Computer simulation studies of microcrystalline cellulose Iβ

Overview of computer modeling of cellulose

Although it has a deceptively simple primary structure, the collective organization of bulk cellulose, particularly as it exists in cellulose fibers in the cell walls of living plants and other organisms, is quite diverse and complex. While some experimental techniques, such as vibrational spectroscopy and diffraction from partially crystalline samples, are able to provide insights into the organization of bulk cellulose, its intrinsic complexity has left many questions still unanswered. For this reason, additional probes of cellulose structure would be highly desirable. With the continuing advances in computer power through massive parallelization, and the steady progress in computer codes and force fields for modeling carbohydrate systems, molecular mechanics simulations have become an attractive means of studying cellulosic systems at the atomic and molecular level. The coming decade will almost certainly see remarkable advances in the understanding of cellulose using such simulations.

Nanostructure of cellulose microfibrils in spruce wood

The structure of cellulose microfibrils in wood is not known in detail, despite the abundance of cellulose in woody biomass and its importance for biology, energy, and engineering. The structure of the microfibrils of spruce wood cellulose was investigated using a range of spectroscopic methods coupled to small-angle neutron and wide-angle X-ray scattering. The scattering data were consistent with 24-chain microfibrils and favored a "rectangular" model with both hydrophobic and hydrophilic surfaces exposed. Disorder in chain packing and hydrogen bonding was shown to increase outwards from the microfibril center. The extent of disorder blurred the distinction between the I alpha and I beta allomorphs. Chains at the surface were distinct in conformation, with high levels of conformational disorder at C-6, less intramolecular hydrogen bonding and more outward-directed hydrogen bonding. Axial disorder could be explained in terms of twisting of the microfibrils, with implications for their biosynthesis.

Ab initio study of structure and interconversion of native cellulose phases

Dispersion-interaction corrected DFT simulations are performed to study the structure of two allomorphs of native cellulose I. Good agreement between theoretical and experimental data is achieved. Two H-bond patterns, previously identified experimentally, are examined for both allomorphs. The transition mechanism for the conversion between the phase I(α) and I(β) is studied by means of constrained relaxations. New metastable intermediate phase occurring on the I(α) → I(β) route is identified, and the corresponding structural data are reported.

A Solvent-Free Coarse Grain Model for Crystalline and Amorphous Cellulose Fibrils

Understanding biomass structure and dynamics on a range of time and length scales is important for the development of cellulosic biofuels. Here, to enable length and time scale extension, we develop a coarse grain (CG) model for molecular dynamics (MD) simulations of cellulose. For this purpose, we use distribution functions from fully atomistic MD simulations as target observables. A single bead per monomer level coarse graining is found to be sufficient to successfully reproduce structural features of crystalline cellulose. Without the use of constraints the CG crystalline fibril is found to remain stable over the maximum simulation length explored in this study (>1 μs). We also extend the CG representation to model fully amorphous cellulose fibrils. This is done by using an atomistic MD simulation of fully solvated individual cellulose chains as a target for developing the corresponding fully amorphous CG force field. Fibril structures with different degrees of crystallinity are obtained using force fields derived using a parameter coupling the crystalline and amorphous potentials. The method provides an accurate and constraint-free approach to derive CG models for cellulose with a wide range of crystallinity, suitable for incorporation into large-scale models of lignocellulosic biomass.

Computerized molecular modeling of carbohydrates

Computerized molecular modeling continues to increase in capability and applicability to carbohydrates. This chapter covers nomenclature and conformational aspects of carbohydrates, perhaps of greater use to carbohydrate-inexperienced computational chemists. Its comments on various methods and studies might be of more use to computation-inexperienced carbohydrate chemists. New work on intrinsic variability of glucose, an overall theme, is described.

Cellulose nanomaterials review: structure, properties and nanocomposites

This critical review provides a processing-structure-property perspective on recent advances in cellulose nanoparticles and composites produced from them. It summarizes cellulose nanoparticles in terms of particle morphology, crystal structure, and properties. Also described are the self-assembly and rheological properties of cellulose nanoparticle suspensions. The methodology of composite processing and resulting properties are fully covered, with an emphasis on neat and high fraction cellulose composites. Additionally, advances in predictive modeling from molecular dynamic simulations of crystalline cellulose to the continuum modeling of composites made with such particles are reviewed (392 references).

