ASPP poster, Charlotte, North Carolina, August, 1995

Dynamic Crystalization of Cellulose I

Susan Cousins and R. Malcolm Brown Jr.
Department of Botany, The University of Texas at Austin, Austin, Tx., 78713


Crystallization of cellulose is interesting because the native form found in higher plants and most other cellulose producing organisms is a metastable crystalline allomorph, cellulose I. Cellulose crystallization has been predominantly analyzed using various fluorescent brightening agents (FBAs) which interrupt glucan chain associations. The most common FBA used, Tinopal LPW™, has given conflicting results in the literature. Previous research on dye altered cellulose has supported two opposing bonding schemes within a glucan minisheet: van der Waals forces and hydrogen bonds. Presented here is morphological an molecular modelling evidence supporting a 3-ply sheet construction for the dye-altered glucan sheets. Energies calculated by molecular mechanics indicate that -1,4 linked glucan minisheets most likely would form by van der Waals forces. A 9.4 Å reflection previously reported has been re-interpreted to provide evidence for folded glucan chains interacting with the dye molecules. Based on the presented approaches and research, a revised mechanism of cellulose I crystallization is proposed.


The most common natural allomorph, cellulose I, consists of microfibrils in which the glucan chains are parallel and extended. Recent studies of cellulose I synthesized in vitro and abiotically are intriguing because cellulose I is the metastable state. Thus, the in vitro and abiotic systems must be mimicking in some way the conditions favored not only for -1,4 linked polymerization, but also for crystallization into a less stable state.

In an abiotic system, cellulose I has been produced by micelles formed from optimized organic solvent and aqueous buffer ratios. This system produced very thin microfibrils only 1-2 glucan chains thick. Such thin microfibrils correlates well with the concept of glucan sheet formation being the first step of crystallization as proposed in various dye altered cellulose studies,,,. Thus, if glucan chain sheets are considered as the first stage of crystallization, it becomes of great interest to determine which bonding scheme is more likely leading to sheet formation.

While the hierarchical levels of cellulose crystallization have been described previously,,, the bonding between the glucan chains in the initial stages of crystallization has been disputed. One model (Fig. 1A), which is more widely accepted, proposes that glucan chains are linked by hydrogen bonds to form two-dimensional sheets. These sheets then associate through van der Waals forces to form a microfibril,. In an alternative model (Fig. 1B), the mini-sheets assemble by van der Waals forces, and the microfibril forms when mini-sheets bind together by hydrogen bonds,. Both models claim support from crystallographic studies of dye-altered cellulose (DAC) synthesized by Acetobacter xylinum in the presence of high concentrations of the fluorescent brightener Tinopal LPW™ (Fig. 2). Differences in interpretation of results may have arisen from different methods of synthesis (shaking vs static conditions). Therefore, the various structural interpretations may be evaluated by repeating the different synthesis conditions and observing the products with electron microscopy and x-ray diffraction.

To further test which bonding is preferred, energy minimization using the molecular mechanics program MM3, was used. This system has been employed to model the various allomorphs of cellulose as well as many other carbohydrates,,. By modelling the glucan minisheets bonded by different forces, the minisheet with the lowest energy would be the most likely to form.

Figure 1. Two alternative models have been proposed to explain the initial stages of cellulose crystallization. Both models suggest glucan chains associate to form glucan sheets which stack to form an aggregate; however, the models differ when considering the source of glucan chain associations, either hydrogen bonds (A) or van der Waals forces (B).

Figure 2. Summary (left) of the two studies which developed opposing models for cellulose I crystallization. Both models claim support from studies of dye-altered cellulose (DAC) synthesized by A. xylinum in the presence of high concentrations of the fluorescent brightener Tinopal LPW™ (right).

Goals of this Study

1) To evaluate two alternatives of cellulose mini-sheet assembly by synthesizing dye altered cellulose (DAC).

2) To determine DAC ultrastructure using x-ray diffraction and electron microscopy.

3) To compare types of bonding in cellulose mini-sheets.


Preparation of DAC under static and shaking conditions

DAC was synthesized from resting cell suspensions incubated for 24 hr in a volume of 100 ml at a final concentration of 1 mM Tinopal LPW™, 40 mM glucose, and 50 mM sodium phosphate buffer (pH 7-8) in the dark at room temperature. The first sample was prepared under static conditions; the second was produced on an orbital shaker at 90 rpm.

X-ray diffraction analysis

Powder patterns were obtained with a Philips PW 1024/30 Debye Scherrer camera using Ni-filtered CuK (1.542 Å) radiation at 35 kV and 25 mA. Samples for powder diffraction were washed in 0.2% NaOH, thoroughly rinsed with deionized water, and observed with TEM before drying. The reflections obtained were compared to those from a control sample of bacterial cellulose synthesized in the absence of Tinopal LPW™ and a control sample of Tinopal LPW™ recrystallized from a 5mM stock solution by air drying.

