Increased enzyme binding to substrate is not necessary for more efficient cellulose hydrolysis

Abstract

Substrate binding is typically one of the rate-limiting steps preceding enzyme catalytic action during homogeneous reactions. However, interfacial-based enzyme catalysis on insoluble crystalline substrates, like cellulose, has additional bottlenecks of individual biopolymer chain decrystallization from the substrate interface followed by its processive depolymerization to soluble sugars. This additional decrystallization step has ramifications on the role of enzyme–substrate binding and its relationship to overall catalytic efficiency. We found that altering the crystalline structure of cellulose from its native allomorph I_β to III_I results in 40–50% lower binding partition coefficient for fungal cellulases, but surprisingly, it enhanced hydrolytic activity on the latter allomorph. We developed a comprehensive kinetic model for processive cellulases acting on insoluble substrates to explain this anomalous finding. Our model predicts that a reduction in the effective binding affinity to the substrate coupled with an increase in the decrystallization procession rate of individual cellulose chains from the substrate surface into the enzyme active site can reproduce our anomalous experimental findings.

Biological conversion of lignocellulosic biomass to fuels and chemicals has attracted tremendous interest because of its potential to address problems associated with climate change, energy security, and rural economic development. However, the transition from a petroleum- to a biomass-based economy is not easily accomplished. Biomass recalcitrance to biological conversion is one of the major hindrances to the production of cheap biofuels (1). Cellulose (a β-1,4-glucose polymer) is the most abundant organic molecule in plant cell walls that is recalcitrant to enzymatic hydrolysis because of its highly self-associated (through hydrogen bonding and stacking forces) and microfibrillar nature. Cellulose fibrils are hydrolyzed by a suite of enzymes called cellulases that can be endo- (cleave midchain glycosidic bonds) or exoactive (processively cleave glycosidic bonds starting at chain ends). Endo- (like endoglucanase I or EG-I; also known as Cel7B) and exocellulases (like cellobiohydrolases I and II or CBH-I and CBH-II, respectively; also known as Cel7A and Cel6A, respectively) are the two major components of aerobic fungal secretomes (e.g., like Trichoderma reesei) active on lignocellulose. However, the inherent disadvantages of processivity to polysaccharide hydrolysis (2) and the high abundance of processive enzymes necessary for efficient lignocellulose hydrolysis (3) suggest the need to better understand and eventually overcome the factors contributing to biomass recalcitrance.

Most Trichoderma cellulases are two domain proteins consisting of a carbohydrate binding module (CBM) and a catalytic domain (CD). CBMs are known to facilitate cellulase binding to cellulose primarily through interactions between the glucopyranose rings and conserved aromatic residues (4). The mechanism of cellulose deconstruction into sugars by processive cellulases can be summarized into several elementary steps (Fig. 1 and SI Appendix, Fig. S1) that are not yet fully elucidated. First, the cellulases bind to the substrate (and possibly disrupt its local ordered crystalline structure) directed through the CBM and/or CD (5). Second, it is hypothesized that individual cellulose chains are decrystallized from the surface (by disruption of the ordered cellulose surface) by the bound cellulase followed by discrete sliding of the chain into the catalytic site tunnel of the exocellulase, eventually leading to the formation of a productive enzyme–substrate complex (1, 6, 7). Recent work on the rate-limiting nature of cellulase–glucan chain complexation to CBH-I activity (6), the variable speed motion of individual CBH-I molecules on cellulose surfaces (8), and the intrinsically variable work needed to decrystallize individual glucan chains as a function of cellulose ultrastructure (7) all lends credence to this hypothesis. Finally, the hydrolysis of the complexed glucan chain is known to proceed to either cellobiose or other glucose oligomers (6).

Fig. 1.

A hypothesized schematic model for crystalline cellulose hydrolysis by processive cellulases like CBH-I. The enzyme (E) is represented by a CBM (triangle) and CD (oval) connected by a linker domain. The cellulose chains are represented by cellobiosyl units (rectangles). The enzyme binds to the cellulose substrate (S) first by either the CBM or CD. Next, an individual cellulose chain is decrystallized from the surface and slides into the enzyme active site through various nonproductive enzyme–substrate complexes (E-S_n). Finally, a productive complex proceeds to hydrolysis of the chain glycosidic bond after release of product (P) cellobiose. The enzyme may continue processing the cellulose chain, get stuck, or eventually desorb from the cellulose surface.

