Acid hydrolysis of bacterial cellulose reveals different modes of synergistic action between cellobiohydrolase I and endoglucanase I


G. Johansson, Department of Biochemistry, University of Uppsala, PO Box 576 S-75 123 Uppsala, Sweden. Fax: + 46 18 552139, Tel.: + 46 18 4714477, E-mail:


Intact and partially acid hydrolyzed cellulose from Acetobacter xylinum were used as model substrates for cellulose hydrolysis by 1,4-β-d-glucan-cellobiohydrolase I (CBH I) and 1,4-β-d-endoglucanase I (EG I) from Trichoderma reesei. A high synergy between CBH I and EG I in simultaneous action was observed with intact bacterial cellulose (BC), but this synergistic effect was rapidly reduced by acid pretreatment of the cellulose. Moreover, a distinct synergistic effect was observed upon sequential endo–exo action on BC, but not on bacterial microcrystalline cellulose (BMCC). A mechanism for endo–exo synergism on crystalline cellulose is proposed where the simultaneous action of the enzymes counteract the decrease of activity caused by undesirable changes in the cellulose surface microstructure.


1, 4-β-d-glucan-cellobiohydrolase Ι


1, 4-β-d-endoglucanase I


β-d-glucoside glucohydrolase from Trichoderma reesei (EC, bacterial cellulose from Acetobacter xylinum


bacterial microcrystalline cellulose from Acetobacter xylinum


crystallinity index


average degree of polymerization


gel-permeation chromatography


leveling-off degree of polymerization


synergy factor based on product formation

The chemical composition of cellulose is simple, consisting of d-glucose residues linked by β-1,4-glucosidic bonds to form a linear polymer. In natural cellulose, the glucan chains have a parallel orientation with all reducing chain ends at one terminus of the crystal [1]. The physical structure and morphology of native cellulose are complex and the fine details of its structure have been difficult to determine experimentally. Although highly crystalline, the structure of cellulose is not uniform. A large body of physical and chemical evidence indicates that native cellulose contains both highly crystalline and less-ordered amorphous or paracrystalline regions [2]. Usually, native cellulose displays about 70% crystallinity. It is generally believed that there exist distinct amorphous parts between crystalline regions in cellulose [3]. The average degree of polymerization (DP) varies between 1000 and 15 000, depending on the source and the preparation method [4].

The cellulose produced by Acetobacter xylinum has recently become a substrate of choice for cellulase studies. This bacterial cellulose (BC) has several advantages in comparison to that of plant origin: it has a more homogeneous structure, higher crystallinity and is available in a never-dried form. Moderate hydrolysis of BC with HCl [5] yields a product with a reduced degree of polymerization and higher crystallinity: bacterial microcrystalline cellulose (BMCC). This highly crystalline cellulose with a simple morphology has been used as a model substrate to study the mechanism of crystal erosion [6,7].

1,4-β-d-Glucan-cellobiohydrolase Ι (CBH I) is typically active on crystalline cellulose and the major product formed is cellobiose [8]. The active site of CBH I has the shape of a 50-Å long tunnel with 10 binding sites for glucose [9], suggesting a processive exo-type action on cellulose [10,11]. The active site of the structurally related 1,4-β-d-endoglucanase I (EG I) has a more open structure, in accordance with its the endoglucanase mode of action [11]. The discussion about endoglucanase–cellobiohydrolase interactions has been extensive. Competitive adsorption, ratio and concentration of enzymes have been considered [12–14] and the requirement for a loose enzyme–enzyme complex has been postulated to be a prerequisite for an attack on the cellulose crystallite [15]. The most widely accepted view is that there is a sequential mechanism for synergistic action whereby endoglucanase initiates the attack on cellulose by forming new chain ends, which then serve as attack points for processive hydrolysis by the endwise-acting cellobiohydrolase [16]. However, a significant endo–exosynergism has recently also been observed in the case of highly crystalline cellulose, which was practically unsusceptible to endoglucanase action [17]. This finding implies the possibility of another, more interactive mechanism for endo–exo action.

