Atomistic, macromolecular model of the Populus secondary cell wall informed by solid-state NMR

Abstract

Plant secondary cell walls (SCWs) are composed of a heterogeneous interplay of three major biopolymers: cellulose, hemicelluloses, and lignin. Details regarding specific intermolecular interactions and higher-order architecture of the SCW superstructure remain ambiguous. Here, we use solid-state nuclear magnetic resonance (ssNMR) measurements to infer refined details about the structural configuration, intermolecular interactions, and relative proximity of all three major biopolymers within air-dried Populus wood. To enhance the utility of these findings and enable evaluation of hypotheses in a physics-based environment in silico, the NMR observables are articulated into an atomistic, macromolecular model for biopolymer assemblies within the plant SCW. Through molecular dynamics simulation, we quantitatively evaluate several variations of atomistic models to determine structural details that are corroborated by ssNMR measurements.

Fig. 1.. 1D ¹H-¹³C MultiCP-MAS ssNMR provides a broad overview of lignified Populus SCWs.

General assignment regions are indicated for the predominant constituent polymers cellulose (gray), xylan (blue), and lignin (green), along with representative structures and their assignments indicated with color-coding. Chemical shift information and peak profiles extracted from a combined set of 1D and 2D ssNMR data were used to inform spectral deconvolution efforts used throughout this work. The raw data are shown in black, and the resultant fit is shown in red. C, cellulose; Xn, xylan; Ac, acetyl; GalA, galacturonic acid (pectin); GlcA, α-d-glucuronic acid; [4OMe]GlcA, (4-O-methyl)-α-d-glucuronic acid.

Fig. 2.. Selective 1D MultiCP-DARR difference data quantify polymer-polymer contacts in the ~1nm length scales.

(A) Selective MultiCP 1D DARR difference pulse sequence. The experiment results in a selective 1D ¹³C spectrum for which only ¹³C signals (sinks) that have exchanged magnetization with the selected resonance (source) during the variable spin-diffusion period τ_m are observed, while all other carbons are not observed because of deconstructive cancelation. (B) Example xylan-selected (Ac^Me site at 22 ppm) 1D MultiCP-DARR difference spectrum (τ_m = 3 s, red trace) is shown next to a nonselective 1D MultiCP-DARR experiment with the same mixing time (relaxation compensated, black trace), which shows all carbons. Comparing the selective and nonselective 1D spectra reveals the relative abundance of ¹³C sites that have or have not received polarization from the selected resonance during the mixing period. (C) Double-difference spectrum at 3 s reveals the ¹³C sites that reside outside of ¹³C-¹³C spin-diffusion range from the selected site during τ_m = 3 s, resembling that of Avicel cellulose (D). (E) Cellulose, xylan, and lignin representative structures with carbon positions, along with their ¹³C spectral assignments indicated on sub-spectra obtained from selective 1D MultiCP-DARR difference data. ¹³C-¹³C spin-diffusion rate constants that describe how quickly magnetization moves from source to sink carbons are indicated with red arrows, imparting distance information. (F) Stacked plot of xylan-selected 1D MultiCP-DARR difference spectra with τ_m ranging from 200 to 5000 ms. Plots are scaled by the intensity of the Xn Ac^Me site. (G and H) Magnetization recovery plots representing calculated xylan-sourced (G) and lignin-sourced (H) through-space contacts at each mixing time out to 5000 ms. Traces from five samples from two biological replicates are shown with thin lines, and their averages are shown with thick lines. Error bars are propagated from the signal to noise.

Fig. 3.. Atomistic SCW periodic model variants and comparison of their proximity analyses with ssNMR spin-diffusion experiments.

(Left) Snapshots of several molecular model variants for the poplar SCW. The naming convention is as follows: The letter refers to the relative arrangement of hemicellulose and lignin, where “a” denotes a model with all xylan bound on cellulose, and no xylan interspersed with lignin; “b” denotes a model with 70% hemicellulose bound to cellulose, and the remaining 30% interspersed in the surrounding lignin-xylan matrix; “d” denotes a model with lignin-bound cellulose and with hemicellulose as the surrounding layer; “e” denotes a model with phase-separated hemicellulose and lignin both bound to cellulose at the top and bottom, respectively; “h” denotes a model with hemicellulose interspersed within the 18-chain elementary cellulose fibril leaving room for increased direct lignin-cellulose interaction. The number after the decimal denotes the number of 18-chain elementary fibrils in the model. b.8 and b.10 are larger systems constructed with 70% of xylan bound to cellulose, and the remaining 30% interspersed with lignin and consist of eight and ten 18-chain cellulose elementary fibrils, respectively. (Right) Calculated percentage of sink atoms within 1 nm of source atoms in the model variants compared to that obtained from ssNMR spin-diffusion experiments depicted on a polar plot. The four dashed lines originating from the center indicate the four ssNMR magnetization metrics, with the experimental values indicated as black dots and the colors indicating the various models.

Fig. 4.. Atomistic model of the lignocellulose assembly within the Populus SCW that best represents ssNMR observables.

(A) Molecular representation of the individual biopolymer constituents. (B) Macromolecular assembly of cellulose, xylan, and lignin. The xylan domains on the surface of cellulose adopt an extended configuration with decorations pointing away from the cellulose surface toward lignin. (C) Visualization of the cellulosic components only, showing two core bundles comprising four 18-chain elementary fibrils. (D) Visualization of the hemicellulose components only shows the nearly complete sheath formed around the cellulose minimizing direct contact between cellulose and lignin; also, some xylan is found between core bundles, isolated from lignin. (E) Lignin interacts predominantly with hemicellulose and displays minimal direct contact with cellulose.

Fig. 5.. Contrasting conformation of xylan polymers with different proximities to cellulose.

(A) Xylan bound to cellulose (orange) adopts conformations that enable acetate groups to consistently orient away from the cellulose surface. Unbound xylose (light blue) adopts a more random orientation of acetate groups. Exemplary xylan polymers are shown in licorice representation, with their acetate groups highlighted as spheres, and cellulose is shown as a white surface. (B) Unbound xylan φ + ψ distributions are not exclusively 3₁ and span a large range of values with a peak at ~170⁰. (C) For xylan chains with similarly oriented acetate groups, a wide distribution of canonical φ + ψ conformations is observed, although a peak ~100⁰ is prominent. (D) θ^2f(O₂-C₅-C₅-O₂) values for the unbound xylan chain are widely distributed corresponding with the nonuniform acetate group orientations. (E) Bound xylan exhibits a narrow θ^2f distribution centered at 0⁰, which highlights the consistent orientation of acetate groups. Definition of canonical φ + ψ and θ^2f(O₂-C₅-C₅-O₂) dihedral measurements is portrayed in (F) and (G), respectively.