Examination of the α-chitin structure and decrystallization thermodynamics at the nanoscale

Chitin is the primary structural material of insect and crustacean exoskeletons and fungal and algal cell walls, and as such it is the one of the most abundant biological materials on Earth. Chitin forms linear polymers of β1,4-linked-N-acetyl-D-glucosamine (GlcNAc), and in Nature, enzyme cocktails deconstruct chitin to GlcNAc. The mechanism of chitin deconstruction, like that of cellulose deconstruction, has been under investigation due to its importance in the global carbon cycle and in production of renewable and sustainable products from biological matter. To further understand the nanoscale properties of chitin, here we simulate crystals of α-chitin, which is the most prevalent form in Nature. We find excellent agreement with the recently reported crystal structure and we report the salient features of the simulations related to crystalline stability. We also compute the thermodynamic work required to peel individual chains from α-chitin surfaces, which a chitinase enzyme must conduct to deconstruct chitin. Compared with previous simulations of native plant cellulose Iβ, α-chitin exhibits higher decrystallization work for chains in the middle of surfaces and similar work for chains on the edges of crystals. Unlike cellulose, the free energy profile is dominated by a single bifurcated hydrogen bond between chains formed by the GlcNAc side chains and the O6 atoms on the primary alcohol group. This study highlights the molecular features of chitin that make it such a tough, recalcitrant material, and provides a key thermodynamic parameter in our quantitative understanding of how enzymes contribute to the turnover of carbohydrates in the biosphere.

Molecular-level origins of biomass recalcitrance: decrystallization free energies for four common cellulose polymorphs

Cellulose is a crystalline polymer of β1,4-D-glucose that is difficult to deconstruct to sugars by enzymes. The recalcitrance of cellulose microfibrils is a function of both the shape of cellulose microfibrils and the intrinsic work required to decrystallize individual chains, the latter of which is calculated here from the surfaces of four crystalline cellulose polymorphs: cellulose Iβ, cellulose Iα, cellulose II, and cellulose III(I). For edge chains, the order of decrystallization work is as follows (from highest to lowest): Iβ, Iα, ΙΙΙ(Ι), and II. For cellulose Iβ, we compare chains from three different locations on the surface and find that an increasing number of intralayer hydrogen bonds (from 0 to 2) increases the intrinsic decrystallization work. From these results, we propose a microkinetic model for the deconstruction of cellulose (and chitin) by processive enzymes, which when taken with a previous study [Horn et al. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 18089] identifies the thermodynamic and kinetic attributes of enzyme and substrate engineering for enhanced cellulose (or chitin) conversion. Overall, this study provides new insights into the molecular interactions that form the structural basis of cellulose, which is the primary building block of plant cell walls, and highlights the need for experimentally determining microfibril shape at the nanometer length scale when comparing conversion rates of cellulose polymorphs by enzymes.

Applications of computational science for understanding enzymatic deconstruction of cellulose

Understanding the molecular-level mechanisms that enzymes employ to deconstruct plant cell walls is a fundamental scientific challenge with significant ramifications for renewable fuel production from biomass. In nature, bacteria and fungi use enzyme cocktails that include processive and non-processive cellulases and hemicellulases to convert cellulose and hemicellulose to soluble sugars. Catalyzed by an accelerated biofuels R&D portfolio, there is now a wealth of new structural and experimental insights related to cellulases and the structure of plant cell walls. From this background, computational approaches commonly used in other fields are now poised to offer insights complementary to experiments designed to probe mechanisms of plant cell wall deconstruction. Here we outline the current status of computational approaches for a collection of critical problems in cellulose deconstruction. We discuss path sampling methods to measure rates of elementary steps of enzyme action, coarse-grained modeling for understanding macromolecular, cellulosomal complexes, methods to screen for enzyme improvements, and studies of cellulose at the molecular level. Overall, simulation is a complementary tool to understand carbohydrate-active enzymes and plant cell walls, which will enable industrial processes for the production of advanced, renewable fuels.

