NMR Assignment of Methyl Groups in Immobilized Proteins Using Multiple-Bond 13C Homonuclear Transfers, Proton Detection, and Very Fast MAS

In nuclear magnetic resonance spectroscopy of proteins, methyl protons play a particular role as extremely sensitive reporters on dynamics, allosteric effects, and protein–protein interactions, accessible even in high-molecular-weight systems approaching 1 MDa. The notorious issue of their chemical shift assignment is addressed here by a joint use of solid-state ¹H-detected methods at very fast (nearly 100 kHz) magic-angle spinning, partial deuteration, and high-magnetic fields. The suitability of a series of RF schemes is evaluated for the efficient coherence transfer across entire ¹³C side chains of methyl-containing residues, which is key for establishing connection between methyl and backbone ¹H resonances. The performance of ten methods for recoupling of either isotropic ¹³C–¹³C scalar or anisotropic dipolar interactions (five variants of TOBSY, FLOPSY, DIPSI, WALTZ, RFDR, and DREAM) is evaluated experimentally at two state-of-the-art magic-angle spinning (55 and 94.5 kHz) and static magnetic field conditions (18.8 and 23.5 T). Model isotopically labeled compounds (alanine and Met-Leu-Phe tripeptide) and ILV-methyl and amide-selectively protonated, and otherwise deuterated chicken α-spectrin SH3 protein are used as convenient reference systems. Spin dynamics simulations in SIMPSON are performed to determine optimal parameters of these RF schemes, up to recently experimentally attained spinning frequencies (200 kHz) and B₀ field strengths (28.2 T). The concept of linearization of ¹³C side chain by appropriate isotope labeling is revisited and showed to significantly increase sensitivity of methyl-to-backbone correlations. A resolution enhancement provided by 4D spectroscopy with non-uniform (sparse) sampling is demonstrated to remove ambiguities in simultaneous resonance assignment of methyl proton and carbon chemical shifts.

1 Introduction

For protein studies by nuclear magnetic resonance (NMR), amide and methyl ¹H resonances are the most commonly exploited. The latter ones are particularly convenient due to the magnetic equivalence of three ¹H spins (thus threefold sensitivity gain), and enhanced longitudinal relaxation, both caused by fast methyl rotation. In solution NMR, multiple-quantum correlations (HMQC) can be employed to select only the slowly relaxing methyl ¹H–¹³C coherences (Tugarinov et al., 2003). When combined with extreme ¹H dilution by deuteration and selective methyl amino acid labeling, the approach allows to study the local dynamics and protein interactions for systems close to MDa molecular weight (MW) (Rosenzweig and Kay 2014; Huang and Kalodimos 2017; Boswell and Latham, 2018).

A prerequisite for the interpretation of NMR data at atomic resolution is a unique, site-specific mapping of chemical shifts to individual atoms. However, conventional resonance assignment strategies for proteins are centered around backbone ¹H resonances (Sattler, Schleucher, and Griesinger 1999). Chemical shifts of methyl spins, which are peripheral with respect to backbone, are clearly challenging to assign in a systematic way, particularly if detached methyl sites are the only nondeuterated spins. Tailored experiments were developed to correlate methyl to amide frequencies; however, in addition to long coherence transfers involved and thus intrinsically low sensitivity, they require full ¹H occupancy of (detected) amide sites (Tugarinov and Kay 2003b). The increased proton density is detrimental to ¹H resolution and sensitivity particularly in large-MW proteins. Alternative strategy relies on correlation to backbone ¹³C′ and ¹³Cα spins (Tugarinov and Kay 2003a), which can be accomplished in the absence of amide ¹H, but the issue of extended coherence transfer pathway persists.

Mutagenesis is commonly employed to address the methyl assignment issue (Amero et al., 2011). In this approach, single point mutations are introduced, and fingerprint ¹H–¹³C HMQC spectra are recorded and compared for as many samples as methyl-containing amino acids. The major disadvantages are the labor and cost of isotope-enriched compounds needed to prepare typically tens of samples. Other pitfalls are ambiguities due to overlap of ¹H–¹³C cross-peaks, or global chemical shift changes induced by mutations (Gorman et al., 2018).

An orthogonal approach is based on the observation of ¹H–¹H methyl proximities (NOEs), which can aid the assignment given the presence of the 3D structure determined using other techniques (Gorman et al., 2018). These data are interpreted either by a spectroscopist or, more effectively, by dedicated algorithms intensively developed over the last few years (Schmidt and Güntert 2012, 2013; Pritišanac et al., 2019; Pritišanac, Alderson, and Güntert 2020). The approach depends critically on the NOESY data quality and performs well mostly for methyl ¹H spins experiencing a dense network of interactions in rigid regions.

Immobilized proteins, such as in amyloid fibrils, sedimented large-MW aggregates, lipid bilayers, or microcrystals, are amenable to solid-state NMR (ssNMR), since this method does not experience high-MW slow-tumbling limitations characteristic of the solution counterpart (Tycko 2011; Miao and Cross 2013; Baker and Baldus 2014; Ladizhansky 2017; Linser 2017; Mandala, Williams, and Hong 2018; van der Wel 2018; Wiegand 2020). The possibility of detection of methyl ¹H resonances at high resolution in ssNMR was investigated already at low magic-angle spinning (MAS) frequencies (up to 20 kHz), when coupled to high to extreme ¹H dilution by deuteration (Agarwal et al., 2006; Agarwal and Reif 2008; Asami, Schmieder, and Reif 2010; Asami et al., 2012; Asami and Reif 2013; Mainz et al., 2013). The advent of fast-spinning (above 40 kHz) MAS probes allowed narrow methyl ¹H lines at significantly higher proton content, either when labile amide ¹H sites are fully reprotonated (Lewandowski et al., 2011; Linser et al., 2011), or when Ile, Leu, and Val residues are also selectively and nonrandomly (100% CH₃ or CHD₂) protonated at methyl sites (Huber et al., 2011; Agarwal et al., 2014; Andreas, et al., 2015a; Kurauskas et al., 2016; Gauto et al., 2019), or when Leu and Val residues are reverse-labeled in an otherwise deuterated matrix (“proton clouds”) (Sinnige et al., 2014). Quite importantly, Schanda and co-workers recently showed that at 55–57 kHz MAS and B₀ = 14.1 T, the ¹³CH₃ isotopomer yields significantly higher sensitivity compared to ¹³CHD₂ labeling, and only at a minor loss of ¹H resolution (Kurauskas et al., 2016). Complete elimination of detrimental sensitivity and resolution effects of strong ¹H–¹H dipole interactions by MAS remains a challenge, particularly in methyl-dense protein regions, and would require yet unavailable MAS rates (above 250 kHz) (Kai Xue et al., 2017; K Xue et al., 2018; Kai Xue et al., 2019, Xue et al., 2020). Nevertheless, even in non-deuterated but relatively small proteins, resolved methyl ¹H–¹³C correlations (of ¹H linewidths of about 150 Hz) were obtained with 100–111 kHz MAS and high static magnetic fields (23.5 T) (Andreas et al., 2016).

There are few examples of de novo assignment of methyl ¹H resonances in ssNMR, and frequently the assignment has been aided by either correlations to previously available ¹³C shifts using either dipolar (Agarwal and Reif 2008; Asami and Reif 2012) or scalar-based (Andreas et al., 2015b) (H)CCH-type spectra, or transferred from solution NMR by comparison to relatively uncomplicated ¹H, ¹³C-CP-HSQC (Huber et al., 2011). In an early work, a strategy was proposed that bases on correlation of methyl resonances to backbone ¹³Cα and C′ spins using (H)CCH-TOBSY experiment, and detection of dilute methyl ¹H spins, but was found challenging for Leu and Ile residues due to low efficiency of multi-¹³C-bond transfers (Agarwal and Reif 2008). A systematic strategy relying on correlations of side-chain ¹³C to backbone ¹⁵N and ¹H, originally proposed by Linser for perdeuterated proteins at slow MAS (25 kHz) (Linser 2011), and more recently at fast MAS (55–60 kHz) with rotor-asynchronous MOCCA mixing (Kulminskaya et al., 2016; Vasa et al., 2018), can be readily adapted for residues selectively 100% reprotonated at methyl sites, but their efficiency with respect to methyl-containing spin systems was not investigated in detail. In the case of non-deuterated (“fully protonated”) proteins, such a long coherence transfer encounters sensitivity limitations, and more practical is a “two-hop” strategy in which side-chain ¹H and ¹³C resonances are first correlated to alpha ¹H and ¹³C spins using (H)CCH and H(C)CH experiments (Andreas et al., 2016; Stanek et al., 2016), and subsequently to amide ¹H and ¹⁵N shifts using a combination of ¹³Cα–¹⁵N–¹H^N and ¹⁵N–¹³Cα–¹Hα correlations (Zhou et al., 2007; Stanek et al., 2016). Both aforementioned approaches were proposed for a general use (for all amino acid residue types), without a particular emphasis on the methyl assignment. In fact, the literature data on non-deuterated microcrystalline protein GB1 (Andreas et al., 2016) show that WALTZ mixing at MAS rate ν_R = 111 kHz and B₀ = 23.5 T leads to an effective ¹³C^x→¹³Cα transfer only in two-, pseudo-two-, and three-aliphatic ¹³C-spin systems (Figures 1A,B), and becomes problematic for terminal spins in pseudo-three (Glu, Gln)- and larger ¹³C spin systems (Figure 1C). Unfortunately, the latter include valine, leucine, and isoleucine, which additionally suffer from branching of ¹³C chain at β- or γ-positions (Figure 1D). This raises questions of feasibility of methyl resonance assignment under less sensitivity-favorable sample or hardware conditions, and calls for a careful optimization of ¹³C mixing scheme at arising new MAS conditions and B₀ fields.

FIGURE 1

FIGURE 1. Transfer efficiencies from side-chain ¹³C (β, γ, … ) to ¹³Cα spin for various residues in microcrystalline U-¹³C,¹⁵N-labeled GB1 protein using ¹³C WALTZ-16 mixing of 15 ms duration, with MAS of ν_R = 111 kHz (in a 0.7-mm rotor) and on a 23.5-T spectrometer. Intensity of ¹³C^X-¹³Cα-¹Hα cross-peaks in (H)CCH spectrum (Andreas et al., 2016; Stanek et al., 2016) was normalized to a sum of 1.0 for each residue, and averaged within each amino acid type. The data were grouped into (A) two- and pseudo-two-, (B) three- and pseudo-three-, and (C) four- and five-aliphatic ¹³C spin systems (disregarding backbone and side-chain carbonyl and aromatic ¹³C spins). (D) Topologies of ¹³C side chains for six amino acid residue types containing a methyl group, from left to right: least to most challenging for methyl resonance assignment.

In this work, we critically assess the efficiency of several ¹³C mixing schemes for methyl ¹H and ¹³C resonance assignment at fast MAS; determine their optimal range of B₀, MAS frequency, and ¹³C RF strength within and beyond currently available regimes; and explore alternative ¹³C labeling to further boost sensitivity of the approach.

2 Materials and Methods

2.1 Sample Preparation

2,3-¹³C (99%)-labeled crystalline alanine was purchased from Eurisotop (France) and manually packed into a Bruker 1.3-mm MAS rotor.

N-formylated microcrystalline uniformly ¹³C,¹⁵N-labeled Met-Leu-Phe tripeptide (“fMLF”) was purchased from Giotto Biotech (Sesto Fiorentino, Italy) and manually packed into Darklands OÜ 0.81 mm MAS rotor.

The plasmid coding for the Src homology 3 (SH3) domain (965–1,025) of chicken α-spectrin (gene SPTAN1, Uniprot P07751) in a pET3a vector was a kind gift of Dr. T. Schubeis (High Field NMR Centre in Lyon, France). The insert was subcloned into a modified pET28a vector that resulted in a construct including an N-terminal His₆-tag followed by a SUMO solubility tag. This allowed us to express His₆-SUMO-SH3 fusion protein.

For preparation of the NMR samples, transformed Escherichia coli BL21 (DE3) cells were grown in M9 D₂O media supplemented with 1 g/L of ¹⁵NH₄Cl (Cortecnet, France) and 3 g/L of (²H,¹³C)-glucose (Cortecnet, France) as the sole nitrogen and carbon sources, respectively, following the established procedure (Tugarinov, Kanelis, and Kay 2006). Labeling of the Ile, Leu, and Val side chains was achieved by the addition of the amino acid precursors 1 h prior to the induction with 1 mM IPTG. For Leu and Val residues, we used two kinds of precursors that yield continuous ¹³C chains from methyl (either Cδ or Cγ) to Cα and C′ atoms, provided that ¹³C-enriched glucose is also employed (Goto et al., 1999; Tugarinov and Kay 2003a). The specific labeling patterns and the origin of particular nuclei are shown in Supplementary Figure S1. SH3 sample with branched side chains of leucine and valine residues (hereafter referred to as “ILV-C5” sample) was prepared using 100 mg/L of 1,2,3,4,4′-¹³C-3-²H-labeled α-ketoisovaleric-acid (sodium salt, Eurisotop, France, catalogue number CDLM-4418-PK) as a Leu and Val precursor. The corresponding sample with linearized ¹³C-side chains of Leu/Val residues (referred to as “ILV-C4” sample) required the addition of 100 mg/L 1,2,3,4-¹³C-3,4′,4′,4′-²H-labeled α-ketoisovaleric acid (sodium salt, Eurisotop, France, catalogue number CDLM-8100-PK). For both samples, 60 mg/L of 1,2,3,4-¹³C-3,3-²H-labeled α-ketobutyric acid (sodium salt, Eurisotop, France, catalogue number CDLM-4611-PK) was used as the precursor of isoleucine residues with uniform ¹³C enrichment (Supplementary Figure S1). The cells were grown at 24°C for 18 h after the induction.

