Proton-Detected Polarization Optimized Experiments (POE) Using Ultrafast Magic Angle Spinning Solid-State NMR: Multi-acquisition of Membrane Protein Spectra

Abstract

Proton-detected solid-state NMR (ssNMR) spectroscopy has dramatically improved the sensitivity and resolution of fast magic angle spinning (MAS) methods. While relatively straightforward for fibers and crystalline samples, the routine application of these techniques to membrane protein samples is still challenging. This is due to the low sensitivity of these samples, which require high lipid:protein ratios to maintain the structural and functional integrity of membrane proteins. We previously introduced a family of novel polarization optimized experiments (POE) that enable to make the best of nuclear polarization to obtain multiple-acquisitions from a single pulse sequence and one receiver. Here, we present the ¹H-detected versions of POE using ultrafast MAS ssNMR. Specifically, we implemented proton detection into our three main POE strategies, H-DUMAS, H-MEIOSIS, and H-MAeSTOSO, achieving the acquisition of up to ten different experiments using a single pulse sequence. We tested these experiments on a model compound N-Acetyl-Val-Leu dipeptide and applied to a six transmembrane acetate transporter, SatP, reconstituted in lipid membranes. These new methods will speed up the spectroscopy of challenging biomacromolecules such as membrane proteins.

Graphical abstract

1. INTRODUCTION

Solid-state NMR spectroscopy is a powerful tool to study the structure and dynamics of biomacromolecules that are recalcitrant to crystallization such as fibrillar and integral membrane proteins^1–11. In the past decades, heteronuclear detected experiments (direct detection) have represented the main approach to determine resonance assignments and obtain internuclear distances for structure calculations^{10, 12–16}. However, direct detection methods with ¹³C or ¹⁵N acquisition are inherently insensitive and require longer experimental times^{10, 17}. On the other hand, inverse detection (i.e., ¹H-detected experiments) has represented the method of choice for solution NMR spectroscopy¹⁸. However, strong ¹H–¹H dipolar couplings, hampered the success of inverse detection for ssNMR applications. This problem has been overcome by the introduction of fast MAS technology, which enabled dramatic sensitivity enhancement for multi-dimensional experiments particularly for microcrystalline and fibrillary protein preparations^19–26. To achieve fast spinning rates, it is necessary to use rotors with smaller diameters which reduces the operative sample volumes to a range of 1–4 μl, lengthening significantly acquisition times. For membrane proteins, this problem is exacerbated by the presence of lipids and bulk water ^{3, 27}. While low lipid-to-protein ratios and hydration levels are often used for ssNMR experiments, it is unlikely that membrane proteins maintain their functional and structural integrity under these conditions.

During the past decade, our group has been developing ¹³C-detected Polarization Optimized experiments (POE) aiming to speed up the spectroscopy of membrane proteins reconstituted with lipid-to-protein ratios close to physiological conditions^{9, 28–36}. Our overall goal has been to adapt ssNMR spectroscopy samples to closely resemble physiological conditions, rather than adapting the samples to the needs of spectroscopy. POE utilize orphan (or undetected) spin operators that are discarded in classical multidimensional NMR experiments and recover them, allowing the acquisition of multiple 2D and 3D experiments in a single pulse sequence^{29–30, 32}. A similar approach has also been utilized for solution NMR experiments (e.g., NOAH sequences) for simultaneous acquisition of 2D spectra of small organic compounds³⁷. Unlike liquid samples, the longitudinal relaxation (T₁) times of solids or solid-like biomacromolecules are significantly longer for ¹⁵N and ¹³C nuclei. Therefore, it is possible to store ¹⁵N and/or ¹³C polarization along the longitudinal axis (N_z or C_z) and subsequently utilize it for multiple acquisitions^{29, 34}. The ease of storing the polarization in solids makes POE (e.g., DUMAS, MEIOSIS, and MAeSTOSO) unique tools for simultaneous acquisition of various types of triple-resonance experiments in ssNMR ³⁸. Another key element for POE is the simultaneous cross-polarization (SIM-CP) scheme that matches the Hartmann-Hahn conditions for ¹H, ¹³C, and ¹⁵N, simultaneously^{29–30, 39}. The beauty of POE comes from its straightforward implementation on commercial spectrometers equipped with only one receiver and Low-E or E-free MAS probes^40–41. As a result, POE are becoming more popular and have been successfully implemented for ¹H-detected experiments, including hybrid pulse sequences with a combination of CP and INEPT polarization transfer periods^{28, 42–46}.

In this work, we further expand our POE family and show the implementation of novel and efficient pulse sequences for multi-acquisitions using ¹H-detection at ultra-fast magic angle spinning rates of up to 65 kHz. Specifically, we present the ¹H-detected versions of the three main POE strategies we previously introduced^{29, 31, 33}, i.e., proton-detected DUMAS, MEIOSIS and MAeSTOSO (H-DUMAS, H-MEIOSIS, and H-MAeSTOSO). While specific experiments are presented in this paper, the DUMAS, MEIOSIS and MAeSTOSO strategies enable one to concatenate several different pulse sequences³². We demonstrate the application of these methods using the standard dipeptide, N-Acetyl-Val-Leu (NAVL), and a six-pass transmembrane protein that is responsible for transporting succinate and acetate in bacteria (succinate-acetate permease or SatP), which was reconstituted in lipid membranes⁴⁷.

2. EXPERIMENTAL

Uniformly ¹³C and ¹⁵N labeled succinate-acetate permease (SatP) was expressed and purified from E. coli bacteria using a SUMO construct as a fusion. The bacteria were cultured in M9 media enriched with ¹³C glucose and ¹⁵N ammonium chloride as reported previously⁴⁷. The final sample consisting of approximately 0.8 mg of SatP was reconstituted in 1:40 protein to lipid ratio with 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) liposomes and packed into a 1.3 mm Bruker MAS rotor using our standard protocols.

