The filamentous fungus Trichoderma reesei produces two cellobiohydrolases (CBHI and CBHII). These, like most other cellulose-degrading enzymes, have a modular structure consisting of a catalytic domain linked to a cellulose-binding domain (CBD). The isolated catalytic domains bind poorly to cellulose and have a much lower activity towards cellulose than the intact enzymes. For the CBDs, no function other than binding to cellulose has been found. We have previously described the reversibility and exchange rate for the binding of the CBD of CBHI to cellulose. In this work, we studied the binding of the CBD of CBHII and showed that it differs markedly from the behaviour of that of CBHI. The apparent binding affinities were similar, but the CBD of CBHII could not be dissociated from cellulose by buffer dilution and did not show a measurable exchange rate. However, desorption could be triggered by shifting the temperature. The CBD of CBHII bound reversibly to chitin. Two variants of the CBHII CBD were made, in which point mutations increased its similarity to the CBD of CBHI. Both variants were found to bind reversibly to cellulose.
Cellulose, the major polysaccharide component of plant cell walls, is degraded in nature by the concerted action of a number of bacterial and fungal organisms . The cellulase enzyme system secreted by the filamentous fungus Trichoderma reesei catalyses the hydrolysis of insoluble crystalline substrate to glucose . This system can be separated by various procedures into three major types of catalytic activity: cellobiohydrolase, endoglucanase and β-glucosidase. Two cellobiohydrolases (also called exoglucanases), CBHI [3,4] and CBHII , account for ≈ 80% of the secreted protein. Like many other cellulases, CBHI and CBHII have a tripartite structure with a catalytic core domain linked by a relatively long O-glycosylated polypeptide to a small cellulose-binding domain (CBD). CBDs from different organisms can be grouped into several families . The CBDs of CBHI (CBDCBHI) and CBHII (CBDCBHII) belong to family I. This family is distinct from the others, being the only one found in fungi and having a protein fold unrelated to the others. Their structure displays two distinct faces, one of which is remarkably flat and the other more rough. The flat face contains several conserved aromatic residues and is responsible for the binding to cellulose [7–9].
The role of CBDs in the activity of many cellulases has been studied extensively [10–13]. Removal of the CBD results in a decreased affinity and hydrolytic activity on crystalline cellulose, but does not generally affect the activity towards soluble substrates. In spite of their apparent importance for the degradation of crystalline cellulose and relatively detailed characterization of their binding properties, an acceptable mechanistic explanation for the function of CBDs and their interplay with the catalytic core during the hydrolysis of cellulose is still lacking.
Because substrate binding will also dictate the mobility of the enzyme on the cellulose surface, the reversibility of binding is a key issue for understanding CBD action. Various studies on fungal cellulases have shown that once adsorbed, the desorption of the enzyme requires drastic conditions and that the binding is at least partly irreversible [14–16]. It has been suggested that the irreversible binding is due to the CBD [17–20]. Irreversible binding has been described in particular detail for some bacterial family-II CBDs [21–23]. However, binding of CBDCBHI, which belongs to family I, has been shown to be completely reversible .
In this study we analysed the off rate of CBDCBHII and found that it differs remarkably from the desorption properties of CBDCBHI binding. We also showed that point mutations in CBDCBHII allowed a switch in desorption behaviour. The results for CBDCBHI have in principle been described previously  but are repeated here because they constitute a critical control for the experiments on CBDCBHII.
Materials and methods
Isolated CBDs were produced by proteolytic cleavage of a double CBD containing the N-terminal CBD of T. reesei CBHII joined by a linker to the C-terminal CBD of CBHI as described in .
Construction of recombinant proteins
The double CBD was produced as described by Linder et al. . All DNA manipulations were performed using standard protocols . A sequence encoding a hexahistidine tag and three residues (PGA) was fused by PCR to the 3′ end of the sequence encoding the double CBD. The construct was inserted downstream of the sequence encoding the PelB signal peptide in the expression vector pKKtac containing a tac promoter and an ampicillin resistance gene .
Two mutants of CBDCBHII were generated by PCR-based mutagenesis , and the polypeptides were produced from the double CBD as described above. In one case Trp7 of CBDCBHII was replaced with Tyr (W7Y-CBDCBHII), and in the other case, the disulfide bridge formed by the Cys5 and Cys22 was deleted by replacing the Cys residues with Thr and Val, respectively (ΔSS3-CBDCBHII).
