Structural Details of Crystalline Cellulose from Higher Plants

Biomacromolecules 2004, 5, 1333-1339

1333

Structural Details of Crystalline Cellulose from Higher Plants Adriana S ˇ turcova´ ,† Isabelle His,† David C. Apperley,‡ Junji Sugiyama,§ and Michael C. Jarvis*,† Chemistry Department, Glasgow University, Glasgow G12 8QQ, Scotland, United Kingdom, EPSRC Solid-state NMR Service, Durham University, Durham DH1 3LE, England, United Kingdom, and Wood Research Institute, Kyoto University, Uji Kyoto 611-0011, Japan Received December 9, 2003; Revised Manuscript Received April 15, 2004

It is commonly assumed that cellulose from higher plants contains the IR and Iβ crystalline allomorphs together with surface and disordered chains. For cellulose IR, the evidence for its presence in higher plants is restricted to the C-4 signals in the solid-state 13C NMR spectrum, which match those of crystalline cellulose IR from algal sources. Algal cellulose IR can be converted to the Iβ form by high-temperature annealing. We used this approach to generate cellulose samples differing in Iβ content from flax fibers and celery collenchyma, which respectively are representative of textile (secondary-wall) and primary-wall cellulose. It was then possible to isolate the detailed spectral contributions of the surface, Iβ and IR-like phases from linear combinations of the observed 13C NMR and FTIR spectra. The 13C NMR spectra resembled those of highly crystalline tunicate or algal cellulose Iβ and IR, with slight differences implying increased disorder and minor conformational discrepancies. The FTIR spectrum of the Iβ form was closely similar to its more crystalline counterparts, but the FTIR spectrum of the IR form was not. In addition to increased bandwith indicative of lower order, it showed substantial differences in the profile of hydroxyl stretching bands. These results confirm that higher plants synthesize cellulose Iβ but show that the IR-like chains, although conformationally quite similar to crystalline algal cellulose IR, sit in a different hydrogen-bonding environment in higher plants. The differences are presumably occasioned by the small diameter of the crystallites and the influence of the crystallite surface on chain packing. Introduction Cellulose accounts for most of the biomass in the plant kingdom. Its biosynthesis is beginning to be understood. Multiple cellulose chains are synthesized in parallel at the cell surface and associate noncovalently to form microfibrils. Distinct biosynthetic enzyme complexes participate in the synthesis of primary-wall and secondary-wall cellulose. Each complex includes the products of at least three CESA (cellulose synthase) genes,1 and a considerable number of further proteins are implicated. Hypotheses on how these enzymes might cooperate in the formation of crystalline cellulose depend on unanswered questions about its crystal structure and about the extent to which this structure is templated by the synthetic complex or formed by selfassembly. Although the cellulose in nature and commerce comes largely from higher plants, most of what we know about the structure of native cellulose is based on algal or tunicate (animal) model systems.2-10 This is because crystallographic structure determination requires crystalline units large enough to contain a substantial number of unit cells in each dimension. Cellulose in the majority of higher plants forms * To whom correspondence should be addressed. Fax +44 141 330 4888. E-mail [email protected]. † Glasgow University. ‡ Durham University. § Kyoto University.

crystalline units that are very long but only about 2-3 nm (4-6 chains) in width.11-13 Crystalline units as narrow as these give diffraction data of low resolution from which no detailed structures can be derived, only approximate unitcell dimensions and limited information on the major lattice spacings.13,14 In the textile celluloses such as cotton, flax, and ramie, the crystalline units are microfibrils about 5 nm thick,15 and crystallography is possible, but the more highly crystalline microfibrils of algal and tunicate cellulose, with lateral dimensions up to 20 nm, are generally preferred.10 Cellulose from algal, tunicate, and bacterial sources is known to contain two crystal forms (allomorphs): cellulose ΙR and Ιβ.16-18 Both are structured from parallel glucan chains in a flat-ribbon conformation, with alternating glucosyl units locked in opposite orientations by intramolecular hydrogen bonds. Further, intermolecular hydrogen bonds hold the chains together edge-to edge in flat sheets (Figure 1). Tunicate cellulose Ιβ has a two-chain monoclinic unit cell in which each chain has 21 screw symmetry, so that all glucosyl residues are identical except that they face alternately in opposite directions.10 Algal cellulose ΙR has a triclinic unit cell5 containing one chain in which alternate glucosyl residues differ slightly in conformation and hydrogen bonding. Thus, in each of the two forms, there are two distinct kinds of glucosyl residue, but these are arranged differently within the crystal lattice. The alternating glucosyl residues of cellulose ΙR differ less than the glucose units in

