Cryo-EM and solid state NMR together provide a more comprehensive structural investigation of protein fibrils

Abstract

The tropomyosin 1 isoform I/C C-terminal domain (Tm1-LC) fibril structure is studied jointly with cryogenic electron microscopy (cryo-EM) and solid state nuclear magnetic resonance (NMR). This study demonstrates the complementary nature of these two structural biology techniques. Chemical shift assignments from solid state NMR are used to determine the secondary structure at the level of individual amino acids, which is faithfully seen in cryo-EM reconstructions. Additionally, solid state NMR demonstrates that the region not observed in the reconstructed cryo-EM density is primarily in a highly mobile random coil conformation rather than adopting multiple rigid conformations. Overall, this study illustrates the benefit of investigations combining cryo-EM and solid state NMR to investigate protein fibril structure.

Keywords: cryogenic electron microscopy; protein fibril; protein structure; solid state nuclear magnetic resonance.

FIGURE 1

Negative stain TEM of fibrils and predictions for the Tm1‐LC tail domain. (a,b) Negative stain TEM micrographs of the Tm1‐LC fibrils. The scale bar measures 200 nm. (c) The sequence from the wild‐type Tm1‐LC domain is in bold and the numbering scheme is relative to the full‐length protein. The N‐terminal Ser‐Tyr tag used for UV quantification is unbolded. Dominant amino acids Ser, Asn, Ala, Thr, and Glu are in red. These five amino acids account for 70% of the sequence. (d) The PASTA 2.0 prediction of disorder and aggregation probability for Tm1‐LC.

FIGURE 2

Cryo‐EM reconstruction of Tm1‐LC fibrils. (a) Model: structural models for the repeating unit of each fibril class. The black bar indicates no model could be built due to low resolution in the density map. Top: view of the cryo‐EM density maps down the helical axis. Side: view of each cryo‐EM electron density map perpendicular to the fibril long axis. (b) Percentage of the total picked particles for each class. See also Figure S2. “Other” particles do not have consensus features. The resolutions noted in the labels are reported by RELION post‐processing. The red asterisk denotes a likely overestimate of resolution for class D due to a lack of secondary structure features. (c) Alignment of residues Y‐S373–S392 for all five monomer models from classes A, B, and C. The N‐terminal SY‐tag is in dark blue. (d) Torsion angles extracted from the cryo‐EM models. Blue represents the ψ angles and red represents the φ angles for each model of the Tm1‐LC monomer. Torsion angles are plotted for residues that are structured across all classes. Figure S5 shows the torsion angle plots by model class for all modeled residues.

FIGURE 3

Solid state NMR of Tm1‐LC fibrils. (a) INEPT‐TOBSY spectrum of the Tm1‐LC fibrils reporting on residues exhibiting fast, large‐amplitude, motions. Resonances with unambiguous residue assignments are labeled in black, tentative resonance assignments for the ambiguous, degenerate residues in the C‐terminal half of Tm1‐LC are labeled in red, and the unassigned resonances are labeled in gray. (b) The aliphatic region of the cross‐polarization based ¹³C–¹³C DARR spectrum recorded with a 50 ms mixing time, reporting on relatively rigid residues in the fibrils. Resonances with unambiguous residue assignments are labeled in black, the unassigned resonances are labeled in gray, and interresidue cross‐peaks between S386 and either N385 or N387 are labeled in cyan.

FIGURE 4

Secondary structure information derived from the solid state NMR measurements. (a) Secondary chemical shifts relative to the random coil chemical shifts obtained from the Poulsen IDP calculator for the assigned residues from the cross‐polarization and scalar coupling based experiments. Red bars represent residues with tentative assignments from INEPT‐TOBSY spectrum. (b) The φ and ψ torsion angle predictions from the TALOS‐N analysis of the assigned chemical shifts from the cross‐polarization experiments. The error bars represent the standard deviations from TALOS‐N.

FIGURE 5

Comparison of secondary structure between solid state NMR and cryo‐EM. (a) Backbone torsion angle comparison represented as the difference between the mean angle across all five modeled Tm1‐LC chains and the solid state NMR predictions from TALOS‐N. The mean and standard deviation of the difference in torsion angles between cryo‐EM and solid state NMR for all ψ and φ angles measured is −10° ± 20°. Each amino acid has a red bar for the φ angle and a blue bar for the ψ angle. The gray regions represent stretches of residues characterized exclusively by solid state NMR measurements. The black error bars are the standard deviations reported by the TALOS‐N prediction. (b) Consensus secondary structure from the cryo‐EM models compared with the secondary structure derived from the solid state NMR secondary chemical shifts and TALOS‐N torsion angle predictions. The dark green arrows represent β‐strand regions, light green arrows represent β‐strand‐like regions in the turn or β‐strands not observed in all cryo‐EM classes, and the solid gray lines represent observed amino acids without well‐defined secondary structure, but includes both rigid and mobile regions of the Tm1‐LC.