5D solid-state NMR spectroscopy for facilitated resonance assignment

Abstract

¹H-detected solid-state NMR spectroscopy has been becoming increasingly popular for the characterization of protein structure, dynamics, and function. Recently, we showed that higher-dimensionality solid-state NMR spectroscopy can aid resonance assignments in large micro-crystalline protein targets to combat ambiguity (Klein et al., Proc. Natl. Acad. Sci. U.S.A. 2022). However, assignments represent both, a time-limiting factor and one of the major practical disadvantages within solid-state NMR studies compared to other structural-biology techniques from a very general perspective. Here, we show that 5D solid-state NMR spectroscopy is not only justified for high-molecular-weight targets but will also be a realistic and practicable method to streamline resonance assignment in small to medium-sized protein targets, which such methodology might not have been expected to be of advantage for. Using a combination of non-uniform sampling and the signal separating algorithm for spectral reconstruction on a deuterated and proton back-exchanged micro-crystalline protein at fast magic-angle spinning, direct amide-to-amide correlations in five dimensions are obtained with competitive sensitivity compatible with common hardware and measurement time commitments. The self-sufficient backbone walks enable efficient assignment with very high confidence and can be combined with higher-dimensionality sidechain-to-backbone correlations from protonated preparations into minimal sets of experiments to be acquired for simultaneous backbone and sidechain assignment. The strategies present themselves as potent alternatives for efficient assignment compared to the traditional assignment approaches in 3D, avoiding user misassignments derived from ambiguity or loss of overview and facilitating automation. This will ease future access to NMR-based characterization for the typical solid-state NMR targets at fast MAS.

Keywords: 5D; Fast magic-angle spinning; Higher dimensionality; Minimal set of experiments; Non-uniform sampling; Proton detection; Resonance assignment; Solid-state NMR.

Fig. 1

H/N overlap as found even for well-behaved targets. (A) 2D hNH spectrum of the SH3 domain of chicken α-spectrin. Despite its small size (62 residues) and high β-sheet content, the HN correlation comprises multiple overlapping peaks, causing ambiguities for any HN-based experiments, e.g., the standard out-and-back experiments. (B) Overlap in 4D HNNH direct amide-to-amide correlations (top), resolved by the additional carbon dimension in a 5D HNcoCANH experiment (bottom row), leaving only unambiguous assignments here

Fig. 2

5D magnetization transfer scheme and pulse sequence for the HNcoCANH using carbon-carbon CP (i) or INEPT (ii). Both versions have been used previously (Andreas et al. ; Klein et al. ; Orton et al. ; Stanek et al. ; Xiang et al. 2015) but are shown here again for completeness. For the INEPT transfer step, using a block similar to the one reported in Barbet-Massin et al. (Barbet-Massin et al. 2013), only hard pulses at a ¹³ C carrier position of 120 ppm were used here. Compare Orton et al., also involving selective pulses (Orton et al. 2020). Phase cycling for HNcoCANH: φ₁ = x,-x; φ₂ = x,x,-x,-x; φ₃ = x,x,x,x,-x,-x,-x,-x; φ_rec = x,-x,-x,x,-x,x,x,-x. Carrier changes are marked by arrows. The Cα carrier position was set to 55 ppm, while 173 ppm was used for carbonyl carbons. Water suppression was achieved by the MISSISSIPPI scheme (Zhou and Rienstra 2008) without gradients, using a pulse length of 20 ms at 10 kHz rf field strength

Fig. 3

5D amide-to-amide backbone walks, illustrated for the HNcoCANH (A) and the HNcaCONH (B) experiment. Both experiments enable a straightforward backbone walk, as they provide direct, well-resolved i to i + 1 connectivities. If sidechain carbon information is desired anyways, an hcaCBcaNH with split CACB transfer can be used instead of an hCANH as the base experiment for 5D reconstruction. The HNcaCONH benefits from the higher resolution of the carbonyl carbons but lacks the possibility of connecting sequential with sidechain information

Fig. 4

Processing of 5D HNcoCANH experiment and information obtained. A) Using SSA for spectral reconstruction, a stack of 2D planes is returned (one for each previously defined 3D peak). B and C) In BSH-CP versions of the amide-to-amide correlations, incomplete magnetization transfers cause additional out-and-back-like correlations with a constant intensity ratio of around 3:1 for non-glycine residues but much higher for glycines (shown as a histogram in B) and as a function of sequence in the SH3 domain in C)