The purification protocol of SH3 protein was modified with respect to the original one (Pauli et al., 2000, 2001; Chevelkov et al., 2006). Instead, a standard protocol involving His-trap affinity column and HiLoad 16/60 Superdex 75 gel filtration column (GE Healthcare) was employed. Briefly, we used 50 mM HEPES, 200 mM NaCl, and 1 mM DTT (pH 7.4) supplemented with 20 mM imidazole as a lysis buffer, and the same buffer including additionally 400 mM imidazole was used to elute a protein from the His-trap column. The cleavage of the His₆-SUMO tag was achieved with a custom-made Ulp1 protease (Reverter and Lima 2009) in a lysis buffer lacking imidazole. The reaction was monitored with the SDS-PAGE and, once completed, the protein was passed through the His-trap column again to remove SUMO as well as His-tagged Ulp1 protease. Cleaved SH3 protein was dialyzed at 4°C overnight against the Superdex 75 running buffer (20 mM citric acid, 150 mM NaCl, pH 3.5), concentrated, and purified on a gel filtration column.

To obtain solid-state NMR protein samples, fractions containing pure SH3 were pooled, concentrated to approximately 10 mg/mL using Vivaspin 3-kDa cutoff centrifugal concentrators (Sartorius) and extensively dialyzed against 100 mM (NH₄)₂SO₄ adjusted to pH 3.5 with sulfuric acid. Finally, to crystallize the protein, ammonia water solution was added dropwise to reach pH 7.5. Obtained turbid solutions were stored in a refrigerator (4°C) for approximately a week for a slow buildup of microcrystals. Crystallization efficiency was estimated to 85% by a spectrophotometric measurement of protein concentration decrease in the supernatant. The suspension was mixed with 1 M CuNa₂EDTA 9:1 v/v (effective c_Cu2+ = 100 mM), and left for impregnation of SH3 crystals with paramagnetic Cu²⁺ ions for 3 days. In the case of the SH3 “ILV-C4” sample, TSP was added to the supernatant at an effective concentration of 10 mM for chemical shift calibration reference. About 2 mg of protein was transferred to a Bruker 1.3-mm MAS rotor by 30 min of ultracentrifugation at an average acceleration of 96,500 g. Each Darklands OÜ 0.81-mm MAS rotor was filled with approximately 0.5 mg of protein in five equal parts by stepwise packing in a tailored ultracentrifuge adapter experiencing an average acceleration of 135,000 g for 30 min.

2.2 Nuclear Magnetic Resonance Spectroscopy

The NMR ¹³C–¹³C 2D correlation experiments on model compounds (alanine and fMLF) were performed using a standard RF irradiation scheme shown in Figure 2A. A collection of homonuclear mixing schemes was employed as detailed below. For each case, a series of 2D spectra with gradually incremented mixing time was acquired, in the range suggested by literature and spin dynamics simulations, if permitted by probe RF circuity.

FIGURE 2

FIGURE 2. RF irradiation schemes of NMR experiments. (A) ¹³C–¹³C 2D correlation experiment with mixing two variants of ¹³C mixing expanded in (B) and (C). All schemes act on the ¹³C magnetization preceded and followed by a z-filter (τ_z) (B) with the exception of DREAM (C), which operates on the ¹³C single-quantum coherence. (D) Pulse diagram for 3D (H)C^MET(CC)(CA)NH, with the optional evolution period for methyl ¹H coherences enclosed in brackets (for the 4D variant). The optional ¹H spin echo immediately before acquisition was only applied for Bruker 1.3-mm probe to eliminate probe ¹H background signal. Open and solid rectangles represent 90° and 180° high-power pulses, respectively, and a solid bell shape denotes a methyl ¹³C-selective Q3 refocusing pulse (Emsley and Bodenhausen 1992) of 942.7 μs duration and peak RF of ν₁ = 2.56 kHz. ¹³C mixing is marked in red, and cross-polarization transfer steps are labeled and color-coded as maroon, purple, and lime green simultaneous irradiation on pairs of respective channels for H→C, C→N, and N→H transfers. ¹H low-power WALTZ (Shaka, Keeler, and Freeman 1983) decoupling is marked in orange, while heteronuclear (¹³C or ¹⁵N) frequency-swept TPPM decoupling (Bennett et al., 1995) is marked in gray. MISSISSPPI solvent suppression (Zhou and Rienstra 2008) is marked in cyan. Changes in ¹³C offset (Ω) are denoted with vertical arrows. Unless indicated otherwise, all pulse phases are set to x. The following phase cycling was employed: ϕ₁ = y, ϕ₂ = y, ϕ₃ = {2 (x), 2 (−x), 2 (y), 2 (−y)}, ϕ₈ = {x, −x}, ϕ_rec. = {x, −x, −x, x, y, −y, −y, y} for 2D experiment with inset (B); ϕ₁ = {y, −y}, ϕ₂ = {2 (x), 2 (−x)}, ϕ₃ = {4 (x), 4 (−x)}, ϕ₈ = {2 (x), 2 (−x)}, ϕ_rec. = {x, −x, x, −x} for 2D DREAM experiment (A + C); ϕ₁ = {4 (y), 4 (−y)}, ϕ₂ = x, ϕ₃ = {x, x, y, y}, ϕ₄ = y, ϕ₅ = {8 (x), 8 −x)}, ϕ₆ = {y, –y}, ϕ₇ = x, ϕ₈ = -y, ϕ_rec. = {x, –x, –x, x, –x, x, x, –x, –x, x, –x, x, –x, –x, x} for 3D experiment (D) with recoupling block (B); and ϕ₁ = {8 (y), 8 (−y)}, ϕ₂ = {4 (x), 4 (−x) }, ϕ₃ = {2 (x), 2 (y), 2 (−x), 2 (−y)}, ϕ₄ = x, ϕ₅ = x, ϕ₆ = {y, −y}, ϕ₇ = {x}, ϕ₈ = {4 (x), 4 (– x)}, ϕ_rec. = {x, –x, –x, x, –x, x, x, –x, –x, x, –x, x, –x, –x, x} for 3D DREAM experiment (D + C). Pulse with phase marked with an asterisk are involved in quadrature detection according to States-TPPI procedure.

The experiments for site-specific assignment of methyl ¹³C and ¹H resonances in SH3 protein that provide correlations to backbone amide ¹⁵N and ¹H chemical shifts [3D (H)C(CC)(CA)NH] were straightforwardly adopted from the literature (Linser 2011) and are shown in Figure 2B. Despite long methyl ¹H transverse relaxation times observed, the experiments employ ¹H–¹³C cross-polarization (CP) instead of INEPT-type transfer, since the latter would suffer from inefficiency of conversion of anti- to in-phase ¹³C coherence caused by concurrent evolution of two (passive) ¹J_CH couplings in CH₃ moieties. Implementations of presented experiments for Bruker spectrometers are freely available from the community-based repository Zenodo as detailed below.

For the site-specific evaluation of experimental performance with various ¹³C mixing schemes, resolution provided by 3D spectra was required, preventing acquisition of mixing time series for SH3 protein. Optimal mixing times were, however, possible to determine using first-increment 1D optimizations prior to 3D data acquisition.

The series of 2D ¹³C–¹³C correlation experiments on 2,3-¹³C-alanine was carried out on a Bruker Avance III spectrometer operating at ¹H and ¹³C frequency of 600.1 and 150.9 MHz, respectively, equipped with a Bruker 1.3-mm H/C/N MAS probehead. The sample has been spun at 55,555 ± 10 Hz and controlled by a Bruker MAS-II unit, without temperature stabilization. ¹H, ¹³C pulses and power during ¹H to ¹³C cross-polarization have been carefully calibrated prior to experiments. Detailed information on pulse lengths, RF amplitude, spectral windows, etc. is provided in Supplementary Table S1.

2D experiments on ¹³C,¹⁵N-fMLF tripeptide and 3D and 4D experiments on two SH3 protein samples were performed on a Bruker Avance III HD spectrometer operating at a ¹H, ¹³C, and ¹⁵N frequency of 799.7, 201.1, and 81.0 MHz, respectively, equipped with a 0.81-mm H/C/N/D MAS probehead developed by Ago Samoson’s group (Darklands OÜ, Estonia). NMR data were acquired at two MAS frequencies, ν_R = 55.5 and 94.5 kHz (98 kHz for fMLF).

3D experiments on “ILV-C5”-labeled SH3 protein were also performed on a Bruker NEO 23.5-T spectrometer (in CRMN Lyon, France) equipped with a Bruker 1.3-mm H/C/N/D MAS probe, and MAS-III and BCU-II spinning and temperature stabilization units.

For the experiments on ¹³C,¹⁵N-fMLF magic angle setting has been set using KBr sample prior to the actual series. In this case, no temperature stabilization device was used. ¹H and ¹³C 90° pulse lengths were carefully calibrated using 1D ¹³C-detected CP experiment.

Prior to experiments on SH3 samples, a careful magnet shimming was performed to maximize reliability of linewidth measurements. The classical shimming protocol employing adamantane sample is very time-consuming in 0.81-mm MAS rotors; thus, a sample of silicon grease was used first, leading to full-width at half-height (FWHH) of ¹H line of 16 Hz at 50 kHz MAS. Subsequently, shim currents were refined using adamantane sample resulting in an FWHH of the low-field ¹³C resonance of 1.1 Hz at 50 kHz MAS under low-power ¹H decoupling.

Magic angle was finely adjusted directly on the protein sample at the target MAS frequency by maximizing the intensity of the ¹H signal in the (H)NH 1D experiment followed by a 5-ms ¹H spin echo. Sample temperature was stabilized to 20°C using Darklands OÜ VT controller. Thermocouple readout was calibrated to sample temperature by measurement of the ²⁰⁷Pb chemical shift of Pb(NO₃)₂ at the exactly same cooling gas flow, spinning speed, and thermocouple target temperature. ¹H, ¹⁵N 90° pulse lengths have been carefully calibrated for each sample and spinning speed using the (H)NH experiment. ¹³C 90° pulse length was calibrated using the (H)CONH experiment. RF amplitude of hard and soft pulses, as well as of decoupling and recoupling schemes was automatically calculated in a pulse program based on widths and powers of reference high-power 90° pulses. ¹H–¹³C, ¹³C–¹⁵N, and ¹⁵N–¹H cross-polarization power was optimized directly using the first increment of the (H)C(DIPSI)(CA)NH experiment, and propagated to all 3D (and 4D) experiments at each sample and spinning condition. Fast recycling (0.3 or 0.4 s interscan delay) was applied to improve sensitivity, taking advantage of longitudinal ¹H relaxation enhancement by paramagnetic Cu²⁺ ions (Ganapathy et al., 1981; Wickramasinghe et al., 2007, 2009). Details on NMR data acquisition for fMLF tripeptide and SH3 protein are provided in Supplementary Tables S1, S3, respectively.

B₀ was not stabilized in either case, but the field drift was monitored using 1D ¹H spectra between experiments and had a negligible effect on data (e.g., a total 11 Hz ¹H downfield drift over 10 days on an 18.8-T spectrometer for “ILV-C5” SH3 sample at 94.5 kHz MAS). The stability of CP conditions was monitored using 1D (H)(CA)(N)H and (H)(C)(DIPSI)(CA)(N)H (for proteins) or ¹³C-detected 1D CP (for fMLF) experiments.

2D and 3D data were Fourier processed in Bruker TopSpin using parameters reported in Supplementary Tables S2, S4. The non-uniformly sampled 4D data were converted using in-house written script bruk2ssa (courtesy of M. Górka), and processed using signal separation algorithm for distortion-free spectral reconstruction (Stanek, Augustyniak, and Koźmiński 2012). All spectra were analyzed in NMRFAM-Sparky (Lee, Tonelli, and Markley 2015).

2.3 Homonuclear ¹³C Mixing Schemes

The key element determining the performance of experiments shown in Figure 2 is ¹³C mixing. For this comparative study, we have made a selection of literature RF designs based on the following criteria: (1) ¹³C should occur primarily between bonded ¹³C spins by recoupling of ¹³C–¹³C scalar (J) or dipolar (D) interactions, (2) RF requirements are acceptable for a typical “fast” MAS probe, and (3) they pose no danger to the hydration of fragile protein samples, e.g., by an extended period of high-power ¹H irradiation. For the last reason, we did not consider second-order recoupling schemes suitable for fast MAS, such as MIRROR (Scholz et al., 2008), PARIS (Weingarth et al., 2009), SHANGHAI (Hu et al., 2011), or CORD/CORD-RFDR (Hou et al., 2013; Lu et al., 2015). In the presence of relaxation, the efficiency of these sequences deteriorates with faster MAS, concomitantly to increasing efficiency of averaging of H–H dipolar interactions. Protein deuteration, which is highly recommended for resolution in large proteins, is also clearly incompatible with second-order recoupling mechanism (De Paëpe, Lewandowski, and Griffin 2008) as it prohibitively dilutes the proton interaction network.

