Proc. NatL Acad. Sci. USAVol. 78, No. 2, pp. 1047-1051, February 1981Cell Biology

Enzymatic hydrolysis of cellulose: Visual characterization of theprocess

(Trichodema reesei/Acetobacter xylinum/high-resolution electron oscopy/endoglucanase/cellobiohydrolase)

ALAN R. WHITE AND R. MALCOLM BROWN, JR. *Department of Botany, University of North Carolina, Chapel Hill, North Carolina 27514

Communicated by John N. Couch, November 3, 1980

ABSTRACT Cellulose from the Gram-neative bacteriumAcetobacter xylinum has been used as a model substrate for visual-izing the action of cellulase enzymes from the fungus Trichodermareesei. High-resolution electron microscopy reveals that A. xy-linum normally produces a ribbon of cellulose that is a compositeof bundles of crystalline microfibrils. Visual patterns of the pro-cess of cellulose degradation have been established. Enzymes areinitially observed bound to the cellulose ribbon. Within 10 min,the ribbon is split along its long axis into bundles of microfibrilswhich are subsequently thinned until they are completely dis-solved within 30 min. Incubations with purified components of thecellulase enzyme system produced less dramatic changes in ribbonstructure. Purified 1,4-(-D-glucan cellobiohydrolase I (D) (EC3.2.1.91) produced no visible change in cellulose structure. Pu-rified endo-1,4-l-D-glucanase IV (EC 3.2.1.4) produced somesplaying of ribbons into microfibril bundles. In both cases, wholeribbons were present even after 60 min of incubation, visuallyconfirming the synergistic mode of action of these enzymes.

Considering the abundance of cellulose, the decomposition ofcellulose is perhaps one of the most common natural degra-dative processes. The importance of understanding this reac-tion becomes apparent when one considers the potential ex-ploitation of cellulose as a renewable energy resource throughits conversion to ethanol (1), the importance of cellulose deg-radation in nutrient cycling, and the possible role of cellulaseaction in the fundamentals of plant cell growth and develop-ment (2). Although the action of cellulases has been extensivelystudied in biochemical terms, little is known about the inter-action of cellulase enzymes with its cellulose substrate at themacromolecular level. This is probably because of the previouslack of a suitable system for visualizing cellulase action and thefailure to exploit high-resolution electron microscopic tech-niques. Cellulase studies to date have characterized the chem-ical composition of cellulase enzymes, the specificities of theirreactions, and certain aspects of their kinetics (3-10); however,biochemical studies are not able to monitor the morphologicalchanges that occur in the cellulose substrate during the processof hydrolysis.

The trivial name cellulase actually refers to a system of threedifferent enzymes whose combined actions lead to the efficientdegradation of cellulose. In a currently accepted scheme ofcellulase action (4, 11, 12), endo-1,4-,&D-glucanase (EC 3.2.1.4)(endoglucanase) randomly cleaves internal glucosidic bondswithin an unbroken glucan chain. The newly created nonreduc-ing chain end then becomes the substrate for 1,4-f3-D-glucancellobiohydrolase (EC 3.2.1.91) (cellobiohydrolase), whichcleaves cellobiose dimers from the glucan chain and releases

them into solution. The hydrolysis of cellulose into the glucoseend product is completed by /3glucosidase (EC 3.2.1.21),which splits cellobiose into glucose monomers. The creation byendoglucanase of nonreducing glucan chain ends that are sitesof catalytic action for cellobiohydrolase leads to a synergism inthe overall rate of cellulose degradation.

In this paper we examine the progressive degradation of bac-terial cellulose with high-resolution electron microscopy andestablish the visual patterns of change in cellulose structure thatoccur during degradation. The cellulose produced by the bac-terium Acetobacter xylinum has proved to be a nearly ideal sub-strate because of its purity, its small size with dimensions ap-proaching those of the cellulase enzymes themselves, and thedisperse nature of this cellulose, which is not consolidated intoa thick cell wall (13-15). We have visualized cellulases free insolution, their initial binding to the substrate, and their ensuingaction upon the substrate.