All the experiments were implemented on a Bruker 600 MHz spectrometer equipped with 1.3 mm fast MAS probe. The CP and SIM-CP spectra reported in Figure 1 were acquired at 12 kHz MAS rate using 20 ms ¹³C and ¹⁵N acquisition times under 100 kHz SPINAL-64 decoupling and a 2 s recycle delay⁴⁸; whereas all the remaining spectra were acquired using a MAS rate (ω_r) of 65 kHz with ¹H acquisition time set to 8.3 ms and recycle delay of 2 s. The heat induced by fast spinning on the sample was compensated by lowering the temperature. Specifically, at 65 kHz the temperature of the RF coil was set to −20 °C, which corresponds to an effective sample temperature of 25 °C as measured from the water resonance position. The RF carrier frequencies for ¹H, ¹³C, and ¹⁵N were set to 4.7, 40, and 110 ppm, respectively. During ¹H acquisition, a 10 kHz WALTZ16 decoupling was applied on ¹³C or ¹⁵N channels⁴⁹. The 90° pulse length on ¹H was set to 1.25 μs, whereas 3 μs 90° pulses were used for ¹³C and ¹⁵N. During CP and SIM-CP, the RF amplitudes on ¹³C and ¹⁵N were set to 40 kHz, and the ¹H RF pulse was linearly ramped from 90 to 100% to obtain maximum signal at n=1 Hartmann-Hahn matching condition⁵⁰. The optimized ¹H amplitudes during CP and SIM-CP periods were 51.2 and 102 kHz at MAS rates of 12 and 65 kHz, respectively. The suppression of the ¹H signals arising from water and lipids was achieved by phase-switched pulses (MISSISSIPPI)⁵¹, with 30 kHz ¹H RF amplitude applied for 200 ms (t_supp). For the 2D experiments, the SIM-CP contact time was set to 750 μs. The durations of back CP transfers from ¹³C to ¹H (T^C_CP) and ¹⁵N to ¹H (T^N_CP) were set to 150 and 200 μs, respectively. The SPECIFIC-CP transfer between ¹⁵N and ¹³CA was obtained by using a linear ramp (85 to 100%) pulse on ¹³C applied at 60 ppm and a constant amplitude pulse on ¹⁵N⁵². The optimized RF amplitudes for SPECIFIC-CP on ¹³C and ¹⁵N were 18.3 and 43.3 kHz, respectively. During t₁ evolution, 10 kHz WALTZ16 decoupling was applied on ¹H. Simultaneous (or parallel) ¹³C and ¹⁵N t₁ evolution periods were synchronized by using 100 ¹³C t₁ increments and two sets (n=2 in Figures 2, 4, 7 and 8) of 50 t₁ increments for ¹⁵N²⁹. The dwell-times for ¹³C and¹⁵ N were set to 100 and 200 μs that corresponds to a maximum t₁ evolution of 10 ms for both ¹³C and ¹⁵N. A four-step phase cycle was applied for all the pulse sequences with ϕ₁= y,-y, ϕ₂= x,x,−x,−x, and receiver phase ϕ_rec=x,-x,-x,x. The t₁ dimension of 2D spectra were acquired in States mode by switching the phase of ϕ* pulse between y and −x. For the 1D data reported in Figure 1, ¹³C CP and SIM-CP spectra were acquired with 32 scans, and ¹⁵N CP and SIM-CP spectra were acquired with 128 scans. ¹³C-detected 1D MEIOSIS spectra of Figure 3 were acquired with 256 scans using 20 ms ¹³C acquisition time under 10 kHz ¹H WALTZ decoupling. Proton 1D spectra in Figure 5 were acquired with 16 scans. The H-DUMAS (Figure 2), H-MAeSTOSO-4 (Figure 7) spectra on SatP were acquired using 16 and 40 scans, respectively, whereas the H-MEIOSIS (Figure 4) spectrum was acquired with 40 scans and two interleaved data sets obtained with ϕ=x and −x. The total experimental times for SatP 2D H-DUMAS, H-MEIOSIS, and H-MAeSTOSO4 spectra in Figures 2, 4, and 7 were 2, 10, and 5.3 hrs, respectively. The 2D H-MAeSTOSO-10 spectra of NAVL in Figure 8 were acquired using 24 scans for a total experimental time of 6.2 hrs. During HH RFDR, ¹H 180° pulse length was set to 2.5 μs. Mixing times for HH RFDR were set to 0.74 or 1.47 ms⁵³, whereas the CC TOCSY mixing (or T^C_WALTZ) time in Figure8 was set to 9.6 ms. All the spectra were processed using NMRpipe scripts as described in references^{34, 54} for multiple acquisition data sets. Both t₁ and t₂ dimensions were processed with 90° shifted sine bell window functions.

Figure 1:

(A) CP and SIM-CP pulse sequences used for quantifying relative signal intensities. For SIM-CP, ¹³C and ¹⁵N spectra were acquired in separate experiments with a single receiver. (B) and (C) Comparison of CP and SIM-CP on SatP membrane protein obtained from integrated intensities of ¹³C and ¹⁵N spectra acquired at 65 kHz and 12 kHz MAS rates, respectively.

Figure 2:

(A) Pulse sequence for two-dimensional H-DUMAS (proton detected DUMAS) experiment for simultaneous acquisition of ¹³C- and ¹⁵N-edited experiments, (H)CH and (H)NH. (B) 1D (H)CH and (H)NH spectra of SatP membrane protein obtained from first increment (t1=0) of H-DUMAS (shown in blue and red), and the corresponding spectra (shown in black) obtained from single acquisition methods using 65 kHz MAS rate. For comparison the 1D spectra shown in blue and red are slightly shifted to the right. (C) Two-dimensional (H)CH and (H)NH spectra of SatP protein acquired using H-DUMAS pulse sequence with n=2 loops for ¹⁵N t1 evolution.

Figure 4:

(A) Two-dimensional H-MEIOSIS (proton detected MEIOSIS) pulse sequence for simultaneous acquisition of two pairs of ¹³C- and ¹⁵N-edited experiments, (H)CH, (H)N(CA)H, (H)NH and (H)CA(N)H. (B) 2D H-MEIOSIS spectra of SatP protein acquired at 65 kHz MAS rate using two loops (n=2) for ¹⁵N t1 evolution. A pair of 2D spectra in the 1^st and 2^nd acquisitions are obtained by adding and subtracting the two data sets recorded with ϕ=x and −x.

Figure 7:

(A) Pulse sequence for the H-MAeSTOSO-4 experiment for simultaneous acquisition of (H)CHH, (H)NHH (RFDR=1.47 ms), H(N)HH (RFDR=0.74ms), and (H)NH experiments. Corresponding 2D spectra on SatP membrane protein are shown in (B). Suppression of ¹H signals from water and lipids is obtained by using t_supp=200ms. Prior to 3^rd and 4^th acquisitions 0.25*t_supp (=50 ms) period was used for eliminating waters signals. (C) 2D spectra (H)NHH and (H)NH obtained from H-MAeSTOSO-4 pulse sequence in the absence of 0.25*t_supp period, showing the residual water signals.

Figure8:

(A) Pulse sequence for the H-MAeSTOSO-10 experiment for simultaneous acquisition of ten 2D spectra. (B) Application of H-MAeSTOSO-10 on the NAVL (N-acetyl-Val-Leu) dipeptide.

Figure 3:

MEIOSIS experiment at fast MAS rates. (A) Pulse sequence for 1D MEIOSIS that records four polarization pathways (CC_res, NCA_trans, NN_res, and CAN_trans) resulting from the first SPECIFIC-CP period τ. (B) Normalized integrated intensities of all four pathways measured for SatP protein by varying the τ mixing times. The polarization in each pathway is normalized with respect to initial value at τ=0. (C) 1D spectra of CC_res, NCA_trans, NN_res, and CAN_trans (color coded accordingly) polarization pathways obtained at τ= 0 and 6.5 ms. (D) 1D spectra at τ=6.5 ms with corresponding intensities normalized with respect to highest intensity spectrum of CC polarization pathway.