The efficiency of cellulase turnover is likely bottlenecked by the rate of substrate decrystallization and productive enzyme–substrate complexation. Altering the cellulose ultrastructure to overcome these bottlenecks should impact the overall hydrolysis kinetics as well (1, 7, 9). Cellulase efficiency is dependent on not only its structure but also the nature of their substrates (2, 10, 11). Recently, it was shown that a nonnative cellulose allomorph (i.e., cellulose III) is more easily digestible by individual or synergistic mixtures of Trichoderma cellulases (8, 12) and requires less work to decrystallize individual glucan chains from its surface than its native form (7). Additionally, this allomorph has been predicted to have a more hydrophilic surface, which could influence its interaction with CBMs (12). The amount of surface-bound cellulases has been shown to be directly correlated to cellulose hydrolysis rates (13–15), but the role of cellulose ultrastructure on the relationship between cellulase binding and activity is not clear. A detailed understanding of processive cellulase kinetics coupling enzyme binding, cellulose chain decrystallization, chain sliding into the active site, and glycosidic bond hydrolysis from an experimental and theoretical perspective is also lacking. Such considerations are not unique to cellulases but are likely common to other interfacial and processive enzymes (e.g., DNA helicases) that need to perform work before the substrate slides into the active site of the enzyme. We follow a two-pronged approach to study this problem. First, we monitor individual cellulase (CBH-I, CBH-II, and EG-I) binding during hydrolysis of defined cellulose substructures, and second, we develop a mechanistic kinetic model coupling enzyme binding, chain decrystallization/sliding, and hydrolysis to simulate conditions that reproduce our general experimental trends.

Results

Cellulase Binding and Hydrolytic Activity on Pure Cellulosic Substrates.

Plant-derived celluloses are known to contain a significant fraction of amorphous cellulose (12), which can selectively be removed by EG-I (and partly be removed by CBH-II) to facilitate exposure of highly crystalline fibrils that are then exclusively depolymerized by processive cellulases (namely CBH-I) (9). Therefore, all cellulosic substrates were hydrolyzed using a defined equimass cellulase mixture comprised of CBH-I, CBH-II, and EG-I (Fig. 2). We have shown previously that the crystallinity indices of celluloses I and III, prepared under certain conditions, are equivalent (12). Thus, exposure of the crystalline regions by polishing of amorphous regions allows us to compare hydrolysis rates as being primarily dependent on the underlying crystalline allomorph structures. Substrate digestibility, during the first 4 h of hydrolysis, was ranked in the following order: amorphous cellulose (AC; 90%) > cellulose III (58%) > cellulose I (43%). Because of an equivalent amorphous background, the differences in hydrolysis rates between celluloses I and III were less apparent during the first hour but became distinct with progression of saccharification time. Near-theoretical glucan conversions were achieved for cellulose I, cellulose III, and amorphous cellulose at 48, 24, and 12 h, respectively (SI Appendix has results on cellulose II).

Fig. 2.

Influence of cellulose ultrastructure on net hydrolysis yield and CBH-I substrate affinity. (A) Total sugar hydrolysis yield and (B) percent of desorbed CBH-I in supernatant for cellulose I, cellulose III, and amorphous cellulose during 48 h of hydrolysis at 50 °C are shown. Solid and dotted lines are a guide to the eye.

The fraction of desorbed CBH-I varied considerably depending on the substrate and saccharification time (Fig. 2B and similar results for CBH-II and EG-I can be seen in SI Appendix, Fig. S2 for all substrates). AC had the highest binding capacity for all cellulases (more than 85% of all three enzymes were bound within 1 h). On the contrary, only 17% CBH-I, 12% CBH-II, and 7% EG-I were bound to cellulose III after 1 h. It is appropriate to compare levels of unbound cellulases for AC after 1 h with both cellulose allomorphs after 4 h, because the extent of glucan conversions (50%) is comparable between the substrates. Even at those time points, similar trends are observed. A greater fraction of the added cellulases was bound to AC and cellulose II (SI Appendix, Fig. S2) compared with cellulose I (1.5- to 2.6-fold greater), which correlates with the higher hydrolysis rates observed for AC and cellulose II. However, this correlation is in striking contrast to observations with cellulose III, which had a lower fraction of bound cellulases but still exhibited an increased hydrolysis rate. During the latter stages of hydrolysis (24–48 h), most of the cellulases are desorbed back into the supernatant (>90% CBH-I and EG-I) as the substrate is depleted (>95% glucan conversion for all substrates at 48 h). However, the fraction of unbound CBH-II decreases slightly, despite increasing substrate solubilization. CBH-II was found to have poorer thermal stability (details in SI Appendix) compared with the other cellulases, which was first highlighted by Reese and Mandel (16), that caused it to denature and precipitate out of solution.