MATERIALS AND METHODS

Enzymes. Purified samples of cellobiohydrolase I (D), en-doglucanase IV, and the complete cellulase enzyme system fromTrichoderma reesei QM9414 were gifts from Ross D. Brown,Jr. (Department of Biochemistry and Molecular Biology, Uni-versity of Florida, Gainesville, FL). Enzyme preparations werepurified as described (6, 8, 12). Release of glucose was deter-mined enzymatically with Worthington Statzyme glucose re-agent. For protein digestions, cellulases were incubated inSigma protease type VI (Pronase P) at pH 7.5 and 370C for 30min. The pH was lowered to 4.8, and the mixture was then in-cubated with cellulose for 30 min. Control cellulases were in-cubated at pH 7.5 and 370C in the absence of Pronase and re-turned to pH 4.8 for incubation with cellulose.Growth Conditions. A. xylinum strain ATCC 23769 from the

American Type Culture Collection was grown in 100 ml of theglucose medium of Hestrin and Schramm (16) at 30'C for 24hr. Pellicles were removed from the air/liquid interface of themedium and soaked in three 20-min changes of cold 50 mMphosphate buffer (pH 7.0) (150 ml per pellicle). Cells and bufferwere released from several cubic centimeters of the washedpellicles by twisting a pellicle around a wooden applicator stick.Cell suspensions were kept at 0C until needed and used with-out dilution. Cells were transferred to micronet-coated elec-tron microscope grids supporting thin carbon films by touchingthe surface of the grid to a drop of the washed cell suspension.

Abbreviations: endoglucanase, endo-1,4-f3-D-glucanase; cellobiohydro-lase, 1,4-f-D-glucan cellobiohydrolase.* To whom reprint requests should be addressed.

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertise-ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact.

1047

Dow

nloa

ded

by g

uest

on

Mar

ch 2

5, 2

021

1048 Cell Biology: White and Brown

.iC as wb- w A; = 'F1 ;-s v _ ,.

1

I. ;* b * ,S. I q - q . , :

I',

A: f _.W; )T --

5_ _

!-* Ad

It,.-;.- .i , ,.

P~~~~~~~~~~~~~~~~~~~~~~

_ b3_

.

FIG. 1. (Legend appears at bottom of following page.)

401--

Proc. Nad Acad. Sci. USA 78 (1981)

Dow

nloa

ded

by g

uest

on

Mar

ch 2

5, 2

021

Proc. NatL Acad. Sci. USA 78 (1981) 1049

Cellulose was generated by floating grids with adhering cellsface down on a drop of glucose incubation medium in a spotplate for 7-10 min at room temperature.

Micronets and Thin Carbon Films. Micronets were manu-factured as described (17). Thin carbon films were produced byindirect carbon evaporation methods similar to those of Whitingand Ottensmeyer (18). Micronet/thin carbon grids were pre-pared individually by cutting 1-mm squares from carbon-coatedmica and floating the thin carbon films onto a water surfacewhile monitoring through a dissecting microscope. Floatingfilms were lifted from the water surface by raising a micronet-covered grid from beneath the film.

Electron Microscopy. Micronet/thin carbon grids with ad-hering cells and cellulose were incubated in separate drops ofcellulase solutions within a moist chamber at room temperature.The cellulase solutions used in these experiments were as fol-lows: (solution A) complete cellulase enzyme system, 1.0 mg/ml in 50 mM citrate buffer (pH 4.8); (solution B) purified en-doglucanase IV, 0.17 mg/ml or 0.5 mg/ml in 50 mM citratebuffer (pH 4.8); and (solution C) purified cellobiohydrolase I(D), 0.5 mg/ml in 50 mM citrate buffer (pH 4.8).

After cellulase incubation, grids were washed with severaldrops of 50 mM citrate buffer and negatively stained with 1%uranyl acetate in 50 mM citrate buffer (pH 4.8). Bacitracin wasused as a spreading agent in the negative stain (19). All prep-arations were examined with a Hitachi HU-l1E electron mi-croscope operating at 75 kV. Micrographs were taken at a mag-nification of X 72,000. For calibration purposes, a 52,000 line/inch grating replica and cationized ferritin molecules wereused.