10.1021/bm034517p CCC: $27.50 © 2004 American Chemical Society Published on Web 06/22/2004

1334

Biomacromolecules, Vol. 5, No. 4, 2004

S ˇ turcova´ et al.

Figure 1. Hydrogen-bonding patterns in cellulose ΙR and Ιβ based on the crystal structures of Nishiyama et al.10,18 Top left and right: the two alternative hydrogen-bond networks in cellulose ΙR.18 Bottom: the dominant hydrogen-bond network in cellulose Ιβ:10 (left) chains at the origin of the unit cell; (right) chains at the center of the unit cell

the two distinct chain types present in cellulose Ιβ. Another distinction between cellulose ΙR and Ιβ lies in the way in which each sheet of parallel chains is stacked above the next, so that the two forms can in principle be interconverted by longitudinal shear.19 Both crystal structures contain a degree of internal disorder in proton positions, with two alternative hydrogen-bonding networks each partly occupied. Spectroscopic methods provide less detailed structural information than crystallography but have the advantage that their resolution is not directly affected by the dimensions of the crystalline units, only by the degree of order within them (although very thin crystals are more exposed to surrounding water and generally show more disorder and molecular motion). Spectroscopy is therefore more useful for studies on cellulose from higher plants. Solid-state NMR and vibrational (infrared or Raman) spectroscopies have been widely applied to cellulose.3,8,20-22 and indeed the two native crystal forms were first distinguished by solid-state NMR16 resolving anomalies that had previously been noted by infrared spectroscopy.23 Likewise the identification of a distinct form of cellulose at crystal surfaces, restricted to higher plants, was based on spectroscopic experiments20,24-25 and would not have been possible starting from diffraction data. Cellulose from a wide range of plants has been shown to give solid-state NMR spectra resembling what would be expected from a mixture of the ΙR, Ιβ, and partially ordered surface forms together with less ordered material in varying proportions.12,24,26-32 This includes cellulose samples from primary cell walls, which in some cases33 have been described as cellulose IV1 on the basis of their poorly resolved diffraction patterns. Based on the NMR evidence, it is commonly assumed that higher plants synthesize both cellulose ΙR and cellulose Ιβ, raising the question whether these might be made by two distinct complexes of cellulose synthases and other enzymes.1

However, the NMR data contain unresolved anomalies32 that led Atalla and VanderHart8 to suggest that higher plants contained no cellulose ΙR at all, only a distorted form of cellulose Ιβ located immediately below the surface of the crystalline units. This controversial view was based principally on discrepancies in the apparent abundance of the ΙRlike form when calculated from the C-1 and C-4 peaks in the 13C NMR spectrum.8,32 Lattice distortion through proximity to the surface would not be unexpected in such narrow crystalline units.34 Indeed it is more difficult to envisage how a regular lattice can arise at all within such small dimensions. The concentric model, with ΙR-like chains sandwiched between the surface monolayer and a Ιβ core, would be consistent with the observation that the NMR signals assigned to both ΙR-like and surface chains are of lower intensity in the thicker microfibrils of cotton, flax, or ramie cellulose, which have lower surface/volume ratios.25 This model would also be consistent with a single type of enzymic complex synthesising primary-wall cellulose, rather than one synthetic complex each for the ΙR and Ιβ forms: presumably the concentric structure would then arise by self-assembly under the influence of any matrix polysaccharides present. It has been suggested8 that the nature of the crystalline phases in the cellulose from higher plants was unlikely to be resolved until the entire 13C NMR spectrum, not just the distinctive C-1 and C-4 peaks, could be used. This has now become possible following the complete assignment of the cellulose ΙR and Ιβ spectra by Kono and co-workers,35,36 who used highly crystalline cellulose from the alga Cladophora. They compared Cladophora cellulose in its native form, rich in the ΙR phase, with the same cellulose after a hydrothermal “annealing” treatment which converts cellulose ΙR to Ιβ.37 Linear combinations of the 13C NMR spectra before and after annealing yielded the spectral contributions of the ΙR and Ιβ forms. We have used the same annealing approach to