Reduction of surface hydrophobicity using a stimulus-responsive polysaccharide

The adsorption of carboxymethyl cellulose (CMC) onto a hydrophobic self-assembled monolayer has been characterized using the quartz crystal microbalance (with dissipation monitoring, QCM-D). Adsorption was studied as a function of initial solution conditions. CMC adsorbs to a greater extent at high ionic strength (10(-1) M KCl as opposed to 10(-2) M KCl) or low pH (3 as opposed to 9). The solution conditions that yielded the lowest initial adsorbed amount (10(-2) M KCl, pH 9) were used as a reference to investigate the response of the adsorbed layer to a switch in solution conditions after adsorption (i.e., to higher ionic strength (10(-1) M KCl) or lower pH (pH 3)). The adsorbed layer released significant amounts of hydration water after each solution switch, as determined by the QCM-D measurements. This expulsion of hydration water was fully reversible. For the two solution switches, reducing the solution pH resulted in a more pronounced change in the amount of hydration water within the adsorbed CMC, accompanied by a distinct conformational change, as determined from a QCM D-f plot. In addition to studying adsorption using QCM-D, the effect of adsorbed CMC on surface hydrophobicity has been investigated using captive bubble contact angle measurements. The effect of the polymer on the contact angle of the surface was seen to be greatest when adsorbed at low pH or at higher ionic strength. CMC was also seen to have a significantly enhanced ability to reduce the surface hydrophobicity after both the ionic strength and pH switches, lowering the advancing water contact angle by 6 and 23° and the receding water contact angle by 10 and 40° for the ionic strength and pH switches, respectively. As with the change in hydration water content, the change in the contact angle of the polymer-coated surface following the solution switches was reversible.

Systematic docking study of the carbohydrate binding module protein of Cel7A with the cellulose Ialpha crystal model

A computer docking study has been carried out on the crystal surfaces of cellulose Ialpha crystal models for the carbohydrate binding module (CBM) protein of the cellobiohydrolase Cel7A produced by Trichoderma reesei. Binding free energy maps between the CBM and the crystal surface were obtained by calculating the noncovalent interactions and the solvation free energy at grid points covering the area of the unit cell dimensions at the crystal surface. The potential maps obtained from grid searches of the hydrophobic (110) crystal surface exhibited two distinct potential wells. These reflected the 2-fold helical symmetry of the cellulose chain and had lower binding energies at the minimum positions than those for the hydrophilic (100) and (010) crystal surfaces. The CBM-cellulose crystal complex models derived from the minimum positions were then subjected to molecular dynamics (MD) simulation under an explicit solvent system. The (110) complex models exhibited larger affinities at the interface than the (100) and (010) ones. The CBM was more stably bound to the (110) surface when it was placed in an antiparallel orientation with respect to the cellulose fiber axis. In the solvated dynamics state, the curved (110) surface resulting from the fiber twist somewhat assisted a complementary fit with the CBM at the interface. In addition to the conventional Generalized Born (GB) method, the three-dimensional reference interaction site model (3D-RISM) theory was adopted to assess a solvent effect for the solvated MD trajectories. Large exothermic values for the noncovalent interactions appeared correlated to and were mostly compensated by endothermic values for the solvation free energy. These gave total binding free energies of -13 to -28 kcal/mol. Results also suggested that the hydrogen bonding scheme was not essential for substrate specificity.

Access to cellulose limits the efficiency of enzymatic hydrolysis: the role of amorphogenesis

The efficient enzymatic saccharification of cellulose at low cellulase (protein) loadings continues to be a challenge for commercialization of a process for bioconversion of lignocellulose to ethanol. Currently, effective pretreatment followed by high enzyme loading is needed to overcome several substrate and enzyme factors that limit rapid and complete hydrolysis of the cellulosic fraction of biomass substrates. One of the major barriers faced by cellulase enzymes is their limited access to much of the cellulose that is buried within the highly ordered and tightly packed fibrillar architecture of the cellulose microfibrils. Rather than a sequential 'shaving' or 'planing' of the cellulose fibrils from the outside, it has been suggested that these inaccessible regions are disrupted or loosened by non-hydrolytic proteins, thereby increasing the cellulose surface area and making it more accessible to the cellulase enzyme complex. This initial stage in enzymatic saccharification of cellulose has been termed amorphogenesis. In this review, we describe the various amorphogenesis-inducing agents that have been suggested, and their possible role in enhancing the enzymatic hydrolysis of cellulose.

The energy landscape for the interaction of the family 1 carbohydrate-binding module and the cellulose surface is altered by hydrolyzed glycosidic bonds

A multiscale simulation model is used to construct potential and free energy surfaces for the carbohydrate-binding module [CBM] from an industrially important cellulase, Trichoderma reesei cellobiohydrolase I, on the hydrophobic face of a coarse-grained cellulose Ibeta polymorph. We predict from computation that the CBM alone exhibits regions of stability on the hydrophobic face of cellulose every 5 and 10 A, corresponding to a glucose unit and a cellobiose unit, respectively. In addition, we predict a new role for the CBM: specifically, that in the presence of hydrolyzed cellulose chain ends, the CBM exerts a thermodynamic driving force to translate away from the free cellulose chain ends. This suggests that the CBM is not only required for binding to cellulose, as has been known for two decades, but also that it has evolved to both assist the enzyme in recognizing a cellulose chain end and exert a driving force on the enzyme during processive hydrolysis of cellulose.