In this study, both the separate, sequential and simultaneous action of purified CBH I and EG I from Trichoderma reesei were investigated using well-defined substrates derived from BC.

Materials and methods

CBH I and EG I were purified from a culture filtrate of T. reesei strain QM 9414 as described in [18]. The final purification was performed by affinity chromatography according to [19] using a cellobiose-based (p-aminobenzyl-1-thio-β-d-cellobioside) affinity column. β-Glucosidase was a gift from G. Henriksson (Department of Pulp and Paper Chemistry and Technology, Royal Institute of Technology, Stockholm, Sweden). The purity of CBH I and EG I were examined by 10% SDS/PAGE. β-Glucosidase activity was determined by monitoring the newly created reducing ends from cellobiose. No formation of soluble sugars as well as new reducing ends on substrate was detected upon the incubation of BC for 2 days with 10 times the β-glucosidase activity used in the kinetic experiments.

BC was a commercially available product CHAOKOH® coconut gel in syrup (Thep. Padung Porn Coconut Co. Ltd, Bangkok, Thailand). All other reagents were of analytical grade.

Cellulose concentration

This was determined by the anthrone/sulfuric acid method using d-glucose as standard and absorption measurements at 585 nm [20].

Concentration of the enzymes

These were determined from absorbance measurements at 280 nm using the molar extinction coefficients 78 800 and 67 000 m−1·cm−1 for CBH I and EG I, respectively.

Degree of polymerization

These were determined in two ways. The first methods was a modified version of the Somogyi–Nelson method for reducing sugars [21,22], where the boiling time was extended from 20 min to 1 h [23]. The cellulose concentration in the sample was kept at 5 mg·mL−1, and boiling was carried out in eppendorf tubes. All samples were centrifuged before absorbance measurements at 510 nm and calibration curves were made using d-glucose as standard. The average DP was calculated by dividing the total number of glucose present in cellulose by the number of reducing ends. The second method used gel permeation chromatography (GPC). For GPC, the cellulose samples were freeze-dried and thereafter dissolved in 10% (w/v) LiCl solution in N,N-dimethylacetamide. After dissolution the sample was diluted with N,N-dimethylacetamide to a LiCl concentration of 0.5%. GPC was performed using Plgel 20 µm Mixed-A column (Polymer Laboratories). The column temperature was 80 °C and the eluant was 0.5% (w/v) LiCl in N,N-dimethylacetamide. The flow rate was 1 mL·min−1. The cellulose concentration in the effluent was monitored using a refractive index detector.

The results obtained by these methods were in good agreement with each other. However, the sensitivity of the Somogy–Nelson method limits its use to substrates having a DP of 400 or less.

Crystallinity determinations

Cellulose suspension (20 mL; 1 mg·mL−1) in 0.05 m NaAc buffer, pH 5.0, was vacuum-filtrated onto a ceramic filter plate (surface area, 1.77 cm2). The samples were allowed to air-dry and placed on an aluminum plate. X-ray diffraction was conducted on rotationally averaged discs in the reflectance mode through a 2Θ range of 5–40° using Cu-Kα radiation. Relative crystallinity indices (CrI) were calculated as [(I020 − Iam)/I020] × 100, where I020 was the intensity at the 020 peak near 22.5° and Iam was the minimum intensity near 2Θ = 18°, between the 020 and 110/1–10 peaks [24].

Preparation of cellulose substrates

BC pellets were cut into smaller pieces and rinsed with running tap water for 2 days. Rinsed BC (1.6 g, as glucose) was boiled with 1% NaOH in 1.8 L total volume under continuous stirring (1200 r.p.m.) for 2 days, replacing the hydroxide when it turned yellow (four exchanges in total). After this alkali treatment, the BC was washed with distilled water until neutral, and then ground in a blender for 1 min. The ground BC was boiled with 1% NaOH in 1.4 L total volume for 2 h and kept overnight under stirring at room temperature, neutralized with acetic acid and pelleted by centrifugation (3000 g, 5 min). The pellet (≈ 200 mL) was resuspended in 1.34 m HCl to 800 mL total volume, heated to boiling (35 min) and allowed to boil under continuous stirring (500 r.p.m.). At selected times, 100-mL samples were withdrawn, neutralized with solid NaOH and pelleted by centrifugation (3000 g, 5 min). The total sugar in the resulting supernatant was determined to quantify the extent of acid hydrolysis. The cellulose pellets were washed seven times by resuspension in excess of distilled water followed twice by resuspension in 200 mL 0.05 m NaAc buffer, pH 5.0, and stored at 5 °C. The BC sample was neutralized immediately after resuspension in 1.34 m HCl and treated similarly.