Fig. 5

Bulk signal intensities for the 5D HNcoCANH and 5D HNcaCONH compared to the hNH intensity. (A) The HNcoCANH maintains approximately 12% of the hNH when used with BSH-CP for homonuclear CC transfers. (B) Using the HNcaCONH with BSH-CP results in ca. 6% of the hNH bulk signal. (C) In case an INEPT block is used, the signal intensity is attenuated by an additional 30%, however, in contrast to the BSH version, the bulk signal represents only the peaks of interest. Also, the difference between CACO and COCA is not observed in the INEPT version

Fig. 6

Influence of the number of NUS points (complex time points) and hence overall experimental time invested on the number of signals, i.e., residues found in the spectrum (A/C) and the signal-to-noise ratio (SNR) (B/C) after reconstruction of the FID recorded with 8 scans (minimal phase cycle). The dashed lines in (A) mark the threshold at which 75% of all 48 possible signals are found, arbitrarily chosen as the limit for a minimal acceptable performance. In (B) the averaged SNR over all identified signals is shown with blue crosses, while the red curve marks the calculated SNR extrapolated down from the SNR measured for 2048 NUS points. The dashed lines mark the SNR reached for the 256 points, for which 75% (36 out of 48) of all signals are found. In (C) the signal-to-noise ratio for the individual residues is shown in dependence of the number of NUS points included for reconstruction. Prolines 21 and 54 and CP inaccessible residues 47 and 48 are marked in gray (problematic both in the roles of residue i or i-1). R21 and N38 are mobile residue always difficult to detect. (Linser et al. 2010a) The data reduction (i.e., reduction of measurement time effectively included in the sparse FID) was implemented by truncating to only the desired subset of time-domain points. As the NUS schedule was randomized before acquisition, the maximum resolution and sampling density weighting is maintained for all subsets. 2048, 1536, 1024, 512, 256, 192, and 64 time-domain complex points correspond to effective experimental times of 52 h, 39 h, 26 h, 13 h, 6.5 h, 5 h, and 1.5 h, respectively

Fig. 7

FLYA assignments using different levels of input data. The classical set of 3D experiments for backbone assignment was used as input for FLYA in (A) and can be seen as a point of orientation for the performance of FLYA. For (B) to E), different combinations of 5D and 3D experiments were used (see the text for more details, HNcoCANH with 24 scans in B) and (C) and with 8 scans in E). F) The 5D HNcoCANH was replaced with a 4D HNcocaNH experiment, yielding slightly worse but similar results. For all runs, the manual assignments were used as a reference. 100 independent runs were performed, using tolerances of 0.1 ppm for ¹ H and 0.5 ppm for ¹³ C and ¹⁵ N, respectively, and ultimately combined into a consensus chemical shift assignment. (A 2D hNH is usually submitted as well, but implies negligible costs, relatively speaking.) Color scheme: Green: In agreement with reference assignment (here the manual assignment). Red: Disagreement with reference assignment. Blue: Additional assignments, not in reference. Black: In reference but not assigned by FLYA. Dark colors generally indicate “strong”/reliable assignments that are found in at least 80% of the independent FLYA runs, lighter colors indicate they have been found less than 80% of the time

Fig. 8

Results of FLYA assignments via an experimental pair of an amide-to-amide experiment and an S2B experiment in comparison to the manual assignments. (A) Sketch of magnetization transfers of the experiments in the set. The orange HNcoCANH was recorded on a 1.3 mm sample, the 4D HCCH (cyan) is an additional pathway available for free while recording the 5D HCCNH (blue); both were simultaneously recorded on a 0.7 mm sample. (B) to E) FLYA results from different combinations of input data as denoted on the right. Although the assignment success is very similar for the different combinations, the best agreement with the manual assignments is achieved through the combination of two 5Ds, likely due to the higher chemical shift redundancy. The tolerances used are 0.1 ppm for ¹ H, 0.8 ppm for ¹³ C, and 0.5 ppm for ¹⁵ N, respectively. Compare Fig. S4 for additional experimental sets. Color scheme as in Fig. 7