Probing Individual Enzyme Binding to Celluloses I and III.

Equilibrium binding studies were carried out for each of the three cellulases at 4 °C (2 h equilibration time) to minimize hydrolytic activity and determine adsorption parameters. Individual enzyme binding to both cellulosic substrates minimizes competitive binding effects that are typical for mixtures of different cellulase types (17). Langmuir single-site (Fig. 3) and Langmuir–Freundlich-type (SI Appendix, Fig. S3) models were fit (correlation coefficients > 0.98) to the data as described before (18). The partition coefficient (α) (Fig. 3) for exocellulases was higher than that observed for endocellulases for both celluloses I and III. However, surprisingly, α was always lower for each cellulase on cellulose III vs. cellulose I, indicating poorer effective affinity of the enzymes to the nonnative allomorph. Estimations of percent bound CBH-I at 50 mg/g glucan loading (roughly equivalent to total protein loading used in the saccharification assays) were 62% and 39% for celluloses I and III, respectively (SI Appendix, Fig. S3). The percent bound CBH-I was lower (by at least twofold) for both substrates during the saccharification assays, likely because of the reduced affinity of cellulases with increased temperature (19) and competition for binding sites between different cellulase families (17). However, for all saccharification times at 50 °C, the general trend was similar to the equilibrium binding results at 4 °C (i.e., reduced fraction of cellulases were bound to cellulose III vs. cellulose I). This result suggests that competition for binding sites between cellulase mixtures and the rapid equilibration times for each saccharification assay time point are not responsible for the reduced binding of cellulases to cellulose III.

Fig. 3.

Equilibrium adsorption of T. reesei cellulases to celluloses I and III. A depicts the functional relationship between bound and free CBH-I concentrations at 4 °C in equilibrium conditions. Langmuir equation (as shown in top-left corner of A) was fit to the experimental data (symbols) as depicted by the dotted/solid lines. B reports Langmuir single-site model adsorption parameters for the binding of CBH-I, CBH-II, and EG-I to celluloses I and III, where n is the maximum enzyme binding capacity of each substrate, K_d is the apparent dissociation constant, and α is the partition coefficient or ratio of n/K_d.

The Langmuir–Freundlich model is analogous to the Hill equation and can be used to describe cooperativity in enzyme binding. Unlike an earlier report (18), all three cellulases displayed increased cooperativity in binding to cellulose I (SI Appendix, Fig. S3) and decreased cooperativity for cellulose III. This finding is not unexpected considering that negative binding cooperativity will likely increase with reduction in the number of available binding sites (>60% decrease) as seen for cellulose III. In contrast to recent work (8), we see a drop in CBH-I binding capacity for cellulose III at 4 °C. However, because the previous study was carried out at 30 °C for 2 h, it is likely that the increased catalytic activity of CBH-I (at 30 °C vs. 4 °C) on cellulose III substantially exposed a greater number of binding sites compared with cellulose I. Nevertheless, as reported earlier (8), the apparent dissociation constant () also decreased by 25% for cellulose III vs. cellulose I as reported in this study. A drop in is possibly representative of the ease in access to surface glucan chains because of the peculiar staircase-type shape of cellulose III surfaces.

Modeling Processive Cellulase Action on Celluloses I and III.

