Article pubs.acs.org/Biomac
Hydration Control of the Mechanical and Dynamical Properties of Cellulose Loukas Petridis,*,† Hugh M. O’Neill,‡,§ Mariah Johnsen,†,∥ Bingxin Fan,⊥ Roland Schulz,†,# Eugene Mamontov,○ Janna Maranas,⊥ Paul Langan,‡,§,△ and Jeremy C. Smith†,# †
UT/ORNL Center for Molecular Biophysics, ‡Biology and Soft Matter Division, §Center for Structural Molecular Biology, and Chemical and Engineering Materials Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, United States ∥ Ripon College, Ripon, Wisconsin 54971, United States ⊥ Department of Chemical Engineering, Pennsylvania State University, University Park, Pennsylvania 16802, United States # Department of Biochemistry and Cellular and Molecular Biology, University of Tennessee, Knoxville, Tennessee 37996, United States △ Department of Chemistry, University of Toledo, Toledo, Ohio 43606, United States ○
S Supporting Information *
ABSTRACT: The mechanical and dynamical properties of cellulose, the most abundant biomolecule on earth, are essential for its function in plant cell walls and advanced biomaterials. Cellulose is almost always found in a hydrated state, and it is therefore important to understand how hydration influences its dynamics and mechanics. Here, the nanosecond-time scale dynamics of cellulose is characterized using dynamic neutron scattering experiments and molecular dynamics (MD) simulation. The experiments reveal that hydrated samples exhibit a higher average mean-square displacement above ∼240 K. The MD simulation reveals that the fluctuations of the surface hydroxymethyl atoms determine the experimental temperature and hydration dependence. The increase in the conformational disorder of the surface hydroxymethyl groups with temperature follows the cellulose persistence length, suggesting a coupling between structural and mechanical properties of the biopolymer. In the MD simulation, 20% hydrated cellulose is more rigid than the dry form, due to more closely packed cellulose chains and water molecules bridging cellulose monomers with hydrogen bonds. This finding may have implications for understanding the origin of strength and rigidity of secondary plant cell walls. The detailed characterization obtained here describes how hydration-dependent increased fluctuations and hydroxymethyl disorder at the cellulose surface lead to enhancement of the rigidity of this important biomolecule.
■
monomers for fermentation to second-generation biofuels9 or other high-value products.7 We consider here how water molecules modulate the mechanics and dynamics of cellulose. Water is known to induce a temperature-dependent dynamic transition in proteins10 and nucleic acids,11 in which the internal biopolymer dynamics become enhanced above a transition temperature of ∼180−240 K. Furthermore, water often renders biomaterials softer. For example, hydration of gelatin films to a water content above 14% significantly decreases the elasticity modulus.12 Given that cellulose is nearly always found in a hydrated state, there is a need for a comprehensive understanding of how hydration affects the dynamical and mechanical properties of the polymer, both for fundamental reasons as well as for fully utilizing its potential as a feedstock for bioproducts, or for the
INTODUCTION Cellulose, an unbranched chain of glucose monomers, constitutes half of plant biomass and is the most abundant biopolymer on earth.1 Individual cellulose glucan chains pack side-by-side to form crystalline microfibrils that provide structural rigidity to plant cell walls. Therefore, the mechanical properties of cellulose are essential to its function. For example, in primary walls, which need to be extensible and strong, individual cellulose fibrils are well separated.2 In contrast, in secondary walls, which have to be strong and not extensible, cellulose is found in oriented bundles of microfibrils.2 Due to its abundance and impressive mechanical properties, its axial elastic modulus is greater than of Kevlar,3 cellulose is a renewable and sustainable source of biomaterials.3,4 Single fibers of bacterial cellulose have a very high Young’s modulus (∼80 GPa) as measured by Raman spectroscopy5 and atomic force microscopy.6 Examples of utilization of the intact biopolymer include cellulose nanoparticle composites,3 thermoplastic polymer blends7 and biomedical devices.8 Alternatively, cellulose can be deconstructed to its constituent glucose © XXXX American Chemical Society