Purification of the CBDs
Strains producing the wild-type and double CBD mutants were cultivated in a 1.5-L Chemap CMF fermentor containing 1 L minimal medium. Aeration and pH were maintained at 1 L·min−1 and 6.9, respectively, throughout the fermentation. Expression was induced with 0.5 mm isopropyl β-d-thiogalactoside when cell growth reached an A600 of 50–60, and cultivation was continued until lysis of the cells occurred (20–30 h). Residual cells were separated from the culture supernatant by centrifugation (6000 g, 15 min). The supernatant was filtered through a 0.45-µm Durapore (Millipore Corp.) membrane and loaded on Ni2+-loaded Chelating Sephadex (HR10/10; Pharmacia Biotech). The column was washed with 50 mm phosphate buffer containing 200 mm NaCl and 50 mm imidazole (pH 7.0), and the protein was eluted with the same buffer but containing 0.5 m imidazole. The eluted peak fractions were loaded on to a Source reversed-phase chromatography (RPC) HR10/10 column (Pharmacia Biotech) equilibrated with 0.1% trifluoroacetic acid in water. Bound protein was eluted with an increasing linear gradient of acetonitrile in 0.1% trifluoroacetic acid. A liquid chromatography system (Äkta; Pharmacia Biotech) was used for this purification step. The purified double CBD was lyophilized.
The lyophilized polypeptide was solubilized in 100 mm Tris buffer, pH 8.0 (2 mg·mL−1), and 5 U of immobilized trypsin (Sigma, T-4019) was added per mg of polypeptide. The solution was incubated overnight at 37 °C, purified on a Source RPC column as described above, and lyophilized. The cleavage products were identified by immunoblot using antibody (CI-89) directed against CBDCBHI and by matrix-assisted laser-desorption mass spectroscopy.
Labelling of CBDs with tritium
[3H]-labelling was performed by reductive methylation essentially as described previously . Lyophilized protein (5–10 mg) was dissolved in 2.4 mL of 100 mm Hepes buffer, pH 7.5. Then 300 µL of 100 mm formaldehyde was added to the solution together with 50 mCi (1.4 µmol) of [3H]NaB3H4 (TRK45; Amersham). The mixture was incubated for 3 h on ice. The reaction was stopped and the labelled protein was separated from other reactants by Source RPC as described above. The specific radioactivities of the different CBDs were: CBDCBHI = 0.36 Ci·mmol−1; CBDCBHII = 4.45 Ci·mmol−1; W7YCBDCBHII = 14.41 Ci·mmol−1; ΔSS3CBDCBHII = 9.84 Ci·mmol−1. Absence of free label was confirmed by analytical RPC.
Determination of binding isotherms
All binding experiments were performed at controlled temperature. Stock solutions of CBDs were prepared by dissolving ≈1 mg lyophilized protein in 5 mL 50 mm sodium acetate buffer, pH 5.0, yielding an ≈ 20 µm solution. Final concentrations were determined by measuring the UV absorbance at 280 nm (εCBDCBHI = 5360 cm−1·m−1; εCBDCBHII = 14 300 cm−1·m−1; εW7YCBDCBHII = 9890 cm−1·m−1;εSS3CBDCBHII = 14 180 cm−1·m−1). Dilutions of the stock solutions were made in the same buffer containing 1% BSA and 50 mm NaCl. Equal volumes of CBD solution and a suspension of bacterial microcrystalline cellulose (BMCC)  (1.7 mg·mL−1) were mixed in a glass tube containing a magnetic stirrer for 1 h. The samples were filtered through a Millex GV13 (0.22-µm) filter (Millipore). The experiments were repeated at different temperatures (4 °C, 20 °C, 34 °C and 50 °C). Binding according to pH was measured using the following buffers: 50 mm glycine, pH 2.5 and 3.5; 50 mm sodium acetate, pH 4.5 and 5.5; 50 mm Hepes, pH 6.5, 7.5 and 8.5; 50 mm glycine, pH 9.5 and 10.5; 50 mm Caps, pH 11.0. CBD amounts in the filtrate were determined by liquid scintillation counting. In each experiment, standard curves of radioactivity against labelled CBD were constructed. The binding isotherms were determined on BMCC, tunicin cellulose (kindly provided by Daicel Co., Osaka, Japan) (2.43 g·L−1) and chitin (Sigma C3387; 7.25 g·L−1).