Based on previous work (8, 9, 12), the differences in the hydrolytic rates on polymorphic cellulose are known to be primarily rate-limited by the processive action of CBH-I on crystalline regions. Therefore, we developed a mechanistic kinetic model for processive cellulases (like CBH-I) to rationalize the observation that cellulases can hydrolyze cellulose III faster than cellulose I, despite a reduced binding partition coefficient for the former. In this model, which is based on experimental evidence (5, 8, 10), there are at least two kinds of nonproductively (NP) bound enzymes—an off-pathway NP complex, in which only the CBM domain is bound to the middle of the cellulose surface chains (Fig. 1 and SI Appendix, Fig. S1), and an on-pathway NP complex, in which both domains are bound individually or together to the cellulose chain ends without being completely productively engaged for catalysis. The effective rates of CBM or CD domain association and dissociation with the substrate can be captured by (where n is the number of available binding sites, is the adsorption rate constant for the domain, and X = CD or CBM) and (the desorption rate constant), respectively. In scenarios in which the adsorption rate constants for both domains are modified by the same amount, we use the generic notation. Next, because the CD active site tunnel of CBH-I is about 5-nm long and has 10 glucosyl binding sites (20), the enzyme is productively engaged only when all 10 sites are occupied. The on-pathway NP-bound enzyme is capable of forming a productive complex by the discrete decrystallization of a single surface chain and sliding of the chain into the active site tunnel one monomer at a time, which is captured by the rate constant (or for reverse chain sliding rate out of the tunnel). Finally, the productive enzyme–substrate complex proceeds to hydrolysis of the bound chain to form cellobiose and could either continue processing the chain or eventually desorb from the surface. A detailed formalization of the model equations, parameters, and additional results is discussed in SI Appendix.

First, we set up this model to establish a baseline trend that captures reported CBH-I activity and binding for cellulose I (Fig. 4 and SI Appendix, Fig. S5). Baseline rate constants were either chosen from the literature or assumed to allow predictions to be consistent with experimental measurements. Next, we apply the model to explore an extensive number of scenarios where the above-mentioned rate constants are varied systematically and the resultant changes in the hydrolysis rate and ratio of cellulose-bound enzymes are monitored. As expected, changing effective adsorption or desorption rates for CD and/or CBM led to proportional changes in hydrolysis rate and fraction of substrate-bound enzyme (i.e., cellulose hydrolysis rate is directly proportional to the percent bound enzyme). Interestingly, increasing increased enzyme turnover without changing the amount of bound enzyme. Moreover, an increase in coupled with a decrease in (with or without a change in ) results in increased hydrolysis rate with a concomitant decrease in the fraction of bound enzyme, thus mimicking the observation seen experimentally for cellulose III vs. cellulose I. We have extensively tested various other scenarios that are summarized in SI Appendix, Figs. S6–S12 and SI Appendix, Tables S3 and S4.

Fig. 4.

Effects of varying different kinetic model parameters on CBH-I hydrolytic action. (A) The hydrolysis rate and (B) the fraction of bound enzyme to cellulose are plotted as a function of time, and different parameters for binding (nk_on or k_off) and chain decrystallization/sliding (k_slide) in the kinetic model were varied separately or simultaneously. Dotted and solid lines are responses from the kinetic model.

Cellulases Catalytic Efficiency on Pure Cellulosic Substrates vs. Lignocelluloses.

Cellulase binding and activity measurements were carried out on ammonia fiber expansion or AFEX treated corn stover (AFCS) to briefly examine the complex role of noncellulosic cell wall polymers (i.e., hemicellulose and lignin) on enzyme binding and their overall catalytic efficiency. Unlike other pretreatments, the bulk composition of lignocellulosic biomass and the crystalline structure of cellulose are mostly unaltered during conventional AFEX pretreatment (1). Without any pretreatment, the cellulose embedded within the plant cell walls has very limited accessibility to cellulases. AFCS was hydrolyzed using an identical ternary cellulase mixture supplemented with endoxylanases and β-xylosidases. Near-theoretical glucan conversion was achieved after 48 h for AFCS (SI Appendix, Fig. S4), but in contrast to cellulose I, most of the cellulases were not desorbed back into the supernatant (<60% recovery of CBH-I). Addition of hemicellulases had marginal influence on cellulase desorption, despite high xylan solubilization, which suggests that hemicellulose is likely not responsible for the partially irreversible cellulase binding to pretreated lignocellulose. Interestingly, for pure cellulosic substrates, the average glucose production rate per unit bound total mass of cellulases (at 4 h) (Fig. 5) was found to vary as follows: cellulose III (0.28 g/L per hour per milligram) > AC (0.15 g/L per hour per milligram) > cellulose I (0.05 g/L per hour per milligram). In contrast, this value was at least 50% lower (0.02–0.03 g/L per hour per milligram) for pretreated lignocellulose compared with cellulose I, suggesting that the relationship between enzyme binding and catalytic efficiency is more complicated for native lignocellulosic biomass.