Figure 5:

(A) Pulse sequence for evaluating the residual ¹³C and ¹⁵N polarization that remains on ¹³C and ¹⁵N at the end of back CP periods (T^C_CP and T^N_CP). (B) Membrane protein SatP spectra acquired in 2^nd and 4^th acquisition periods by varying T^C_WALTZ and T^N_WALTZ decoupling periods. (C) Transferred and residual polarization of ¹³C–¹H and ¹⁵N–¹H back CP periods measured by varying T^C_CP and T^N_CP back CP periods.

To avoid the sharing of WALTZ16 pulse profile, we recommend using different names for each WALTZ16 period in the pulse sequences. Bruker TOPSPIN pulse programming allows up to nine CPD names (CPD1, CPD2…..CPD9) that can be recalled during the decoupling periods of a pulse program. For example, in Figures 5A and 5B the duration of T^C_WALTZ on ¹³C (or T^N_WALTZ on ¹⁵N) channels was varied from 0 to 10.2 ms followed by back-CP and another ¹³C (or ¹⁵N) WALTZ16 period. If the two ¹³C WALTZ16 decoupling sequences (1’st and 2^nd acquisitions of Figure 5A) on ¹³C channel are used in a pulse sequence with the same name, then the shape or pulse profile of WALTZ16 is shared between two periods depending on the length of each period. An alternative way is to have the WALTZ16 sequence to complete an entire cycle, setting ‘n’ cycles of WALTZ16 (= 2.4, 4.8, 7.2, and 9.6 ms). In this case, the WALTZ16 profile for two periods do not overlap.

3. RESULTS

3.1. Simultaneous cross-polarization (SIM-CP)

In POE, the initial ¹³C and ¹⁵N polarization is obtained from SIM-CP, where both ¹³C and ¹⁵N RF amplitudes (ω) are simultaneously Hartmann-Hahn matched to ¹H RF (ω_H)^{29, 32}. For a given MAS rate (ω_r), the ¹³C and ¹⁵N RF amplitudes satisfy the matching condition ω_H=n·ω_r ±ω, where n=1 and 2 are known as 1^st and 2^nd sideband Hartmann-Hahn matching conditions. Figure 1 shows the systematic comparison of CP and SIM-CP intensities for the membrane protein SatP at fast (65 kHz) and slow (12 kHz) MAS rates using separate ¹³C- and ¹⁵N-detected experiments (Figure 1A) with a single receiver. At 65 kHz MAS rate, heteronuclear ¹H decoupling was obtained by 10 kHz WALTZ16 pulses; whereas at 12 kHz MAS rate the ¹H RF was set to 100 kHz for SPINAL64 proton decoupling. During CP and SIM-CP contact time (t), ¹³C and ¹⁵N RF amplitudes (ω) were set to 40 kHz, and ¹H RF was optimized to satisfy the first or n=1 sideband matching condition (ω_r+ ω), which corresponds to 52 kHz and 105 kHz at MAS rates of 12 and 65 kHz, respectively. The ¹H radio frequency (RF) was linearly ramped from 90 to 100%. The integrated intensities for ¹³C and ¹⁵N spectra were calculated for spectral regions spanning from 0 to 80 ppm and 95 to 140 ppm, respectively. As shown in Figure 1B and C, the ¹³C intensities for SIM-CP follow a trend similar to the conventional CP for both 65 and 12 kHz MAS rates. The average intensity loss with the ¹⁵N SIM-CP experiments are approximately 15 and 8% for 12 and 65 kHz spinning rates, respectively, with contact times ranging from 0.3 to 0.8 ms. Note that the data reported in Figure 1 are recorded under identical sample conditions using a 1.3 mm MAS probe. The effective sample temperatures for SatP at 12 and 65 kHz were set to 25 °C and monitored using the ¹H peak position of the water signal. The relative intensities of CP and SIM-CP spectra at 12 kHz MAS are in close agreement with our previous results obtained for crystalline and membrane protein samples using a 3.2 mm probe with MAS rates of 8 and 12 kHz^29–30. Interestingly the ¹⁵N SIM-CP scheme is more efficient at faster MAS rates. At contact times greater than 1 ms, the performance of SIM-CP at 65 kHz MAS gradually decreases with an average intensity loss of approximately 15% with respect to CP. Due to the complexity of ¹H–¹H dipolar coupling network, it is difficult to quantify the CP and SIM-CP dynamics. However, it is possible to speculate that the increased efficiency of ¹⁵N SIM-CP at lower contact times and higher MAS rates may originate from more localized polarization transfer from the directly bonded amide protons to the ¹⁵N atoms. In fact, at 65 kHz MAS rate we observe a more oscillatory buildup for the ¹⁵N polarization (Figure 1B). In contrast, at 12 kHz MAS rate the oscillations of the ¹⁵N polarization are significantly damped by the strong ¹H–¹H dipolar couplings (Figure 1C).

3.2. ¹H-detected DUMAS (H-DUMAS): simultaneous acquisition of (H)CH and (H)NH experiments

In the original implementation of DUMAS (DUal acquisition Magic Angle Spinning) experiment²⁹, the ¹³C and ¹⁵N polarization from SIM-CP was used to acquire 2D ¹³C- and ¹⁵N-edited experiments such as DARR and NCA with ¹³C detection at slow MAS rates (8–12 kHz). Analogously, the H-DUMAS experiment enables the acquisition of 2D ¹H-detected ¹³C- and ¹⁵N- edited experiments (Figure 2A). After SIM-CP transfer from ¹H to ¹³C and ¹⁵N, a pair of 90° pulses on ¹³C and ¹⁵N flips the polarization to the z-direction followed by a water suppression period. A 90° pulse on ¹³C brings the polarization to the transverse plane, which evolves during t₁ followed by a back CP period (T^C_CP) from ¹³C to ¹H. The ¹H magnetization is then detected during the first t₂ acquisition that records a 2D (H)CH experiment (blue). After the first acquisition, a 90° pulse on ¹⁵N flips the polarization to the transverse plane, which evolves during ¹⁵N t₁ followed by an ¹⁵N–¹H back CP (T^N_CP) and a second t₂ acquisition period that records a 2D (H)NH spectrum. During the t₁ and t₂ evolution periods, heteronuclear decoupling is achieved by a WALTZ16 pulse sequence⁴⁹.