Measurement of non-equilibrium desorption
A series of identical cellulose-CBD mixtures (200 µL) were incubated for 1 h allowing them to reach equilibrium. The concentration of soluble CBD was determined for two of the samples. The effect of dilution was studied by adding 1 mL of 50 mm sodium acetate/1% BSA buffer, pH 5.0, to the rest of the series. The samples were filtered at different time points and the CBD concentrations measured to determine how fast the new equilibrium was established. The experiment was repeated using different initial concentrations. The desorption was tested using BMCC, tunicin cellulose and chitin.
Exchange rate at equilibrium
To determine the exchange rate at equilibrium, bound [3H]-labelled CBD was allowed to compete for binding with unlabelled CBD and vice versa. BMCC suspension (100 µL; 1.69 g·L−1) was mixed with 100 µL of the [3H]-labelled CBD solution. When equilibrium was reached, the same amount of unlabelled CBD was added. The concentration of the added CBD was equal to the concentration of the free [3H]-labelled CBD. In this way, the equilibrium as a whole was not affected. Fitting of the experimental points to a one-phase exponential curve yielded the rate constant. The experiments were also performed in the reverse order so that the unlabelled CBD was first adsorbed and the [3H]-labelled CBD was then added.
Desorption according to temperature shift
Sets of triplicate samples containing different ratios of cellulose and CBD were left to equilibrate for 1 h at 4 °C. One of the samples was filtered to determine the extent of binding, the other two were transferred to 34 °C and left again for 1 h before filtering. The same experiment was performed but the temperatures were applied in reverse order. Incubations and filtrations were carried out in thermostatically controlled incubation rooms set at the indicated temperatures. The experiments performed at 50 °C were carried out in a water bath, and materials and reactants were kept in a 50°C oven.
Adsorption properties of CBDCBHII and comparison with CBDCBHI
Binding isotherms for each CBD were determined on BMCC by assaying unbound [3H]-labelled CBD after binding had reached a steady-state. For both CBDs, it was possible to calculate partition coefficients based on the slope of the linear (initial) part of their isotherms (Table 1) [17,31]. Binding was complete in 10 min of incubation, and did not increase upon extending incubation up to 2 h.
Table 1. Apparent partition coefficients of the studied CBDs on BMCC at 20 °C, pH 5.0.
Partition coefficient (L·g−1)
1.95 ± 0.2
2.53 ± 0.2
0.33 ± 0.2
1.23 ± 0.3
The CBD preparations were checked for the absence of nonbinding labelled material that might alter the partition coefficient. The supernatants from a series of isotherm points were recovered and used again in a second series with fresh cellulose. The first and second series gave identical isotherm curves, showing that the CBD in the supernatant after the first adsorption is fully functional (data not shown). Thus, errors in the apparent partition coefficient due to heterogeneity of the CBD preparations could be ruled out.
Like CBDCBHI, the affinity of CBDCBHII for BMCC increased with decreasing temperature (Fig. 1).The binding of both CBDCBHI and CBDCBHII varied by less than 10% over the pH range 2.5–11 (data not shown). With a concentration range between 0.1 and 2.5 g·L−1, the amount of cellulose in the adsorption experiment did not affect the partition coefficient. Adding 20% ethanol or 5% dimethyl sulfoxide resulted in decreased CBD binding (data not shown), consistent with the binding interaction containing a hydrophobic component.
Desorption of CBDs by dilution and competiton
In accordance with previous studies, adsorbed CBDCBHI reached a new equilibrium position within 10 min when the steady-state was disturbed by dilution with buffer. In contrast, CBDCBHII had not dissociated detectably from the cellulose even 8 days after the dilution (Fig. 2). Thus, in the case of CBDCBHII the ‘ascending’ isotherm was different from the ‘descending’ isotherm. This behaviour is expected if binding is at least partially irreversible . The same result was observed when the experiment was repeated at temperatures ranging from 4 to 50 °C and pH ranging from 2.5 to 11, or when 5% dimethyl sulfoxide or 20% ethanol was added to the incubation mixture (data not shown).
Because CBDCBHII also binds to chitin , the reversibility of the binding to this substrate was checked by dilution of samples at equilibrium. As shown in Fig. 2, CBDCBHII exhibits the same desorption behaviour as CBDCBHI.
Detachment of the CBDs from the binding surface was also assayed in a competition experiment by measuring the exchange rate at steady-state between [3H]-labelled CBD and unlabelled CBD. As shown in Fig. 3 and Table 2, exchange between adsorbed and free CBDCBHII was not detected at steady-state, whereas CBDCBHI was readily exchanged, as shown in previous studies .
Table 2. Measured half time of the exchange on the cellulose surface at equilibrium. Experiments were performed on BMCC at 20 °C, pH 5.0.