Computational simulations of the Trichoderma reesei cellobiohydrolase I acting on microcrystalline cellulose Ibeta: the enzyme-substrate complex

Cellobiohydrolases are the dominant components of the commercially relevant Trichoderma reesei cellulase system. Although natural cellulases can totally hydrolyze crystalline cellulose to soluble sugars, the current enzyme loadings and long digestion times required render these enzymes less than cost effective for biomass conversion processes. It is clear that cellobiohydrolases must be improved via protein engineering to reduce processing costs. To better understand cellobiohydrolase function, new simulations have been conducted using charmm of cellobiohydrolase I (CBH I) from T.reesei interacting with a model segment (cellodextrin) of a cellulose microfibril in which one chain from the substrate has been placed into the active site tunnel mimicking the hypothesized configuration prior to final substrate docking (i.e., the +1 and +2 sites are unoccupied), which is also the structure following a catalytic bond scission. No tendency was found for the protein to dissociate from or translate along the substrate surface during this initial simulation, nor to align with the direction of the cellulose chains. However, a tendency for the decrystallized cellodextrin to partially re-anneal into the cellulose surface hints that the arbitrary starting configuration selected was not ideal.

The stability of cellulose: a statistical perspective from a coarse-grained model of hydrogen-bond networks

A critical roadblock to the production of biofuels from lignocellulosic biomass is the efficient degradation of crystalline microfibrils of cellulose to glucose. A microscopic understanding of how different physical conditions affect the overall stability of the crystalline structure of microfibrils could facilitate the design of more effective protocols for their degradation. One of the essential physical interactions that stabilizes microfibrils is a network of hydrogen (H) bonds: both intrachain H-bonds between neighboring monomers of a single cellulose polymer chain and interchain H-bonds between adjacent chains. We construct a statistical mechanical model of cellulose assembly at the resolution of explicit hydrogen-bond networks. Using the transfer matrix method, the partition function and the subsequent statistical properties are evaluated. With the help of this lattice-based model, we capture the plasticity of the H-bond network in cellulose due to frustration and redundancy in the placement of H-bonds. This plasticity is responsible for the stability of cellulose over a wide range of temperatures. Stable intrachain and interchain H-bonds are identified as a function of temperature that could possibly be manipulated toward rational destruction of crystalline cellulose.

Potential of biofilm-based biofuel production

Biofilm technology has been extensively applied to wastewater treatment, but its potential application in biofuel production has not been explored. Current technologies of converting lignocellulose materials to biofuel are hampered by costly processing steps in pretreatment, saccharification, and product recovery. Biofilms may have a potential to improve efficiency of these processes. Advantages of biofilms include concentration of cell-associated hydrolytic enzymes at the biofilm-substrate interface to increase reaction rates, a layered microbial structure in which multiple species may sequentially convert complex substrates and coferment hexose and pentose as hydrolysates diffuse outward, and the possibility of fungal-bacterial symbioses that allow simultaneous delignification and saccharification. More importantly, the confined microenvironment within a biofilm selectively rewards cells with better phenotypes conferred from intercellular gene or signal exchange, a process which is absent in suspended cultures. The immobilized property of biofilm, especially when affixed to a membrane, simplifies the separation of biofuel from its producer and promotes retention of biomass for continued reaction in the fermenter. Highly consolidated bioprocessing, including delignification, saccharification, fermentation, and separation in a single reactor, may be possible through the application of biofilm technology. To date, solid-state fermentation is the only biofuel process to which the advantages of biofilms have been applied, even though it has received limited attention and improvements. The transfer of biofilm technology from environmental engineering has the potential to spur great innovations in the optimization of biofuel production.

