the pulse scheme shown in Figure 1B. The pulse sequence correlates alpha ¹H and amide ¹H with either ¹³Cα or ¹⁵N amide nuclei in a peptide plane. The pulse sequence is similar to that in Figure 1A, except that a spectral editing double cross-polarization between ¹³C and ¹⁵N nuclei is included between the initial and final CP. Both ¹³C and ¹⁵N chemical shift evolutions occur after initial CP. The signals are encoded in the indirect dimensions, and a correlation spectrum of ¹H^N/¹⁵N(¹³Cα)/¹Hα and ¹Hα/¹³Cα (¹⁵N)/¹H^N results.

The spectra shown in Figures 2 and 3 were obtained from a polycrystalline sample of uniformly ¹³C and ¹⁵N labeled Met-Leu-Phe (MLF) peptide utilizing the pulse sequences diagrammed in Figure 1. One-dimensional ¹H NMR spectra are shown in Figure 2A–C. Simultaneous observation of amide and aliphatic ¹H signals in Figure 2A resulted from using both ¹³C and ¹⁵N editing. The ¹⁵N edited amide ¹H NMR spectrum is shown in Figure 2B. Similarly the ¹³C edited aliphatic ¹H NMR spectrum in Figure 2C. Notably, all of the magnetization was recovered with simultaneous cross-polarization. Following the one-dimensional experiments, heteronuclear correlation of ¹H with ¹³C and ¹⁵N nuclei was obtained by evolving both nuclei simultaneously in the indirect dimension. This is demonstrated with the two-dimensional spectrum shown in Figure 2D, which contains cross-peaks for both ¹³C and ¹⁵N nuclei. The spectrum was obtained with the ¹³C and ¹⁵N magnetization locked with the same phase during Hartmann-Hahn CP. In order to distinguish the chemical shifts by nucleus, two separate data sets were collected. The first experiment was performed using the spin-lock pulses with the same phase. The second data set was acquired with 180° phase shifted ¹⁵N spin lock pulses. This resulted in a spectrum where the ¹³C and ¹⁵N cross-peaks are opposite in phase. By adding and subtracting the two data sets, two separate ¹H/¹⁵N and ¹H/¹³C heteronuclear correlation spectra are obtained, as shown in Figure 2E and F, respectively. The spectral width/dwell time was kept constant for chemical shift evolution in the indirect dimensions for both ¹³C and ¹⁵N nuclei. However, sensitivity was found to be independent of the dwell times used for each nucleus. A three-dimensional spectrum obtained using different dwell times is shown in Figure 2G. The ¹³C and ¹⁵N spin-lock pulses differed by 180° in phase and the ¹⁵N dwell time was set to twice that used for the detection of ¹³C signals. The cube shows a total of six cross-peaks correlating amide ¹H, with alpha ¹H, and with ¹⁵N amide or ¹³Cα chemical shift frequencies. The color codes indicate the signals acquired with opposite phases. The blue and green colors indicate ¹H^N/¹⁵N/¹Hα and ¹Hα/¹³Cα/¹H^N chemical shift correlation and were generated using ¹H^N (t₂) > ¹⁵N (t₁) > ¹³Cα> ¹Hα(t₃) and ¹Hα(t₂) > ¹³Cα(t₁) > N > ¹H^N (t₃) coherence transfer pathways, respectively. Two separate three-dimensional experiments were performed in a similar fashion to that used for the two-dimensional experiments described above. By addition and subtraction of the three-dimensional spectra, separate chemical shift correlation spectra are obtained. The 180° phase shifted spectra are obtained by changing the initial CP irradiation phases followed by chemical shift evolution in the indirect dimension. It is also possible to obtain the phase shifted ¹H spectrum with direct detection by changing the phases of the final CP irradiations by 180°.

Figure 2.

One-, two-, and three- dimensional spectra of uniformly ¹³C and ¹⁵N labeled polycrystalline Met-Leu-Phe tripeptide obtained with 60 kHz MAS. A.-C. One-dimensional ¹H spectra obtained using simultaneous CP. Four scans were co-added for each spectrum. A. ¹³C and ¹⁵N-edited ¹H spectrum. B. ¹⁵N -edited spectrum. C. ¹³C edited spectrum. For visual comparison, the one dimensional spectrum from A is plotted in grey in the background. D and E, two-dimensional heteronuclear correlation spectra obtained for a total number of 4 scans. G. Three-dimensional correlation spectrum obtained using the pulse scheme shown in Figure 1C. The color codes are blue for positive and green for negative cross-peaks for ¹Hα/¹³Cα (N)/¹H^N and ¹Hn/N(Cα)/¹Hα correlations, respectively. The three-dimensional data was acquired with a total number of 16 scans; 4 second recycle delay; and evolution periods of 11 ms (t3) ¹H, 1.5 ms (t2) ¹³C, and 2.5 ms (t1) ¹⁵N. An equal number of data points were linear predicted during data processing.