Hydrolysis kinetics

Enzymatic hydrolysis experiments were carried out in 1.5-mL eppendorf tubes by incubating 0.5 mL of cellulose suspension (1 mg·mL−1) in 0.05 m NaAc buffer, pH 5.0, with 1 µm enzyme at 25 °C without agitation. Agitation was omitted to better mimic the natural conditions. The reaction was initiated by addition of enzyme followed by vortex mixing for 2 s and stopped at selected times by addition of 53 µL 1.0 m NaOH (final pH 12.5). The cellulose residue was pelleted by centrifugation (16 000 g, 5 min) and the concentration of the total sugar in the supernatant was determined. Each time point was analyzed for total sugar at least in triplicate and was represented by the average. Deviations from the average never exceeded 10% and differences between repeated hydrolysis experiments were even smaller. In synergy experiments, the reaction was initiated by addition of enzyme mixture giving a final concentration of 1 µm for each enzyme component if not otherwise stated. All enzyme preparations were supplied with an excess of β-glucosidase activity to avoid product inhibition by cellobiose. Synergy factors based on product formation (SFp) were calculated according to the equation:

SFp=pmix/ Σpi

where pmix is the amount of soluble sugars produced by the enzyme mixture and Σpi is the sum of soluble sugars produced by the individual components.

Prehydrolysis experiments

These were performed in two steps:

Step 1. Cellulose suspension (0.5 mL; 1 mg·mL−1) in 0.05 m NaAc buffer, pH 5.0 was incubated with 0.05 µm EG I or 1 µm CBH I at 25 °C without agitation. At specified times the reaction was stopped with 53 µL 1.0 m NaOH (final pH 12.5) to inactivate and desorb the adsorbed enzyme [25]. The cellulose was separated by centrifugation (10 000 g, 2 min) and washed four times by resuspension in 1.0 mL fresh buffer under continuous vortex mixing followed by centrifugation. Finally, the volume was adjusted with buffer to the initial conditions and the washed cellulose was used in a manner similar to the substrate in the second step. Controls containing only buffer and cellulose were treated similarly.

Step 2. Prehydrolyzed cellulose from step 1 was hydrolyzed with 1 µm CBH I or 1 µm EG I as described in the hydrolysis experiments.

Data treatment

Empirical functions were calculated using the approximate formula for soluble product formation:

pta(1 – e-bt)+c(1 – e-dt)(1)

Activities at any given time within the studied interval were calculated by differentiation of the product formation function with respect to time.


where pt is the soluble sugar (as glucose, mm); t is time (min); a, b, c and d are empirical constants and C is the total concentration of cellulose initially present (as glucose, mm).

The relation between the synergistic product formation V and EG I concentration was approximated by the saturation function

VS·[EG I]/(H+[EG I])(3)

where [EG I] is concentration of the EG I (µm), S is the maximum synergistic product formation (as glucose, mm), and H represents [EG I] needed for half-maximum synergy (µm).