Structural Details of Crystalline Cellulose

study the crystalline fractions of cellulose from celery collenchyma and flax phloem fibers, representative examples of primary-wall and textile celluloses.25 In addition to solidstate NMR, which provides mainly conformational information, we used polarized FTIR spectroscopy which is informative about patterns of hydrogen bonding.17 Experimental Section Plant Materials. Flax (Linum usitatissimum L.) was grown at the Glasgow Botanic Garden. Phloem fibers were isolated from main stems of mature flax plants by dissection. Collenchyma strands were isloated from celery (Apium graVeolens L.) petioles as previously described.25 Phloem and collenchyma fibers were separated into single cells by the following sequence of acid hydrolysis and extraction: (1) 1 M H2SO4 for 30 min at 100 °C; 0.1 M NaOH for 1 h at 20 °C; 0.1 M HOAc for 10 min at 20 °C. For hydrothermal treatment of plant celluloses, the purified and dried materials were introduced into a glass test tube containing a solution of 0.1 N NaOH in water. The tube was inserted into a stainless steel reactor vessel, which was immersed in an oil bath where the temperature was kept at 260 °C. After 30 min the vessel was cooled under tap water, and the tube was recovered. The resulting annealed plant materials were then thoroughly washed with distilled water and finally freezedried. NMR Spectroscopy. The NMR experiments were conducted after addition of 20% of H2O (by mass) to cellulose samples isolated from flax phloem and celery collenchyma fibers. This level of hydration exceeds the capacity of the cellulose samples to bind water, as shown by a narrow signal assignable to free water in the 1H NMR spectrum. There were slight losses of water during each experiment due to MAS-generated centrifugal force but the 1H2O signal still indicated the presence of free water at the end of the experiment. Spectra were obtained on a Varian Unity Inova spectrometer operating at 75.430 MHz for 13C. CP-MAS 13C spectra were obtained under conditions designed to maximize resolution, with high-power (100 kHz) phase-modulated proton decoupling (TPPM) and linear-ramped CP.38 The CP contact time was 1 ms and recycle time 1 s. A 4 mm MAS probe was used giving sample spin-rates of 10 kHz. Spectral referencing was to tetramethylsilane carried out indirectly by setting the high-frequency 13C adamantane signal to 38.4 ppm. The surface cellulose spectrum was obtained as described25 by scaled subtraction of the flax cellulose spectrum from the celery cellulose spectrum. The spectra associated with interior chains of flax and of celery cellulose were then obtained by scaled subtraction of the surface cellulose spectrum from each of the original spectra. The scaling factors were adjusted to reduce the integrated relative signal intensity to zero in the 87-91 ppm region of the surfacechain spectrum and in the 82-86 ppm region of the interiorchain spectrum. The interior-chain spectra were then decomposed into IR-like and Iβ-like components such that the relative intensity of the 90-91 ppm shoulder specific for cellulose IR was equal to zero in the spectrum of the Iβ-like component.