Figure 3.

¹³C- and ¹H- detected spectra of MLF acquired with 60 kHz MAS. A. Variable amplitude ¹³C spectra of SPECIFIC-CP between ¹⁵N and ¹³Cα at fixed 25 kHz ¹⁵N amplitude. ¹³C and ¹⁵N are irradiated at 55 ppm and 120 ppm respectively. B. Same as A, but ¹³C irradiated at 115 ppm and ¹⁵N amplitude at 15 kHz (2.8 watts). Envelopes on top are drawn to visually guide the magnetization transfer to ¹³Cα and ¹³C′. C. Optimal 1D spectra obtained from A. D. Same as C, but for ¹³C amplitude at 38 watts. E. ¹⁵N-edited and ¹H-detected ¹H^N/¹³C two-dimensional correlation spectrum. F. ¹H-detected ¹³C/¹⁵N/¹H^N three-dimensional correlation spectrum. Both A. and B. were obtained with 16 scans, and 4 second recycle delays. The three-dimensional spectrum was reconstructed from 10% sampling density of a 400 (t2)/64 (t₁) complex data matrix. The sampling schedule was generated in TOPSPIN 3.2 by incorporating T₂ and J coupling values for ¹³C nuclei. The spectrum was reconstructed using the compressed sensing package in TOPSPIN.

¹³C-detected experiments optimized to quantify the magnetization transfer between ¹³C and ¹⁵N at 60 kHz MAS are illustrated in Figure 3. The experiment was carried out using a tailored double CP (DCP) pulse sequence where ¹H magnetization is transferred to ¹⁵N and subsequently transferred to ¹³C for detection. During the DCP, no ¹H RF irradiation is applied, and the¹³C spectrum is irradiated at 115 ppm, 55 ppm, or 175 ppm. For specific transfer of ¹⁵N magnetization to either ¹³Cα or ¹³C′, we optimized the experimental conditions in a similar manner as described in an earlier publication [10]. We fixed the ¹⁵N and ¹³C RF amplitudes at 25 kHz and 35 kHz and performed the DCP with the ¹³C resonance frequency at 55 ppm for ¹³Cα and 175 ppm for ¹³C′. The RF amplitude applied to ¹³C was varied to find the maximum sensitivity. Magnetization transfer for ¹³Cα only is shown in Figure 3A. For simultaneous magnetization transfer to ¹³Cα and ¹³C′, we optimized the experiment with fixed RF amplitudes of 15 kHz for ¹⁵N and 75 kHz for ¹³C irradiated at 115 ppm. This is demonstrated in Figure 3B.

One-dimensional results for the MLF tripeptide powder sample, demonstrating the optimal polarization transfer to either carbon or simultaneous selective transfer of magnetization to ¹³Cα and ¹³C′, are shown in Figure 3C and D, respectively. For comparison, a spectrum obtained by ¹H to ¹³C direct CP (with a 1 ms mix time) was also obtained (data not shown). All experiments were carried out under similar conditions and the data were processed in the same manner. ¹⁵N-edited two-dimensional ¹³C/¹H^N heteronuclear correlation is shown in Figure 3E. The experiment was carried out using the pulse sequence shown in Figure 1A, except that no ¹⁵N chemical shift evolution was incorporated. The peaks are marked for resonances assigned to ¹³Cα and ¹³C′ chemical shift frequencies of the three amino acid residues.

A three-dimensional experiment correlating the ¹³C/¹⁵N/¹H^N chemical shift frequencies is shown in Figure 3F. The spectrum was obtained using the pulse sequence in Figure 1A and incorporating a 10% non-uniform sampling schedule generated in TOPSPIN and processed using compressed sensing provided in the same software package. The spectrum was reconstructed for a total evolution period of 11 ms ¹H (t₃), 6 ms ¹³C (t₂) and 12 ms ¹⁵N (t₁).