Heterogeneous hydrolysis of BC by HCl

Here, the BC was boiled in 1 m HCl and both the formation of soluble sugars and changes in substrate crystallinity as well as in DP were measured. The time-curve of acid hydrolysis has a characteristic pattern with a lag phase followed by rapid formation of soluble sugars, which quickly slows down (Fig. 1). Further hydrolysis proceeds at a more constant, although still decreasing, rate. After 12 h of hydrolysis, 5.2% of the substrate became dissolved. The changes in crystallinity of the BC accompanying the hydrolysis were followed by analyzing the X-ray diffraction patterns of residual substrates. CrI (see Materials and methods) for BC was calculated to be 87.9%. Acid hydrolysis leads to an increase in crystallinity, finally reaching 92.4% in the case of BMCC. Figure 1 demonstrates that the changes in CrI and the weight loss of the substrate are comparable in magnitude, although the relation between these parameters was not linear and clearly reveals a saturation effect. The changes in DP of the celluloses were measured by Somogyi–Nelson and GPC analyses of residual substrates. The DP of the BC was estimated to be about 2600 glucose units and was rapidly reduced by acidic treatment, levelling off at a degree of polymerization (LODP) at about 110 glucose residues. Some parameters of the cellulose samples are listed in Table 1.

Figure 1.

Acidic hydrolysis of bacterial cellulose, time course and changes in DPn. BC (4 mg·mL−1) was hydrolysed by boiling in 1M HCl under continuous stirring. (◊) weight loss; (▪) DPn data from GPC analysis. The inset shows the increase in crystallinity as a function of the reaction coordinate.

Table 1. Parameters of cellulose samples. BC, appearance common to bacterial cellulose (large fibrous bundles having a slimy consistency showing heterogeneity of the suspension at the macroscopic level); BMCC, habit common to bacterial microcrystalline cellulose: suspension with a milky consistency is homogeneous at the macroscopic level. The designation of the samples presented in this table is used throughout the paper.
  1. a. Data from GPC analysis; b  relative activities based on product formation after 30 min hydrolysis; c concentration of both components was 1 µm; d 0.5 h product formation by CBH I on EG I-pretreated cellulose is divided by the 0.5 h product formation by CBH I on nonpretreated cellulose.

Time in 1 m HCl (min)0102540180760
Weight loss by:      
 CrI (%) 87.990.391.392.4
DP (glucose unit) a2620283212151132114
Relative activity      
 CBH I (%)b37691008266
 EG I (%)b95535791100
 CBH I/EG I (%)b,c10056503933
Maximum SFp7.
Maximum effect of pretreatment with EG Id2.121.091.13

Enzymatic hydrolysis of cellulose

EG I. The hydrolysis of different celluloses by EG I shows a similar pattern of product formation. The only evident difference occurs in the initial stage of the reaction where ≈ 60% lower activity was observed for the samples (acid-treated samples are designated by A; the figure following this refers to the number of minutes the sample was treated with acid) A25 and A40 (Fig. 2). Therefore, a moderate treatment of BC with HCl leads to a pronounced decrease in initial activity, whereas the following acidic treatment raises the initial activities to the level of BC. For all cellulose samples, a strong retardation of the hydrolysis rate was observed: the activity remaining after 30 min of hydrolysis was only 4–8% of the initial rate (Fig. 3).

Figure 2.

Time course of hydrolysis of different celluloses by EG I. Cellulose (1 mg·mL−1) was incubated with 1 µm EG I in 0.05 m NaAc buffer, pH 5.0 at 25 °C in the presence of excess β-glucosidase activity. Sample (◆) BC (▵) A25 (●) A40 (□) A180 (○) BMCC (for properties of the samples see Table 1). Solid lines are according to the approximate Eqn (1).

Figure 3.

Retardation patterns during hydrolysis of BC by CBH I and EG I at different molar ratios. BC (1 mg·mL−1) was incubated with 1 µm CBH I and EG I (1 µm– 1 nm) at different molar ratios in 0.05M NaAc buffer, pH 5.0, at 25 °C in the presence of excess β-glucosidase activity. Initial velocities based on 3min. activity measurements. Data points are calculated using approximate Eqn (2). EG I/CBH I was (◆) 1 (▵) 0.05 (●) 0.01 (○) 0.001 (▪) pure CBH I (□) pure EG I.

Like treatment with acid, hydrolysis of BC with EG I led to a decrease of DP which, however, was much less dramatic and appeared to level off at much higher DP values (> 400 glucose units).