Fig. 5.

Comparison of average glucose production rate (in grams per liter per hour) per unit bound total cellulases (milligrams of CBH-I, CBH-II, and EG-I) for different cellulosic substrates and pretreated lignocellulose as a function of hydrolysis time. Solid and dotted lines are a guide to the eye.

Discussion

Nonproductive binding of cellulases with lignocellulose has been implicated as an important factor limiting efficient biocatalysis (1, 10, 11). Lowering the kinetic barriers to desorption of nonproductively or irreversibly bound cellulases is one possible way to desorb these unengaged enzymes. In nature, one suggested mechanism to desorb catalytically inactive enzymes from the surface of cellulose is competitive binding of other enzymes or enzyme synergism (9, 17). Laboratory research has shown that targeted engineering of the CBM can disengage nonproductively bound enzymes (21). Another strategy would be to modify the properties of the binding surface of crystalline cellulose. We have shown recently that altering the ultrastructure of cellulose can modify its surface morphology and its inherent surface chemical properties (12).

Crystalline cellulose can exist in different allomorphic states (e.g., cellulose I_α, I_β, II, and III_I) that differ in the hydrogen bonding and stacking interaction patterns holding the cellulose chains together (22). Cellulose I_β is the major naturally occurring allomorph found in higher plants that is recalcitrant to enzymatic hydrolysis (12, 23). Dissolving native cellulose into phosphoric acid allows disruption of its crystalline fibrillar nature, which on precipitation with water, results in the formation of AC (24). An alternative approach to selectively reorganize cellulose ultrastructure without reducing its crystallinity is to treat it with liquid ammonia followed by its removal using an anhydrous solvent to form cellulose III (12, 25). Recent work has shown that forming cellulose III can enhance its enzymatic hydrolysis rates by up to fivefold vs. native cellulose (8, 12). Our current study has additionally shown that cellulose III can lower the enzyme-loading requirements by at least fivefold to achieve hydrolysis yields comparable with cellulose I (SI Appendix). The weaker intrasheet hydrogen bonding is one possible reason responsible for the increased glucan chain structural flexibility and lower thermodynamic barrier to individual chain extraction from the surface of cellulose III (7, 12). Therefore, this unique allomorph provides us with an opportunity to investigate the mechanistic action of cellulases in a way that is not possible with native crystalline or AC.

The current paradigm suggests that cellulase catalytic efficiency is directly correlated to the amount of total cellulose surface-bound enzymes (13–15, 19). However, although this paradigm has been shown to hold true for native crystalline or AC, it was not observed in the case of cellulose III. AC is known to have at least an order of magnitude greater solvent-accessible specific surface area (SSA) compared with cellulose I (26). Along with disruption of the crystalline morphology, the availability of higher SSA for AC would increase the total available cellulase binding sites, thus further enhancing its hydrolysis rates. Then, the rate-limiting steps to cellulase catalysis for AC would be primarily because of the thermodynamic barriers of individual cellulose chain threading, hydrolysis, and product expulsion from the enzyme active site (7). However, in the case of crystalline cellulose (for cellulose I or III), the decrystallization free energy required to remove individual chains from the crystal surface would be an additional thermodynamic barrier that would contribute to the greater apparent activation energy and hence, slower enzyme kinetics (compared with AC). Interestingly, the initial (1–4 h) average hydrolysis rate per unit bound cellulases (Fig. 5) for cellulose III was two- to fivefold greater than the rates for both cellulose I and AC. This result is surprising considering that the crystallinity indices (12), SSA, and total available cellulose chain-reducing ends (details in SI Appendix) for celluloses I and III are comparable. These findings highlight that in addition to total percent bound enzyme and availability of SSA, differences in individual cellulose chain decrystallization free energies and the intrinsic kinetics associated with the processivity of bound cellulases also impact enzyme catalytic efficiency.