The experiments were also applied to a protein sample to demonstrate the breadth of their utility. For this purpose, we used the structural form of the coat protein in fd bacteriophage. NMR signals were obtained from a concentrated solution of uniformly ¹³C and ¹⁵N labeled virus particles. The virus particles are ~90% by weight coat protein subunits, thus the individual signals come from the symmetrically arranged coat proteins. Like wild type fd bacteriophage proteins [28], the mutant Y21M fd coat protein used in these experiments is a 50 amino acid long polypeptide chain with an alpha helical conformation [5]. At 60 kHz MAS, simultaneous CP and ¹H detection experiments showed essentially complete polarization transfer similar to that observed for the polycrystalline tripeptide sample. However, due to spectral overlap, only ~30 individual cross-peaks could be identified in the two-dimensional ¹H/¹⁵N heteronuclear single quantum correlation spectrum shown in Figure 4A. Similarly only ~20 cross-peaks could be resolved in the ¹Hα/¹³Cαcorrelation spectrum in Figure 4D. Both spectra were acquired using simultaneous-CP with the pulse sequence shown in Figure 1B. Nearly all protein signals are resolved in the three-dimensional spectra. Two separate three-dimensional correlation spectra such as ¹H^N/¹⁵N/¹Hα and ¹Hα/¹³Cα/¹H^N were obtained simultaneously using the pulse sequence diagrammed in Figure 1C. Two-dimensional slices correlating ¹H/¹³C and ¹H/¹⁵N shifts at various ¹H chemical shifts are shown in Figure 4. ¹⁵N/¹H^N chemical shift correlation in Figure 4B and C were extracted at 4.2 ppm and 3.8 ppm ¹H chemical shift frequencies, respectively. Similarly ¹Hα/¹³Cα two-dimensional planes were obtained at 8.5 ppm and 7.6 ppm ¹H^N chemical shift frequencies. Simultaneous observation of NCO and NCA correlation with ¹H^N detection was carried out using the pulse sequence depicted in Figure 1B. The experiment was carried out in a similar manner to that used for MLF in Figure 3. The two-dimensional plane correlating ¹H^N with ¹³C shifts in Figure 4G was obtained from the three-dimensional data at 109.7ppm ¹⁵N chemical shift.

Figure 4.

Two- and three-dimensional spectra of uniformly ¹³C and ¹⁵N labeled fd coat protein in bacteriophage particles obtained with 60 kHz MAS. A. Two-dimensional ¹H/¹⁵N heteronuclear correlation spectrum. B. and C. ¹H/¹⁵N two-dimensional slices obtained at 4.2 and 3.8 ppm ¹H chemical shifts, respectively. D. Two-dimensional ¹H/¹³C heteronuclear correlation spectrum. E. and F. Two-dimensional slices obtained at 8.5 and 7.6 ppm ¹H chemical shifts. A. and D. were obtained using the pulse sequence shown in Figure 1B without incorporating DCP editing pulses. Three-dimensional experiments were carried out using the pulse scheme shown in Figure 1B with a total number of 16 scans, 4 s recycle delay; total evolution periods of 10 ms (t₃) ¹H; 3 ms (t₂) ¹H; 7.2 ms (t₁) ¹⁵N, and 3.6 ms (t_1′) ¹³C. Equal numbers of data points were linear predicted during data processing.

Discussion

60 kHz MAS enables pulse sequences for simultaneous observation of ¹³C and ¹⁵N signals with ¹H detection to be performed. Several pulse sequences were developed with protein structure determination in mind. In particular, in order to measure ¹H chemical shifts in a sequential manner, multiple coherence pathways are used in combination.

Simultaneous cross-polarization transfer enables measurement of the chemical shifts of hydrogens bonded to amide nitrogens and aliphatic carbons. Three resolved amide nitrogen ¹H and three alpha carbon ¹H resonances are observed in spectra obtained from one-dimensional experiments performed on a polycrystalline sample of the tripeptide MLF, as shown in Figure 2 A–C. Significantly, efficient polarization transfer was achieved using SIM-CP by matching the ¹³C and ¹⁵N RF fields to the (n-1) Hartmann-Hahn resonance condition with ¹H.

The two-dimensional correlation spectra in Figure 2D show one-bond correlations of ¹⁵N/¹H^N and ¹³C/¹H resonances. The residue specific chemical shift assignments were obtained from previously reported ¹³C-detected experiments [1]. The pulse sequence in Figure 1C enables two separate three-dimensional spectra to be obtained from a single experiment. In particular, these experiments enable the simultaneous observation of ¹³C and ¹⁵N chemical shifts correlated to amide ¹H and alpha ¹H, in particular ¹Hn/N(Cα)/¹Hα and ¹Hα/¹³Cα (N)/¹H^N. The advantage of using bidirectional polarization transfer is that amide ¹H^N are correlated selectively with the ¹³C and/or ¹⁵N, and ¹Hα within a residue. This is a prerequisite to assign backbone ¹H resonances before they can be correlated to other ¹H for assignments and distance measurements.

The experiments simultaneously measure the amide ¹H and aliphatic ¹H resonance frequencies and have the advantage over conventional methods of using amide ¹H-detection only to assign back bone resonances and promote distance measurements through homonuclear spin exchange [14]. This requires multiple step coherence transfers and may cause sensitivity losses for samples that undergo fast relaxation. Secondly, simultaneous ¹³C and ¹⁵N editing experiments may be useful for lifting the ambiguity of resonance assignments for heavily crowded spectra, such as the case shown for fd coat protein.