53 ± 11
6 ± 1.5
45 ± 8
Desorption resulting from temperature shift
Desorption was, however, observed when the binding steady-state was altered by a temperature shift (Fig. 4). Changing the temperature from 4 °C to 34 °C caused desorption of both CBDCBHI and CBDCBHII, so that the final amount of cellulose-bound CBD decreased to the point that was expected from the ‘on’ isotherm at 34 °C. Performing the experiment in reverse order correspondingly increased the amount of bound CBD.
Binding and desorption properties of CBDCBHII mutants
CBDCBHI is shaped like a wedge with an irregular triple-stranded β sheet as the major secondary-structure element (Fig. 5B). Three tyrosyl residues are aligned on one face of the wedge and constitute the binding surface . The predicted structure of CBDCBHII is quite similar, but has some significant differences . On the cellulose-binding face, the Tyr5 residue of CBDCBHI is replaced by a Trp residue (Trp7) in CBDCBHII (Fig. 5A). Another structural difference is the number of disulfide bridges. CBDCBHI contains only two, but according to the sequence and the modelled structure, it can be predicted that CBDCBHII contains three. This has been confirmed by MS measurements. The N-terminus of CBDCBHII is kept in tight contact with the rest of the structure through this third disulfide bridge between Cys3 and Cys20, which correspond to Thr1 and Val18, respectively, in CBDCBHI (Fig. 5B,C).
Two mutants of CBDCBHII were constructed in order to investigate how they affect the off rate of the polypeptide. In the W7Y-CBDCBHII mutant, the Trp7 residue of CBDCBHII was changed into Tyr, making the binding surface more similar to that of CBDCBHI. In the ΔSS3-CBDCBHII mutant, the third cystine bridge found in CBDCBHII was removed. The two Cys residues were replaced by Thr and Val, as in CBDCBHI. Both mutants were verified by MS. The experimental masses (W7Y-CBDCBHII = 5775.8 Da; ΔSS3-CBDCBHII = 5795.3 Da) differed from the theoretical ones by less than one atomic mass unit.
The binding of the two mutants was investigated with the same methods as the wild-types. The partition coefficients of the mutants are shown in Table 1. Both exhibited a lower affinity than either of the wild-types. As compared with CBDCBHII, the affinity was 1/2 for ΔSS3-CBDCBHII and about 1/8 for W7Y-CBDCBHII. For both mutants, binding shifted to a new equilibrium upon dilution, in clear contrast with wild-type CBDCBHII (Fig. 2). The desorption of both W7Y-CBDCBHII and ΔSS3-CBDCBHII was faster than the desorption of CBDCBHI. These results were confirmed in experiments in which the labelled and unlabelled forms of each mutant were allowed to exchange at equilibrium (Fig. 4). The exchange rate for the W7Y-CBDCBHII mutant was particularly fast (Table 2).
The aim of this study was to analyse the desorption and exchange rate of CBDCBHII binding to cellulose using the techniques previously used for CBDCBHI. At first glance, CBDCBHI and CBDCBHII have similar binding properties. The values of their apparent partition coefficients are quite close, and binding to cellulose is similarly affected by temperature and pH. However, experiments in this study showed two significant differences in the interaction. (a) When a CBDCBHII–cellulose mixture was diluted at apparent equilibrium (steady-state), the system did not adjust to a new equilibrium on the isotherm. Instead, the bound CBD was not released to return to a point on the isotherm even after a long incubation period (Fig. 2). (b) When exchange rates between labelled and unlabelled protein at the cellulose surface were studied, the CBDCBHII did not show a measurable exchange rate, whereas the exchange rate of CBDCBHI allows equilibrium within 300 s (Fig. 3). From these two observations we conclude that the binding of CBDCBHII is not compatible with a reversible binding model, in contrast with CBDCBHI, as has been demonstrated previously .
The observed behaviour of the CBDCBHII presents a paradox. Binding assays were performed after incubation times that were long enough for the proportion of total CBDCBHII bound to cellulose to remain unchanged, which would suggest that equilibrium was reached. However, dilution of the free CBDCBHII indicated that bound CBDCBHII could not be dissociated. This suggests that the binding is not controlled by thermodynamic equilibrium. As binding of CBDCBHII does not continue until the unbound CBDCBHII is exhausted, it means that binding and desorption are kinetically controlled. In the equilibrium exchange experiments, not only did bound labelled CBDCBHII fail to dissociate on dilution with cold CBDCBHII, but labelled CBDCBHII also failed to bind when added to substrate that had been preincubated with cold CBDCBHII. In apparent contradiction, bound CBDCBHII was able to desorb from cellulose when the temperature of equilibrated mixtures was increased, and binding of CBDCBHII increased when the temperature was decreased.