Received: August 12, 2014 Revised: October 16, 2014
A
dx.doi.org/10.1021/bm5011849 | Biomacromolecules XXXX, XXX, XXX−XXX
Biomacromolecules
Article
incoherent scattering length and position at time t of atom k, respectively, and N is the total number of atoms in the system. To account for finite instrument resolution (Δω), the elastic incoherent neutron scattering intensity is taken as
development of advanced materials. The hydration dependence of cellulose dynamics has been studied with dielectric spectroscopy,13 which probes the 1 μs to 10 s orientational motion of groups that have a permanent dipole moment, that is, hydroxyl groups in the case of cellulose. It was found that, compared to a vacuum-dried sample, addition of 0.05 w/w water reduces the mobility of the dominant low temperature (T < 273 K) relaxation. However, the molecular origin of this relaxation is not agreed upon, as it has been attributed to different types of motion: a segmental motion of the polymer chain13a or localized side-chain movement.13b Another study used solid-state nuclear magnetic resonance (NMR) spectroscopy to investigate the rigidity of dry and hydrated onion cell walls.14 It was inferred that the rigidity of the cellulose component of the cell walls is not significantly affected by hydration.14 Here, quasi-elastic incoherent dynamic neutron scattering, which directly probes the motions of nonlabile hydrogen atoms on the ps-ns time scale, is employed to study the dependence of cellulose dynamics on temperature and hydration. Complementary atomistic molecular dynamics (MD) simulations allow decomposition of the experimental signal into motions of chemically equivalent types of atoms. We are thus able to furnish a molecular-level description of how these motions change as a function of temperature and hydration. The simulations are further employed to determine how the rigidity of cellulose changes with hydration, finding that hydrating cellulose to 20% stiffens the fiber. The molecular-level information gained from this combined experimental and theoretical approach provides a detailed description of cellulose dynamics and a microscopic mechanism of how it affects its structural and mechanical properties.
■
Δω
Sel(Q ) ≈
⎛ Q 2 Δr 2 ⎞ Sel(Q , T ) ≈ exp⎜− ⎟ Sel(Q , T = 4K) 6 ⎝ ⎠
1 2π
∫ dte
1 N
∑ bk 2⟨e−iQ[rk(t) − rk(0)]⟩ k
(3)
where the wave-vector Q ≪ (6/Δr ) . Indicative plots of the fit are shown in Figure S2 in the Supporting Information. The MSD can be also independently directly calculated from MD simulations. Molecular Dynamics Simulations. The simulated system contains four aligned cellulose Iα17 fibers, the predominant allomorph of bacterial cellulose, each consisting of 36 chains with a chain length of 80 glucose monomers. The shape of the fibers is “diamond”, with the hydrophilic (100) and (010) crystallographic faces predominantly exposed to the solvent.1b Therefore, the interaction of water with cellulose hydrophobic surfaces is of secondary importance in the present context. The fibers were solvated at two different degrees of hydration to match the experiments: h = 0.05, similar to the dry sample, and h = 0.20 g of water per g of cellulose. The CHARMM C36 carbohydrate force field18 and the TIP4P-EW19 water model were employed. The simulations were performed with the program GROMACS 4.5.5.20 using a time step of 2 fs. Bonds involving hydrogen atoms were constrained using the LINCS21 algorithm (fourth order with one iteration), and for water the SETTLE algorithm was used.22 Neighbor searching was performed every 10 steps. Periodic boundary conditions were employed, and the PME algorithm23 was used for electrostatic interactions. A cutoff of 12 Å was used for the neighbor searching and real-space electrostatics. For the van der Waals interactions a switch function was used for distances 9−10 Å. Temperature coupling was performed with the V-rescale algorithm24 (τ = 0.1 fs) and pressure coupling with the Parrinello− Rahman algorithm25 (τ = 1 fs). Each system was simulated at a pressure of 105 Pa and at 16 temperatures: from T = 153 to 293 K in 10 K increments and at 298 K. Simulations were 11 ns long at each temperature. Data from the last 10 ns of simulation at each temperature were analyzed. The magnitudes of the MSDs were determined from the simulations as ⟨Δr2⟩ = ⟨[r(t) − r(0)]2⟩, where r(t) is the