Biomacromolecules, Vol. 5, No. 4, 2004 1335

FTIR Microscopy. The acid-hydrolyzed and subsequently alkali-extracted cellulose samples were infiltrated with 20% gelatine solution in phosphate-buffered saline containing NaN2 at 37 °C for several hours and then maintained at room temperature for several days. The annealed samples were similarly infiltrated with gelatine. The infiltrated samples were cryosectioned, and the sections were placed on BaF2 windows. The nominal section thickness was 1 µm. FTIR spectra were obtained on a Nicolet Nexus Spectrometer equipped with a Nicolet Continuum microscope attachment, with a liquid-nitrogen-cooled MCT detector and a single ZnSe wire grid polarizer. The scanning parameters used were as follows: resolution, 4 cm-1; number of scans, 128. Aperture sizes varied according to the dimensions of the section but were always at least 25 µm in each dimension to minimize distortion of the spectra by scattering effects. Vapor-phase deuteration was as described.38 Results and Discussion Solid-State 13C NMR Spectroscopy. The solid-state 13C NMR spectra from both flax and celery cellulose showed clear changes due to the annealing process (Figure 2). The changes in the shape of the C-1 and C-4 peak envelopes were consistent with a large increase in the proportion of cellulose Ιβ at the expense of cellulose ΙR. The 84 and 62 ppm peaks assignable to the surface chains25 were reduced in intensity by annealing of celery cellulose, but not in the case of flax cellulose. The intensity of these peaks in annealed celery cellulose was close to that observed for flax cellulose before or after annealing. As previously described,25 the greater diameter (and hence volume:surface ratio) of flax fibers allowed the spectrum of the surface chains to be obtained by subtraction of the flax cellulose from the celery cellulose spectrum. The two initial spectra were scaled to equalize the total intensity of the 89 ppm group of peaks from cellulose Ι, so that in this region, the intensity of the calculated spectrum of the surface chains was zero. In the present experiment, this procedure gave a clean spectrum (Figure 2) similar to that obtained previously25 from surface cellulose. This allowed the spectrum of the interior chains in each sample to be obtained by scaled subtraction of the surface spectrum from each of the total spectra in turn. For both flax and celery cellulose, the IR-like and Ιβ spectra were then obtained by subtracting the annealed interior-chain spectrum from the native interior-chain spectrum. The difference was taken to represent the ΙR-like phase lost during the annealing process. The spectra thus derived from flax and celery cellulose were similar (Figure 3), supporting the validity of this quite complex procedure. The spectra obtained for the IR-like and Iβ phases were clearly different from one another whether they were derived from the interior chains of flax cellulose or of celery cellulose. There was a close, but not exact, match with the ΙR and Ιβ 13C NMR spectra of highly crystalline cellulose from Cladophora35 and Halocynthia36 provided that a correction was made for the different chemical shift reference used in these studies (Figure 3). The Cladophora and

1336

Biomacromolecules, Vol. 5, No. 4, 2004

S ˇ turcova´ et al.

Figure 2. Solid-state 13C NMR spectra of cellulose from flax fiber cells and celery collenchyma cells, in their native state and after hydrothermal “annealing” which converts cellulose IR to Iβ. The difference spectrum (celery-flax) is dominated by signals from crystallite-surface cellulose reflecting the greater surface: volume ratio of the thinner celery collenchyma crystallites.

Figure 3. Spectra from the interior chains of flax and celery cellulose decomposed into components comparable with the ΙR and Ιβ phases. The ΙR-like spectrum corresponds to the material lost on annealing, whereas the Ιβ-like spectrum corresponds to the material present in greater abundance after annealing. The line spectra show the chemical shifts (peak positions) similarly derived for highly crystalline cellulose ΙR and Ιβ from Cladophora.35

