Helix conformations in 7TM membrane proteins determined using oriented-sample solid-state NMR with multiple residue-specific 15N labeling

Abstract

Oriented solid-state NMR in combination with multiple-residue-specific (15)N labeling and extensive numerical spectral analysis is proposed to determine helix conformations of large membrane proteins in native membranes. The method is demonstrated on uniaxially oriented samples of (15)N-methionine, -valine, and -glycine-labeled bacteriorhopsin in native purple membranes. Experimental two-dimensional (1)H-(15)N dipole-dipole coupling versus (15)N chemical shift spectra for all samples are analyzed numerically to establish combined constraints on the orientation of the seven transmembrane helices relative to the membrane bilayer normal. Since the method does not depend on specific resonance assignments and proves robust toward nonidealities in the sample alignment, it may be generally feasible for the study of conformational arrangement and function-induced conformation changes of large integral membrane proteins.

FIGURE 1

(a) Structure of the M1 and M2 TM helices of the Ca²⁺ ATPase (43). (b) Helical wheel plots of the same helices as in a, with the Ala residues highlighted. (c, e, and g) Simulated PISEMA spectra employing ideal helices with helix-tilt (τ) and rotational-pitch (ρ) angles τ^M1 = 13.92°, ρ^M1 = −294°, and τ^M2 = 185.9°, ρ^M2 = −88° corresponding to an [¹⁵N]Ala-labeled version of the M1, M2 fragment of the Ca²⁺ ATPase. (c) Ideal simulation. (e and g) The same simulation as in c, assuming fluctuations of up to ±5 ppm/±500 Hz (e), and ±10 ppm/±1000 Hz (g) for the resonance positions in the ¹⁵N chemical shift/¹H-¹⁵N dipolar coupling dimensions. The crosses in c, e, and g represent the resonance positions, and the contour plots represent the spectra resulting from applying line broadening of 10 ppm/1000 Hz in the ¹⁵N chemical shift/¹H-¹⁵N dipolar coupling dimensions. (d, f, and h) Simulated PISEMA spectra and corresponding resonance positions resulting from the 100 best of 500 independent deconvolutions of the broadened spectra in c, e, and g, respectively.

FIGURE 2

(a) Schematic representation of the secondary structure of bR based on the 1C3W crystal structure (44) highlighting the Gly, Val, and Met residues. (b) Helical wheel plots of the seven TM helices illustrating the diversity of the labeling patterns for the helices achieved by considering a combination of [¹⁵N]Gly, [¹⁵N]Val, and [¹⁵N]Met labeling.

FIGURE 3

PISEMA spectra of bR. (a) Simulated powder spectrum employing chemical shift and dipolar coupling parameters of δ_iso = 120 ppm, δ_aniso = 99 ppm, η_σ = 0.21, and b_IS = 9940 Hz. (b–g) Experimental (b, d, and f) and simulated (c, e, and g) PISEMA spectra of (b and c) [¹⁵N]Gly bR, (d and e) [¹⁵N]Val bR, and (f and g) [¹⁵N]Met bR. In all spectra, the dipolar dimension was corrected for the theoretical scaling factor In the experimental spectra, we observe that minor peaks around zero frequency in the ¹H-¹⁵N dipole-dipole dimension, in particular in the loop region (∼60 ppm in the ¹⁵N chemical shift dimension), are experimental artifacts.

FIGURE 4

(a) Sorted RMS deviations between experimental and simulated PISEMA spectra including data from [¹⁵N]Gly bR, [¹⁵N]Val bR, and [¹⁵N]Met bR. (b–h) Helix-tilt angles resulting from the 250 optimizations yielding the lowest RMS deviations (i.e., those to the left of the vertical dashed line in a), along with plots of the tilt-angle distributions (black lines) and best-fit Gaussian profiles (dashed gray lines), for helices A–G, respectively.

FIGURE 5

Helix-tilt distributions resulting from optimizations employing only the data listed in the leftmost column for the seven helices (A–G). Missing plots are due to the lack of data for the particular helix, and gray plots are those with only one ¹⁵N label in the particular helix.