2.3.1 TOBSY

TOBSY was designed based on average Hamiltonian theory (AHT) and symmetry properties of particular interactions (Levitt 2007), imposing that chemical shift anisotropy (CSA), isotropic chemical shifts (CS), and dipolar interaction terms are suppressed in the zeroth- and first-order Hamiltonians, and the zeroth-order term stems only from the isotropic scalar interaction (Baldus and Meier 1996; Hardy, Verel, and Meier 2001). In a general design denoted CN^ν_n (Edén and Levitt 1999), a primitive multi-pulse block C is repeated N times over n rotor cycles with the gradual phase incrementation in Δϕ = 2πν/N steps (Figure 3A). Low-power TOBSY schemes suitable for fast MAS employ POST primitive block (90_ϕ360_ϕ+π270_ϕ), and use N = 9, ν = 1, and n = 9p ± 3 (where p is an integer), e.g., C9¹₂₁, C9¹₂₄, and C9¹₃₀, to retain only assumed interaction (Tan et al., 2018). The schemes with particular n differ in robustness to dipolar C–H interactions and susceptibility to CSA (in higher-order AHT terms). TOBSY requires an RF strength of ν₁ = 2N/n ν_R and thus is readily applicable in typical fast MAS probes for n > 18. Here, we employed sequences with n = 24, 30, 33, 39, and 48, with mixing time either varied between 0 and 50 ms (in 2D series) or optimized for maximum transfer (in 3D experiments).

FIGURE 3

FIGURE 3. RF schemes of ¹³C mixings compared in this study. (A) TOBSY C9¹_n variants with variable n. (B) WALTZ-16, a representative of TOCSY class. The primitive element is composed of phase-alternated 90, 180, 270, and 360° pulses, depicted as rectangles color-coded from light gray to black. (C,D) Transfer schemes primarily recoupling ¹³C dipolar interactions: (C) finite-pulse RFDR and (D) DREAM.

2.3.2 TOCSY

Isotropic mixing schemes developed for ¹³C mixing in solution NMR (Bax, Clore, and Gronenborn 1990; Kay et al., 1993), hereafter referred to using a general term TOCSY, retain only the Hamiltonian term associated with scalar (J) interaction between ¹³C nuclei (in the presence of the overall tumbling). For solids under MAS, all isotropic interactions are preserved; thus, these RF schemes induce the evolution of J_CC interactions while suppressing the evolution of J_HC and isotropic ¹³C chemical shifts. The proper treatment of anisotropic interactions is not ensured at all; however, at sufficiently fast MAS, their contribution is supposedly small. Simple arguments advocated for the use of WALTZ-16 (Shaka, Keeler, and Freeman 1983) mixing synchronized with rotation (τ₉₀ = τ_R; Figure 3B), and basic properties were verified using 2-¹³C-spin SIMPSON simulations (Andreas et al., 2016). Classical TOCSY mixing schemes that are robust with respect to large chemical shift offsets, namely, DIPSI-3 (Shaka, Lee, and Pines 1988) and FLOPSY-16 (Kadkhodaie et al., 1991), were also employed here. In solution NMR, TOCSY designs are ranked according to the figure of merit, i.e., the bandwidth with respect to applied RF strength, but such a ranking is of limited relevance here since MAS probes can easily generate sufficient ν₁. Although not required, we retained rotor synchronization (ν₁ = ¼ ν_R) in experiments for three selected TOCSY sequences, but tested other RF conditions in spin dynamics simulations. Mixing time was varied (for alanine and fMLF) or optimized in 1D experiments (for proteins) in steps of 188.448, 217.32, and 96.0 τ_R (assuming τ_R = τ^C₉₀) for FLOPSY, DIPSI, and WALTZ, respectively.

2.3.3 RFDR

Finite-pulse ¹³C RFDR (Bennett et al., 1998) recouples dipolar ¹³C interactions in the first-order average Hamiltonian by the application of high-power π pulse each rotor cycle (Figure 3C). Although the sequence is known to suffer from dipolar truncation (Griffin 1998), this is not necessarily a disadvantage for intraresidue ¹³C transfers. RFDR requires ν₁ > ½ ν_R for pulses partially covering the mixing time; in practice, ν₁ on the order of ν_R is preferred. The ratio of τ₁₈₀/τ_R determines the scaling factor of dipolar interaction (nominally of about 2 kHz for a bonded ¹³C–¹³C spin pair), and in this study, we used ν₁ = 150 (for fMLF) and 100 kHz (otherwise). The following is the phase cycle (xy8) of the π pulse: x, y, x, y, y, x, y, x, as suggested previously (Shen et al., 2012). Mixing time was optimized in steps of multiples of 8τ_R between 0 and 50 ms.

2.3.4 DREAM

DREAM recouples dipolar ¹³C interactions in the first-order average Hamiltonian using an adiabatic pulse for introducing HORROR condition (ν₁ = ½ ν_R, Figure 3D) (Verel et al., 1998; Verel, Ernst, and Meier 2001). DREAM is commonly employed with slow (Pauli et al., 2001) and fast MAS (Penzel et al., 2015) to trigger ¹³Cα→¹³Cβ or ¹³C′→¹³Cα coherence transfer, for which RF and offset conditions can be straightforwardly determined. Transfers in multi-spin systems were studied and conditions were optimized at “slow” MAS (Westfeld et al., 2012); however, the proper order of HORROR conditions cannot in general be satisfied due to characteristic chemical shifts in ¹³C side chains. Additionally, by its nature, DREAM cross-peak intensity is negative with respect to the origin coherence, which potentially causes destructive interferences within a spin system or due to spectral overlaps. The RF shape was defined as usual, $v\left(t\right)=\overline{v_1}+d^{eff}\text{tan}\left(\frac{2}{\tau }{\text{tan}}^{-1} \left(\Delta /d^{eff}\right)\left(t-\tau /2\right)\right)$, where $d^{eff}$, Δ, τ, and $\overline{v_1}$ denote the effective dipolar coupling (averaged over crystal orientations and decreased due to dynamics), modulation depth, mixing time, and average RF strength, respectively. Directionality of the transfer was selected here with increasing RF over mixing time. We employed the parameters recommended for DREAM adiabatic RF modulation at fast MAS, namely, the modulation depth of 1/5 ν_R, dipolar C–C coupling of 1 kHz, and average RF ${\overline{v}}_1$ to nominally ½ ν_R, which potentially allows DREAM to cover the entire aliphatic ¹³C band at typical high fields (e.g., 18.8 T). In addition to the mixing time (varied up to 10 or 25 ms for alanine, or fMLF and SH3, respectively), ¹³C offset for DREAM was also optimized for best ¹³C^met→¹³Cα transfer in the 2D series for alanine and fMLF, and using 1D first-FID of sequence shown in Figure 2D.

2.4 Spin Dynamics Simulations Using SIMPSON

Spin dynamics simulations have been performed using SIMPSON (version 4.0.0c) (Bak, Rasmussen, and Nielsen 2011; Tošner et al., 2014). Powder averaging has been performed with 1,848 {α, β, γ} Euler angles that describe the orientation of the molecule in the rotor frame. A total of 168 {α, β} angle pairs have been selected using REPULSION algorithms (Bak and Nielsen 1997) and 11 γ angles have been regularly sampled from 0 to 360°. Spin systems have been generated using the SIMMOL package (Bak et al., 2002).

For simulations in the case of a model four-spin system (Hα–Cα–Cβ–Hβ), dipolar spin couplings have been generated based on distances in L-alanine structure (ref. code LALNIN61 in CCS database), and only α and β carbons and protons have been considered. ¹J_CC and ¹J_HC couplings have been set to 33 and 145 Hz, respectively. In addition, only one proton from the CH₃ group has been considered with the H–C dipolar coupling value reduced by three due to fast rotation around the C–C axis. Other relevant parameters of the spin system are reported in the Supplementary Material.

For simulation in the case of C₆ and C₅ spin systems, dipolar couplings have been calculated for the geometry of leucine-8 in the X-ray structure of chicken SH3 protein (PDB 1SHG). Experimentally determined chemical shifts in fMLF have been assumed, and CSA parameters were calculated in Gaussian. ¹J_CC couplings were set to 50 Hz for the ¹³Cα–¹³C′ pair and to 33 Hz between aliphatic ¹³C spins. ²J_CC was set to 3 Hz, and longer-distance J couplings were neglected. Examples of SIMPSON input files are provided in the Supplementary Material.

3 Results and Discussion

3.1 Homonuclear Mixing in Model Two-¹³C Spin Systems

Efficiency of ¹³C mixing varies in general with MAS frequency ν_R, RF strength ν₁ (if not imposed by ν_R), B₀ field (through scaling of isotropic chemical shift differences and CSA), presence of ¹H–¹³C interactions, and extent of dynamics as well as other design-specific parameters detailed above. To mitigate the complexity of the problem, one usually refers to model two-¹³C spin systems, such as, e.g., in the acetate anion. Here, we used 2,3-¹³C₂,¹⁵N-labeled crystalline alanine as a closer analogue of a protein residue, additionally possessing a methyl group, and avoiding interference with large CSA of the (unlabeled) carbonyl carbon. The efficiency of ¹³Cβ→¹³Cα transfer was quantified by observation of cross-peak intensity in a τ_mix series of 2D ¹³C–¹³C spectra at fast MAS (ν_R = 55.5 kHz) and moderate B₀ field (14.1 T).

The results shown in Figure 4 confirm the high efficiency of selected RF designs. Also, fundamental differences between them exhibit the following: TOBSY and TOCSY sequences recouple J_CC in a time close to 1/(2J) (scaling factors might apply to the effective coupling constant, e.g., for FLOPSY). RFDR initially shows rapid oscillations with frequency related to a scaled dipolar coupling (D_CC ≈ 2 kHz), stabilizes at approximately τ_mix = 5–10 ms, and then decays due to π pulse imperfections and incoherent effects. DREAM shows a steady buildup, reaches a plateau at τ_mix ≈ 5 ms, and then decays slowly mostly due to ¹³C T_1ρ-like relaxation. The results are generally consistent with the literature, also confirming the increasing robustness of TOBSY C9¹_n to D_HC interactions with increasing n (Figure 4A) (Tan et al., 2018). Although RFDR experimentally showed highest performance, observed differences between considered schemes are rather minor.

FIGURE 4

FIGURE 4. (A–D) Performance of ¹³C mixing schemes observed experimentally in ¹³C–¹³C 2D spectra on 2,3-¹³C₂-labeled alanine: (A) TOBSY C9¹_n with n = 21, 24, 33, 39, and 48 are labeled and color-coded as purple, green, cyan, red, and orange curves, respectively; (B) FLOPSY-16 (in green) and WALTZ-16 (in purple); (C) RFDR at ν₁ = 100 kHz; (D) DREAM. Cross-peak intensities were normalized to the most intense point in the RFDR series. Experimental points are connected with lines for eye guidance only. (E–H) Buildup of transferred ¹³C magnetization (absolute transfer efficiency) in a two ¹³C- and two ¹H-spin system simulated in SIMPSON. (E) TOBSY C9¹_n with n = 24, 30, 33, 39, and 48 are labeled and color-coded as green, dark blue, cyan, red, and orange curves, respectively; (F) FLOPSY-16 (in green), DIPSI-3 (in light blue) and WALTZ-16 (in purple); (G) RFDR at ν₁ = 100 (in purple) and 160 kHz (in green); (H) DREAM. Gray boxes indicate the mixing time ranges sampled experimentally on alanine.

Spin dynamics simulations are instrumental for the in-depth understanding of complex RF schemes applied under new MAS conditions. Here, we resorted to SIMPSON as a generally agreed simulation platform (Tošner et al., 2014), and attempted first to reproduce experimental results. We modeled alanine using a minimal system consisting of two ¹³C and ¹H spins as described above. Results shown in Figures 4E–H reproduce well the experimental buildup curves, optimal mixing times, and scaling factors of recoupled interactions. Differences in efficiency between TOBSY C9¹_n sequences with n = 48, 39, 33, and 24 (Figure 4E) are more pronounced than in the experiment (Figure 4A). C9¹₄₈ allows for an almost complete transfer for τ_mix ≈ 16 ms. For longer mixing times, recoupling of higher-order AHT terms occurs (mostly the cross-terms between ¹³C isotropic and anisotropic chemical shifts, and D_HC interactions), leading to coherence dephasing and decreased intensity of subsequent maximums. Somewhat surprisingly, all TOCSY sequences perform excellently (Figure 4F), despite no deliberate treatment of anisotropic interactions. RFDR performance is relatively worse than in the experiment, and equalizes ¹³C magnetization between both coupled ¹³C spins. Supposedly, incoherent effects and B₁ field inhomogeneity in the MAS coil affect RFDR to the smallest extent among the considered mixing types. In simulations, DREAM does not show an optimum, but a steadily increasing coherence transfer.

Having validated the simulation platform, we attempted to differentiate ¹³C mixing schemes with respect to destructive interferences (cross-terms) with interactions that are field-dependent (isotropic and anisotropic CS), and proton-content-dependent (D_H-C interactions). All combinations of selectively “activated” interactions were probed, and the effects of most relevant cross-terms are summarized in Supplementary Figure S2. We confirmed that TOBSY C9¹_n sequences with n > 30 are quite robust to D_HC, with (J_CC, D_HC), (D_CC, D_HC), and (D_CC, CSA) being the primary cross-terms responsible for non-ideal performance. We additionally confirmed that TOBSY is robust with respect to an increasing chemical shift difference (or B₀ field), e.g., up to ΔΩ = 15 kHz for C9¹₂₄ at ν_R = 55 kHz (and v₁ = 2/3 ν_R). Among TOCSY sequences, FLOPSY appeared to be most susceptible to (J_CC, D_HC) interferences (or, in other words, to the presence of protons). Also, all TOCSY designs are less affected by (D_CC, D_HC) terms than TOBSY and very robust to CSA (Supplementary Figure S2). Thus, WALTZ and DIPSI are good candidates for high B₀ field measurements, and on systems with high proton density. Within the initial buildup, RFDR is extremely robust to all interaction interferences, and in fact even enhanced by recoupling of J_CC interactions.

Additionally, we verified the susceptibility of TOBSY and TOCSY sequences to B₁ field inhomogeneity, which can be substantial in MAS RF coils [as large as 20% in the active volume (Tošner et al., 2018)]. As shown in Supplementary Figure S3, TOBSY sequences only tolerate maximum 5% RF deviations (or miscalibration), and TOCSY designs, DIPSI in particular, are significantly more robust with this respect.