Halocynthia cellulose spectra were in general slightly better resolved, as might be expected from the high crystallinity of these samples. Flax and celery cellulose IR also appeared to differ slightly from Cladophora cellulose IR within the complex group of overlapping signals assigned to C-2 and C-5. Splitting of the C-6 signal in the Ιβ spectra from both flax and celery was a little more evident than with Halocynthia cellulose. The higher plant sources appeared to have some intensity upfield of the main C-6 region and thus assignable to the gg and gt exocyclic conformations,25 although disentanglement of the spectral contributions of the surface and interior chains was less conclusive in this spectral region. The NMR evidence, therefore, is consistent with structures very like the crystalline ΙR and Ιβ phases found in the more crystalline cellulose microfibrils of lower plants, but with less order and minor conformational differences, particularly for the ΙR phase. However, solid-state NMR is most effective

in yielding information on chain conformations and is less informative about how the chains are hydrogen-bonded to one another within the lattice. Polarized FTIR Spectroscopy. The vibrational (FTIR or Raman) spectra of carbohydrates contain much structural information. The hydroxyl stretching region of the spectrum is particularly useful for elucidating hydrogen-bonding patterns because, in favourable cases, each distinct hydroxyl group gives a single stretching band at a frequency that decreases with increasing strength of hydrogen bonding.39 In principle, coupling between vibrational modes in spatial and energetic proximity can introduce extra complexity in the spectra, but in practice, this problem seems much less severe in the hydroxyl stretching region of the spectrum than at lower frequencies.39,40 Polarized radiation has been used qualitatively for many years to distinguish between hydroxyl groups that are longitudinally and transversely oriented with respect to the fiber axis.3,5,41 This approach was recently

Structural Details of Crystalline Cellulose

Biomacromolecules, Vol. 5, No. 4, 2004 1337

Figure 4. FTIR spectra of cellulose isolated from celery collenchyma cells before and after hydrothermal annealing (above), with vaporphase deuteration of surface chains which removes their hydroxyl stretching bands from 3200 to 3500 cm-1 to the 2500 cm-1 region. (Below) Polarized spectra of the annealed cellulose with the electric vector of the incident radiation either parallel or transverse to the longitudinal axis of the cells, which corresponds to the mean chain axis.

extended to allow quantitative determination of orientation.40 In the vibrational spectra of cellulose, interference from surface chains can be avoided by deuteration.23 This converts the surface -OH groups to -OD, moving the corresponding stretching bands to lower frequency by a factor of 1.34.42 In contrast with the data analysis from our NMR experiments, therefore, no subtraction of flax from celery spectra is needed to remove the contribution of the surface chains. The polarized FTIR spectra of deuterated celery collenchyma cellulose, before and after annealing, are shown in Figure 4. These generally resembled published spectra from more crystalline forms of cellulose3,5,41 but the resolution was somewhat lower. Also two distinctive shoulders that have been used to identify the ΙR and Ιβ forms, at 3240 and 3270 cm-1, respectively, were not evident: it was unclear at this stage whether they were absent or too broad to be visible. To compare the cellulose Ιβ spectrum in more detail with the higher-resolution published spectra from crystalline celluloses of algal origin, the O-H stretching region (32503500 cm-1) was deconvoluted into Gaussian components. This procedure allows band positions to be compared between spectra when the bandwidths and the extent of overlap differ. The initial peak-fit parameters were set to those of the six O-H stretching bands identified for annealed Valonia cellulose.41 Since each glucosyl unit contains three hydroxyl groups, deconvolution into six O-H stretching bands is consistent with the two-chain unit cell of cellulose Ιβ. The peak positions changed very little during statistical fitting, remaining within the 5 cm-1 rounding uncertainty of the values reported41 (Figure 5), but the resolution was considerably lower than for annealed Valonia cellulose. The reduced resolution was responsible for the fact that the 3270 cm-1 shoulder was not visually evident before deconvolution. This level of agreement between the modeled and observed Ιβ spectra shows that hydrogen bonding in the Ιβ fraction of annealed celery cellulose conformed closely to the pattern in annealed Valonia cellulose41 and presumably conformed

Figure 5. Deconvolution of the hydroxyl stretching region of the FTIR spectra shown in Figure 4 from celery collenchyma cellulose. Above, non-polarized; below, polarized. The spectra were deconvoluted into six Gaussian components centered on the frequencies identified for annealed Valonia cellulose (cellulose Iβ),41 with statistical fitting of bandwidth and intensity. Only the intensity of each band was allowed to vary between the polarized spectra.