RESULTS

Cellulose Substrate. Knowledge of the macromolecular or-ganization of cellulose in A. xylinum is a prerequisite for inter-pretation of the progressive enzymatic hydrolysis of cellulose.Negative staining of A. xylinum cellulose revealed about 50-803.0- to 3.5-nm microfibrils aggregated into ribbons between 40and 60 nm wide. The ribbons were twisted with a periodicityof 1 ,Am (Fig. la). Several of these microfibrils may have co-alesced into discrete bundles that subsequently aggregated intoa full ribbon (Fig. ld). The ribbon is thus composed of severallevels of hydrogen-bonded structures. Glucan chains crystallizeinto microfibrils, intermicrofibrillar hydrogen bonding holdsindividual microfibrils into bundles, and these bundles are hy-drogen bonded along their surfaces to form the ribbon (20).Bundles of microfibrils occasionally separated from the ribbonand exhibited twisting, similar to the composite ribbon. Theminimal width of a separated bundle was 3 nm, and the maximalwidth was usually a multiple of 3 nm (Fig. 1 c and e).

Initial Binding of Enzyme to Substrate. In samples incu-bated for several seconds in cellulase solutions, the ribbonstructure was obscured by a coating of particles, with diameters

ranging from 3 to 7 nm, bound to the ribbon surface (Fig. if).Negative staining of the enzyme system alone revealed particleswithin the same diameter range (Figs. 1 and 2). Mean diam-eters (±SEM) of bound and unbound particles were 5.4 ± 0.2nm and 5.4 ± 0.1 nm, respectively.

Fig. 3c illustrates the appearance of individual enzyme par-ticles bound to the cellulose surface. Although it is difficult toassign exact three-dimensional shapes to the enzymes, it ap-pears that these particles are approximately spherical. At thelevel of resolution determined by the negative stain proceduresused, it has not been possible to resolve surface features of theindividual particles.

Grids were also incubated in cellulases that had been exposedto protease. Normally, ribbons were completely degradedwithin 30 min of exposure to the complete cellulase system.However, cellulose ribbons were still present after 60 min ofincubation with protease-treated cellulase. In control experi-ments, cellulases incubated at pH 7.5 and returned to pH 4.8degraded ribbons within 30 min.

Degradation by Complete Cellulase System. Incubation ofA. xylinum cellulose in solutions of T. reesei cellulases for var-ious lengths of time resulted in the establishment of certainpatterns of degradation. Ribbons that were incubated in cel-lulases for 5 min were coated with particles (Fig. lg). One ofthe first distinct changes in ribbon structure was the splayingof ribbons into bundles of microfibrils. After 10 min of exposureto cellulases, splaying into microfibril bundles became moreprominent (Fig. lh). Bends and splits within the bundles werenumerous. Some fragmentation of ribbons and their bundleswas observed during these intermediate stages of degradation(Fig. lj).

At 20 min of cellulase incubation, only a few bundles of mi-crofibrils or amorphous fragments were observed (Fig. 1k). Theremaining strands were low in contrast, indicating that littlenegative stain was trapped by these remnants. Amorphous frag-ments, which may represent the final stages of degradationprior to complete solubilization, were frequently associatedwith the remaining strands (Fig. 11). Statzyme reagent indicatedthat approximately 3 mg of glucose per dl was liberated after30 min of cellulase incubation. No recognizable ribbon bundles,microfibrils, or fragments were found in incubations longer than20 min.

Degradation by Purified Cellulase Components. A. xylinumcellulose incubated in purified cellobiohydrolase I (D) up to 30min exhibited no change in ribbon structure, except that rib-bons were coated with particles (Fig. 3a). Incubation of cel-lulose in purified endoglucanase IV at a protein concentrationof 0.17 mg/ml produced no observable breakdown of ribbonstructure after 30 min. A higher protein concentration of 0.5mg/ml did produce some splaying of microfibril bundles duringthe first 30 min of incubation (Fig. 3b). Numerous recognizableribbons were observed even after 60 min of incubation in both