However, multiple limitations of spin dynamics simulations must be considered when their results are interpreted quantitatively: (1) no incoherent effects (relaxation) are included; (2) structural variability of dipolar interactions, chemical shifts, and CSA are tedious to replicate; and (3) pulse transients, limited short-term rotation stability, and other experimental deficiencies are neglected.

3.2 Homonuclear ¹³C Transfer Across Multi-Spin Side Chains in Model Systems

Subsequently, we performed an analogous series of 2D ¹³C–¹³C correlation spectra on a sample of crystalline tripeptide fMLF, where the leucine residue serves as a realistic and convenient reference multi-spin system containing a methyl group. Spectral excerpts in Figure 5 demonstrate that, while one-bond transfers remain effective in all cases, efficiencies of multi-bond ones are low and differ dramatically between particular RF schemes (see Supplementary Figure S4 for mixing time dependencies). In contrast to the case of alanine, RFDR performs worst in this respect, emphasizing the limited utility of simplistic two-¹³C spin systems for evaluation of ¹³C mixing performance.

FIGURE 5

FIGURE 5. Strips from 2D ¹³C–¹³C spectra recorded for fMLF at ν_R = 55.5 kHz on an 18.8-T spectrometer. Presented are diagonal and cross-peaks originating from Leu ¹³Cδ₁ magnetization (prior to ¹³C mixing). Spectrum diagonal is indicated with a dashed line. Spectra with a mixing time optimal for Cδ₁→ Cα transfer are shown: 34.6, 32.0, 21.4, 25.3, and 36.3 ms for TOBSY C9¹_n, with n = 24, 30, 33, 39, and 48, respectively (A–E); 23.7, 27.4, and 20.7 for FLOPSY-16 (F), DIPSI-3 (G), and WALTZ-16 (H), respectively; and 11.2 and 14 ms for RFDR (ν₁ = 150 kHz) (I) and DREAM (J), respectively.

It was noted previously in both solution (Tugarinov and Kay 2003a) and solid-state NMR (Agarwal and Reif 2008) that a combination of a branched ¹³C chain, large number of ¹³C spins, and large chemical shift differences makes the leucine spin system extremely challenging for any ¹³C mixing. In the following, we will deliberately focus on the most demanding Cδ₁→Cα transfer as the sensitivity-limiting step, regardless of which side-chain assignment strategy mentioned above [the one-step side-chain to backbone, (H)C(CC)(CA)NH, or the two-step approach using (H)CCH and (H)NCAHA spectra] is employed. (Another potentially relevant transfer Cδ₁→C′ is expected to be even more problematic due to a large CSA and distinct chemical shift.)

The entire experimental series was repeated at very fast spinning conditions (ν_R = 98 kHz) to probe sensitivity of selected mixing schemes to coherent and incoherent effects of dipolar C–H and H–H interactions. Absolute efficiencies of Cδ₁→Cα transfer were quantified as described in Supplementary Material, and presented in Figure 6. Strikingly, all RF schemes greatly benefit from increased MAS rate (up to a factor of 4.5–5 for three TOCSY representatives). At ν_R = 55.5 kHz DREAM performs best, and the most efficient scheme at ν_R = 98 kHz is DIPSI, with TOBSY C9¹₄₈ being close to the leaders at both MAS frequencies. In both cases, RFDR and TOBSY C9¹_n with n < 30 remain evident outliers.

FIGURE 6

FIGURE 6. Maximum absolute efficiency of Cδ₁ → Cα magnetization transfer observed in leucine residue of fMLF at B₀ = 18.8 T field and at ν_R = 55.5 (blue bars, left in each pair) and 98 kHz (tan bars, right in each pair).

Given such a dramatic response to MAS frequency, we attempted to verify capabilities of spin dynamics simulation for reproducing the behavior of a complex, but extremely useful leucine spin system, and possibly extrapolate it to different spinning and field conditions. Unfortunately, despite intense optimization of SIMPSON routines (e.g., the reuse of propagators) and the use of the state-of-the-art high-performance clusters, inclusion of any meaningful number of ¹H spins (L ≈ 5–6) turned out unfeasible due to the exponential scaling of required computational resources (4^L). Nevertheless, many features of the transfers were quite well reproduced in the exclusively ¹³C 6-spin system, in particular (1) the optimal Cδ₁ → Cα mixing times (Supplementary Figure S5), (2) the shape of build-up curves for individual transfers (Supplementary Figure S6), and (3) poor performance of RFDR (≈5%) at both spinning speeds. The results also corroborate TOBSY C9¹₄₈ and DIPSI as the performance leaders at ν_R ≈ 100 kHz (Supplementary Figure S7). The exact quantification of DREAM poses difficulties due to the absence of an optimum in simulations, but efficiency at τ_mix = 25 ms, as used in the experiments, places this mixing scheme closer to the best than to mediocre designs.

The absence of protons in the spin system used in simulations resulted in almost identical performance of all TOBSY schemes at ν_R = 100 kHz, which clearly disagrees with the experiments. Likely for the same reason, the experimentally observed relative order of performance of TOCSY sequences was not reproduced. To some extent, one can still rely on the analysis of cross-term importance for four spin system (Supplementary Figure S2) to predict the impact of protons also in leucine residues. However, absolute efficiencies predicted in simulations (Supplementary Figure S7) and observed experimentally (Figure 6) show discrepancies from roughly 25% (for TOBSY C9¹₄₈ and DIPSI) to more than 100% (e.g., for FLOPSY and WALTZ) at ν_R = 98–100 kHz, and even larger ones at the lower spinning speed (ν_R = 55.5 kHz). These disagreements arise not only due to the (unaccounted) coherent effects of ¹H–¹³C interactions, but also due to ¹³C relaxation, suggesting the ultimate need of experimental verification for a quantitative comparison at specific ν_R and B₀.

Despite apparent limitations, SIMPSON simulations still provide upper limits for transfer efficiencies, which is useful to predict performance trends at different experimental conditions. For example, ν_R dependence sampled between 30 and 200 kHz (Supplementary Figure S8) shows an optimal MAS range of 80–100 kHz for TOBSY and TOCSY schemes. The penalty observed at ν_R > 100 kHz is surprising, and in fact related to RF strength (ν₁), which was fixed in a constant proportion to ν_R. To isolate ν₁ dependence without traversing through recoupling conditions, we performed additional simulations at constant ν₁ = 55.5 kHz and ν₁ = ¼ ν_R, but indirectly varied RF bandwidth of sequences via modulation of resonance frequencies and CSA with the strength of B₀ field (Figure 7). At high field (above ν_0,H = 800–1,000 MHz), most sequences show insufficient ¹³C bandwidth, thus decreasing the performance of a multibond transfer. At low field, the transfer efficiency degrades as well, with a very similar behavior to that observed in Supplementary Figure S8 for large rotation rates ν_R. This suggests that the effect actually relates only to the effective ¹³C bandwidth, and we confirmed by observation of C′ magnetization that an excessive RF strength is detrimental to ¹³Cα magnetization due to the concurrent drainage to C′.

FIGURE 7

FIGURE 7. Efficiency of Cδ₁→Cα transfer in a ¹³C₆ spin system simulated in SIMPSON for a range of magnetic fields B₀ (corresponding to ¹H frequency between 50 and 1,300 MHz), MAS frequency of 55.5 kHz, and a constant RF strength of ν₁ = ¼ ν_R ≈ 13.9 kHz. Note that, in essence, the dependence on ¹³C bandwidth (ν₁ expressed in ¹³C ppm) of a particular mixing is investigated here. For each B₀ point, a full buildup was performed, and an optimum was picked at the mixing times that are reported in Supplementary Figure S9. All RF schemes except the particularly computationally demanding DREAM were evaluated. Blue boxes indicate the B₀ field range available for ¹H-detected MAS NMR at present.

This leads to a paradox that increased MAS rates might not necessarily be beneficial if RF strength is bound to ν_R by the mixing design. Contrary to the case of TOBSY, TOCSY schemes tolerate variable ν₁/ν_R ratios in a wide range below ¼ (Supplementary Figure S10), which gives an additional flexibility for their application at moderate-to-high fields and very fast spinning. The simulation does not account for RF-dependent ¹³C T_1ρ, and very small RF should obviously also be avoided, particularly in systems showing increased microsecond local dynamics.

3.3 Methyl Resonance Assignment in Proteins by Correlation to Backbone Spins

Results obtained for fMLF strongly suggest proton dilution for the effective ¹³C mixing in extended spin systems in proteins. Fortunately, with fast MAS, this can be accomplished without compromising the occupancy of methyl ¹H sites, e.g., by selective labeling of Ile, Leu, and Val residues from suitable precursors coupled to expression in D₂O media (Tugarinov and Kay 2004). Importantly, ¹³C (and ²H)-enriched glucose must be used to preserve the continuity of Ile and Leu ¹³C side chains (Supplementary Figure S1), since α and β of the former and α and carbonyl carbons of the latter residue are incorporated from the medium rather than from the precursor (Lundström et al., 2007).

To demonstrate the efficacy of methyl ¹H assignment, we resorted to a model SH3 domain of α-spectrin in a microcrystalline state. Given the selective ¹H labeling, a magnetization transfer from methyl to backbone amide ¹H (and not a reverse one) is strongly preferred for the sensitivity reasons. In addition to 3-fold larger occupancy of methyl ¹H, the initial polarization is enhanced by fast T₁ relaxation of methyl protons, while NMR signal detection is performed on well-dispersed, generally narrower, and, thus, more sensitive amide ¹H protons. Naturally, a protocol for complete reprotonation of labile amide ¹H sites after expression in D₂O is a prerequisite for this approach. While extensive deuteration is not a necessity for proteins as simple as SH3 at fast MAS (>50 kHz) and high field (here 18.8 T), it would certainly be required for resolution and sensitivity reasons in more challenging applications. We employed a slightly adapted literature RF irradiation scheme (shown already in Figure 2D), with various ¹³C mixings as in the previous cases.

3.3.1 Joint Effect of the Rotation Frequency and RF Strength

For SH3 sample we acquired 3D spectra with site-specific resolution at a single optimized mixing time for each ¹³C mixing. Two spinning conditions, ν_R = 55.5 and 94.5 kHz, were applied using the same probe and rotor for the maximum cohesion of the data, and the representative strips are presented in Figure 8 (for ν_R = 94.5 kHz).

FIGURE 8

FIGURE 8. Representative strips from 3D (H)C(CC)(CA)NH experiment for SH3 protein recorded at MAS frequency of 94.5 kHz and on an 18.8-T spectrometer, using the following mixing schemes at their optimal mixing time (see Supplementary Table S3): (A–C) TOBSY C9¹_n, with n = 33, 39, and 48, respectively, (D) FLOPSY-16, (E) DIPSI-3, (F) WALTZ-16, (G) RFDR (ν₁ = 100 kHz), and (H) DREAM. All planes show the cross-section at δ₂ (¹⁵N) = 125.4 ppm, and show two cross-peaks for Leu-34 residue (L34 ¹H^N,¹⁵N intra-residue correlations to δ₁ and δ₂ ¹³C). The most intense spectra (panels C–F) additionally show cross-peaks of Val-46 residue due to partial overlap of L34 and V46 ¹H^N and ¹⁵N frequencies.

The experiment inherently shows only the transferred signal, thus poses difficulties to the rigorous quantification of ¹³C transfer efficiency. Nevertheless, we attempted to correct for uneven efficiency of the remaining part of the pulse sequence (notably CP steps) at different MAS conditions by normalizing the cross-peak intensities to those observed in the 3D (H)CANH experiment, which shares majority of the coherence pathway. The observed relative signal-to-noise ratio was averaged over 24 individual strong and resolved correlations. Results shown in Figure 9 (for all residue types) and Supplementary Figure S11 (separately for Val, Leu, and Ile) show a dramatic increase of ¹³C mixing efficiency with fast MAS (94.5 w. r. t. 55.5 kHz), with a similar effect (a factor of 3.2–3.5 for TOCSY and C9¹₄₈) to that observed on nondeuterated fMLF. Despite a dilute network of ¹H interactions in the SH3 sample, and averaging also over valine and isoleucine residues, the best-performing RF schemes are virtually the same as for the leucine residue in fMLF (Figure 6), namely, DREAM at 55.5 kHz MAS, and DREAM, TOBSY C9¹₄₈, and three TOCSY sequences at 94.5 kHz MAS. The relevance of ¹H–¹H and ¹H–¹³C interactions might be due to the fact that, despite partial sample deuteration, the local ¹H environment of leucine, valine, and isoleucine residues is quite dense, with only 4 out of 11, 2 of 9, and 7 of 11 protons replaced by deuterons, respectively. As for the overall proton density, 14 ILV residues (out of 62 residues in SH3) contain 81 protons at methyl sites. These, in addition to 116 backbone and side-chain labile protons, yield a moderate rather than a low protonation level (approximately 38%).

FIGURE 9

FIGURE 9. Sensitivity (S/N per unit time) of methyl-to-amide correlations in 3D (H)C(CC)(CA)NH spectra normalized to sensitivity of respective intraresidue correlations in 3D (H)CANH spectra, recorded for SH3 protein at a MAS frequency of 55.5 (blue bars, left in each pair) and 94.5 kHz (tan bars, right in each pair) on an 18.8-T spectrometer using various ¹³C mixing schemes. Relative intensity of 24 cross-peaks [for 6 valine (all except Val-46), 1 isoleucine, and 6 leucine residues] was averaged with equal weights. Data for TOBSY C9¹₂₄ and C9¹₃₀ were not collected at ν_R = 94.5 kHz. The error bars reflect the standard deviation (scatter) of values observed for a set of residues, not the experimental error of average sensitivity.