also in most respects to the dominant hydrogen-bonding network in the published structure of Halocynthia cellulose Ιβ.10 The reduced resolution was consistent with increased disorder around the mean structure in celery cellulose Ιβ. Subtraction of the scaled spectra as for NMR allowed the contributions of the Ιβ and ΙR-like components to be separated. These were markedly different from one another and from the O-D stretching bands of the surface chains, emphasising the different hydrogen-bonding patterns of these three cellulose constituents (Figure 6). The ΙR-like spectra could not be modeled using the published band positions for Valonia cellulose ΙR.17 The longitudinally polarized ΙRlike spectrum included a small feature at 3245 cm-1 which probably corresponds to the 3240 cm-1 band used as a marker for the ΙR form,42,43 but its intensity was very low. This spectrum also had low intensity at 3345 cm-1 relative to 3410 cm-1, a feature that has been observed in the spectra of mechanically degraded celluloses of low crystallinity.42 Since the ΙR-like spectrum was derived by a process of spectral subtraction, the possibility of artifacts introduced by the subtraction process must be considered. Incorrect choice of scaling factors would lead to the inclusion of positive or negative peaks corresponding to those of the Ιβ spectrum. These were not observed. Subtraction of two spectra baseline-corrected in different ways can, in some circumstances, lead to “peaks” that are artifacts. The use of standardized, linear baseline correction eliminated this possibility here. The linear baseline model is likely to be an oversimplification where there is scattering from physical features of the sample with dimensions similar to the

1338

Biomacromolecules, Vol. 5, No. 4, 2004

S ˇ turcova´ et al.

Figure 6. Hydroxyl stretching region of FTIR spectra of interior-chain components of deuterated celery collenchyma cellulose. The spectra show components depleted (left) and enriched (right) by annealing, which were taken to correspond respectively to IR-like and Iβ cellulose and were derived from linear combinations of the spectra before and after annealing. The Iβ spectral component accounts for most of the intensity in the spectrum of annealed celery cellulose (Figure 5).

wavelength of radiation used, as was the case for the celery fibers. In diffuse reflectance spectroscopy scattering can generate additional peaks through the Christiansen effect, although this phenomenon has not been described in the transmission mode used for our experiments.45 There is also a question whether the deuteration procedure discriminates cleanly between all of the hydroxyl groups of the surface chains and all those of the crystallite interior.46 For these reasons the subtraction spectra in Figure 6 must be treated with some caution, but the magnitude of the differences from the O-H stretching modes of cellulose from Valonia5 implies a genuine difference in the pattern of hydrogen bonding. The structure of cellulose ΙR from Glaucocystis18 contains two alternative hydrogen-bonding networks, occupied in the ratio 0.55:0.45. Thus, neither network is dominant in cellulose ΙR, unlike cellulose Ιβ. It is possible that the observed FTIR spectral differences between the ΙR-like phase of celery cellulose and cellulose ΙR from Valonia5 are due to a difference in the relative occupancy of two such hydrogen-bonding networks, compounded perhaps by interference from surface chains. Considering both the NMR and the FTIR data, it can be concluded that celery and flax fibers (representative of primary-wall and secondary-wall cellulose) contain one form of cellulose that is similar to tunicate cellulose Ιβ but rather less ordered; and another form that resembles algal cellulose ΙR in conformation but differs significantly in the pattern of hydrogen bonding that holds the chains together. The data presented here are not consistent with the suggestion8 that the ΙR-like cellulose of higher plants is merely a distorted form of cellulose Ιβ. Its 13C NMR spectrum is much closer to that of algal cellulose ΙR.35 It has also been suggested8 that this phase was located just