An important feature of the DUMAS experiments is the distinct t₁ evolution periods for ¹³C and ¹⁵N²⁹, which enables to optimize ¹³C and ¹⁵N t₁ evolution parameters independently. Therefore, the t₁ evolution time for the ¹³C-edited (H)CH experiment was set with a maximum evolution time of 10 ms, with a dwell time of 100 μs and 100 increments. On the other hand, the t₁ evolution time for the ¹⁵N-edited (H)NH experiment was divided into two identical t₁ evolution periods with 50 increments each corresponding to a maximum t₁ evolution period of 10 ms and a dwell time of 200 μs (n=2, Figure 2A). As described previously^{29, 34, 38}, the two data sets for the (H)NH experiment are added post acquisition during the data processing. In this manner, the effective number of scans for the (H)NH is double that of the (H)CH experiment. Figure 2B displays the 1D spectra of SatP obtained from the first increment (t₁ = 0) of the H-DUMAS-(H)CH-(H)NH experiment shown in blue and red, respectively. As a comparison, the single-acquisition spectra obtained from the conventional (H)CH and (H)NH pulse sequences are overlaid in black. As expected, the intensity of the H-DUMAS-(H)CH spectrum is identical to the conventional (H)CH. On the other hand, for the H-DUMAS-(H)NH spectrum shows an intensity loss of approximately 5%, which is expected from the lower performance of the SIM-CP scheme. Figure 2C shows the application of the 2D H-DUMAS-(H)CH-(H)NH experiment to SatP acquired simultaneously at 65 kHz for a total experimental time of ~2 hours. Although ¹H resolution at 65 kHz MAS rate is still challenging for larger proteins such as SatP, we anticipate that it is possible to further improve sensitivity and resolution using a combination of 3D experiments, higher MAS rates (100–110 kHz) as well as sparse protein deuteration⁵⁵. Note that the acquisition of 2D (H)NH and (H)CH spectra at fast MAS rates can also be obtained using single acquisition as reported in references^{42, 44} at slow and fast MAS rates. The H-DUMAS experiment is more flexible and enables to choose different T^C_CP and T^N_CP back CP periods and optimize sensitivity and resolution of the ¹H spectra. Similar to ¹³C detected DUMAS experiments, H-DUMAS can easily be extended to the third dimension with homo- or hetero-nuclear (CC, HH or NC) mixing times.

3.3. ¹H detected MEIOSIS (H-MEIOSIS): simultaneous acquisition of (H)CH, (H)N(CA)H, (H)CA(N)H and (H)NH experiments

We previously demonstrated that when a given ¹³C- or ¹⁵N-edited pulse sequence includes NC or CN polarization transfer (SPECIFIC-CP) periods, it is possible to nest into it three additional experiments using the MEIOSIS (Multiple ExperIments via Orphan SpIn operatorS) strategy³¹. To implement this for proton detection, we first investigated the relative intensities of the polarization pathways using ¹³C-detected MEIOSIS experiments at fast MAS rates (Figure 3A). In this case, the preparation period (SIM-CP) is followed by SPECIFIC-CP that creates two bidirectional polarization transfer pathways from ¹⁵N to ¹³C and vice versa named NCA_trans and CAN_trans, respectively. The residual ¹³C and ¹⁵N polarization pathways (i.e., CC_res and NN_res) remain on ¹³C and ¹⁵N at the end of SPECIFIC-CP period³¹. The polarization from the CC_res and NCA_trans pathways is recorded in the first acquisition; whereas the ¹⁵N polarization from NN_res and CAN_trans pathways is transferred to ¹³C via SPECIFIC-CP and is recorded in a second acquisition period. For each acquisition, the decoding of two pathways is obtained by altering the phase ϕ of ¹⁵N SPECIFIC-CP pulse between x and −x, and the resultant data sets are stored in two separate files. The phase ϕ inverts the transferred polarization (CN_trans and NC_trans,), whereas the CC_res and NN_res are not affected by flipping ϕ³¹. In the first acquisition, CC_res and NCA_trans are decoded by adding and subtracting the two data sets recorded with ϕ= x and −x. Similarly, addition and subtraction of two data sets of the second acquisition decodes both CAN_trans and NN_res pathways.

Figure 3B shows the 1D normalized intensities of SatP protein spectra for all four pathways measured at 65 kHz. SPECIFIC-CP mixing times (τ) were arrayed from 0 to 8 ms. The optimal NCA and CAN transfer is obtained at a mixing time of 6.5 ms. Figure 3C shows the 1D spectra corresponding to four pathways obtained at 0 and 6.5 ms SPECIFIC-CP mixing times. The residual ¹³C and ¹⁵N polarization at optimal NCA or CAN mixing time (~ 6.5 ms) corresponds to approximately 52 and 40% of the normalized signal as previously shown for a 3.2 mm probe at slower MAS rates³¹. Note that the T_1ρ relaxation times for ¹³C and ¹⁵N during SPECIFIC-CP are relatively longer at faster MAS rates and the transfer efficiency for NCA and CAN can vary significantly based on the MAS rates. As a result, the extent of residual polarization depends on the net effect of T_1ρ and SPECIFIC-CP transfer efficiencies for a given MAS rate.

The spectra of Figure 3 were recorded using 256 scans with ϕ phase switching between x and −x, which corresponds to 512 effective scans. If the signal-to-noise achieved with ‘n’ scans is ‘S_o’, then k * S_o signal-to-noise requires ‘k² * n’ scans. Therefore, using conventional single acquisition methods (τ=0) to obtain 52% (k=0.52) ¹³C CP signal it would take 128 scans (k²*512), which corresponds to approximately one fourth of the experimental time. Similarly, to obtain 40% (k=0.4) NCA signal, it would require 82 scans (k²*512), which corresponds to one sixth of the experimental time. The experimental time for 1D MEIOSIS using 512 scans and τ=6.5 ms is 18 minutes (Figure 3C). Using conventional, single acquisition methods, NCA and CAN 1D spectra require 36 minutes, whereas to obtain 52% ¹³C CP and 40% NCA signal it would take 4.5 and 3 minutes respectively. In other words, to achieve similar signal intensities, it would require approximately 43.5 minutes with conventional methods; whereas the MEIOSIS experiment would only take 18 minutes, with a net time saving of ~59%. Note that in Figures 3B and 3C, each of the four pathways are normalized separately to show gains and losses of transferred and residual polarization intensities with respect to the initial value at τ=0. For setting up the 2D MEIOSIS experiment, it is important to consider the relative intensities of the four 1D spectra. Figure 3D shows the relative intensities of four MEIOSIS pathways for a SPECIFIC-CP period of 6.5 ms calculated by integrating the intensity values between 45 to 75 ppm. The residual CC pathway shows the highest intensity, which was normalized to 100%, and the relative intensities of NCA, NN, and CNC spectra, corresponding to 41%, 18%, and 17% respectively.