CBH I. The activity of CBH I is significantly influenced by acid pretreatment of the BC (Fig. 4). The lowest initial activity was observed on BC and the highest on sample A40. More extended acid pretreatment leads to a decrease of initial enzymatic activity. In contrast, after 2 h of enzymatic hydrolysis the activities on all celluloses are nearly equal. Subsequent enzymatic hydrolysis leads to increased activity on BC. A strong retardation of CBH I activity was typical for all cellulose samples (Fig. 8).

Figure 4.

Time course of hydrolysis of different celluloses by CBH I. Cellulose (1 mg·mL−1) was incubated with 1 µm CBH I in 0.05 m NaAc buffer, pH 5.0, at 25 °C in the presence of excess β-glucosidase activity. Sample (◆) BC (▵) A25 (●) A40 (□) A180 (○) BMCC (for properties of the samples see Table 1). Solid lines are according to the approximate Eqn (1).

Figure 8.

Normalized synergistic product formation during hydrolysis of (◆) BC and (□) BMCC at different EG I/CBH I molar ratios. Cellulose (1 mg·mL−1) was incubated with 1 µm CBH I and EG I (1–1000 nm) in 0.05 m NaAc buffer, pH 5.0, at 25 °C in the presence of excess β-glucosidase activity for 0.5 h. The inset shows the measured product formation under same conditions. Solid lines are according to the approximate Eqn (3).

CBH I/EG I. When equimolar amounts of the enzymes are present, the synergistic effect is evident for all cellulose samples. The product formation rate is highest for BC (nearly 70% of the substrate was degraded during 4 h) and is much less for acid-pretreated celluloses (Fig. 5). However, the decrease in product formation rate was not proportional to the extent of acidic pretreatment of BC and was, relatively speaking, higher at a low extent of preconversion (Table 1).

Figure 5.

Time course of hydrolysis of different celluloses by CBH I and EG I. Cellulose (1 mg·mL−1) was incubated with 1 µm CBH I and 1 µm EG I in 0.05M NaAc buffer, pH 5.0, at 25 °C in presence of excess β-glucosidase activity. Sample (◆) BC (▵) A25 (●) A40 (□) A180 (○) BMCC (for properties of the samples see Table 1). Solid lines are according to the approximate Eqn (1).

Synergistic action was characterized in terms of synergy factors. The time-dependence of SFp values reveals that the initial synergistic effect is always less than the maximal one (Fig. 6) due to the high initial rates of separately acting enzymes. The pattern was clearly pronounced for the BC but rather weak for the acid-pretreated cellulose. SFp values on BMCC-resembling celluloses stay at the level of 1.5.

Figure 6.

Synergy factors (SFp) based on product formation as a function of time for the hydrolysis of different celluloses by CBH I and EG I. Cellulose (1 mg·mL−1) was incubated with 1 µm CBH I and 1 µm EG I in 0.05 M NaAc buffer, pH 5.0, at 25 °C in presence of excess β-glucosidase activity. SFp values are calculated by dividing the product formation during synergistic action by the sum of the products formed during separate action using approximate Eqn (1). Sample (◆) BC (▵) A25 (●) A40 (□) A180 (○) BMCC (for properties of the samples see Table 1).

The synergy at different molar ratios of CBH I/EG I was investigated both on BC and BMCC as substrates. The concentration of CBH I was kept at 1 µm while that of EG I was varied between 1 µm and 1 nm. In all cases, product formation during the first 0.5 h remains almost linear after the initial retardation stage, except with 1 nm EG I (Figs 3 and 7). The retardation of hydrolysis of BC is distinctly dependent on the CBH I/EG I ratio. Both at an equimolar ratio and at a 20-fold excess of CBH I, the relatively high initial activity was retained for a comparatively long time. In the case of a 100-fold excess of CBH I, almost 60% of the initial activity was lost after the first 3 min and no further retardation was observed. However, when CBH I was used in a 1000-fold excess, the retardation of hydrolysis did not differ from that observed for pure CBH I (Fig. 3).

Figure 7.