Compared to fMLF, spins in SH3 protein undergo higher-amplitude dynamics, and certainly there is a larger contribution of incoherent effects, such as leading to ¹³C T₂ and T_1ρ relaxation and ¹H spin-diffusion. T_1ρ relaxation is further suppressed by stronger RF at increased MAS frequency (particularly for DREAM, which effectively spin-locks the ¹³C coherence), as it is bound to ν_R for all sequences except RFDR. As for the coherent effects, increased MAS frequency also more effectively averages D_CH interactions (i.e., a fast rotation suppresses second-order AHT cross-terms, deleterious, e.g., for TOBSY with low n). Definitely, there is an additional S/N benefit from line narrowing of amide ¹H resonances at 94.5 kHz MAS (separable under a few of assumptions), but, as described below, it is estimated to contribute no more than 20% to the overall S/N gain.

In view of the apparent relevance of H–C and H–H interactions in deuterated SH3, additional efficiency gains are expected with the use of high-power ¹H decoupling (ν_1,H > 3ν_R) during ¹³C mixing; however, this likely endangers sample hydration and thus spectral quality. Overall, fast MAS seems to be an attractive route to amplify the signal, and, in our comparison, the gain in efficiency well compensates the 2- to 3-fold sensitivity losses entailed by active volume reduction from a 1.3-mm rotor (optimal for ν_R ≈ 55 kHz) to a 0.7- or 0.81-mm rotor (ν_R ≈ 100 kHz). We speculate that for methyl resonance assignment using the presented approach, nondeuterated proteins would benefit from fast MAS to an even larger extent.

3.3.2 Effect of the Static Magnetic Field

Our experiments were performed at a typical high-field spectrometer (18.8 T) used commonly in protein studies by ¹H-detected MAS. Stronger B₀ fields are increasingly accessible; therefore, it is relevant to explore the performance of ¹³C mixing at such conditions. For example, for a transition to a 23.5-T field, spin dynamics simulations on a six-¹³C spin system suggest only a minor and negative impact on transfer efficiency (results in Figure 7 are summarized for two considered B₀ fields in Supplementary Figure S12). For experimental verification, we performed a full series of 3D experiments on SH3 protein (from the same crystallization batch) in a 1.3-mm rotor spun at ν_R = 55.5 kHz on a 1,000-MHz ¹H spectrometer. Data were analyzed with a site-specific resolution and averaged over all reliable cross-peaks of Leu, Val, and Ile residues as mentioned above. Also, as before, we normalized the obtained S/N ratios of cross-peaks with respect to peaks in 3D (H)CANH to eliminate the effects of larger Boltzmann polarization, rotor active volume, possibly different sample packing density, and specific RF coil efficiency (see Supplementary Material). Results compared in Figure 10 illustrate that only selected ¹³C mixing schemes, notably, FLOPSY, RFDR, and DIPSI, actually profit from higher B₀ field. Note that for consistency with the data acquired at 18.8 T, we did not optimize RF strength (ν₁) for TOCSY sequences (in the range below ¼ ν_R), despite potential gains in efficiency (see Figure 7 and Supplementary Figure S10). The rationale for observed efficiency changes are again coherent and incoherent effects of H–H and H–C interactions, which evade proper treatment in our spin dynamics simulations.

FIGURE 10

FIGURE 10. Efficiency of various ¹³C mixing schemes compared between 18.8- and 23.5-T fields [purple (left) and blue (right) bars, respectively]. Presented are sensitivity of correlation peaks in 3D (H)C(CC)(CA)NH spectra of SH3 protein obtained at 55.5 kHz MAS, with respect to sensitivity of correlations in (H)CANH experiment at both conditions. Relative sensitivity is averaged equally over 24 cross-peaks of Val, Leu, and Ile residues. Data for TOBSY C9¹₃₉ were not collected at 23.5 T field. The error bars reflect the standard deviation (scatter) of values observed for a set of residues, not the experimental error of average sensitivity.

3.3.3 Linearization of the ¹³C Side Chain

One of the reasons for a low transfer efficiency between methyl and alpha ¹³C spins using J_CC-based isotropic mixing in Leu, Val, and Ile is a branched topology of the ¹³C chain. A 2-fold degradation of FLOPSY-8 efficiency for Ile residues was identified by Kay and co-workers, and remedied with relay-type (COSY) experiments in which ¹³C coherence is sequentially transferred with a careful manipulation of individual ¹³C spins using very frequency selective pulses (Tugarinov and Kay 2003b). This approach is not immediately applicable in MAS NMR, since required transfer delays are prohibitively long compared to ¹³C T₂ lifetimes. For Leu residues, the overlap between chemical shifts of δ and γ ¹³C precludes the elimination of detrimental passive coupling to the second methyl ¹³C; thus, in solution NMR, the issue was addressed by a tailored amino acid labeling, in which one of the methyl groups was labeled as ¹²CD₃ (Tugarinov and Kay 2003a). Since the labeling of the α-ketoisovalerate (precursor) is not stereospecific, a 1:1 mixture of Leu (¹³CH₃, ¹²CD₃) and Leu (¹²CD₃,¹³CH₃) isotopomers is obtained (and analogously for valines), entailing a 50% loss of methyl ¹H occupancy. Nevertheless, this sensitivity loss was compensated by larger efficiency of complex out-and-back transfer schemes proposed by Kay and co-workers for methyl ¹H assignment in solution.

To verify the utility of the linearization of ¹³C side chains of Leu and Val residues for MAS NMR, we first performed SIMPSON simulations, and compared the behavior of 5-¹³C spin (linear) and 6-¹³C spin (branched) spin systems that mimic the leucine residue with respective isotope labeling patterns (Supplementary Figures S13 and Supplementary Figure S14). Indeed, for all J-based isotropic mixing schemes (TOBSY and TOCSY), a 2- to 3-fold efficiency gain is predicted at ν_R = 100 kHz, and comparable at ν_R = 55.5 kHz. It is thus expected that, in this case, the 50% loss of initial signal is at least compensated by increased Cδ→Cα (Cγ→Cα for valines) transfer efficiency. Dipolar-based mixing (RFDR and DREAM) marginally profits from the simplification of ¹³C chain, likely only due to dispersion of ¹³C magnetization over a smaller number of ¹³C spins.

The sample of SH3 with linearized side chains of Leu and Val [formally {I (δ₁), L (¹³CH₃, ¹²CD₃), V (¹³CH₃,¹²CD₃)} U-(¹⁵N, ¹³C, ²H, ¹H^N)-labeled, with Ile-¹³C₆, Leu-¹³C₅,Val-¹³C₄, here referred to as “ILV-C4” for brevity] was prepared by following carefully the expression protocol of the previous (“ILV-C5”, in fact Ile-¹³C₆, Leu-¹³C₆, and Val-¹³C₅) sample, but using the 1,2,3,4-¹³C-3,4′,4′,4′-²H-labeled α-ketoisovalerate as a precursor (Tugarinov and Kay 2003a). The microcrystals were paramagnetically doped with EDTA-chelated Cu^II ions, and transferred into MAS 0.81-mm rotor with comparable density (see ¹³C 1D spectra of both samples in Supplementary Figure S15A). As expected, the methyl ¹H signal in direct-excitation spectra decreased approximately twice (Supplementary Figure S15B). We repeated the entire series of 3D (H)C(CC)(CA)NH experiments at both spinning frequencies ν_R = 55.5 and 94.5 kHz, using virtually the same previous experimental settings. Differences in CP efficiency and rotor packing density were compensated by normalization to the respective intraresidue peak intensities in (H)CANH spectra. Finally, for each condition, the relative cross-peak intensities that approximate Cδ (or Cγ)→Cα transfer efficiencies were averaged over 24 correlation peaks. The comparison in Figure 11 provides evidence that transfer efficiency increased well beyond the factor of 2, yielding improved sensitivity for “ILV-C4” sample (with the exception of FLOPSY), and unexpectedly also manifested for dipolar-based mixings, for which a total sensitivity loss was expected based on SIMPSON simulations. Despite the 50% dilution of methyl ¹H spins, “ILV-C4”-labeling clearly yields superior results, with DIPSI, WALTZ, and TOBSY C9¹₄₈ as the methods of first choice. Interestingly, if the stereospecifically labeled ¹³C₄-α-ketoisovalerate was commercially available, S/N ratios could double, corresponding to a further sensitivity gain (i.e., per square root of time) of √2 (accounting for the need for two separate acquisition series).

FIGURE 11

FIGURE 11. Relative efficiency of ¹³C mixing schemes compared between two SH3 protein samples: with branched (“ILV-C5”, tan filled bars) and linear (“ILV-C4”, red filled bars) ¹³C spin systems of leucine and valine residues, quantified using 24 cross-peak intensities in 3D (H)C(CC)(CA)NH spectra recorded at a MAS frequency of 94.5 kHz on an 18.8-T spectrometer. Data for 55.5 kHz MAS is shown as blue and cyan bars for “ILV-C5” and “ILV-C4” SH3 samples, respectively. Sensitivity of each cross-peak was normalized to respective intra-residue peak in 3D (H)CANH spectrum at each sample and spinning condition, and subsequently averaged over Val, Leu, and Ile residues.

The unexpectedly high gains observed with “ILV-C4” labeling can be explained by (1) decreased proton density, particularly in the local environments of leucine and valine ¹³C spin systems, and (2) increased detection sensitivity due to amide ¹H line narrowing. The relevance of the first effect on the efficiency of ¹³C transfer can be appreciated based on the MAS frequency dependence discussed above for the “ILV-C5” sample. Here, we illustrated the effect by measurement of methyl ¹H linewidth change upon additional proton dilution, as observed in ¹H, ¹³C-CP-HSQC spectra as cross-peak raw full-width at half-height. Indeed, a 2-fold linewidth reduction is observed at ν_R = 55.5 kHz (Figure 12A and Supplementary Figure S16A, B), surpassing the effect of faster rotation for the “ILV-C5” sample (the ratio of 1.3 between 55.5 and 94.5 kHz). For the “ILV-C4” sample, both spinning conditions lead to similar linewidths (46 ± 11 Hz) with only few exceptions.

FIGURE 12

FIGURE 12. Linewidths (full widths at half height) of (A) methyl and (B) backbone amide ¹H resonances observed in 2D ¹³C-¹H and ¹⁵N–¹H CP-HSQC spectra, respectively, for “ILV-C4” (at ν_R = 55.5 kHz in cyan, and at ν_R = 94.5 kHz in red) and “ILV-C5” (at ν_R = 55.5 kHz in blue, and at ν_R = 94.5 kHz in tan) SH3 samples on an 18.8-T spectrometer. Linewidths observed in each case were averaged over all resonances, and the error bars indicate the standard deviation (the dispersion of values observed for all residues), not the precision of linewidth determination.

The second effect was carefully quantified by ¹H amide linewidth measurements in fingerprint ¹H, ¹⁵N-CP-HSQC spectra (also without application of a window function in ¹H dimension). As expected, proton dilution in the methyl side chain of Leu and Val residues has a relatively smaller impact on amide ¹H coherence lifetimes (Figure 12B). At ν_R = 94.5 kHz, both samples show similar linewidths of 40 Hz (see Supplementary Figures S16C,D for a per-peak comparison), but approximately 9% difference is observed at ν_R = 55.5 kHz. The line-narrowing effect of rotation frequency increase is of a factor of 1.4 and 1.3 for “ILV-C5” and “ILV-C4” samples, respectively. Overall, the use of “ILV-C4” labeling is beneficial for resolution of ¹H, ¹³C methyl resonances, and sensitivity of methyl-to-amide correlations.

3.3.4 Assignment of Protons: 4D Correlations

As shown in Figure 13, even in proteins as small as SH3, a single 3D (¹³C^met-edited) spectrum can result in massive ambiguities. For example, V53 γ₂, V58 γ₂, L8 δ₁, and L34 δ₁ ¹H assignment cannot be established from (H)C(CC)(CA)NH spectrum alone, since these methyl sites exhibit close ¹³C shifts (highlighted by a dashed line in ¹³C-CP-HSQC, panel C). The joint analysis of a pair of 3D (H)C(CC)(CA)NH and H(C)(CC)(CA)NH spectra could help to correctly match ¹H to ¹³C frequencies within a spin system (given sufficient resolution of each spectrum); however, if both methyl sites in a residue show either close ¹H or close ¹³C shifts, the ambiguity remains (e.g., in Val-58). The stereospecific labeling, i.e., using leucine and valine pro-S-C₄ and pro-R-C₄ precursors (yet unavailable for the presented approach) would be a costly solution.

FIGURE 13

FIGURE 13. Resolving ambiguity of methyl ¹H and ¹³C resonance assignment using 4D HC(DIPSI)(CA)NH experiment, demonstrated on the “ILV-C4” sample of SH3 protein, spun at 94.5 kHz on an 18.8-T spectrometer. Navigation in the 4D spectral matrix is based on peak positions in 2D ¹H, ¹⁵N-CP-HSQC (A), allowing to display intraresidue pairs of correlations in ¹H, ¹³C-cross sections of the 4D cube (B). Obtained methyl resonance assignments for four representative residues (L8, L34, V53, and V58, color-coded) are readily transferable to 2D ¹H, ¹³C-CP-HSQC spectrum (C). In (A), amide correlations of all Leu, Val, and Ile residues are labeled. All correlations except for residue Val-46 were observed in the 4D experiment.