Additionally, the bidirectional polarization transfer between ¹³C and ¹⁵N has the advantage of correlating amide ¹⁵N and ¹³Cα chemical shifts within a residue with the ¹³C′ chemical shift of the preceding residue. This step of spectral editing/selective magnetization transfer allows the experiment to be used for sequential resonance assignments in proteins. As observed, ¹³C-detected experiments show nearly 67% and 33% magnetization transfer from ¹⁵N to ¹³C′ and to ¹³Cα, respectively. We also observed that the magnetization transfer for the current optimization surprisingly showed 30–50% higher sensitivity compared to the previous optimization method by Barbet-Masin et al. [10] and under similar experimental conditions. Inter- and intra- residue correlation spectra are shown in Figure 3. It is clear from the well-resolved two- and three-dimensional spectra of MLF, which show the amide nitrogen ¹H of leucine correlated with the Leu ¹³Cα and Met ¹³C′ chemical shifts. In the same spectra, the Phe ¹H^N resonance is correlated with its ¹³Cα and the Leu ¹³C′ chemical shifts. The cross-peaks are marked in Figure 3E by residue type. The three-dimensional data correlating amide ¹⁵N and ¹³C resonances were acquired using a 10% non-uniform sampling schedule in order to reduce the amount of time needed to perform the experiment. The successfully reconstructed spectrum is shown in Figure 3F. The spectrum showed increase in sensitivity and resolution when compared to the linear method of data acquisition. This was achieved due to the long evolution periods that enabled the line narrowing and improved signal-to-noise ratios. We also performed homonuclear ¹³C/¹³C mixing prior to heteronuclear ¹³C/¹⁵N mixing to simultaneously obtain intra- and inter- residue correlations such as ¹H^N/NCACO and ¹H^N/NCOCA in a single experiment (data not shown). We note that similar results performed at moderate spinning rate have been recently reported [11].

We also applied the experiments to the 50-residue fd coat protein in intact bacteriophage particles. The heteronuclear correlation spectra in Figure 4A and D contain ~30 distinguishable ¹⁵N/¹H^N cross-peaks and ~20¹³C/¹H cross-peaks, respectively. The relatively narrow spectral dispersion among the alpha hydrogen resonances prevents the two-dimensional from providing unambiguous resolution and assignments. In order to achieve adequate resolution, two separate three-dimensional spectra were acquired in a single experiment using the pulse sequence shown in Figure 1. Two two-dimensional slices correlating amide ¹H chemical shifts with ¹⁵N chemical shifts obtained at 4.2 ppm and 3.8 ppm ¹H chemical shifts are shown in Figure 4B and C, respectively. Similarly, two-dimensional slices correlating alpha carbon ¹H resonances with ¹³C chemical shifts obtained at 8.5 ppm and 7.6 ppm are shown in Figure 4E and F, respectively. We also performed the intra- and inter- residue ¹³C correlation experiment for fd coat protein similar to the experiment performed on MLF shown in Figure 3D. A representative two-dimensional slice obtained at 109.7 ppm ¹⁵N chemical shift is shown in Figure 4G. The spectrum shows Gly23 ¹H^N correlation with Gly23 ¹³Cα and Ile22 ¹³C′. This experiment is similar to the CANCO type correlation experiment routinely used for resonance assignment in proteins, but with the additional advantage of correlating ¹H^N chemical shifts. As seen from the representative slices, we were able to resolve nearly all resonance peaks and able to assign the proton, carbon and nitrogen chemical shifts for the fd coat protein back bone.

Here we demonstrate that pulse sequences utilizing simultaneous cross-polarization, bidirectional coherence transfer pathways and non-uniform sampling can be used to perform multi-dimensional experiments rapidly and efficiently. We have applied these experiments to fully protonated and uniformly ¹³C and ¹⁵N labeled polycrystalline tripeptide and hydrated bacteriophage samples. Multiple chemical shift correlations are observed using simultaneous CP and bidirectional coherence transfer techniques. Bidirectional CP between inter- and intra- residue ¹³C and ¹⁵N nuclei is efficient under fast MAS and can be exploited further in sequential resonance assignment experiments.

Highlights.

• Bidirectional ¹³C-¹⁵N transfers at 60 kHz

• Simultaneous inter- and intra- residue transfers with ¹H irradiation

• Simultaneous acquisition of two three-dimensional spectra with one receiver

• ¹H assignments based on bidirectional ¹³C-¹⁵N transfers