Similar to ¹³C-detected MEIOSIS experiments at slow MAS rates, the four polarization pathways can be used to design various combinations of H-MEIOSIS experiments at fast MAS rates³⁸. Figure 4A shows one example of two-dimensional H-MEIOSIS, where ¹³C and ¹⁵N t₁ evolution is followed by SPECIFIC-CP to create CC_res, NCA_trans, CAN_trans, and NN_res pathways to record (H)CH and (H)N(CA)H spectra in the first acquisition, and (H)CA(N)H and (H)NH spectra in the second acquisition. The ¹³C and ¹⁵N t₁ evolutions of H-MEIOSIS are synchronized by using two loops (n=2) for ¹⁵N t₁ evolution. By addition and subtraction of two data sets acquired with phase ϕ set to x and −x, it is possible to decode the two pathways for the first and second acquisitions. Figure 4B shows four 2D spectra [(H)CH, (H)N(CA)H, (H)CA(N)H, and (H)NH] of SatP acquired simultaneously using H-MEIOSIS. The total experimental time was approximately 10 hours. The 2D spectra (H)CA(N)H and (H)N(CA)H are obtained from transferred polarization pathways (CN_trans and NC_trans) under optimal SPECIFIC-CP conditions; while the 2D (H)CH and (H)NH spectra are obtained from 40–50% orphan polarization. Thus the relative intensities of (H)CH and (H)NH spectra are 40–50% lower than the corresponding H-DUMAS or single acquisition spectra. Note that the peaks in the 2D (H)N(CA)H and (H)CA(N)H experiments have lower intensity due to NCA (or CAN) transfer period, and thus require higher number of scans. The latter produces sufficient residual polarization that enables the acquisition of the additional (H)CH and (H)NH spectra. In this way, using the H-MEIOSIS strategy we combined a pair of low sensitive experiments [(H)N(CA)H and (H)CA(N)H] with another pair of high sensitive experiments [(H)NH and (H)CH].

3.4. How much residual ¹³C and ¹⁵N polarization remains after back CP transfer periods? Can it be used for additional experiments?

To quantify the residual ¹³C and ¹⁵N polarization at the end of back CP periods (T^C_CP and T^N_CP), we modified the H-DUMAS experiment. The 1D pulse sequence reported in Figure 5A consists of four acquisition periods, where the first and third FIDs (orange and green) correspond to H-DUMAS polarization pathways obtained from SIM-CP. After ¹³C–¹H and ¹⁵N–¹H back CP periods, a 90° pulse on ¹³C and ¹⁵N stores the residual polarization along the z-direction followed by ¹³C and ¹⁵N WALTZ16 decoupling periods, T^C_WALTZ and T^N_WALTZ, respectively. After a WALTZ16 decoupling period, the longitudinal ¹³C and ¹⁵N residual polarization is transferred to ¹H by applying a 90° pulse followed by another ¹³C–¹H and ¹⁵N–¹H back CP period that records aliphatic (H^C) and amide (H^N) proton spectra. In this way, it is possible to quantify the net residual polarization by varying T_CP and T_WALTZ decoupling periods. Typically, for fully protonated samples under 40–100 kHz MAS rates the back-CP are set to 100 to 400 μs and the ¹³C and ¹⁵N WALTZ16 decoupling periods are set between 7 and 10 ms. The H-DUMAS and H-MEIOSIS spectra reported in Figures 2 and 4 were obtained using ¹³C–¹H and ¹⁵N–¹H back CP periods set to 150 and 200 μs respectively; whereas t₂ WALTZ16 decoupling on ¹³C and ¹⁵N was set to 8.3 ms.

We first evaluated the effect of WALTZ16 decoupling pulses on ¹³C and ¹⁵N residual z-polarization during T_WALTZ (T^C_WALTZ and T_WALTZ). Figure 5B shows H^C (blue) and H^N (red) 1D spectra of SatP protein acquired at various T_WALTZ periods with T_CP (T^C_CP and T^N_CP) set to zero. Note that the second back-CP periods prior to 3^rd and 4^th acquisitions were set to a fixed value of 200 μs. The RF amplitudes of both ¹³C and ¹⁵N WALTZ16 pulses were set to 10 kHz, corresponding to a 25 μs 90° pulse. The WALTZ16 decoupling is based on composite rotation and consists of ninety-six 90° pulses with four phase switched Q blocks, namely, $\text{Q\hspace{0.17em}}\overline{\text{Q}}\overline{\text{Q}}\text{\hspace{0.17em}Q}$ ⁴⁹. The total time of WALTZ16 cycle corresponds to 2.4 ms (= 96 × 25 μs). Whereas each Q unit is a spin inversion sequence with twenty-four 90° pulses that corresponds to 600 μs (=24×25μs). The patterns of the 1D spectra in Figure 5B show a remarkable agreement with the WALTZ16 cycle. The latter demonstrates that the residual ¹³C and ¹⁵N z-polarization can be preserved at the end each Q unit (i.e., at multiple values of 600 μs) as well as 2.4, 4.8, 7.2, and 9.6 ms WALTZ16 periods, which correspond to 1, 2, 3, and 4 WALTZ16 cycles. Therefore, the residual ¹³C and ¹⁵N polarization can be recovered even in the presence of ¹³C and ¹⁵N decoupling pulses by synchronizing the total length of T_WALTZ to multiples of the (n/4) of cycle time, where ‘n’ is an integer. As shown in Figure 5B, the loss of ¹⁵N signal intensity is only 10% even after a WALTZ16 period of 9.6 ms. On the other hand, the WALTZ16 period on ¹³C acts as homonuclear TOCSY mixing, which causes ~ 50% of signal loss for T^C_WALTZ of 9.6 ms.

We then evaluated the effect of back-CP or T_CP (T^C_CP and T^N_CP) periods on the four pathways, namely, ¹³C–¹H and ¹⁵N–¹H transferred polarization pathways (orange and green, Figure 5A) detected in the 1^st and 3^rd acquisitions, and ¹³C and ¹⁵N residual pathways (blue and red) detected in 2^nd and 4^th acquisitions. The four acquisition periods were set to 8.3 ms. To preserve the residual polarization after 1^st and 3^rd acquisitions, the WALTZ16 decoupling was set on for another 1.3 ms (τ₁ in Figure 5A) to complete four WALTZ16 cycles for a total duration of T^C_waltz and T^N_WALTZ periods of 9.6 ms. When T_CP is set to zero, no signal was observed for the 1^st and 3^rd acquisitions, due to the absence of ¹³C–¹H and ¹⁵N–¹H back CP transfer (Figure 5C). On the other hand, when T_CP is zero, full ¹³C and ¹⁵N polarization is used for residual pathways, showing maximum signal intensities. At higher T_CP values, efficient ¹³C–¹H and ¹⁵N–¹H transfers lead to maximum intensities for 1^st and 3^rd acquisitions, whereas residual polarization (2^nd and 4^th acquisitions) gradually decreases. Typically, for U-¹³C,¹⁵N labeled proteins selective transfer from ¹³C to ¹H is is obtained by using 100 to 200 μs back CP period. Whereas ¹⁵N to ¹H back CP transfer is obtained by using the T_CP values between 200 to 500 μs²¹. As shown in Figure 4C, at T^C_CP value of 150 μs, we observed approximately 10% ¹³C residual polarization. On the other hand, we observed 54 and 27% ¹⁵N residual polarization for T^N_CP values of 200 and 300 μs, respectively. Figure 6 shows the normalized intensities for the H^C and H^N spectra illustrated in Figure 5C. The transferred and residual polarization from the CH back-CP are CH_trans and CH_res, respectively. Whereas NH back-CP transferred and residual polarization are represented by NH_trans and NH_res, respectively. Note that the efficiency of transferred and residual polarization pathways of back-CP periods depends on sample conditions as well as the MAS rate. The NH back-CP periods of 400 to 600 μs were reported for crystalline proteins with fast MAS rates and perdeuteration sample conditions. On the other hand, for SatP protein, the back-CP transfer efficiency reached a plateau at 300 μs.