FIG. 1 (on preceding page). (a) A. xylinum cell with attached cellulose ribbon. Note the periodic twisting of the ribbon. (x 13,000; bar = 1 ,um.)(b) High magnification view of the cellulose ribbon. Note the 3-nm-wide striations visible on the upper half of the ribbon surface. (x 200,000; bar= 100 nm.) (c) Bundle of microfibrils that is separated from the composite ribbon. The maximal width of the bundle is 6 nm and the minimal widthis 3 nm. (x 200,000; bar = 100 nm.) (d) Composite ribbon showing a twist and surface striations arising from the irregular coalescence of bundlesof microfibrils. (x200,000; bar = 100 nm.) (e) A ribbon and a twisting, separated microfibrillar bundle that hasa maximal width of 12.5 nm anda minimal width of 3 nm. (x200,000; bar = 100 nm.) (f) Ribbon incubated for several seconds in a cellulase solution. Note the coating of boundparticles as compared to d. (x200,000; bar = 100 nm.) (g) Ribbon exhibiting bundle splaying after 5 min of cellulase incubation. (X200,000; bar= 100 nm.) (h) More pronounced bundle splaying after 10 min of cellulase incubation. Note the sharp bend in the upper splayed bundle. (X 200,000;bar = 100 nm.) (i) Negatively stained preparation of the complete cellulase enzyme system showing individual particles. The mean particle diameteris 5.4 nm. (X350,000; bar = 50 nm.) (j) Ribbon that is fragmenting instead of exhibiting bundle splaying after 15 min of cellulase incubation.(x90,000; bar = 200 nm.) (k) Microfibrillar bundle remaining after 20 min of cellulase incubation. (x200,000; bar = 100 nm.) (1) Bundle and as-sociated amorphous fragments after 20 min of incubation. Note the low contrast and tapering of the bundle towards the right. (x 200,000; bar =100 nm.)

Cell Biology: White and Brown

Dow

nloa

ded

by g

uest

on

Mar

ch 2

5, 2

021

1050 Cell Biology: White and Brown

.Y 60(a

40(U

1220

I A I B

3 4 5 6 7 3 4 5 6 7Particle Diameter (nm)

FIG. 2. (A) Particle diameters measured from negatively stainedpreparations of the complete cellulase enzyme system alone. (B) Di-ameters of particles bound to cellulose ribbon surfaces after short in-cubations in cellulases. Arrows indicate the mean particle diameterin each case.

purified cellobiohydrolase and purified endoglucanase; how-ever, no ribbons were observed after 30-min incubations withboth purified enzymes simultaneously present.

DISCUSSIONCellulose is a unique substrate for an enzyme because it is alarge, crystalline, insoluble macromolecule. Whereas most en-zymes have soluble substrates, the cellulase enzymes must dif-fuse to the surface of their insoluble substrates before catalyticactivity is possible. The use of thin carbon films, negative stain-ing, and high-resolution electron microscopy have permittedthe visualization of the macromolecular fine structure of cel-lulose, the binding of cellulases to cellulose, and changes in thefine structure of cellulose as it is degraded.

Several lines of evidence presented indicate that the particlesobserved on the surface of cellulose ribbons are, in fact, cel-lulase enzymes. (i) Release of glucose into solution during in-cubation of cellulose ribbons in cellulase solutions is concurrentwith the disappearance of observable cellulose. (ii) Particlesappear on the surface of cellulose ribbons immediately afterthey are introduced into cellulase solutions, and those particlesremain even after washing with buffer. This indicates that theparticles are tightly bound to the cellulose surface. (iii) Thereis an overlap in the size range of the particles observed in neg-atively stained cellulase solutions and the particles observed onthe cellulose surface (Fig. 2). Furthermore, the mean particlesizes are almost identical. Reported molecular weights for thevarious cellulase enzymes are 28,000 for the ,B-glucosidase,

37,000-52,000 for endoglucanase, and 53,000 for cellobiohy-drolase (6, 8, 21), so some variation in observed enzyme di-ameter would be expected. (iv) Incubation of cellulases withnonspecific proteases results in a loss of enzyme activity. Rib-bons are present even after 60-min incubations in protease-treated cellulases.The general pattern of cellulose degradation begins with an