Our equilibrium binding assays have shown that there is a decrease in the partition coefficient (α = n/K_d or nk_on/k_off) for cellulases on cellulose III. Cellulase binding to cellulose I is known to be driven by strong hydrophobic interactions between the glucopyranose rings and planar aromatic residues on the CBM binding interface (4, 21). Therefore, a likely reason for the reduced effective binding of cellulases (i.e., lower α-value) to cellulose III could be because of the relatively lower hydrophobicity of the cellulose III fibril surfaces (12). This suspicion was confirmed by the 37% increase in water activity (a_W) (SI Appendix) for cellulose III vs. cellulose I fibrils, validating our molecular dynamic simulation-based predictions (12). The increase in a_W reflects the more hydrophilic nature of cellulose III fibril surfaces, which is likely unfavorable to binding of type A CBMs that target hydrophobic cellulose surfaces specifically.

Our kinetic model indicates that changing nk_on or k_off alone supports the current paradigm (i.e., hydrolysis rate is directly proportional to the fraction of bound enzymes). The lower hydrolysis rate per unit bound enzymes seen for cellulose I suggests that a larger fraction of enzyme-bound complexes with cellulose I could be, in part, because of greater nonproductive binding. This hypothesis is supported by recent work on CBH-I that showed that a significant fraction of the enzymes is bound nonproductively to cellulose I (10, 11). This work suggested that the increased fraction of NP-bound CBH-I was likely because of a lower k_off for the immobilized enzymes. Restarting cellulose hydrolysis can recover the initial hydrolysis rate, which suggests that the decrease of hydrolysis rate was because of the enzymes getting immobilized irreversibly (10, 27). Cellulases can readily bind to hydrophobic regions in the middle of the cellulose surface (off-pathway binding) but must be able to dissociate or diffuse along the surface before the CD can bind to an available cellulose chain end (on-pathway binding). Our model shows that increasing k_off increases the rate at which off-pathway nonproductive complexes can dissociate from the cellulose surface, leading to a faster build up of productively engaged enzymes and hence, leading to more efficient hydrolysis but with a concomitant increase in the fraction of bound enzymes (details in SI Appendix). Although this result supports the hypothesis proposed in refs. 10 and 11, it does not simulate our experimental findings (i.e., lower α-value but increased hydrolysis rate) with cellulose III.

Interestingly, the model suggests that an increase in k_slide coupled together with a reduction in α reproduces experimental trends, where the hydrolysis rate increased even with a reduced fraction of bound enzymes. Currently, there are no reported techniques that can measure k_slide for processive cellulases acting on crystalline cellulose, but our work suggests that this rate constant is likely an important and ignored parameter that influences intrinsic cellulase kinetics. In our model, k_slide is a coarse-grained rate constant that includes multiple elementary steps, namely decrystallization of individual chains (breaking of hydrogen bonds and pulling out of chains) and enzyme procession (i.e., sliding of the chain into the enzyme active site by breaking interactions at current binding sites to move one glucosyl unit and reform interactions at new sites). Recent molecular simulations have shown that the intrinsic work required to decrystallize 10 glucosyl units (corresponding to the length of a CBH-I active site tunnel) from the surface of cellulose III is about 1–2 kcal/mol per cellobiosyl-unit lower than cellulose I. Then, if we assume that the decrystallization rate is alone limiting among all of the different microscopic rates coarse-grained within the k_slide parameter, we predict that k_slide of CBH-I would be higher for cellulose III.

Nonproductive binding can also stall processive enzyme action as supported by recent high-speed atomic force microscopy (AFM) based imaging of CBH-I motion on cellulose I surfaces (5). It was found that CBH-I was more likely to get stalled on cellulose I surfaces in traffic jams unlike cellulose III (8). This difference is possibly because of the increased k_slide for CBH-I on the latter substrate that prevents significant NP stalling of cellulases. A fivefold higher k_off was reported for CBH-I on the more hydrophilic amorphous cellulose vs. cellulose I (10). This result suggests that reduced stalling of processive cellulases could be further facilitated by a higher k_off for cellulases on cellulose III. Type A CBMs have been sometimes found to disrupt cellulose ultrastructure, which was reported in a recent review article (1). However, the impact of these domains on k_slide has never before been determined and could offer avenues to engineer more efficient cellulases. Additional experiments exploring the complex interactions of cellulases and their individual domains with various cellulosic substrates of defined morphology are necessary to further shed light on this matter.