Alternatively, one can resort to high-resolution 4D spectroscopy with non-uniform sampling, increasingly popular for MAS NMR (Huber et al., 2011; Linser et al., 2011; Paramasivam et al., 2012; Linser et al., 2014; Xiang et al., 2014; Fraga et al., 2017; Sergeyev et al., 2017; Marchanka et al., 2018; Vasa et al., 2018). Indeed, the use of a 4D HC(CC)(CA)NH spectrum (Figure 13B) allows disambiguating the assignment as demonstrated for L34 and V58 spin systems (close ¹H shifts), which additionally suffer from overlap of L34 δ₁ and V58 γ₂ peaks in ¹H, ¹³C-CP-HSQC. Given that a 4D NUS spectrum can be acquired in a similar time to a pair of 3D ones, this approach is recommended for all but the smallest proteins.

In the presented 4D NUS experiment, the average S/N of cross-peaks, normalized to 24 h of acquisition, was 25 ± 7 (Supplementary Figure S17). One can safely interpolate that a minimal S/N ratio of 10 is obtained in 3.8 h for SH3 (62 aa). If we limit the discussion to the class of microcrystalline samples of comparable spectral properties (CP efficiency and linewidths) and crystal packing density, S/N of a single peak scales inversely with the number of amino acids K (to account for the smaller number of molecules in a rotor). Thus, the time needed for the same minimal S/N ratio scales as K², yielding reasonable acquisition times of 0.42, 1.7, 3.7, 6.7, 10.4, and 15 days needed at 18.8 T for proteins of K = 100, 200, 300, 400, 500, and 600 residues, respectively. It is noteworthy that these times would be approximately 2- and 3.4-fold shorter, respectively, on 23.5- and 28.2-T (1,000 and 1,200 MHz ¹H) spectrometers available nowadays (assuming B₀^3/2 scaling of S/N ratio). Our results thus suggest feasibility of site-specific assignment of methyl resonances in sizable proteins for which the crystallization conditions and deuteration protocol have been already established, and can potentially be of great value to further solution or solid-state NMR studies. Whether methyl chemical shifts are readily transferable to other sample conditions is yet to be investigated. However, provided no significant alteration to protein fold, aliphatic ¹³C should not experience large deviations as demonstrated recently for maltose binding protein (Stanek et al., 2020).

4 Conclusion

We showed that a careful selection of ¹³C homonuclear mixing allows one to obtain sensitive correlations of methyl-to-amide ¹H chemical shifts under fast MAS and high B₀ field conditions. For highly deuterated proteins on an 18.8-T spectrometer, the best performance was obtained with DIPSI-3, WALTZ-16, and TOBSY C9¹₄₈ at ν_R = 94.5 kHz, and with DREAM at ν_R = 55.5 kHz. Dramatic improvement of the multi-bond (methyl to alpha ¹³C) transfer efficiency was observed upon increase of MAS frequency from 55.5 to 94.5 kHz, which is attributed to the suppression of incoherent effects of H–H and H–C dipolar interactions. Further significant performance increase was obtained by the linearization of ¹³C side chains of leucine and valine residues, with additional gains in resolution of methyl ¹H resonances. We demonstrated that unambiguous assignment of methyl ¹H and ¹³C resonances is feasible for microcrystalline proteins by a combination of protein deuteration, paramagnetic T₁ relaxation enhancement, suitable ¹³C isotope labeling, ultra-fast MAS, and 4D spectroscopy with non-uniform sampling. These findings pave the way for efficient assignment complementary to the labor-intensive mutagenesis-based strategy, with protein mass limitations largely mitigated by the favorable scaling of sensitivity in MAS NMR.

Data Availability Statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://doi.org/10.5281/zenodo.5911897.

Author Contributions

PP and JS designed and performed the research and analyzed data. RA prepared protein samples. M-LO, KV, and AK designed, constructed, and tested the MAS 100-kHz system (probe, rotation, and temperature controllers and rotor tools) under supervision of AS. PP, RA, and JS wrote the manuscript.

Funding

This research was funded by the Polish National Science Centre with grant No. 2019/33/B/ST4/02021. JS and PP thank the Polish National Agency for Academic Exchange for the generous support of Polish Return programme (Contract No. PPN/PPO/2018/1/00098). The computational resources for this work were provided by the Polish Infrastructure for Supporting Computational Science in the European Research Space (PLGRID, grant ID: plgnmrsi2, plgnmrsi3). We gratefully acknowledge financial support from the TGIR-RMN-THC Fr3050 CNRS for the access to the 23.5-T NMR spectrometer in CRMN (Lyon, France).

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors, and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Acknowledgments

We are grateful to Tobias Schubeis (CRMN Lyon) for sharing the SH3 plasmid and the original sample preparation protocol. The help of Tanguy Le Marchand in setting up the experiments on the 23.5-T spectrometer (CRMN Lyon) is greatly appreciated. The authors thank Morgane Callon (ETH Zurich) and Alons Lends (CBMN—IECB Bordeaux) for hints on rotor packing. We gratefully acknowledge M.J. Potrzebowski for providing access to the 14.1-T spectrometer at CBMM PAS in Łódź (Poland). The authors wish to express gratitude to Rector of Tallin University of Technology for financial support, and Darklands OÜ for technical support. We would like to thank W. Koźmiński for technical advice and maintenance of NMR system at CNBCh of University of Warsaw.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmolb.2022.828785/full#supplementary-material

References

Agarwal, V., Diehl, A., Skrynnikov, N., and Reif, B. (2006). High Resolution ¹H Detected ¹H,¹³C Correlation Spectra in MAS Solid-State NMR Using Deuterated Proteins with Selective ¹H,²H Isotopic Labeling of Methyl Groups. J. Am. Chem. Soc. 128 (39), 12620–12621. doi:10.1021/ja064379m

PubMed Abstract | CrossRef Full Text | Google Scholar

Agarwal, V., Penzel, S., Szekely, K., Cadalbert, R., Testori, E., Oss, A., et al. (2014). De Novo 3D Structure Determination from Sub-milligram Protein Samples by Solid-State 100 kHz MAS NMR Spectroscopy. Angew. Chem. Int. Ed. 53 (45), 12253–12256. doi:10.1002/anie.201405730

PubMed Abstract | CrossRef Full Text | Google Scholar

Agarwal, V., and Reif, B. (2008). Residual Methyl Protonation in Perdeuterated Proteins for Multi-Dimensional Correlation Experiments in MAS Solid-State NMR Spectroscopy. J. Magn. Reson. 194 (1), 16–24. doi:10.1016/j.jmr.2008.05.021

PubMed Abstract | CrossRef Full Text | Google Scholar

Amero, C., Asunción Durá, M., Noirclerc-Savoye, M., Perollier, A., Gallet, B., Plevin, M. J., et al. (2011). A Systematic Mutagenesis-Driven Strategy for Site-Resolved NMR Studies of Supramolecular Assemblies. J. Biomol. NMR 50 (3), 229–236. doi:10.1007/s10858-011-9513-5

PubMed Abstract | CrossRef Full Text | Google Scholar

Andreas, L. B., Jaudzems, K., Stanek, J., Lalli, D., Bertarello, A., Le Marchand, T., et al. (2016). Structure of Fully Protonated Proteins by Proton-Detected Magic-Angle Spinning NMR. Proc. Natl. Acad. Sci. USA 113 (33), 9187–9192. doi:10.1073/pnas.1602248113

PubMed Abstract | CrossRef Full Text | Google Scholar

Andreas, L. B., Le Marchand, T., Jaudzems, K., and Pintacuda, G. (2015b). High-Resolution Proton-Detected NMR of Proteins at Very Fast MAS. J. Magn. Reson. 253, 36–49. doi:10.1016/j.jmr.2015.01.003

PubMed Abstract | CrossRef Full Text | Google Scholar

Andreas, L. B., Reese, M., Eddy, M. T., Gelev, V., Ni, Q. Z., Miller, E. A., et al. (2015a). Structure and Mechanism of the Influenza A M218-60 Dimer of Dimers. J. Am. Chem. Soc. 137 (47), 14877–14886. doi:10.1021/jacs.5b04802

PubMed Abstract | CrossRef Full Text | Google Scholar

Asami, S., and Reif, B. (2012). Assignment Strategies for Aliphatic Protons in the Solid-State in Randomly Protonated Proteins. J. Biomol. NMR 52 (1), 31–39. doi:10.1007/s10858-011-9591-4

PubMed Abstract | CrossRef Full Text | Google Scholar

Asami, S., and Reif, B. (2013). Proton-Detected Solid-State NMR Spectroscopy at Aliphatic Sites: Application to Crystalline Systems. Acc. Chem. Res. 46 (9), 2089–2097. doi:10.1021/ar400063y

PubMed Abstract | CrossRef Full Text | Google Scholar

Asami, S., Schmieder, P., and Reif, B. (2010). High Resolution ¹H-Detected Solid-State NMR Spectroscopy of Protein Aliphatic Resonances: Access to Tertiary Structure Information. J. Am. Chem. Soc. 132 (43), 15133–15135. doi:10.1021/ja106170h

PubMed Abstract | CrossRef Full Text | Google Scholar

Asami, S., Szekely, K., Schanda, P., Meier, B. H., and Reif, B. (2012). Optimal Degree of Protonation for ¹H Detection of Aliphatic Sites in Randomly Deuterated Proteins as a Function of the MAS Frequency. J. Biomol. NMR 54 (2), 155–168. doi:10.1007/s10858-012-9659-9

PubMed Abstract | CrossRef Full Text | Google Scholar

Bak, M., and Nielsen, N. C. (1997). REPULSION, A Novel Approach to Efficient Powder Averaging in Solid-State NMR. J. Magn. Reson. 125 (1), 132–139. doi:10.1006/jmre.1996.1087

PubMed Abstract | CrossRef Full Text | Google Scholar

Bak, M., Rasmussen, J. T., and Nielsen, N. C. (2011). SIMPSON: A General Simulation Program for Solid-State NMR Spectroscopy. J. Magn. Reson. 213 (2), 366–400. doi:10.1016/j.jmr.2011.09.008

PubMed Abstract | CrossRef Full Text | Google Scholar

Bak, M., Schultz, R., Vosegaard, T., and Nielsen, N. C. (2002). Specification and Visualization of Anisotropic Interaction Tensors in Polypeptides and Numerical Simulations in Biological Solid-State NMR. J. Magn. Reson. 154 (1), 28–45. doi:10.1006/jmre.2001.2454

PubMed Abstract | CrossRef Full Text | Google Scholar

Baker, L. A., and Baldus, M. (2014). Characterization of Membrane Protein Function by Solid-State NMR Spectroscopy. Curr. Opin. Struct. Biol. 27 (August), 48–55. doi:10.1016/j.sbi.2014.03.009

PubMed Abstract | CrossRef Full Text | Google Scholar

Baldus, M., and Meier, B. H. (1996). Total Correlation Spectroscopy in the Solid State. The Use of Scalar Couplings to Determine the Through-Bond Connectivity. J. Magn. Reson. Ser. A 121 (1), 65–69. doi:10.1006/jmra.1996.0137

CrossRef Full Text | Google Scholar

Bax, A., Clore, G. M., and Gronenborn, A. M. (1990). ¹H¹H Correlation via Isotropic Mixing of ¹³C Magnetization, a New Three-Dimensional Approach for Assigning ¹H and ¹³C Spectra of ¹³C-Enriched Proteins. J. Magn. Reson. (1969) 88 (2), 425–431. doi:10.1016/0022-2364(90)90202-K

CrossRef Full Text | Google Scholar

Bennett, A. E., Rienstra, C. M., Auger, M., Lakshmi, K. V., and Griffin, R. G. (1995). Heteronuclear Decoupling in Rotating Solids. J. Chem. Phys. 103 (16), 6951–6958. doi:10.1063/1.470372

CrossRef Full Text | Google Scholar

Bennett, A. E., Rienstra, C. M., Griffiths, J. M., Zhen, W., Lansbury, P. T., and Griffin, R. G. (1998). Homonuclear Radio Frequency-Driven Recoupling in Rotating Solids. J. Chem. Phys. 108 (22), 9463–9479. doi:10.1063/1.476420

CrossRef Full Text | Google Scholar

Boswell, Z. K., and Latham, M. P. (2018). Methyl-Based NMR Spectroscopy Methods for Uncovering Structural Dynamics in Large Proteins and Protein Complexes. Biochemistry 58 (3), 144–155. doi:10.1021/acs.biochem.8b00953

PubMed Abstract | CrossRef Full Text | Google Scholar

Chevelkov, V., Rehbein, K., Diehl, A., and Reif, B. (2006). Ultrahigh Resolution in Proton Solid-State NMR Spectroscopy at High Levels of Deuteration. Angew. Chem. Int. Ed. 45 (23), 3878–3881. doi:10.1002/anie.200600328

CrossRef Full Text | Google Scholar

De Paëpe, G., Lewandowski, J. R., and Griffin, R. G. (2008). Spin Dynamics in the Modulation Frame: Application to Homonuclear Recoupling in Magic Angle Spinning Solid-State NMR. J. Chem. Phys. 128 (12), 124503. doi:10.1063/1.2834732

PubMed Abstract | CrossRef Full Text | Google Scholar

Edén, M., and Levitt, M. H. (1999). Pulse Sequence Symmetries in the Nuclear Magnetic Resonance of Spinning Solids: Application to Heteronuclear Decoupling. J. Chem. Phys. 111 (4), 1511–1519. doi:10.1063/1.479410

CrossRef Full Text | Google Scholar

Emsley, L., and Bodenhausen, G. (1992). Optimization of Shaped Selective Pulses for NMR Using a Quaternion Description of Their Overall Propagators. J. Magn. Reson. (1969) 97 (1), 135–148. doi:10.1016/0022-2364(92)90242-Y