Time course of hydrolysis of bacterial cellulose by CBH I and EG I at different molar ratios. BC (1 mg·mL−1) was incubated with 1 µm CBH I and EG I (1–1000 nm) at different molar ratios in 0.05 m NaAc buffer, pH 5.0 at 25 °C in the presence of excess β-glucosidase activity. EG I/CBH I was (◆) 1 (▵) 0.05 (●) 0.01 (○) 0.001 (▪) pure CBH I. Solid lines are according to the approximate Eqn (1).

The normalized increase of synergistic product formation plotted against the concentration of EG I clearly reveals saturation with EG I (Fig. 8). The EG I concentrations needed for half-maximal synergism were calculated by Eqn (3)to be 0.03 µm and 0.19 µm for BC and BMCC, respectively.

In the prehydrolysis experiments, EG I and CBH I were applied sequentially. Here, the substrate was first treated with 0.05 µm EG I for various times. After the removal of EG I by alkali and subsequent washing the activity of CBH I on the cellulose was determined. The activity of CBH I is definitely higher on EG I pretreated BC and is dependent on the time of pretreatment. The effect of pretreatment with EG I is less for acid-pretreated celluloses (Table 1). Generally, the increase of CBH I activity levels off after a certain degree of substrate pretreatment with EG I for all substrates, but takes a much longer time in case of BC (Fig. 9).

Figure 9.

Hydrolysis of different celluloses by CBH I following pretreatment with EG I. Cellulose (1 mg·mL−1) was prehydrolyzed with 0.05 µm EG I for various times. After removal of EG I and washing the residual cellulose (1 mg·mL−1) was incubated with 1 µm CBH I in 0.05 m NaAc buffer, pH 5.0, at 25 °C in the presence of excess β-glucosidase activity for 0.5 h. Sample (◆) BC (●) A40 (○) A180 (for properties of the samples see Table 1).

Substrate pretreatment with the opposite order of enzyme application was investigated on BC. The pretreatment of BC with 1 µm CBH I for 2 h resulted in a remarkable increase in the EG I activity (Fig. 10).

Figure 10.

Hydrolysis of intact and CBH I pretreated BC by EG I. BC (1 mg·mL−1) was prehydrolyzed with 1 µm CBH I for 2 h. After removal of CBH I and washing the residual cellulose (0.1–2 mg·mL−1) was incubated with 1 µm EG I in 0.05 m NaAc buffer, pH 5.0, at 25 °C.


Heterogeneous acidic hydrolysis of BC proceeds, as expected, in accordance with the accepted idea that the rate of glucose formation depends primarily on the accessibility of the glucosidic bonds [26]. The initial attack takes place in amorphous regions where the glucosidic bonds are most accessible and causes a gradual increase of the cellulose crystallinity during hydrolysis (Fig. 1). The final weight loss of the substrate of 5.2% during the acidic hydrolysis is accompanied by a 4.5% increase in CrI, which is consistent with a preference for the amorphous regions in acidic degradation. Note that in our experiment, the acid hydrolysis of BC displays a lag phase for soluble sugar formation (Fig. 1), which has, indeed, been predicted by a Monte Carlo simulation [27] but has not earlier been confirmed experimentally. The explanation for this lag phase is that the earliest random cleavages of the cellulose chains are too distant from each other to release soluble sugars. The process of hydrolysis is here manifested only by a rapid decrease of the DP, as the longer oligosaccharides formed after cleavages still remain noncovalently bound to the bulk of the cellulose. This accounts for more than 90% of the total decrease in DP observed before the end of the lag phase (Fig. 1). An accelerated formation of soluble sugars begins after a certain number of cleavages when the probability for rupture of a glucosidic bond close to an already cleaved one becomes higher. Further acid hydrolysis, which takes place when the average chain size is close to the LODP, does, however, have an influence on some macroscopic parameters as well as on the accessibility to enzymes (Table 1). Indeed, the rupture of remaining amorphous connections between crystallites can be a crucial step in the change of cellulose properties, as it drastically reduces the sizes of the cellulose particles. At the same time, the accompanying changes in average DP or CrI are virtually undetectable.