Figure 6:

Normalized integrated intensities of transferred and residual polarization pathways of ¹³C–¹H and ¹⁵N–¹H back CP obtained from the spectra reported in Figure 5C using the pulse sequences of Figure 5A.

3.5. ¹H detected MAeSTOSO-4 (H-MAeSTOSO-4): simultaneous acquisition of four experiments

The ¹³C and ¹⁵N residual polarization of the back-CP periods can be used to implement the experiments based on the MAeSTOSO (Multiple Acquisitions via Sequential Transfer of Orphan Spin polarizatiOn) strategy³³. The 2D pulse sequence for H-MAESTOSO-4 is obtained by extending the 2D H-DUMAS pulse sequence for four acquisition periods as shown in Figure 7A. As in H-DUMAS, the ¹³C and ¹⁵N polarization from SIM-CP is used for recording ¹³C- and ¹⁵N-edited experiments in the 1^st and 2^nd acquisitions, respectively, where a ¹H–¹H RFDR period was incorporated for homonuclear transfer between ¹H nuclei prior to the second acquisition period. The resulting signal in the 1^st and 2^nd acquisitions corresponds to the (H)CHH and (H)NHH 2D spectra, respectively. The residual ¹⁵N polarization after the first ¹⁵N–¹H back-CP (T^N_CP) is stored along the z-direction by applying a ¹⁵N 90° pulse followed by WALTZ16 decoupling. As shown in Figure 5, the total length of the T^N_WALTZ period can be adjusted to 9.6 ms to retrieve the residual ¹⁵N polarization. Similarly, in the H-MAeSTOSO-4 pulse sequence the T^N_WALTZ period is set to 9.6 ms. After the T^N_WALTZ period, the ¹⁵N residual polarization is transferred to ¹H followed by a ¹H–¹H RFDR transfer ⁵³ that records another (H)NHH experiment in the third t₂ acquisition. After the 3^rd acquisition, the ¹⁵N residual polarization resulting from second ¹⁵N–¹H back CP is transferred to ¹H followed by a 4^th acquisition that records a (H)NH 2D experiment. At T^N_CP of 200us, The residual ¹⁵N polarization after the first ¹⁵N–¹H CP is approximately 54%, which is utilized in the 3^rd acquisition of the H-MAESTOSO-4 sequence. On the other hand, the residual polarization after the second ¹⁵N–¹H CP becomes approximately 29% (54% of 54% of the polarization), which is recorded in the 4^th acquisition period. Note that in Figure 7, for T^N_CP value of 300 μs, the signal intensities of the 3^rd and 4^th acquisitions become 27 and 7 %, respectively, due to lower residual polarization (Figure 5c). To increase the signal intensities of 3^rd and 4^th acquisitions, a lower T^N_CP duration (equal to 200 μs) was used, which reduces the polarization transfer for the 2^nd acquisition.

Figure 7B shows the SatP 2D spectra (H)CHH (RFDR=1.47 ms), (H)NHH (RFDR=1.47 ms), (H)NHH (RFDR=0.74 ms), and (H)NH acquired simultaneously using the H-MAeSTOSO-4 pulse sequence (Figure 7A) with a total experimental time of 5.3 hours. At 1.47 ms RFDR mixing, intense cross peaks were observed between H^N and H^C groups, whereas 0.74 ms mixing resulted in more resolved peaks, particularly for the aliphatic proton region (0 to 5 ppm). The HH cross peaks in the ¹⁵N-edited spectra recorded at two mixing times can be compared with the reference (H)NH spectrum recorded in the 4^th acquisition. During the execution of these pulse sequences, some of the water signal can interfere with the 2D spectra acquisition. Therefore, the spectra in Figure 7B were acquired with two additional water suppression periods (0.25*t_supp in Figure 7A) set to 50 ms, prior to second and third ¹⁵N–¹H back CP periods. As shown in Figure 7C, in the absence of second and third water suppression periods (0.25*t_supp in Figure 7A) we observed a weak water peak signal in (H)NHH (RFDR=0.74 ms) and (H)NH 2D spectra acquired in 3^rd and 4^th acquisitions respectively. Note that the solvent suppression element (t_supp) can also be incorporated prior to T^N_CP periods, which may improve water suppression as well as lipid signal intensities⁵⁷. Similarly, a constant time version can be used keeping the total number of ¹H pulses constant for all t₁ increments, which will improve water suppression^{19, 21}.

3.6. ¹H detected MAeSTOSO-10 (H-MAeSTOSO-10): simultaneous acquisition of ten experiments

By exploiting the ¹³C–¹H and ¹⁵N–¹H back CP residual polarization pathways, we also designed the H-MAeSTOSO-10 sequence for simultaneous acquisition of ten 2D spectra using five acquisition periods. The H-MAeSTOSO-10 (Figure8) experiment records one pair of ¹³C- and 2D ¹⁵N-edited spectra in each of the five acquisition periods by addition and subtraction of data sets recorded with phase ϕ = x and −x. The 1^st and 3^rd acquisitions are respectively obtained from H-MEIOSIS pathways (CC_res, NCA_trans, CAN_trans, and NN_res) followed by HH RFDR mixing periods to record (H)CHH and (H)N(CA)HH spectra in the 1^st acquisition, and (H)CA(N)HH and (H)NHH spectra in the 3^rd acquisition. During the 1^st acquisition, two residual ¹³C polarization pathways of first ¹³C–¹H back CP (T^C_CP) period are stored along the z-direction by synchronizing the T^C_WALTZ period to 9.6 ms. Unlike ¹⁵N, at fast MAS rates ¹³C z-polarization undergoes TOCSY ¹³C–¹³C homonuclear transfer under ¹³C WALTZ pulses (T^C_WALTZ period). After a T^C_WALTZ period the ¹³C polarization is flipped to the transverse plane by a 90° pulse and then transferred to ¹H by another ¹³C–¹H back CP followed by the 2^nd acquisition. The corresponding ¹³C- and ¹⁵N-edited 2D spectra of 2^nd acquisition are (H)CCH and (H)N(CAC)H, where CC TOCSY mixing is achieved during ¹³C T^C_WALTZ period⁵⁷. The two ¹⁵N residual polarization pathways of the first ¹⁵N–¹H back-CP period (T^N_CP) are recovered after the 3^rd acquisition by setting the T^N_WALTZ period to 9.6 ms. A 90° pulse is then applied on ¹⁵N followed by a second ¹⁵N–¹H back-CP and HH RFDR mixing that records (H)CA(N)HH and (H)NHH 2D spectra in the 4^th acquisition. Similarly, two residual ¹⁵N polarization pathways resulting from a second ¹⁵N–¹H back-CP (T^N_CP) are recovered for acquiring a fifth pair of 2D ¹³C- and ¹⁵N-edited spectra (H)CA(N)H and (H)NH in the 5^th acquisition period.