Finally, cellulose is found naturally in plant cell walls in close association with lignin and hemicellulose (1). It is known that cellulases can nonproductively bind to lignin through hydrophobic interactions through either the CBM or the exposed CD active site aromatic residues (28–30). We find that, although equivalent glucan hydrolysis yields for pretreated lignocellulose and model cellulosic substrates were achieved within 48 h, less than 25–60% of cellulases were desorbed back into solution for the former. Because both cellulose and hemicellulose solubilizations were extensive, the partially irreversible binding of cellulases is mostly related to the presence of lignin. CBH-II is sensitive to thermal denaturation (16), and thus, the presence of hydrophobic lignin could further accelerate its heat-induced denaturation and precipitation. The initial average glucose production rate per unit bound cellulases for pretreated lignocellulose (Fig. 5 and SI Appendix, Fig. S2) is significantly lower than even cellulose I and about an order of magnitude lower than cellulose III. These results highlight that increased cellulase binding to cellulose or lignocellulose does not necessarily correlate with improved activity per unit bound enzymes.

Conclusion

We find that altering the crystalline structure of plant-derived cellulose I to III can lower fungal cellulase loading by up to fivefold to achieve comparable saccharification yields, despite a reduced effective enzyme affinity for the latter substrate. We confirmed that the crystallinity indices, fibril surface area, and available cellulose chain-reducing ends for both substrates are similar. Therefore, reduced cellulase affinity is not a result of the above-mentioned factors but is likely because of the increased fibril surface hydrophilicity as confirmed by water activity measurements. A mechanistic kinetic model coupling enzyme binding, cellulose surface chain extraction, and glycosidic bond hydrolysis was developed to explain these nonintuitive findings. The enhanced catalysis of cellulose III is attributed to the favorable individual cellulose chain decrystallization free energies and intrinsic kinetics associated with the processivity of bound cellulases as captured by a coarse-grained parameter called k_slide in our model. Future research efforts are needed to better characterize k_slide and its molecular underpinnings, because it will open up avenues to engineer more efficient cellulases in concert with both native and nonnative (i.e., chemically pretreated) celluloses.

Materials and Methods

Biomass Substrates, Cellulases Purification, Enzymatic Hydrolysis, and Enzyme Adsorption Assays.

Details on the model cellulosic and lignocellulosic substrates and chemical treatment methods used are provided in SI Appendix. Accellerase 1000 from Genencor (Danisco) was used to isolate CBH-I, CBH-II, and EG-I (3). Protein concentration was determined colorimetrically using the Pierce (Pierce Biotechnology) bicinchoninic acid assay using BSA as the standard. All hydrolysis experiments were performed in 2.2-mL deep-well microplates (Greiner) at 10 g/L glucan loading along with 50 mM (pH 4.8) citrate buffer in total reaction volume of 500 μL. CBH-I, CBH-II, and EG-I were loaded at 15 mg/g glucan (corresponding to 0.15 mg/mL) each along with 2 mg/g glucan loading of purified β-glucosidase (3). For pretreated lignocellulose, additional endoxylanase (5 mg/g glucan) and β-xylosidase (at 2 mg/g glucan) were included (3). The microplates were incubated at 50 °C with shaking at 250 rpm. The supernatant was separated from the insoluble solids by filtering through a 0.45-μm low-protein binding hydrophilic filter (Millipore) for protein and sugar analysis. Glucose and/or xylose concentration within the hydrolyzate were analyzed by HPLC (31). Enzyme binding assays at 4 °C were carried out as reported elsewhere (12). All experiments were carried out in triplicates, with SDs less than 5% of the mean values reported.

Quantitation of CBH-I, CBH-II, and EG-I in Supernatant.

Detailed methodology for individual cellulase quantification is available elsewhere (32). The differences in cellulase isoelectric points allow them to be bound to an anion exchange column (Mono Q, GE Healthcare) and eluted as individual components by applying a linear gradient of 1 M NaCl at pH 7.5. The concentration of individual enzymes was correlated to the eluted protein UV trace peak area detected at 280 nm and calculated using the Unicorn 5.11 software (GE Healthcare). Before injecting the hydrolyzate (originally at pH 4.8) into the column, a gel filtration step was applied to remove UV-sensitive low-molecular weight components and simultaneously perform buffer exchange (to pH 7.5) (32).

Mechanistic Kinetic Modeling of Processive Cellulase Activity.

The model formalization and related additional results are described in SI Appendix. The coupled equations were solved using MATLAB 7.4 (MathWorks Inc), and the algorithms used are available on request.