Molecular Mechanics

The energies of various associations of glucan chains were calculated with the molecular mechanics program MM3(92), obtained from the Quantum Chemistry Program Exchange. The program was executed on an IBM-PC compatible 486 computer after it was compiled with Microsoft Power Station FORTRAN by Paul Vercellotti. Input files for MM3 were created from Chem-X files with a utility supplied by Dr. Alfred D. French, as was another utility for converting MM3 output files into Chem-X files.

The original coordinates for atoms in the various mini-sheets of cellotetraoses were generated using the crystal packing subroutine of Chem-X, April 1993 version (developed and distributed by Chemical Design, Ltd.). The mini-sheets for cellulose I were based on the monoclinic, two-chain unit cell dimensions of Woodcock and Sarko. The mini-sheets for cellulose I were generated manually following the specifications of the Aabloo and French model using the single chain triclinic unit cell of Sugiyama et al.. The glucan mini-sheets were then minimized with the block diagonal matrix method of MM3(92) at = 4 and = 80. Termination of optimization occurred when the energy change after five iterations was less than 0.00008 kcal/atom, or 0.0278 kcal for these models. Energies for each mini-sheet were compared to energies for an aggregate of 4 glucan chains and for a single glucan chain multiplied by 4.


X-ray Diffraction analysis

X-ray powder diffraction for bacterial cellulose (Fig. 3A) produced reflections at 6.1 Å, 5.4 Å, 3.9 Å, and 2.6 Å, characteristic of the expected cellulose I allomorph. The crystallized Tinopal LPW™ (Fig. 3B) produced sharp reflections at 2.9 Å and 2.0 Å as well as a very line broadened reflection at 4.0 Å. The sample of glucan dye sheets synthesized under static conditions gave a single 4.0 Å reflection (Fig. 3C). The sample of glucan dye sheets synthesized under shaking conditions gave several sharp reflections (Fig. 3D), including the 4.0 Å and the disputed 9.4 Å reflections. Furthermore, the morphology of this sample was radically different. While the sample of glucan dye sheets synthesized under static conditions appeared as overlapping non-microfibrillar sheets (Fig. 4A), the sample which produced the 9.4 Å reflection appeared as fibrillar material extending perpendicularly from the longitudinal axis of the cells (Fig. 4B).

Figure 3. X-ray diffraction analysis of glucan dye sheets. (A) Powder pattern of unaltered bacterial cellulose as a cellulose I control. (a=6.1 Å; b=5.4 Å; c=3.9 Å; d=2.6 Å) (B) Powder pattern of Tinopal LPW™ crystallized from aqueous solution. Notice the line broadening of the 4.0 Å reflection. (e=4.0 Å; f=2.9 Å; g=2.0 Å) (C) Powder pattern of glucan dye sheets synthesized under static conditions. Notice the line sharpening of the 4.0 Å reflection when compared to the same reflection in the crystallized Tinopal LPW™. (h=4.0 Å) (D) Powder pattern of glucan dye sheets synthesized under shaking conditions. (i=9.4 Å; j=4.6 Å; k=4.0 Å; l=3.1 Å; m=2.6 Å; n=2.4 Å; o=1.5 Å; several other low intensity reflections at 3.7 Å, 2.2 Å, 2.1 Å, and 1.2 Å are not labeled)

Figure 4. Morphology of DAC samples. (A) Electron micrograph of DAC synthesized under static conditions; note the non-microfibrillar, sheet-like, nature of the material. Compare morphology to the diffraction patterns in Fig. 3C. (B) Electron micrograph of DAC synthesized under shaking conditions; note the microfibrillar characteristic of the material. Compare morphology to diffraction pattern in Fig.. 3D. Bars =

Theoretical Modelling of Glucan-Dye Interactions

Tinopal LPW™ molecules can interact with glucan chains by van der Waals forces or hydrogen bonding. Five alternatives are theoretically possible for forming sheets (Fig. 5).

Figure 5. The theoretical alternatives for interactions between glucan chains and Tinopal LPW™ molecules representing different molecular structures for DAC. (A) hydrogen bonded glucan sheets alternating with hydrogen bonded dye molecules (B) van der Waals associated glucan sheets alternating with hydrogen bonded dye molecules (C) alternating glucan chains and dye molecules within sheets (D) hydrogen bonded glucan sheets alternating with van der Waals associated dye molecules (E) van der Waals associated glucan sheets alternating with van der Waals associated dye molecules

Narrowing the Alternatives

Molecular Mechanics

For models of cellulose I and I (Fig. 6), the energy calculations indicate a general trend that hydrogen-bonded glucan mini-sheets have much higher energies than mini-sheets associated by van der Waals forces (Tables I and II). The range of energy differences between each successive mini-sheet model was about 10 kcal, suggesting a sufficiently notable difference between models. Furthermore, cellulose I planes had lower energies than the corresponding planes for cellulose I. This provides another good indication of the merit of these models because cellulose I has been shown experimentally to be of lower energy than cellulose I,. Energy differences also were correlated with the quantity of inter-chain associations. The optimized single cellotetraoses had the highest energies when multiplied four times to increase the number of atoms for equal comparison with the other models (no inter-chain associations), and the mini-aggregates had the lowest energies (both types of inter-chain associations).