Figure8B shows ten 2D spectra of the NAVL dipeptide acquired simultaneously using H-MAeSTOSO-10 pulse sequence in Figure8A. Depending on the signal intensities of the 1^st increment (t₁ = 0), one can choose appropriate durations for HH RFDR or CC TOCSY mixing periods. In general, experiments with longer mixing times are less sensitive and require relatively higher spectral intensities compared to lower mixing times experiments. The signal intensities of H-MEIOSIS pathways recorded in the 1^st and 3^rd acquisition periods are relatively higher, hence a longer RFDR mixing time (1.47 ms) was used prior to 1^st and 3^rd acquisitions spectra as reported in Figure8B. The 4^th acquisition uses ¹⁵N residual signal from first ¹⁵N–¹H back CP, hence only 0.74 ms RFDR mixing time was used prior to the 4^th acquisition. On the other hand, the 5^th acquisition uses weak residual polarization after two ¹⁵N–¹H back CP periods, hence the spectra in the 5^th acquisition were recorded without RFDR mixing that gives (H)NH and (H)CA(N)H 2D data sets. For the spectra acquired in the 2^nd acquisition, the TOCSY mixing or WALTZ16 period (T^C_WALTZ) was set to 9.6 ms. Note that the TOCSY mixing can also be increased by setting the T^C_WALTZ period to multiples of 600 μs, as shown in Figure 5. The 2D spectra, (H)NH and (H)CA(N)H, recorded in 5^th acquisition are used as a reference to identify the cross peaks in the spectra acquired in the first four acquisition periods with RFDR and TOCSY mixing periods. The CC TOCSY transfer is mediated by the J-coupling Hamiltonian that achieves intra-residue CC transfer in the 2D spectra [(H)C(C)H and (H)N(CAC)H] as shown in Figure8B. At a RFDR mixing time of 1.47 ms, both ¹³C- and ¹⁵N-edited 2D spectra showed both intra- and inter-residue HH cross peaks between Val and Leu protons. On the other hand, more selective intra-reside HH transfer was observed with a RFDR mixing period of 0.74 ms. As expected, the TOCSY transfer only showed intra-residue CC correlations. The ¹H peak at 12.5 corresponds to carboxyl-terminal group of leucine⁵⁸. Note that the HH transfer efficiency can vary depending on the sample conditions and MAS rates. Although we used RFDR mixing periods for HH transfer, H-MAeSTOSO sequences can easily be modified with other mixing schemes⁵⁹. For example, DQ-SQ pulse sequences using BABA-XY16 recoupling is shown to be useful for assigning the proton spectra at 100 kHz MAS rate^60–61. The H-MAeSTOSO-10 spectra (Figure8) on NAVL were acquired using 24 scans per t₁ increment. Although four scans were sufficient to see most of the cross peaks, we used 24 scans to detect the lowest intensity cross-peaks (Vγ1 and Vγ2) in the (H)CHH and (H)C(C)H spectra. Of course, the application of H-MAeSTOSO-10 to proteins requires highly sensitive samples and careful optimization of the mixing periods. Note that in this case, long-range correlations might not be feasible due to limited signal-to-noise ratio.

4. DISCUSSION

Solid-state NMR experiments utilize 5–10% of the total experimental time for pulse sequence execution and acquisition of the signal (FID), while for 90–95% of the time the spectrometer is idle (2–4 seconds) waiting for the spin system to return to the equilibrium through longitudinal relaxation, T₁. Classical ssNMR methods are designed for single acquisition period, where each 2D or 3D experiment is acquired using a single coherence transfer pathway. The introduction of POE with the DUMAS, MEIOSIS, and MAeSTOSO strategies made it possible to recover orphan spin operators and utilize the full potential of nuclear spin polarization generated by CP^{29–34, 38, 46–47}. Using these strategies, we have successfully acquired two to eight 2D and 3D spectra using ¹³C detected multiple acquisition experiments. The expansion of the POE family to ¹H-detection experiments and ultrafast magic angle spinning is a stepping-stone to advance the potential of ssNMR to study challenging systems such as membrane proteins.

A key element in the POE pulse sequences is the preparation of both ¹³C and ¹⁵N polarization pathways using SIM-CP that satisfies the Hartmann-Hahn condition simultaneously for ¹³C and ¹⁵N with respect to the ¹H spin bath. We have found that the performance of SIM-CP at a 65kHz MAS rate is more efficient compared to slow MAS rates. Other elements of POE involve the recovery of both transferred and residual polarization pathways during NC SPECIFIC-CP (τ), and ¹³C–¹H and ¹⁵N–¹H back-CP periods (T^C_CH and T^N_NH). MEIOSIS and H-MAeSTOSO 2D pulse sequences were implemented using optimal parameters for SPECIFIC-CP transfer periods as described in Figure 3. Thus, the sensitivity of transferred polarization pathways is similar to conventional single acquisition experiments. On the other hand, the sensitivity of residual polarization pathways depends on the number of CP or SPECIFIC-CP periods, as well the efficiency of back CP periods. Figures 3 and 5 quantifies the signal intensities of transferred and residual polarization pathways. Both transferred and residual polarization pathways of NC SPECIFIC-CP, NH back-CP, and CH back-CP were successfully used for the simultaneous acquisition of four to ten two-dimensional experiments using H-MAeSTOSO-4 and HMAeSTOSO-10 experiments. Although, in this work the 2D ¹³C-edited spectra were acquired with T^C_CP set to 150 μs, one could use lower T^C_CP values to increase the residual polarization. For instance, as shown by Rienstra and co-workers, spectral overlap in (H)CH spectra can be reduced by using as low as a 50 μs back CP (T^C_CP) transfer period²¹. Note that the shorter back-CP periods achieve selective transfer at the cost of lower ¹H signal intensities for transfer pathways ¹³C to ¹H and ¹⁵N to ¹H. As shown in Figure 5C, at a 50 μs T^C_CP period one can retain up to 26% ¹³C residual polarization even in the presence of WALTZ16 pulses. WALTZ16 decoupling is routinely used for both protonated as well as perdeuterated protein samples at a range of MAS rates of 40 to 110 kHz^{24, 55, 57}. Note that in the absence of ¹³C and ¹⁵N WALTZ decoupling pulses it is possible to retain higher residual polarization, but such an approach leads to lower sensitivity and resolution.