Acid hydrolysis of less exposed glucosidic bonds on the crystalline surface occurs at a significantly slower rate [28] and is not revealed by a change of DP or CrI. However, the random erosion of the crystalline surface is probably the reason for the observed reaggregation of microfibril bundles [29] and for changes in the accessibility of cellulose to enzymes as well.

We have here used samples of acid-hydrolyzed BC with different degrees of hydrolysis as a set of the well-defined substrates for characterization of the mode of action of cellulolytic enzymes (Fig. 11).

Figure 11.

Cartoon depicting hydrolysis of cellulose by CBH I and EG I. The cellulose microfibril is composed of 12 × 18 chains [33]. The relative content of amorphous parts and the length of the crystallites (LODP) are derived from our experiments. Only a longitudinal section of the microfibril is shown. R = reducing end. Crystalline parts are represented by parallel lines and dotted regions are amorphous. The dimensions and shapes of the enzymes are estimated from X-ray data [10,34]. Catalytic domains of nonproductively bound enzymes are marked with a cross. The different sample designations are listed at the left side of the figure (for properties of the samples see Table 1) The bar represents 10 nm.

A common feature of hydrolysis by EG I on all of these substrates was an extremely vigorous initial stage of hydrolysis followed by a rapid retardation in rates to an almost negligible degree of further degradation (Figs 2 and 3). The attack of EG I as an endoenzyme has been believed to be most prominent on the more amorphous regions [11]. A typical pattern of hydrolysis can thus be interpreted as a rapid ‘exhaustion’ of the most accessible amorphous regions. The hydrolysis by EG I is, to some extent, similar to the hydrolysis by acid, as in both cases the amorphous regions are attacked preferentially. We cannot, however, obtain a BMCC-like product from BC by EG I treatment. Hydrolysis of BC by EG I leads to a much less pronounced decrease in DP. The reason for this is clear considering that EG I molecules and the amorphous regions between crystallites are comparable in size (Fig. 11). Cellulose chains located at the center of the amorphous parts of the microfibril can thus only become accessible to EG I after extensive removal of the upper layers of the crystal. This model is supported by observations that the decrease in DP during cellulase action is often accompanied by a substantial loss of mass [30–32]. Unlike EG I, low molecular mass acidic agents (H3O+) can probably penetrate the amorphous parts, leading to a rapid decrease in DP and to the complete separation of the crystallites. It is important to remember that, after longer acidic prehydrolysis of BC, the activity of EG I becomes higher again (Table 1). The acid-eroded crystalline surface probably offers new targets, which are readily accessible for hydrolysis by EG I.

The activity of CBH I as a typical exo-enzyme is dependent on the availability of chain ends on the substrate. It is plausible that those chain ends where glucose rings are not very tightly involved in hydrogen bonding to the bulk of the crystal, the so-called ‘loose ends’, are most suitable for CBH I to start its processive action. The initial acid treatment of BC rapidly increases the amount of cellulose ends, which may gradually dissociate from the bulk of the crystal, giving rise to such ‘loose ends’; this correlates well with the observed increase of CBH I activity on acid-hydrolyzed cellulose (Table 1). The importance of accessible chain ends for the activity of CBH I on BC is also confirmed by the experiments where the pretreatment of BC with EG I led to an increase in the activity of CBH I (Fig. 9). The average length of the cellulose chain is probably another factor that affects the activity of CBH I; the optimal chain length being a little bit less than the processivity index for the enzyme. If a chain is too short, the full processivity can not be realized, whereas chains that are too long will frequently result in residual chains that remain on the cellulose and can hinder the hydrolysis of adjacent cellulose chains [25]. The decrease in the efficiency of CBH I after prolonged acidic treatment of cellulose (Table 1) may actually be a combined effect of shorter chains and randomly eroded crystalline surface, as both factors reduce the apparent processivity. Moreover, one can imagine that partially acid-hydrolyzed BC (samples A25 and A40) can be a better substrate than BMCC because the remaining amorphous connections between crystallites allows an enzyme molecule that has just finished its processive action on the upstream crystallite to ‘catch’ the reducing ends of the downstream crystallite.