CrossRef Full Text | Google Scholar

Fraga, H., Arnaud, C. A., Gauto, D. F., Audin, M., Kurauskas, V., Macek, P., et al. (2017). Solid‐State NMR H–N–(C)–H and H–N–C–C 3D/4D Correlation Experiments for Resonance Assignment of Large Proteins. ChemPhysChem 18 (19), 2697–2703. doi:10.1002/cphc.201700572

PubMed Abstract | CrossRef Full Text | Google Scholar

Ganapathy, S., Naito, A., and McDowell, C. A. (1981). Paramagnetic Doping as an Aid in Obtaining High-Resolution Carbon-13 NMR Spectra of Biomolecules in the Solid State. J. Am. Chem. Soc. 103 (20), 6011–6015. doi:10.1021/ja00410a003

CrossRef Full Text | Google Scholar

Gauto, D. F., Estrozi, L. F., Schwieters, C. D., Effantin, G., Macek, P., Sounier, R., et al. (2019). Integrated NMR and Cryo-EM Atomic-Resolution Structure Determination of a Half-Megadalton Enzyme Complex. Nat. Commun. 10 (1), 2697. doi:10.1038/s41467-019-10490-9

PubMed Abstract | CrossRef Full Text | Google Scholar

Gorman, S. D., Sahu, D., O'Rourke, K. F., and Boehr, D. D. (2018). Assigning Methyl Resonances for Protein Solution-State NMR Studies. Methods 148 (September), 88–99. doi:10.1016/j.ymeth.2018.06.010

PubMed Abstract | CrossRef Full Text | Google Scholar

Goto, N. K., Gardner, K. H., Mueller, G. A., Willis, R. C., and Kay, L. E. (1999). A Robust and Cost-Effective Method for the Production of Val, Leu, Ile (δ1) Methyl-Protonated ¹⁵N-, ¹³C-, ²H-Labeled Proteins. J. Biomol. NMR 13 (4), 369–374. doi:10.1023/A:1008393201236

PubMed Abstract | CrossRef Full Text | Google Scholar

Griffin, R. (1998). Dipolar Recoupling in MAS Spectra of Biological Solids. Nat. Struct. Mol. Biol. 5 (7), 508–512. doi:10.1038/749

PubMed Abstract | CrossRef Full Text | Google Scholar

Hardy, E. H., Verel, R., and Meier, B. H. (2001). Fast MAS Total Through-Bond Correlation Spectroscopy. J. Magn. Reson. 148 (2), 459–464. doi:10.1006/jmre.2000.2258

PubMed Abstract | CrossRef Full Text | Google Scholar

Hou, G., Yan, S., Trébosc, J., Amoureux, J.-P., and Polenova, T. (2013). Broadband Homonuclear Correlation Spectroscopy Driven by Combined R2_n^v Sequences under Fast Magic Angle Spinning for NMR Structural Analysis of Organic and Biological Solids. J. Magn. Reson. 232, 18–30. doi:10.1016/j.jmr.2013.04.009

PubMed Abstract | CrossRef Full Text | Google Scholar

Hu, B., Lafon, O., Trébosc, J., Chen, Q., and Amoureux, J.-P. (2011). Broad-Band Homo-Nuclear Correlations Assisted by ¹H Irradiation for Bio-Molecules in Very High Magnetic Field at Fast and Ultra-fast MAS Frequencies. J. Magn. Reson. 212 (2), 320–329. doi:10.1016/j.jmr.2011.07.011

PubMed Abstract | CrossRef Full Text | Google Scholar

Huang, C., and Kalodimos, C. G. (2017). Structures of Large Protein Complexes Determined by Nuclear Magnetic Resonance Spectroscopy. Annu. Rev. Biophys. 46 (1), 317–336. doi:10.1146/annurev-biophys-070816-033701

PubMed Abstract | CrossRef Full Text | Google Scholar

Huber, M., Hiller, S., Schanda, P., Ernst, M., Böckmann, A., Verel, R., et al. (2011). A Proton-Detected 4D Solid-State NMR Experiment for Protein Structure Determination. ChemPhysChem 12 (5), 915–918. doi:10.1002/cphc.201100062

PubMed Abstract | CrossRef Full Text | Google Scholar

Kadkhodaie, M., Rivas, O., Tan, M., Mohebbi, A., and Shaka, A. J. (1991). Broadband Homonuclear Cross Polarization Using Flip-Flop Spectroscopy. J. Magn. Reson. (1969) 91 (2), 437–443. doi:10.1016/0022-2364(91)90210-K

CrossRef Full Text | Google Scholar

Kay, L. E., Xu, G. Y., Singer, A. U., Muhandiram, D. R., and Forman-Kay, J. D. (1993). A Gradient-Enhanced HCCH-TOCSY Experiment for Recording Side-Chain ¹H and ¹³C Correlations in H₂O Samples of Proteins. J. Magn. Reson. Ser. B 101 (3), 333–337. doi:10.1006/jmrb.1993.1053

CrossRef Full Text | Google Scholar

Kulminskaya, N., Vasa, S. K., Giller, K., Becker, S., Kwan, A., Sunde, M., et al. (2016). Access to Side-Chain Carbon Information in Deuterated Solids under Fast MAS through Non-rotor-synchronized Mixing. Chem. Commun. 52 (2), 268–271. doi:10.1039/C5CC07345F

PubMed Abstract | CrossRef Full Text | Google Scholar

Kurauskas, V., Crublet, E., Macek, P., Kerfah, R., Gauto, D. F., Boisbouvier, J., et al. (2016). Sensitive Proton-Detected Solid-State NMR Spectroscopy of Large Proteins with Selective CH3 Labelling: Application to the 50S Ribosome Subunit. Chem. Commun. 52 (61), 9558–9561. doi:10.1039/C6CC04484K

CrossRef Full Text | Google Scholar

Ladizhansky, V. (2017). Applications of Solid-State NMR to Membrane Proteins. Biochim. Biophys. Acta (Bba) - Proteins Proteomics 1865 (11), 1577–1586. doi:10.1016/j.bbapap.2017.07.004

CrossRef Full Text | Google Scholar

Lee, W., Tonelli, M., and Markley, J. L. (2015). NMRFAM-SPARKY: Enhanced Software for Biomolecular NMR Spectroscopy. Bioinformatics 31 (8), 1325–1327. doi:10.1093/bioinformatics/btu830

PubMed Abstract | CrossRef Full Text | Google Scholar

Levitt, M. H. (2007). “eMagRes,”. Editors R. K. Harris, and R. L. Wasylishen (Chichester, UK: John Wiley & Sons), 9. doi:10.1002/9780470034590

CrossRef Full Text | Google Scholar

Lewandowski, J. R., Dumez, J.-N., Akbey, Ü., Lange, S., Emsley, L., and Oschkinat, H. (2011). Enhanced Resolution and Coherence Lifetimes in the Solid-State NMR Spectroscopy of Perdeuterated Proteins under Ultrafast Magic-Angle Spinning. J. Phys. Chem. Lett. 2 (17), 2205–2211. doi:10.1021/jz200844n

CrossRef Full Text | Google Scholar

Linser, R., Bardiaux, B., Andreas, L. B., Hyberts, S. G., Morris, V. K., Pintacuda, G., et al. (2014). Solid-State NMR Structure Determination from Diagonal-Compensated, Sparsely Nonuniform-Sampled 4D Proton-Proton Restraints. J. Am. Chem. Soc. 136 (31), 11002–11010. doi:10.1021/ja504603g

PubMed Abstract | CrossRef Full Text | Google Scholar

Linser, R., Bardiaux, B., Higman, V., Fink, U., and Reif, B. (2011). Structure Calculation from Unambiguous Long-Range Amide and Methyl ¹H−¹H Distance Restraints for a Microcrystalline Protein with MAS Solid-State NMR Spectroscopy. J. Am. Chem. Soc. 133 (15), 5905–5912. doi:10.1021/ja110222h

PubMed Abstract | CrossRef Full Text | Google Scholar

Linser, R. (2011). Side-Chain to Backbone Correlations from Solid-State NMR of Perdeuterated Proteins through Combined Excitation and Long-Range Magnetization Transfers. J. Biomol. NMR 51 (3), 221–226. doi:10.1007/s10858-011-9531-3

PubMed Abstract | CrossRef Full Text | Google Scholar

Linser, R. (2017). Solid-State NMR Spectroscopic Trends for Supramolecular Assemblies and Protein Aggregates. Solid State. Nucl. Magn. Reson. 87 (August), 45–53. doi:10.1016/j.ssnmr.2017.08.003

PubMed Abstract | CrossRef Full Text | Google Scholar

Lu, X., Guo, C., Hou, G., and Polenova, T. (2015). Combined Zero-Quantum and Spin-Diffusion Mixing for Efficient Homonuclear Correlation Spectroscopy under Fast MAS: Broadband Recoupling and Detection of Long-Range Correlations. J. Biomol. NMR 61 (1), 7–20. doi:10.1007/s10858-014-9875-6

PubMed Abstract | CrossRef Full Text | Google Scholar

Lundström, P., Teilum, K., Carstensen, T., Bezsonova, I., Wiesner, S., Hansen, D. F., et al. (2007). Fractional ¹³C Enrichment of Isolated Carbons Using [1-¹³C]- or [2-¹³C]-Glucose Facilitates the Accurate Measurement of Dynamics at Backbone Cα and Side-Chain Methyl Positions in Proteins. J. Biomol. NMR 38 (3), 199–212. doi:10.1007/s10858-007-9158-6

PubMed Abstract | CrossRef Full Text | Google Scholar

Mainz, A., Religa, T. L., Sprangers, R., Linser, R., Kay, L. E., and Reif, B. (2013). NMR Spectroscopy of Soluble Protein Complexes at One Mega-Dalton and beyond. Angew. Chem. Int. Ed. 52 (33), 8746–8751. doi:10.1002/anie.201301215

CrossRef Full Text | Google Scholar

Mandala, V. S., Williams, J. K., and Hong, M. (2018). Structure and Dynamics of Membrane Proteins from Solid-State NMR. Annu. Rev. Biophys. 47 (1), 201–222. doi:10.1146/annurev-biophys-070816-033712

PubMed Abstract | CrossRef Full Text | Google Scholar

Marchanka, A., Stanek, J., Pintacuda, G., and Carlomagno, T. (2018). Rapid Access to RNA Resonances by Proton-Detected Solid-State NMR at >100 kHz MAS. Chem. Commun. 54 (65), 8972–8975. doi:10.1039/c8cc04437f

CrossRef Full Text | Google Scholar

Miao, Y., and Cross, T. A. (2013). Solid State NMR and Protein-Protein Interactions in Membranes. Curr. Opin. Struct. Biol. 23 (6), 919–928. doi:10.1016/j.sbi.2013.08.004

PubMed Abstract | CrossRef Full Text | Google Scholar

Paramasivam, S., Suiter, C. L., Hou, G., Sun, S., Palmer, M., Hoch, J. C., et al. (2012). Enhanced Sensitivity by Nonuniform Sampling Enables Multidimensional MAS NMR Spectroscopy of Protein Assemblies. J. Phys. Chem. B 116 (25), 7416–7427. doi:10.1021/jp3032786

PubMed Abstract | CrossRef Full Text | Google Scholar

Pauli, J., Baldus, M., Van Rossum, B., De Groot, H., and Oschkinat, H. (2001). Backbone and Side-Chain ¹³C and ¹⁵N Signal Assignments of the α-Spectrin SH3 Domain by Magic Angle Spinning Solid-State NMR at 17.6 Tesla. ChemBioChem 2 (4), 272–281. doi:10.1002/1439-7633(20010401)2:4<272::aid-cbic272>3.0.co;2-2

PubMed Abstract | CrossRef Full Text | Google Scholar

Pauli, J., van Rossum, B., Förster, H., de Groot, H. J. M., and Oschkinat, H. (2000). Sample Optimization and Identification of Signal Patterns of Amino Acid Side Chains in 2D RFDR Spectra of the α-Spectrin SH3 Domain. J. Magn. Reson. 143 (2), 411–416. doi:10.1006/jmre.2000.2029

PubMed Abstract | CrossRef Full Text | Google Scholar

Penzel, S., Smith, A. A., Agarwal, V., Hunkeler, A., Org, M.-L., Samoson, A., et al. (2015). Mai-Liis Org, Ago Samoson, Anja Böckmann, Matthias Ernst, and Beat H MeierProtein Resonance Assignment at MAS Frequencies Approaching 100 kHz: A Quantitative Comparison of J-Coupling and Dipolar-Coupling-Based Transfer Methods. J. Biomol. NMR 63 (2), 165–186. doi:10.1007/s10858-015-9975-y

PubMed Abstract | CrossRef Full Text | Google Scholar

Pritišanac, I., Alderson, T. R., and Güntert, P. (2020). Automated Assignment of Methyl NMR Spectra from Large Proteins. Prog. Nucl. Magn. Reson. Spectrosc. 118-119 (June), 54–73. doi:10.1016/j.pnmrs.2020.04.001

PubMed Abstract | CrossRef Full Text | Google Scholar

Pritišanac, I., Würz, J. M., Alderson, T. R., and Güntert, P. (2019). Automatic Structure-Based NMR Methyl Resonance Assignment in Large Proteins. Nat. Commun. 10 (1), 1–12. doi:10.1038/s41467-019-12837-8

PubMed Abstract | CrossRef Full Text | Google Scholar

Reverter, D., and Lima, C. D. (2009). Preparation of SUMO Proteases and Kinetic Analysis Using Endogenous Substrates. Methods Mol. Biolin 497, 225–239. doi:10.1007/978-1-59745-566-4_15

PubMed Abstract | CrossRef Full Text | Google Scholar

Rosenzweig, R., and Kay, L. E. (2014). Bringing Dynamic Molecular Machines into Focus by Methyl-TROSY NMR. Annu. Rev. Biochem. 83 (1), 291–315. doi:10.1146/annurev-biochem-060713-035829