atomic position at time t. All cellulose hydroxyl hydrogen atoms were considered exchanged. A buried atom was defined as having zero solvent accessible surface area at the midpoint of the trajectory, t = 6 ns. Quasi-elastic spectra were computed using SASSENA,26 having assigned the deuterium scattering length density to all water and cellulose exchangeable hydrogen atoms. A geometric criterion was employed to determine the presence of hydrogen bonds (HB), in which two molecules are considered bonded if their donor−acceptor separation is less than 3.5 Å and the donorhydrogen-acceptor angle greater than 150°. The dynamic behavior of the hydrogen bonds was examined by defining a binary function f(t) equal to 1 when a hydrogen bond is present at time t and 0 otherwise. The autocorrelation function C(t) of f(t): 2 2
Sample Preparation. The procedure for growth of Acetobacter xylinus sub sp. sucrofermentans (ATCC 700178) in minimal media and purification of cellulose has been described previously.15 For the neutron scattering experiments, the cellulose pellicles were dispersed by grinding in a commercial blender for 5 min. The cellulose was then suspended in H2O and repeatedly dried by lyophilization and isobarically hydrated in a sealed dish with D2O. Neutron Scattering Experiments. Neutron scattering experiments were performed on the backscattering BASIS spectrometer16 at the Spallation Neutron Source at Oak Ridge National Laboratory on two powder samples of bacterial cellulose: dry and hydrated at h=0.20 g D2O per g cellulose. The energy resolution of BASIS is 1.75 μeV half-width half-maximum (HWHM), corresponding to ℏ/HWHM ≈ 400 ps, where ℏ is the Planck constant divided by 2π. Raw data were normalized to a vanadium standard and corrected for empty cell contributions. Analysis of Neutron Scattering Experiments. Incoherent neutron scattering directly determines atomic fluctuations in biomolecules on the ps-ns time scale, and has been used widely to study the dynamics of various biological systems.10a As shown in Figure S1 in the Supporting Information, the scattering is dominated by the signal of the cellulose hydrogen atoms, as their incoherent scattering cross section is >40× larger than that of other atoms (including deuterium). The experimentally determined quantity is the incoherent dynamic structure factor S(Q,ω):
S(Q , ω) =
(2)
An important quantity derived from Sel is the average atomic meansquare displacement (MSD), which quantifies the amplitude of atomic motions in cellulose. The experimental average MSD (⟨Δr2⟩) of the cellulose nonexchangeable hydrogen atoms is be derived by fitting the normalized elastic intensity using a Gaussian approximation:
METHODS
−iωt
∫−Δω dωS(Q , ω)
C(t ) =
f (0)f (t )⟩ f2
(4)
was computed and averaged over all possible time origins and similar hydrogen bond types.27 The persistence length (lp) of a cellulose chain was defined via the bending angle θ(s) between the vectors tangent to two glucose monomers whose contour length separation is s:
(1)
where ℏQ and ℏω are the momentum and energy transferred from the scattered neutron to the sample, respectively, bk and rk(t) are the B
dx.doi.org/10.1021/bm5011849 | Biomacromolecules XXXX, XXX, XXX−XXX
Biomacromolecules ⎛ s⎞ ⟨cos θ(s)⟩ = ni ·ni + s = exp⎜⎜ − ⎟⎟ ⎝ lp ⎠
Article
increase in their MSD, indicating the underlying atomic fluctuations are not affected by hydration. However, a dynamical transition, indicated by a steep increase of MSD at T > 210 K is observed only in the hydrated sample only. This transition arises from the onset of large-amplitude, anharmonic dynamics. The cellulose dynamical transition is different to its glass transition, the latter occurring at ∼500 K for dry cellulose.28 Furthermore, the observed increase in MSD of cellulose on hydration is considerably smaller than for proteins and nucleic acids,10c,29 and the reason for this can be determined from molecular dynamics (MD) simulation. The MSDs in Figure 1 represent the average motions of four types of nonexchangeable cellulose hydrogen atoms. We make a distinction between those hydrogens bonded to the ring and hydroxymethyl carbons, see the inset of Figure 1, and these two atom types can be further subdivided into those exposed to the solvent and those buried in the cellulose core. To decompose the average MSD into contributions from these four types of hydrogen atom, we conducted atomistic