A synergistic action between endo- and exo-type enzymes has been explained by the production of the new chain ends by an endo-enzyme [16]. As the EG I is expected to attack the amorphous parts of the substrate, the synergistic effect should be more evident on the less acid-hydrolyzed substrates. Indeed, the highest effect of synergy was observed on BC (SFp reaching 7.8) and decreased rapidly when the cellulose had been pretreated with acid (Fig. 6). A twofold increase of CBH I activity is also, in that case, achieved by sequential treatment of BC with EG I followed by CBH I (Table 1, Fig. 9). The pretreatment of cellulose with acid can be regarded here as almost equivalent to pretreatment with EG I. Indeed, the decrease of synergism on acid-pretreated celluloses is accompanied by an increase in the individual CBH I activity on that cellulose. However, it must be noted that the ‘synergistic’ effect observed upon sequential treatment of BC is far from the real effect of synergy.

A synergistic action leading to a 1.7- to 1.8-fold increase in degradation rate was observed also with BMCC-type substrates. A simple production of new chain ends is not likely to account for this synergism as pretreatment of BMCC by EG I has practically no effect on the activity of CBH I (Table 1, Fig. 9). Furthermore, BMCC should not contain EG I-susceptible amorphous part in significant amounts (Table 1, Fig. 1). In addition, an almost seven times higher EG I/CBH I ratio is needed for half-maximal synergy on BMCC than on BC (Fig. 8). These observations point to a more interactive mechanism for endo–exo synergism based on simultaneous action of these two enzymes. The usual way to explain such phenomena is to assume the formation of a partial complex between enzymes on the cellulose surface, which behaves differently from the individual components acting separately [15]. However, using known experimental systems, it is impossible to corroborate directly the existence of these loose in situ complexes.

We propose here a mechanistic explanation for interactive synergism based on the role of cellulose changes during hydrolysis. It has been shown that a strong retardation of BMCC hydrolysis by CBH I is caused by enzyme-generated alteration of the cellulose surface [25]. After a certain number of hydrolytic actions by a processive enzyme the erosion of the cellulose surface has created a state where randomly left solitary chains may form obstacles for the hydrolysis of chains in the layer beneath (Fig. 12). The crucial function of EG I in this new model should lie in its ability to attack these solitary chains, thus facilitating further processive action of CBH I. This model is also in accordance with the data shown in Figs 3 and 7, where the presence of EG I clearly counteracts the dramatic retardation observed for CBH I acting alone. EG I can here support CBH I as a coexisting ‘scavenging’ agent that efficiently removes the isolated residual chains on the ‘eroded’ crystalline surface. In the same context, the erosion effect by CBH I should indeed make a crystalline cellulose more susceptible to the EG I, as confirmed by our experiments (Fig. 10).

Figure 12.

Erosion of the cellulose surface by a processive exoenzyme.

It must be emphasized that, despite a rather modest synergistic effect, BMCC serves as a good model substrate for a thorough investigation of endo–exo synergism. The mechanism whereby EG I produces new chain ends is not revealed on an almost nonamorphous substrate like BMCC. On more complex substrates with both amorphous and crystalline parts, such as BC, the continuous interplay between the ‘new-end-producing’ and the ‘scavenging’ mechanisms is required for a full synergistic effect.


This work is dedicated to Dr Veljo Sild, whose tragic death occurred during the preparation of this paper. We thank Dr Christina Gustafsson from the Swedish Pulp & Paper Institiute for help in running GPC, Dr Jaan Aruväli from University of Tartu for help in conducting X-ray diffraction and to Dr David Eaker for linguistic revision. Economical support was provided by the Swedish Research Council for Engineering Sciences (230 96-789), Swedish Natural Research Council (K-AA/KU 02475-0306) and Estonian Science Foundation.


  1. Enzymes: CBH I, 1, 4-β-d-glucan-cellobiohydrolase Ι from Trichoderma reesei (EC; EG I, 1, 4-β-d-endoglucanase I from Trichoderma reesei (EC; β-glucosidase, β-d-glucoside glucohydrolase from Trichoderma reesei (EC