The unusual behaviour of the CBDCBHII could perhaps be described by a two-step model. The first step would involve a simple reversible adsorption, with the rate constants k+1 for adsorption and a k–1 for desorption, as in the equation:
The first transient reversible complex would in a second step be converted into an irreversible complex with no detectable dissociation rate. The second step might involve a conformational change in the CBD or the cellulose, or some surface-penetration phenomenon, for example the CBD ‘digging’ into the cellulose. The observed temperature-induced desorption could be due to a possible temperature-induced reversal of the change occurring in the second step. The second step would not occur for CBDCBHI or the mutants of CBDCBHII on cellulose or the binding of CBDCBHII on chitin, explaining their reversible binding. This could indicate that the second component is an extra acquired specific property, implying some added benefit for the CBHII enzyme as a whole. However, a comparison of CBHI and CHBII enzymes indicates a very similar role for the CBD in both . This theory has, however, some serious shortcomings. We did not observe a continuous increase in the amount of bound CBDCBHII, which would be expected because the CBDbound, reversible should be depleted as the k2 reaction proceeds and thus drive the k1/k–1 reaction further to the right. Similarly, the theory does not explain why there is no exchange in the equilibrium exchange experiments.
An alternative hypothesis could be that a global change would occur at cellulose surface upon binding of CBDCBHII, thereby preventing any desorption or further binding of CBD. As the effect is observed at CBD concentrations well below the amount required to saturate the substrate, it would seem that some long-range interactions are involved, perhaps through perturbation of the surface layer of the crystal lattice. Whatever changes are responsible for the shift to an irreversible mode of binding, these must be labile, as reversibility is at least transiently restored upon changing the temperature. However, because the irreversibility effect was observed at a very wide concentration range, also as low as nanomolar, it is very difficult to see how this low amount of CBD could provide the necessary energy for such a large conformational change. The change should also be evident in other ways such as the shape of the binding isotherms or visible changes in the substrate, but these are not observed.
Irreversible binding to BMCC has also been reported for the CBD of the mixed function exoglucanase/xylanase Cex (CBDCex) from Cellulomonas fimi[21–23]. CBDCex which also fails to dissociate from cellulose, although the binding isotherm suggests saturable binding with a measurable affinity constant. Irreversible binding does not imply complete immobilization on the surface of the substrate, which would hardly be compatible with the efficiency of the enzyme. Fluorescence recovery after local photobleaching of cellulose-bound fluorescently labelled CBDCex showed that bound fluorescent molecules from unbleached areas were able to diffuse laterally into the bleached patches, although unbound fluorescent CBD failed to bind in the centre of the patches .
The two mutants of CBDCBHII aimed at the most obvious differences between the CBDs. Both mutations made the binding reversible. The W7Y mutation led to a dramatic loss of affinity, while the apparent partition coefficient of the ΔSS3 mutant was only twofold lower than that of the wild-type CBDCBHII. Thus, as observed with CBHI, large differences in desorption rates do not correlate with large changes in apparent binding constants. Which structural parameter is critical for irreversible binding behaviour is not known at present. A possible factor may be the rigidity of the structure. The loss of a disulfide bridge in the ΔSS3 mutant would be expected to increase the flexibility of the structure. Indeed, molecular-dynamics simulation predicted that the structure of CBDCBHII should be more rigid than that of CBDCBHI owing to the presence of an additional disulfide bridge . The W7Y mutation may have the same effect, as it affects a residue the homologue of which in CBDCBHI is known to play a role in the structural framework at the N-terminus of the polypeptide .
The two modes of binding displayed by CBDCBHI and CBDCBHII reflect a significant difference in the way they interact with the substrate. This new variable appears to be largely distinct from true or apparent binding affinity for cellulose. Thus, it will be of interest to discover how it is correlated with the functional properties of CBDs with respect to cellulose hydrolysis.
We thank Dr Pierre Béguin, Pasteur Institute, France, for helpful discussions and revisions of the manuscript, Dr Arja Lappalainen for providing some of the double CBD protein, and Professor Tuula T. Teeri, Professor Liisa Viikari and Dr Johanna Buchert for their kind support. This work was supported by a fellowship from the Ministère Français de l’Enseignement Supérieur et de la Recherche and by a grant from the Academy of Finland.