One of the objectives of this work was to investigate the extent of the orphan spin-polarization pathways using optimal experimental conditions. We found that approximately 30 to 50% ¹⁵N residual polarization was obtained from ¹⁵N–¹H back CP periods; while ¹³C residual polarization drops to 10% due to CC TOCSY mixing during WALTZ16 pulses (Figure8). Note that the efficiency of CC TOCSY mixing can be improved under fast MAS rates. In fact, intra-residue CC correlations were obtained by WALTZ pulse sequence on fully protonated membrane protein at 100 kHz MAS rate⁵⁷. Also, the residual polarization from back-CP periods can also be preserved by applying other decoupling schemes in place of WALTZ that use back to back 180° pulses. In fact, we have recently developed a hybrid 2D and 3D TEDOR-NCX experiments in which the ¹⁵N z-polarization is stored under XY-4 ¹⁵N decoupling with 180° pulses⁴⁷. Similarly, continuous wave (CW) decoupling can also be used to preserve the residual polarization for simultaneous acquisition of multiple experiments^{45, 62–63}. In this work, we used fully protonated samples with ¹H acquisition times set to 8.3 ms. As reported by Pintacuda and co-workers^{57, 64}, even at 100 kHz MAS rate, ¹H acquisition times of 10 ms are sufficient for fully protonated protein samples. However, proton resolution can be further improved by using fast MAS rates together with protein perdeuteration, which increases the ¹H T₂ relaxation times. For samples with longer T₂ relaxation, ¹H acquisition times are typically set to 20 to 30 ms. Note that for longer acquisition times the residual polarization may be lower, decreasing the sensitivity for the subsequent acquisitions. The pulse sequences shown here lay the groundwork for future investigation of transferred and residual polarization pathways under various sample conditions, including ¹H dilution or perdeuteration as well as different decoupling conditions.

In principle, there are multiple combinations of pulse sequences that are possible for each of the H-DUMAS, H-MEIOSIS, and H-MAeSTOSO strategies by incorporating various homonuclear pulse sequence blocks with different mixing times³⁸. However, to best optimize the experimental time it is recommended to record the first increment (t₁ = 0) and then choose the appropriate homonuclear recoupling scheme and mixing times. Of course, as the number of back-CP periods increases, the signal decreases due to the lower amount of residual polarization. A key for the successful application of MAeSTOSO experiments is the use of larger number of scans to accumulate sufficient residual polarization to drive the 4^th and 5^th acquisitions³³. To this extent, it is important to code the low-sensitivity experiments, which require a larger number of scans and enable the accumulation of residual polarization, during the first two acquisitions of the pulse scheme and the more sensitive experiments for the following acquisitions. In this way, it is possible to accumulate sufficient residual polarization to drive the acquisition of the other experiments. The power of these experimental strategies were tested on the NAVL model peptide, as well as the SatP membrane protein. As shown in Figure8, for small peptides with resolved peaks, sequential assignment can be obtained from a single H-MAeSTOSO-10 (or H-MAeSTOSO-4) experiment with multiple mixing periods. For larger proteins such as SatP, the ¹H resolution in the 2D spectra is still challenging for resonance assignment. Nevertheless, these 2D fingerprints are useful in setting up the corresponding 3D experiments. This new class of POE are not alternatives to the existing approaches to speed up data acquisition; rather it provides a clever strategy to concatenate multiple experiments into a single pulse sequence to optimize the timing of the spectrometers and speed up data acquisition through the acquisition of multiple 2D and 3D experiments. We also anticipate further applications of POE at fast MAS rates (e.g., 100 kHz) using other sensitivity enhancement approaches such as DNP, PRE, and non-uniform sampling^65–67. In fact, recent work by Pintacuda, Reif, and Meier groups demonstrates the feasibility of resonance assignment of large proteins using 100–110 kHz MAS rates^{55, 68–70}.

Pines and Waugh were the first to implement multiple acquisition solid-state NMR, also known as proton enhanced experiments, using a rich ¹H spin bath of unlabeled organic molecules⁶³. POE pulse sequences are built on a similar philosophy with two basic observations in ssNMR, (1) the rich ¹H spin bath enables the simultaneous generation of ¹³C and ¹⁵N polarization pathways, and (2) storage of ¹⁵N z-polarization which is unique for solid samples where the T1 relaxation times are significantly longer ²⁹. In the previous POE pulse sequences, we utilized ¹³C and ¹⁵N residual polarization pathways resulting from NC cross polarization periods via MEIOSIS and MAeSTOSO experiments at slow spinning speeds. In this work, we systematically used similar experimental protocols with respect to slow MAS experiments to understand the relative efficiency of SIM-CP and SPECIFIC-CP residual polarization pathways at fast MAS rates. Furthermore, we also designed new strategies for recovering residual polarization from ¹³C–¹H or ¹⁵N–¹H back CP transfer periods. The ‘afterglow’ pulse sequence developed by Traaseth lab also exploits residual ¹⁵N polarization of SPECIFIC-CP and simultaneously records two 2D spectra NCA and NCO using CP as a preparation period⁷¹. Similarly, the residual ¹⁵N polarization resulting from hetero-nuclear NC mixing sequences has also been utilized at fast MAS rates⁴³. Proton detected H-DUMAS and H-MEIOSIS experiments are indeed extensions of its ¹³C-detected counterparts namely, DUMAS and MEIOSIS, and other versions have also been proposed by other research groups using proton detection experiments at slow and fast MAS rates^42–44.

Finally, we would like to point out that POE can also be modified for dual receiver experiments using ¹³C and ¹H detection. In fact simultaneous acquisition of ¹H and ¹³C detected dual receiver experiments have already been reported by other research groups⁷². Multi-nuclear acquisition experiments can also be implemented using single receiver systems as demonstrated in UTOPIA experiment⁷³. From a technical point of view, at slow MAS rates POE may cause RF heating due to high power decoupling (~100 kHz), hence, it is strongly recommended to use low-E or E-free probes MAS probes³⁴. On the other hand, the low power decoupling (~10 kHz) at fast MAS rates make the POE more promising for multi-acquisition pulse sequences.

5. CONCLUSIONS

In this work, we have developed a new set of ¹H detected POE (Polarization Optimized experiments) that simultaneously acquire two to ten two-dimensional experiments. POE represent a general strategy for multiple acquisitions of almost all types of triple-resonance ssNMR experiments at slow and fast MAS rates. Proton detected H-DUMAS and H-MEIOSIS experiments are simple extensions of its ¹³C-detected counterparts namely, DUMAS and MEIOSIS. On the other hand, MAeSTOSO experiments require synchronized WALTZ decoupling sequences to preserve the residual polarization pathways. The proposed methods can be extended to theeand four-dimensional experiments together with multi-receiver technology. These new class of experiments are not alternatives to the existing approaches to speed up data acquisition; rather it provides a clever strategy to concatenate multiple experiments into a single pulse sequence to optimize the timing of the spectrometers and speed up data acquisition through the acquisition of multiple 2D and 3D experiments. When applied in concert with other fast acquisition and sensitivity enhancement techniques (e.g., Dynamic Nuclear Polarization, Paramagnetic Relaxation Enhancements, etc.) this approach can further push the boundaries of ssNMR applications to structural biology.

Highlights.

• New polarization optimized experiments for proton detection of biomacromolecules

• Proton detection enables up to 10 experiments acquired with one pulse sequence

• Application to six transmembrane protein transporter SaTP