initial splitting of the ribbon along its long axis into bundles ofmicrofibrils, followed by a thinning of these bundles until theyare dissolved (Fig. 1). The specificities of the individual cel-lulase enzymes are known (7, 12). Endoglucanase cleaves un-broken glucan chains, whereas cellobiohydrolase requires afree, nonreducing glucan chain end for substantial catalytic ac-tivity. This suggests that in the initial stages of degradation,endoglucanase would be the most active enzyme because fewnonreducing ends are available for cellobiohydrolase to actupon. Bundles appear when ribbons are treated with endog-lucanase only, but not when ribbons are treated with cello-biohydrolase only (Fig. 3 a and b). Endoglucanase must providethe primary degradative action by hydrolyzing,glucosidic bondson the microfibril surfaces, thereby disrupting the organizationof glucan chains required to maintain hydrogen bonding be-tween bundles. This leads to the observed splaying. During thisprocess, cellobiohydrolase undoubtedly begins to cleave cel-lobiose from newly created chain ends, eventually eroding awaythe original crystalline surface, completely disrupting the forcesthat hold the microfibril bundles into the ribbon configuration,and exposing new surfaces of glucan chains. Cellobiohydrolasethus accelerates the breakdown of ribbons into bundles. Theinitial splaying of bundles in the presence of cellulase supportsthe morphological suggestion (Fig. 1 b-e) that ribbons are com-posed of discrete bundles of microfibrils, where the weakestbonds in the composite ribbon may well be the hydrogen bondsbetween these bundles (20). Furthermore, bundles do not seemto slip apart into individual microfibrils, perhaps because thehydrogen bonding within bundles is stronger than that betweenbundles. Splayed bundles are thinned down by the combinedactions of endoglucanase and cellobiohydrolase, which resultin the gradual erosion of the outer surfaces of the microfibrils.

Our results visually confirm the widely noted synergistic ac-tion of endoglucanase and cellobiohydrolase (10, 11, 22). Cel-lulose is completely dissolved in less than 30 min when the com-plete cellulase enzyme system or reconstituted mixtures ofpurified endoglucanase and cellobiohydrolase are present in the

FIG. 3. (a) Cellulose ribbon incubated for 30 min in purified cellobiohydrolase I (D). Note that particles are bound to the ribbon surface, butgeneral ribbon structure has not changed. (x 200,000; bar = 100 nm.) (b) Ribbon incubated for 30 min in purified endoglucanase IV. This ribbonexhibits some bundle splaying; however, whole ribbons are still present. (x200,000; bar = 100 nm.) (c) High-magnification view of a microfibrilbundle. Circles represent possible cellulase enzyme molecules bound to the cellulose substrate. (x510,000; bar = 50 nm.)

Proc. Nad Acad. Sci. USA 78 (1981)

Dow

nloa

ded

by g

uest

on

Mar

ch 2

5, 2

021

Proc. NatL Acad. Sci. USA 78 (1981) 1051

incubation solution. However, when grids are incubated withonly endoglucanase or cellobiohydrolase, ribbons are still ob-served even after 60 min of exposure to enzymes. Cellobio-hydrolase had no noticeable effect on ribbon structure duringthese long incubations. This is not surprising because thereshould be relatively few nonreducing ends exposed on the sur-face of normal A. xylinum ribbons. Endoglucanase producessome splaying of ribbons into microfibril bundles (Fig. 3b), butwhole ribbons are still present after 60 min of incubation. Overthe relatively long period of 30-60 min, endoglucanase maybreak enough glucosidic bonds to weaken the microfibril sur-faces that are holding the bundles into the ribbon configuration,and splaying occurs. This observation is consistent with thefindings of Wood and McCrae (10) that endoglucanase increasesthe capacity for uptake of alkali and reduces the tensile strengthof cellulose. However, although endoglucanase produces visi-ble changes in ribbon structure, the degradative action of thisenzyme when used alone is slow compared to the synergisticdegradation that occurs when endoglucanase and cellobiohy-drolase are simultaneously present.

Occasionally, discrete fragments and breaks along whole rib-bons have been observed (Fig. lj), suggesting that in some casessplaying may not be inherent to ribbon degradation. Fragmen-tation could result either from locally elevated concentrationsof endoglucanase or weak areas in the ribbon itself which aredegraded faster than adjacent areas (23, 24). Fragmentation,however, seems to be an exception to the generally observedpattern of degradation.The model cellulose from A. xylinum has allowed the visu-

alization of cellulases bound to the substrate and has providedsome unique observations of the progressive events of cellulaseenzymatic activity. Our observations are consistent with themechanisms of cellulase action proposed on the basis of exten-sive biochemical data. Although our techniques may have ap-proached the limits of resolution attainable by negative stain-ing, an informative picture of the more general mode ofcellulase action is beginning to appear. With our techniques,we have examined bundles of cellulose microfibrils in wall frag-ments of growing corn roots and have found progressive deg-radation similar to the splaying of bundles in A. xylinum (25).