MD simulations of cellulose fibers hydrated to h = 0.05, to mimic the dry sample, and h = 0.20. The overall temperature dependence MSDs derived from the MD simulations, calculated at each temperature at t = 400 ps (the resolution time of the spectrometer employed in the experiments), closely match the neutron scattering results, see Figure 1B,C. Furthermore, in all simulations, the MSD of the hydroxymethyl hydrogens is always higher than that of the ring hydrogen atoms. Furthermore, those ring and hydroxymethyl hydrogens exposed to the solvent have higher MSD values at h = 0.20 than those in the core. Higher enhancement of mobility upon hydration for the surface groups relative to those in the core is also found in lignin, another abundant plant cell wall polymer.30 A strong temperature dependence of the MSD is found only in the h = 0.20 simulation and only for the exposed hydroxymethyl hydrogens, which account for only 1.5% of the total nonexchangeable hydrogen atoms in the system. In contrast, the average MSD of buried ring hydrogen atoms, which comprise 70% of the nonexchangeable H, is nearly independent of temperature (see Table S1 in the Supporting Information). Therefore, the weaker experimental dynamical transition in cellulose relative to other biopolymers arises from the relatively small proportion of nonexchangeable hydrogens in the solvent-exposed side chain hydroxymethyl groups. Hydroxymethyl Surface Disorder. We now focus on examining in detail the dynamics of the hydroxymethyl group. This involves the population of three low-energy rotameric conformations of the dihedral angle ϕ = C4−C5−C6−O6, see Figure 2. These three states are labeled GT, TG, and GG, referring to the gauche or trans position of O6 relative to both O5 and C4, corresponding to ϕ = −180°, −60°, and +60°, respectively.31 The increased fluctuations of the exposed hydroxymethyl hydrogens arises from the activation of three types of motion: rigid-body translation of the entire CH3OH group, anharmonic jumps between rotameric states and localized motion inside rotameric states, the last including intrawell dihedral rotation, angle bending, and bond stretching. To decouple the first from the latter two, the MSD of the C6 atom was subtracted from the total hydrogen MSD, see Figure 3. The resulting ΔMSD has a strong temperature and hydration dependence, thus demonstrating that non rigid-body motions contribute to the increased fluctuations shown in Figure 1.
(5)
where ni is a unit vector connecting atoms C1 and C4 of monomer i (see inset of Figure 5) and ⟨···⟩ indicate time and ensemble averaging.
■
RESULTS AND DISCUSSION Heterogeneous Cellulose Dynamics. Elastic incoherent neutron scattering experiments were conducted on two powder samples of bacterial cellulose: dry and hydration level h = 0.20 g D2O per g cellulose. The experiments probe predominantly the motion of nonexchangeable hydrogen atoms of cellulose, see Figure S1 in the Supporting Information. The experimental atomic mean-square displacement (MSD), a quantitative measure of the average magnitude of atomic motions, of the two cellulose samples is shown in Figure 1A. Up to ∼210 K both dry and hydrated cellulose display a relatively small
Figure 1. (A) Mean square displacement (MSD) as a function of temperature obtained from neutron scattering experiments by fitting eq 3 of dry and hydrated cellulose. (B) and (C) MSD obtained from MD simulation at a time lag of 400 ps. The simulation-derived MSD is decomposed into contributions from solvent-exposed (“e-”) and buried (“b-”) hydroxymethyl and chain atoms, see inset in A. C
dx.doi.org/10.1021/bm5011849 | Biomacromolecules XXXX, XXX, XXX−XXX
Biomacromolecules
Article
surface chains exhibit increasing disorder with temperature, core chains (inset) are found only in TG. Therefore, the increase with T in ΔMSD of the buried hydrogens does not arise from jumps between rotameric chains and is thus purely due to local motion in one rotameric state. Assuming that local intrawell motion is the same for atoms on the surface and in the core, we can therefore infer that the difference between ΔMSD of buried and exposed hydrogen atoms at h = 0.20 is due to jumps between rotameric states. The above considerations allow decomposition of the hydroxymethyl hydrogen MSD into its three contributions, see the inset of Figure 3. Although rigidbody translation dominates the MSD at temperatures
