Structure of fully protonated proteins by proton-detected magic-angle spinning NMR

Abstract

Protein structure determination by proton-detected magic-angle spinning (MAS) NMR has focused on highly deuterated samples, in which only a small number of protons are introduced and observation of signals from side chains is extremely limited. Here, we show in two fully protonated proteins that, at 100-kHz MAS and above, spectral resolution is high enough to detect resolved correlations from amide and side-chain protons of all residue types, and to reliably measure a dense network of (1)H-(1)H proximities that define a protein structure. The high data quality allowed the correct identification of internuclear distance restraints encoded in 3D spectra with automated data analysis, resulting in accurate, unbiased, and fast structure determination. Additionally, we find that narrower proton resonance lines, longer coherence lifetimes, and improved magnetization transfer offset the reduced sample size at 100-kHz spinning and above. Less than 2 weeks of experiment time and a single 0.5-mg sample was sufficient for the acquisition of all data necessary for backbone and side-chain resonance assignment and unsupervised structure determination. We expect the technique to pave the way for atomic-resolution structure analysis applicable to a wide range of proteins.

Keywords: NMR spectroscopy; magic-angle spinning; protein structures; proton detection; viral nucleocapsids.

Fig. 1.

MAS NMR structures of GB1 (PDB ID code 5JXV; Top) and AP205CP (PDB ID code 5JZR; Bottom). Ribbon diagram of the 10 lowest-energy conformers for GB1 (A) and dimeric AP205CP (E), with monomers colored in cyan and tan. The approximate location of 1 of the 90 dimers in the cryo-EM electron density is also illustrated (53). The long-range proton–proton contacts measured in this study are depicted as dark blue lines onto the lowest-energy conformers of the two proteins: contacts between amide protons (B and F), contacts among amide and ILV-methyl–labeled protons (C and G), and all proton–proton contacts (D and H).

Fig. 2.

High-resolution ¹H-detected spectra recorded on fully protonated microcrystalline GB1 and AP205CP. (A) ¹⁵N-¹H CP-HSQC of GB1 at 60 kHz (black), and at 111.111-kHz MAS (red). (B) Hα-Cα (Bottom) and methyl regions (Top) of a ¹³C-¹H CP-HSQC of GB1 at 111.111-kHz MAS. (C) ¹⁵N-¹H CP-HSQC of AP205CP acquired on a fully protonated sample at 100-kHz MAS (red), and on a perdeuterated sample, exchanged in 100% H₂O, at 60-kHz MAS (black). The exchange-protected residues are labeled in the spectrum and colored on the NMR structure in the Inset. (D) Hα-Cα (Bottom) and methyl regions (Top) of a ¹³C-¹H CP-HSQC of AP205CP at 100-kHz MAS.

Fig. 3.

(A and C) ¹³C-¹³C projection and (B and D) selected strips of the (H)CCH spectrum of GB1 (A and B) and AP205CP (C and D). GB1 and AP205CP spectra were run at a MAS frequency of 111 or 100 kHz, respectively, with WALTZ-16 mixing applied for 15 ms at 27.8 kHz or 14.4 ms at 25 kHz of radiofrequency field.

Fig. S1.

In A, the β-strand arrangement in AP205CP. Residues identified by TALOS+ as β-sheet secondary structure are labeled in red. Black arrows indicate contacts present in the (H)CHH and (H)NHH spectra. When the cross-peak could be identified from both directions (the majority of cases), a double-headed arrow is drawn. Dashed lines indicate the hydrogen bonds entered in the structure calculation. Intermolecular hydrogen bonds are colored in blue. In B, the 10 lowest-energy conformers of a test calculation with an incorrect arrangement of one of the β-strands. Two of the hydrogen bonds (residues 6–124 and 4–128) are entered as intramolecular instead of intermolecular. The bundle is not well defined, and the helices are severely bent, indicating disagreement with the experimental data.

Fig. 4.

(A and B) Representative restraints from the (H)CHH spectra are displayed on the lowest-energy NMR structures of GB1 (A) and AP205CP (B). The color of these restraints indicates the corresponding strip from the spectra shown in C and D. In the case of AP205CP (D), the labels of intermolecular cross-peaks are underlined. The RFDR mixing time was 0.5 ms for GB1 and 1.0 ms for AP205CP.

Fig. S2.

Comparison of optimized ¹⁵N-¹H CP-HSQC spectra acquired either in a 1.3-mm probe at 60 kHz on perdeuterated GB1 (red) or in the 0.7-mm probe at 111 kHz on protonated GB1 (blue). Both spectra were acquired on the 1-GHz spectrometer, and 100 Hz of exponential line broadening was applied.

Fig. S3.

Comparison of sensitivity of AP205CP in the 1.3- and 0.7-mm probes. In A, 1D ¹⁵N-¹H CP-HSQC spectra acquired either in a 1.3-mm probe at 60-kHz MAS with perdeuterated AP205CP (red), a 1.3-mm probe at 60 kHz with protonated AP205CP (green), or in the 0.7-mm probe at 100 kHz and protonated AP205CP (blue). Spectra were acquired on the 1-GHz spectrometer, and no line broadening was applied. In B, 1D ¹³C-¹H CP-HSQC spectra of protonated AP205CP acquired either in a 1.3-mm probe at 60-kHz MAS (green) or in a 0.7-mm probe (blue).

Fig. S4.

Improvement in sensitivity due to improved linewidth, as a function of the acquisition time in an indirect dimension. The x axis is shown in units of the new T₂* (T_2n*).

Fig. S5.

Simulation of the transfer efficiency of WALTZ-16 sequence at 111-kHz MAS and an external magnetic field of 1 GHz. The applied nutation frequency is one-quarter of the rotor frequency. The spin system corresponds to two ¹³C spins. The scalar coupling constant was set to 50 Hz, and the dipolar coupling constant was chosen to correspond to a spin–spin distance of 1.5 Å. In A, the isotropic chemical shifts were 10 and −10 ppm and the CSA was neglected (aliphatic–aliphatic transfer with little offset). In B, the isotropic chemical shifts were larger, 20 and −20 ppm, and the CSA was again neglected (aliphatic–aliphatic transfer with larger offset). In C, the isotropic chemical shifts were 10 and −10 ppm, and the chemical shift anisotropy and asymmetry were set to 80 ppm and 0.9, respectively, for both spins (aromatic–aromatic transfer). Finally, in D, the isotropic chemical shifts were 80 and −10 ppm, and the chemical shift anisotropy and asymmetry were set to 80 ppm and 0.9, respectively, for the first spin only (aromatic–aliphatic transfer). In all cases, only scalar or dipolar coupling was switched on for “J” and “D” curves, respectively; both were switched on for “J + D” curves. The “ø” curves show no transfer (both J and D off). The simulations were performed by using SIMPSON software (70).