In conclusion, it now seems that the application of low-dosedarkfield electron microscopy to cellulases and their associationwith cellulose is feasible. By use of these methods, resolutionbetter than 0.5 nm has been obtained, and the surface regionsof enzymes have been analyzed (26). Successful observationsof cellulase action in this resolution range would undoubtedly

help to clarify some of the long-standing problems of the mech-anism of cellulase action not obtainable with conventional elec-tron microscopic or biochemical techniques.

We thank Richard Santos for expert technical assistance; Dr. Ross D.Brown, Jr. for supplying the cellulase enzymes; Susan Sizemore for artwork and photographic assistance; and Candace H. Haigler and Dr.Moshe Benziman for valuable discussions and suggestions. Part of thiswork is being submitted by A.R.W. for the Doctorate Degree in Botany,University of North Carolina, Chapel Hill.

1. Wilkie, C. R. (1975) Cellulose As a Chemical and Energy Resource(Wiley, New York).

2. MacLachlan, G. A. (1976) AppI Polym. Symp. 28, 645-658.3. Halliwell, G. & Griffin, M. (1973) Biochem. J. 135, 587-594.4. Reese, E. T. (1975) Biological Transformation of Wood, ed.

Liese, W. (Springer, Berlin), pp. 165-181.5. Gum, E. K. & Brown, R. D., Jr. (1976) Biochim. Biophys. Acta

446, 371-386.6. Gum, E. K. & Brown, R. D., Jr. (1977) Biochim. Biophys. Acta

492, 225-231.7. Shoemaker, S. P. & Brown, R. D., Jr. (1978) Biochim. Biophys.

Acta 523, 133-146.8. Shoemaker, S. P. & Brown, R. D., Jr. (1978) Biochim. Biophys.

Acta 523, 147-161.9. Eriksson, K.-E. (1978) Biotechnol Bioeng. 20, 317-332.

10. Wood, T. M. & McCrae, S. I. (1978) Biochem. J. 171, 61-72.11. Wood, T. M. & McCrae, S. I. (1972) Biochem. J. 128, 1183-1192.12. Gritzali, M. & Brown, R. D., Jr. (1979) Adv. Chem. Ser. 181,

237-260.13. Brown, R. M., Jr., Willison, J. H. M. & Richardson, C. L. (1976)

Proc. NatL Acad. Sci. USA 73, 4565-4569.14. Haigler, C. H., Brown, R. M., Jr. & Benziman, M. (1980) Science

210, 903.-906.15. Benziman, M., Haigler, C. H., Brown, R. M., Jr., White, A. R.

& Cooper, K. M. (1980) Proc. NatLAcad. Sci. USA 77, 6678-6682.16. Hestrin, S. & Schramm, M. (1954) Biochem. J. 58, 345-352.17. Pease, D. C. (1975) Micron 6, 85-92.18. Whiting, R. F. & Ottensmeyer, F. P. (1972) J. Mol Biol. 67,

173-181.19. Gregory, D. W. & Pirie, B. J. S. (1973) J. Microsc. (Oxford) 99,

251-267.20. Blackwell, J. & Kolpak, F. J. (1976) Appl Polym. Symp. 28,

751-761.21. Iwasaki, T., Hayashi, K. & Funatsu, M. (1964)J. Biochem. (Tokyo)

55, 209-212.22. Mandels, M. & Reese, E. T. (1964) Dev. Ind. Microbiol 5, 5-20.23. Cowling, E. B. & Brown, W. (1969) Adv. Chem. Ser. 95, 152-187.24. Fan, L. T., Lee, Y.-H. & Beardmore, D. H. (1980) Biotechnol

Bioeng. 22, 177-199.25. Mueller, S. C. & Brown, R. M., Jr. (1980)1. Cell Biol 84, 315-326.26. Ottensmeyer, F. P., Bazett-Jones, D. B., Hewitt, J. & Price, G.

B. (1978) Ultramicroscopy 3, 303-313.

Cell Biology: White and Brown

Dow

nloa

ded

by g

uest

on

Mar

ch 2

5, 2

021