Modelling polymer interactions of the 'molecular Velcro' type in wood under mechanical stress

Trees withstand wind and snow loads by synthesising wood that varies greatly in mechanical properties: flexible in twigs and in the stem of the sapling, and rigid in the outer part of the mature stem. The 'molecular Velcro' model of Keckes et al. [2003. Cell-wall recovery after irreversible deformation of wood. Nat. Mater. 2, 810-814] permits the simulation of the tensile properties of water-saturated wood as found in living trees. A basic feature of this model is the presence of non-covalent interactions between hemicellulose chains attached to adjacent cellulose microfibrils, which are disrupted above a threshold level of interfibrillar shear. However, other evidence does not confirm the importance of hemicellulose-hemicellulose association in the cohesion of the interfibrillar matrix. Here, we present an alternative model in which hemicellulose chains bridging continuously from one microfibril aggregate (macrofibril) to the next provide most of the cohesion. We show that such hemicellulose bridges exist and that the stripping of the bridging chains from the cellulose surfaces under the tensile stress component normal to the macrofibrils can provide an alternative triggering mechanism for shear deformation between one macrofibril and the next. When one macrofibril then slides past another, a domain of the wood cell wall can extend but simultaneously it twists until the spacing between macrofibrils is reduced again and contact through hemicelluloses bridges is restored. Overall deformation therefore takes place through a series of local stick-slip events involving temporary twisting of small domains within the wood cell wall. Modelled load-deformation curves for this modified 'molecular Velcro' model are similar, although not identical, to those for the original model. However, the mechanism is different and more consistent with current views of the structure of wood cell walls, providing a framework within which the developmental control of rigidity in wood synthesised in different parts of a tree may be considered.

Thermal Response in Crystalline Iβ Cellulose: A Molecular Dynamics Study

The influence of temperature on structure and properties of the cellulose Ibeta crystal was studied by molecular dynamics simulations with the GROMOS 45a4 force-field. At 300 K, the modeled crystal agreed reasonably with several sets of experimental data, including crystal density, corresponding packing and crystal unit cell dimensions, chain conformation parameters, hydrogen bonds, Young's modulus, and thermal expansion coefficient at room temperature. At high-temperature (500 K), the cellulose chains remained in sheets, despite differences in the fine details compared to the room-temperature structure. The density decreased while the a and b cell parameters expanded by 7.4% and 6%, respectively, and the c parameter (chain axis) slightly contracted by 0.5%. Cell angles alpha and beta divided into two populations. The hydroxymethyl groups mainly adopted the gt orientation, and the hydrogen-bonding pattern thereby changed. One intrachain hydrogen bond, O2'H2'...O6, disappeared and consequently the Young's modulus decreased by 25%. A transition pathway between the low- and high-temperature structures has been proposed, with an initial step being an increased intersheet separation, which allowed every second cellulose chain to rotate around its helix axis by about 30 degrees . Second, all hydroxymethyl groups changed their orientations, from tg to gg (rotated chains) and from tg to gt (non-rotated chains). When temperature was further increased, the rotated chains returned to their original orientation and their hydroxymethyl groups again changed their conformation, from gg to gt. A transition temperature of about 450 K was suggested; however, the transition seems to be more gradual than sudden. The simulated data on temperature-induced changes in crystal unit cell dimensions and the hydrogen-bonding pattern also compared well with experimental results.

Molecular modeling suggests induced fit of Family I carbohydrate-binding modules with a broken-chain cellulose surface

Cellobiohydrolases are the most effective single component of fungal cellulase systems; however, their molecular mode of action on cellulose is not well understood. These enzymes act to detach and hydrolyze cellodextrin chains from crystalline cellulose in a processive manner, and the carbohydrate-binding module (CBM) is thought to play an important role in this process. Understanding the interactions between the CBM and cellulose at the molecular level can assist greatly in formulating selective mutagenesis experiments to confirm the function of the CBM. Computational molecular dynamics was used to investigate the interaction of the CBM from Trichoderma reesei cellobiohydrolase I with a model of the (1,0,0) cellulose surface modified to display a broken chain. Initially, the CBM was located in different positions relative to the reducing end of this break, and during the simulations it appeared to translate freely and randomly across the cellulose surface, which is consistent with its role in processivity. Another important finding is that the reducing end of a cellulose chain appears to induce a conformational change in the CBM. Simulations show that the tyrosine residues on the hydrophobic surface of the CBM, Y5, Y31 and Y32 align with the cellulose chain adjacent to the reducing end and, importantly, that the fourth tyrosine residue in the CBM (Y13) moves from its internal position to form van der Waals interactions with the cellulose surface. As a consequence of this induced change near the surface, the CBM straddles the reducing end of the broken chain. Interestingly, all four aromatic residues are highly conserved in Family I CBM, and thus this recognition mechanism may be universal to this family.