under the surface monolayer of chains, surrounding a core of cellulose Ιβ. Although this arrangement is not contradicted by our data, on balance it seems easier to construct models of primary-wall and secondary-wall cellulose in which all of the limited space under the surface monolayer of chains is occupied with either Ιβ or ΙR-like cellulose. Even then, a majority of these core chains must be in contact with at least one surface chain rather than surrounded by a continuous Ιβ or ΙR lattice. It is thus not surprising that the FTIR data indicate patterns of hydrogen bonding that differ from those in the established crystalline phases. Indeed it is more remarkable that the similarity is relatively close for cellulose Ιβ. There is evidence that the ΙR-like component of primarywall cellulose can be partially converted to cellulose Ιβ under relatively mild conditions, without recourse to the high temperatures of the annealing process.47,48 Should the terms Ιβ and ΙR be used to describe the cellulose from higher plants? The designations of the cellulose allomorphs are crystallographic in origin, which raises no problems for cellulose Ιβ since flax and ramie cellulose give diffraction patterns that are recognisably of the Ιβ type.17 For cellulose ΙR, though, the question is unlikely to be resolved by crystallographic methods because no higher-plant source of cellulose ΙR has large enough crystallites to diffract well. The designation cellulose IV1 has been used to describe primary-wall celluloses on the basis of their diffraction patterns,33 but it is not yet clear how this should be understood in relation to the spectroscopic evidence. For the past two decades, the 13C NMR spectra, and particularly the C-4 signals, have been used to extend the identification of cellulose ΙR to higher plants.12,16,24,28-31 The limitations of this nomenclature are evident from our experiments, but it has become so well established that

Structural Details of Crystalline Cellulose

introducing a new designation would be excessively confusing. However, if the term ΙR is retained to describe higherplant cellulose giving a ΙR-like NMR spectrum, it must be recognized that although the chains appear to be similar in conformation the manner in which they are packed together may differ from that described in algal models. Such differences are particularly relevant to hypotheses about the relationship between the geometry of the membrane-bound enzyme systems that synthesize cellulose and the structure of the cellulose synthesized1. Acknowledgment. This work was financially supported by BBSRC and EPSRC. References and Notes (1) Taylor, N. G.; Howells, R. M.; Huttly, A. K.; Vickers, K.; Turner, S. R. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 1450. (2) Preston, R. D.; Cronshaw, J. Nature 1958, 181, 248. (3) Liang, C. Y.; Marchessault, R. H. J. Polym. Sci. 1959, 37, 385. (4) Revol, J. F. Carbohydr. Polym. 1982, 2, 123. (5) Sugiyama, J.; Vuong, R.; Chanzy, H. Macromolecules 1991, 24, 4168. (6) Helbert, W.; Nishiyama, Y.; Okano, T.; Sugiyama, J. J. Struct. Biol. 1998, 124, 42. (7) Nishiyama, Y.; Okano, T.; Langan, P.; Chanzy, H. Int. J. Biol. Macromol. 1999, 26, 279. (8) Atalla, R. H.; VanderHart, D. L. Solid State Nucl. Magn. Reson. 1999, 15, 1. (9) Baker, A. A.; Helbert, W.; Sugiyama, J.; Miles, M. J. Biophys. J. 2000, 79, 1139. (10) Nishiyama, Y.; Langan, P.; Chanzy, H. J. Am. Chem. Soc. 2002, 124, 9074. (11) Jakob, H. F.; Fengel, D.; Tschegg, S. E.; Fratzl, P. Macromolecules 1995, 28, 8782. (12) Ha, M. A.; Apperley, D. C.; Evans, B. W.; Huxham, I. M.; Jardine, W. G.; Vie¨tor, R. J.; Reis, D.; Vian, B.; Jarvis, M. C. Plant J. 1998, 16, 183. (13) Mu¨ller, M.; Hori, R.; Itoh, T.; Sugiyama, J. Biomacromolecules 2002, 3, 182. (14) Him, J. L. K.; Chanzy, H.; Mu¨ller, M.; Putaux, J. L.; Imai, T.; Bulone, V. J. Biol. Chem. 2002, 277, 36931. (15) Fink, H. P.; Walenta, E.; Kunze, J. Papier 1999, 53, 534. (16) Atalla, R. H.; VanderHart, D. L. Science 1984, 223, 283. (17) Sugiyama, J.; Persson, J.; Chanzy, H. Macromolecules 1991, 24, 2461.