PubMed Abstract | CrossRef Full Text | Google Scholar

Sattler, M., Schleucher, J., and Griesinger, C. (1999). Heteronuclear Multidimensional NMR Experiments for the Structure Determination of Proteins in Solution Employing Pulsed Field Gradients. Prog. Nucl. Magn. Reson. Spectrosc. 34 (2), 93–158. doi:10.1016/S0079-6565(98)00025-9

CrossRef Full Text | Google Scholar

Schmidt, E., and Güntert, P. (2012). A New Algorithm for Reliable and General NMR Resonance Assignment. J. Am. Chem. Soc. 134 (30), 12817–12829. doi:10.1021/ja305091n

PubMed Abstract | CrossRef Full Text | Google Scholar

Schmidt, E., and Güntert, P. (2013). Reliability of Exclusively NOESY-Based Automated Resonance Assignment and Structure Determination of Proteins. J. Biomol. NMR 57 (2), 193–204. doi:10.1007/s10858-013-9779-x

PubMed Abstract | CrossRef Full Text | Google Scholar

Scholz, I., Huber, M., Manolikas, T., Meier, B. H., and Ernst, M. (2008). MIRROR Recoupling and its Application to Spin Diffusion under Fast Magic-Angle Spinning. Chem. Phys. Lett. 460 (1), 278–283. doi:10.1016/j.cplett.2008.05.058

CrossRef Full Text | Google Scholar

Sergeyev, I. V., Itin, B., Rogawski, R., Day, L. A., and McDermott, A. E. (2017). Efficient Assignment and NMR Analysis of an Intact Virus Using Sequential Side-Chain Correlations and DNP Sensitization. Proc. Natl. Acad. Sci. USA 114 (20), 5171–5176. doi:10.1073/pnas.1701484114

PubMed Abstract | CrossRef Full Text | Google Scholar

Shaka, A. J., Keeler, J., and Freeman, R. (1983). Evaluation of a New Broadband Decoupling Sequence: WALTZ-16. J. Magn. Reson. (1969) 53 (2), 313–340. doi:10.1016/0022-2364(83)90035-5

CrossRef Full Text | Google Scholar

Shaka, A. J., Lee, C. J., and Pines, A. (1988). Iterative Schemes for Bilinear Operators; Application to Spin Decoupling. J. Magn. Reson. (1969) 77 (2), 274–293. doi:10.1016/0022-2364(88)90178-3

CrossRef Full Text | Google Scholar

Shen, M., Hu, B., Lafon, O., Trébosc, J., Chen, Q., and Amoureux, J.-P. (2012). Broadband Finite-Pulse Radio-Frequency-Driven Recoupling (Fp-RFDR) with (XY8)41 Super-cycling for Homo-Nuclear Correlations in Very High Magnetic Fields at Fast and Ultra-fast MAS Frequencies. J. Magn. Reson. 223 (October), 107–119. doi:10.1016/j.jmr.2012.07.013

PubMed Abstract | CrossRef Full Text | Google Scholar

Sinnige, T., Daniëls, M., Baldus, M., and Weingarth, M. (2014). Proton Clouds to Measure Long-Range Contacts between Nonexchangeable Side Chain Protons in Solid-State NMR. J. Am. Chem. Soc. 136 (12), 4452–4455. doi:10.1021/ja412870m

PubMed Abstract | CrossRef Full Text | Google Scholar

Stanek, J., Andreas, L. B., Jaudzems, K., Cala, D., Lalli, D., Bertarello, A., et al. (2016). NMR Spectroscopic Assignment of Backbone and Side-Chain Protons in Fully Protonated Proteins: Microcrystals, Sedimented Assemblies, and Amyloid Fibrils. Angew. Chem. Int. Ed. 55 (50), 15504–15509. doi:10.1002/anie.201607084

PubMed Abstract | CrossRef Full Text | Google Scholar

Stanek, J., Augustyniak, R., and Koźmiński, W. (2012). Suppression of Sampling Artefacts in High-Resolution Four-Dimensional NMR Spectra Using Signal Separation Algorithm. J. Magn. Reson. 214 (January), 91–102. doi:10.1016/j.jmr.2011.10.009

PubMed Abstract | CrossRef Full Text | Google Scholar

Stanek, J., Schubeis, T., Paluch, P., Güntert, P., Andreas, L. B., and Pintacuda, G. (2020). Automated Backbone NMR Resonance Assignment of Large Proteins Using Redundant Linking from a Single Simultaneous Acquisition. J. Am. Chem. Soc. 142 (12), 5793–5799. doi:10.1021/jacs.0c00251

PubMed Abstract | CrossRef Full Text | Google Scholar

Tan, K. O., Agarwal, V., Lakomek, N.-A., Penzel, S., Meier, B. H., and Ernst, M. (2018). Efficient Low-Power TOBSY Sequences for Fast MAS. Solid State. Nucl. Magn. Reson. 89, 27–34. doi:10.1016/j.ssnmr.2017.11.003

PubMed Abstract | CrossRef Full Text | Google Scholar

Tošner, Z., Andersen, R., Stevensson, B., Edén, M., Nielsen, N. C., and Vosegaard, T. (2014). Computer-Intensive Simulation of Solid-State NMR Experiments Using SIMPSON. J. Magn. Reson. 246 (September), 79–93. doi:10.1016/j.jmr.2014.07.002

PubMed Abstract | CrossRef Full Text | Google Scholar

Tošner, Z., Sarkar, R., Becker-Baldus, J., Glaubitz, C., Wegner, S., Engelke, F., et al. (2018). Overcoming Volume Selectivity of Dipolar Recoupling in Biological Solid-State NMR Spectroscopy. Angew. Chem. Int. Ed. 57 (44), 14514–14518. doi:10.1002/anie.201805002

CrossRef Full Text | Google Scholar

Tugarinov, V., Hwang, P. M., Ollerenshaw, J. E., and Kay, L. E. (2003). Cross-Correlated Relaxation Enhanced ¹H−¹³C NMR Spectroscopy of Methyl Groups in Very High Molecular Weight Proteins and Protein Complexes. J. Am. Chem. Soc. 125 (34), 10420–10428. doi:10.1021/ja030153x

PubMed Abstract | CrossRef Full Text | Google Scholar

Tugarinov, V., Kanelis, V., and Kay, L. E. (2006). Isotope Labeling Strategies for the Study of High-Molecular-Weight Proteins by Solution NMR Spectroscopy. Nat. Protoc. 1 (2), 749–754. doi:10.1038/nprot.2006.101

PubMed Abstract | CrossRef Full Text | Google Scholar

Tugarinov, V., and Kay, L. E. (2004). An Isotope Labeling Strategy for Methyl TROSY Spectroscopy. J. Biomol. NMR 28 (2), 165–172. doi:10.1023/B:JNMR.0000013824.93994.1f

PubMed Abstract | CrossRef Full Text | Google Scholar

Tugarinov, V., and Kay, L. E. (2003a). Ile, Leu, and Val Methyl Assignments of the 723-Residue Malate Synthase G Using a New Labeling Strategy and Novel NMR Methods. J. Am. Chem. Soc. 125 (45), 13868–13878. doi:10.1021/ja030345s

PubMed Abstract | CrossRef Full Text | Google Scholar

Tugarinov, V., and Kay, L. E. (2003b). Side Chain Assignments of Ile δ1 Methyl Groups in High Molecular Weight Proteins: an Application to a 46 ns Tumbling Molecule. J. Am. Chem. Soc. 125 (19), 5701–5706. doi:10.1021/ja021452+

PubMed Abstract | CrossRef Full Text | Google Scholar

Tycko, R. (2011). Solid-State NMR Studies of Amyloid Fibril Structure. Annu. Rev. Phys. Chem. 62 (1), 279–299. doi:10.1146/annurev-physchem-032210-103539

PubMed Abstract | CrossRef Full Text | Google Scholar

van der Wel, P. C. A. (2018). New Applications of Solid-State NMR in Structural Biology. Emerging Top. Life Sci. 2 (1), 57–67. doi:10.1042/ETLS20170088

CrossRef Full Text | Google Scholar

Vasa, S. K., Singh, H., Rovó, P., Linser, R., and Linser, R. (2018). Dynamics and Interactions of a 29 KDa Human Enzyme Studied by Solid-State NMR. J. Phys. Chem. Lett. 9 (6), 1307–1311. doi:10.1021/acs.jpclett.8b00110

PubMed Abstract | CrossRef Full Text | Google Scholar

Verel, R., Baldus, M., Ernst, M., Beat, H., and Meier, H. (1998). A Homonuclear Spin-Pair Filter for Solid-State NMR Based on Adiabatic-Passage Techniques. Chem. Phys. Lett. 287 (3–4), 421–428. doi:10.1016/S0009-2614(98)00172-9

CrossRef Full Text | Google Scholar

Verel, R., Ernst, M., and Meier, B. H. (2001). Adiabatic Dipolar Recoupling in Solid-State NMR: The DREAM Scheme. J. Magn. Reson. 150 (1), 81–99. doi:10.1006/jmre.2001.2310

PubMed Abstract | CrossRef Full Text | Google Scholar

Weingarth, M., Demco, D. E., Bodenhausen, G., and Tekely, P. (2009). Improved Magnetization Transfer in Solid-State NMR with Fast Magic Angle Spinning. Chem. Phys. Lett. 469 (4), 342–348. doi:10.1016/j.cplett.2008.12.084

CrossRef Full Text | Google Scholar

Westfeld, T., Verel, R., Ernst, M., Böckmann, A., and Meier, B. H. (2012). Properties of the DREAM Scheme and its Optimization for Application to Proteins. J. Biomol. NMR 53 (2), 103–112. doi:10.1007/s10858-012-9627-4

PubMed Abstract | CrossRef Full Text | Google Scholar

Wickramasinghe, N. P., Kotecha, M., Samoson, A., Past, J., and Ishii, Y. (2007). Sensitivity Enhancement in ¹³C Solid-State NMR of Protein Microcrystals by Use of Paramagnetic Metal Ions for Optimizing ¹H T₁ Relaxation. J. Magn. Reson. 184 (2), 350–356. doi:10.1016/j.jmr.2006.10.012

PubMed Abstract | CrossRef Full Text | Google Scholar

Wickramasinghe, N. P., Parthasarathy, S., Jones, C. R., Bhardwaj, C., Long, F., Kotecha, M., et al. (2009). Nanomole-Scale Protein Solid-State NMR by Breaking Intrinsic ¹H T₁ Boundaries. Nat. Methods 6 (3), 215–218. doi:10.1038/nmeth.1300

PubMed Abstract | CrossRef Full Text | Google Scholar

Wiegand, T. (2020). A Solid-State NMR Tool Box for the Investigation of ATP-Fueled Protein Engines. Prog. Nucl. Magn. Reson. Spectrosc. 117 (April), 1–32. doi:10.1016/j.pnmrs.2020.02.001

PubMed Abstract | CrossRef Full Text | Google Scholar

Xiang, S., Chevelkov, V., Becker, S., and Lange, A. (2014). Towards Automatic Protein Backbone Assignment Using Proton-Detected 4D Solid-State NMR Data. J. Biomol. NMR 60 (2–3), 85–90. doi:10.1007/s10858-014-9859-6

PubMed Abstract | CrossRef Full Text | Google Scholar

Xue, K., Motz, C., Asami, S., Decker, V., Wegner, S., Tosner, Z., et al. (2018). Magic-Angle Spinning Frequencies beyond 300 kHz Are Necessary to Yield Maximum Sensitivity in Selectively Methyl Protonated Protein Samples in Solid-State NMR. J. Phys. Chem. C 122, 16437–16442. doi:10.1021/acs.jpcc.8b05600

CrossRef Full Text | Google Scholar

Xue, K., Sarkar, R., Lalli, D., Koch, B., Pintacuda, G., Tosner, Z., et al. (2020). Impact of Magnetic Field Strength on Resolution and Sensitivity of Proton Resonances in Biological Solids. J. Phys. Chem. C 124 (41), 22631–22637. doi:10.1021/acs.jpcc.0c05407

CrossRef Full Text | Google Scholar

Xue, K., Sarkar, R., Motz, C., Asami, S., Camargo, D. C. R., Decker, V., et al. (2017). Limits of Resolution and Sensitivity of Proton Detected MAS Solid-State NMR Experiments at 111 kHz in Deuterated and Protonated Proteins. Sci. Rep. 7 (1), 7444. doi:10.1038/s41598-017-07253-1

PubMed Abstract | CrossRef Full Text | Google Scholar

Xue, K., Sarkar, R., Tosner, Z., Lalli, D., Motz, C., Koch, B., et al. (2019). MAS Dependent Sensitivity of Different Isotopomers in Selectively Methyl Protonated Protein Samples in Solid State NMR. J. Biomol. NMR 73 (10–11), 625–631. doi:10.1007/s10858-019-00274-0

PubMed Abstract | CrossRef Full Text | Google Scholar

Zhou, D. H., and Rienstra, C. M. (2008). High-Performance Solvent Suppression for Proton Detected Solid-State NMR. J. Magn. Reson. 192 (1), 167–172. doi:10.1016/j.jmr.2008.01.012

PubMed Abstract | CrossRef Full Text | Google Scholar

Zhou, D. H., Shah, G., Cormos, M., Mullen, C., Sandoz, D., and Rienstra, C. M. (2007). Proton-Detected Solid-State NMR Spectroscopy of Fully Protonated Proteins at 40 kHz Magic-Angle Spinning. J. Am. Chem. Soc. 129 (38), 11791–11801. doi:10.1021/ja073462m

PubMed Abstract | CrossRef Full Text | Google Scholar