Figure 6. Cellulose I is divided into two suballomorphs, I (left) and I (right), based on rotational changes around the C6-C5 bond. The various glucan mini-sheet planes are labelled.


Structure of DAC

The DAC produced under shaking conditions is similar in ultrastructure to the cellulose II band material. Shaking is known to produce band-like material. If dye molecules were present during that process, individual dye molecules could intercalate between glucan folds to form a true cellulose-dye complex (Fig. 7). Comparison of the diffraction data with the alternative interactions between glucan sheets and dye stacks supports a three-layer construction for glucan dye sheets formed under static conditions. This three-layer construction also is supported by the flexibility and thinness of the material. Assembly of glucan dye sheets could be initiated by the formation of glucan minisheets being coated on both the upper and lower surfaces by dye stacks upon extrusion from the TC subunits. These coated minisheets could interact laterally to form an extended glucan dye sheet. The extended glucan dye sheet could then helically coil to reduce strain; thus forming a tube (Fig. 8).

Figure 7. Schematic diagram of DAC synthesized under shaking conditions. Dye molecules intercalate between the speculated folded sheets of cellulose II. Note that the observed 9.4 Å spacing would occur between the stacks of sheets (bracket = 9.4 Å spacing from Tinopal stack to Tinopal stack). This diagram is not to scale.

Figure 8. Schematic diagram of DAC synthesized under static conditions. Stacks of Tinopal LPW™ molecules associated by van der Waals forces (grey ovals) saturate the upper and lower surfaces of glucan minisheets (black rectangles) as they are extruded from the TC subunits.

Bonding Between Glucan Chains

The formation of glucan sheets through van der Waals forces during synthesis of native cellulose is supported by our energy minimization analyses. While the strength of a single van der Waals association is much reduced in comparison with a single hydrogen bond, the sheer number of van der Waals associations could make up the difference in energy calculated from the molecular mechanics analysis. Furthermore, the sum of energy contributed by the van der Waals forces is greater than the sum of energy contributed by the hydrogen bonds. It may be argued that = 80 is unfavorably high; however, van der Waals associated glucan chains were still favored at = 4, which stresses hydrogen bonding. The first glucan mini-sheets most likely to form would be along the 110 plane for cellulose I and along the 010 plane for cellulose I.

Mechanism of Cellulose Crystallization

In the native crystallization of cellulose I, we theorize that at least three sequential steps are involved (Fig. 9): (1) the formation of mini-sheets held together by van der Waals forces just after extrusion from the enzyme's catalytic site; (2) the association of hydrogen bonded mini-sheets to form a mini-crystal as the mini-sheets pass from the interior of the enzyme complex subunit to the true exterior of the cell; (3) the convergence of mini-crystals from different enzyme complex subunits to form a crystalline microfibril.

Figure 9. A proposed model for stages of microfibril formation. 0) Glucose monomers are polymerized enzymatically from catalytic sites in enzyme complex subunits to form glucan chains. 1) The glucan chains associate via van der Waals forces to form mini-sheets. 2) Mini-sheets associate and hydrogen bond to form mini-crystals. 3) Several minicrystals then associate to form a crystalline microfibril.


1. DAC synthesized under shaking conditions produces fibrillar, pre-cellulose II, material.

2. DAC produced under static conditions forms sheet-like, pre-cellulose I, material.

3. DAC studies support a crystallization model favoring van der Waals associated glucan chains as the initial stage of in vivo cellulose I crystallization.

4. Energy minimization studies favor van der Waals forces initiating glucan chain crystallization leading to native cellulose.

5. We propose that native cellulose I crystallization is a 3 step process: (1) glucan mini-sheets form via van der Waals forces; (2) mini-crystals form from hydrogen bonded mini-sheets; (3) mini-crystals converge to form the crystalline microfibri1.


Special thanks are due to Dr. Alfred D. French for assistance with energy calculations and for discussions of their implications. Appreciation is due to Mr. Richard Santos for technical assistance. This research was funded in part by Welch Grant # F-1217 and the Johnson and Johnson Centennial Chair to Dr. R. Malcolm Brown, Jr.

20 August 1995

^ Up to Posters Page