Biomacromolecules, Vol. 5, No. 4, 2004 1339 (18) Nishiyama, Y.; Sugiyama, J.; Chanzy, H.; Langan, P. J. Am. Chem. Soc. 2003, 125, 14300-14306. (19) Hardy, B. J.; Sarko, A. Polymer 1996, 37, 1833. (20) Earl, W. L.; VanderHart, D. L. J. Am. Chem. Soc. 1980, 102, 3251. (21) Horii, F.; Hirai, A.; Kitamaru, R. Polym. Bull. 1983, 10, 357. (22) Atalla, R. H. J. Appl. Polym. Sci. Appl. Polym. Symp. 1983, 37, 295. (23) Marrinan, H. J.; Mann, J. J. Polym. Sci. 1956, 21, 301. (24) Newman, R. H. Holzforschung 1998, 52, 157. (25) Vie¨tor, R. J.; Newman, R. H.; Ha, M. A.; Apperley, D. C.; Jarvis, M. C. Plant J. 2002, 30, 721. (26) Foster, T. J.; Ablett, S.; McCann, M. C.; Gidley, M. J. Biopolymers 1996, 39, 51. (27) Newman, R. H. Holzforschung 1999, 53, 335. (28) Newman, R. H. Solid State Nucl. Magn. Reson. 1999, 15, 21. (29) Newman, R. H. Cellulose 1997, 4, 269. (30) Newman, R. H.; Davies, L. M.; Harris, P. J. Plant Physiol. 1996, 111, 475. (31) Smith, B. G.; Harris, P. J.; Melton, L. D.; Newman, R. H. Plant Cell Physiol. 1998, 39, 711. (32) Wickholm, K.; Larsson, P. T.; Iversen, T. Carbohydr. Res. 1998, 312, 123. (33) Chanzy, H.; Imada, K.; Mollard, A.; Vuong, R.; Barnoud, F. Protoplasma 1979, 100, 303. (34) Hult, E. L.; Iversen, T.; Sugiyama, J. Cellulose 2003, 10, 103. (35) Kono, H.; Erata, T.; Takai, M. Macromolecules 2003, 36, 5131. (36) Kono, H.; Yunoki, S.; Shikano, T.; Fujiwara, M.; Erata, T.; Takai, M. J. Am. Chem. Soc. 2002, 124, 7506. (37) Sugiyama, J.; Okano, T.; Yamamoto, H.; Horii, F. Macromolecules 1990, 23, 3196. (38) Ha, M. A.; MacKinnon, I. M.; Sturcova, A.; Apperley, D. C.; McCann, M. C.; Turner, S. R.; Jarvis, M. C. Phytochemistry 2002, 61, 7. (39) Cael, J. J.; Gardner, K. H.; Koenig, J. L.; Blackwell, J. J. Chem. Phys. 1975, 62, 1145. (40) Sˇ turcova´, A.; His, I.; Wess, T. J.; Cameron G.; Jarvis, M. C. Biomacromolecules 2003, 4, 1589-1595. (41) Mare´chal, Y.; Chanzy, H. J. Mol. Struct. 2000, 523, 183. (42) Fengel, D. Holzforschung 1992, 46, 283. (43) Michell, A. J. Carbohydr. Res. 1993, 241, 47. (44) Tsuboi, M. J. Polym. Sci. 1957, 25, 159. (45) Le Bras, A.; Erard, S. Planet. Space Sci. 2003, 51, 281. (46) Ioelovitch, M.; Gordeev, M. Acta Polym. 1994, 45, 121. (47) Davies, L. M.; Harris, P. J.; Newman, R. H. Carbohydr. Res. 2002, 337, 587. (48) Rondeau-Mouro, C.; Bouchet, B.; Pontoire, B.; Robert, P.; Mazoyer, J.; Buleon, A. Carbohydr. Polym. 2003, 53, 241.

BM034517P