Liquefaction of lignocellulose at high-solids concentrations

To improve process economics of the lignocellulose to ethanol process a reactor system for enzymatic liquefaction and saccharification at high-solids concentrations was developed. The technology is based on free fall mixing employing a horizontally placed drum with a horizontal rotating shaft mounted with paddlers for mixing. Enzymatic liquefaction and saccharification of pretreated wheat straw was tested with up to 40% (w/w) initial DM. In less than 10 h, the structure of the material was changed from intact straw particles (length 1-5 cm) into a paste/liquid that could be pumped. Tests revealed no significant effect of mixing speed in the range 3.3-11.5 rpm on the glucose conversion after 24 h and ethanol yield after subsequent fermentation for 48 h. Low-power inputs for mixing are therefore possible. Liquefaction and saccharification for 96 h using an enzyme loading of 7 FPU/g.DM and 40% DM resulted in a glucose concentration of 86 g/kg. Experiments conducted at 2%-40% (w/w) initial DM revealed that cellulose and hemicellulose conversion decreased almost linearly with increasing DM. Performing the experiments as simultaneous saccharification and fermentation also revealed a decrease in ethanol yield at increasing initial DM. Saccharomyces cerevisiae was capable of fermenting hydrolysates up to 40% DM. The highest ethanol concentration, 48 g/kg, was obtained using 35% (w/w) DM. Liquefaction of biomass with this reactor system unlocks the possibility of 10% (w/w) ethanol in the fermentation broth in future lignocellulose to ethanol plants.

Exhaustive crystal structure search and crystal modeling of β-chitin

An exhaustive search of the crystal structure of beta-chitin was carried out by simultaneously optimizing all the structural parameters based on published X-ray diffraction data and stereochemical criteria. The most probable structure was characterized by a parallel-up chain polarity, a gg orientation of hydroxymethyl groups and an intermolecular hydrogen bond along the a-axis, which essentially reproduced the original structure proposed by Gardner and Blackwell. The proposed crystal structure was subsequently subjected to crystal modeling using the AMBER force field. The probable orientation of hydroxyl groups and their motional behaviors is proposed based on calculations for the crystal models identified. Solvated crystal models exhibited a slightly deformed structure with the formation of appreciable numbers of hydrogen bonds along the b-axis.

Biomass recalcitrance: engineering plants and enzymes for biofuels production

Lignocellulosic biomass has long been recognized as a potential sustainable source of mixed sugars for fermentation to biofuels and other biomaterials. Several technologies have been developed during the past 80 years that allow this conversion process to occur, and the clear objective now is to make this process cost-competitive in today's markets. Here, we consider the natural resistance of plant cell walls to microbial and enzymatic deconstruction, collectively known as "biomass recalcitrance." It is this property of plants that is largely responsible for the high cost of lignocellulose conversion. To achieve sustainable energy production, it will be necessary to overcome the chemical and structural properties that have evolved in biomass to prevent its disassembly.

Molecular Dynamics Simulations of Solvated Crystal Models of Cellulose Iα and IIII

Swelling behaviors of cellulose I(alpha) and III(I) crystals have been studied using molecular dynamics simulations of the solvated finite-crystal models. The typical crystal models consisted of 48 x 10-mer chains. For the cellulose I(alpha) crystal, models consisting of different numbers of chains and chain lengths were also studied. The structural features of the swollen crystal models, including the cellulose I(beta) crystal model reported previously, were compared. A distinct right-handed twist was observed for models of the native cellulose crystals (cellulose I(alpha) and I(beta)), with a greater amount of twisting observed for the I(alpha) crystal model. Although the amount of twist decreased with increasing dimensions of the cellulose I(alpha) crystal model, the relative changes in twist angle suggest that considerable twist would arise in a crystal model of the actual dimensions. In contrast to the swelling behavior of crystal models of the native cellulose, the cellulose III(I) crystal model exhibited local, gradual disordering at the corner of the reducing end. Comparison of the lattice energies indicated that the cellulose chains of the I(beta) crystal were packed in the most stable fashion, whereas those of the I(alpha) and III(I) crystals were in a metastable state, which is consistent with the crystallization behaviors observed. Upon heating of the native cellulose crystal models, the chain sheets of the I(alpha) model showed a continuous increase in twist angle, suggesting weaker intersheet interactions in this model. The swollen crystal models of cellulose I(alpha) and III(I) reproduce well the representative structural features observed in the corresponding crystal structures. The crystal model twist thus characterizes the swelling behavior of the native cellulose crystal models, which seems to be related to the insolubility of the crystals.