Thermostable endoglucanases in the liquefaction of hydrothermally pretreated wheat straw

Novel transportation fuels produced from renewable resources hold increasing promise as environmentally friendly alternatives to petroleum-derived vehicle fuels. Advanced biofuels from non-food plant materials, such as dedicated crops or byproducts of existing agricultural production, are of particular interest, as they represent an ecologically and ethically more acceptable solution [1]. Technologies for the conversion of lignocellulosic substrates to ethanol are being tested in pilot (for example, by SEKAB in Sweden and BioGasol/DTU in Denmark) and demonstration (for example, Abengoa in Spain and Inbicon in Denmark) scales [2]. Nevertheless, the production of lignocellulosic ethanol still needs technological improvements to become economically viable.

Intensive research carried out during the past two decades in the field of lignocellulose pretreatment has lead to the successful development of efficient physicochemical technologies that can open up the complex structure of lignocellulosic substrates, and thus make cellulose embedded in the matrix of hemicelluloses and lignin more accessible to biological conversion [3, 4]. Dedicated enzymes are also available on the market to convert both cellulose and hemicellulose to monomeric sugars [5–7], which can be readily taken up and converted to ethanol by, for example, yeasts in anaerobic fermentations.

One of the single most important process parameters that affect the efficiency of the conversion process is the solid content used. Low substrate loadings are connected with large processing volumes, and result in low product concentrations in the broth, leading to increased capital (large equipment size) and energy costs (mixing, heat transition, product recovery). Until recently, dry matter (DM) content percentage of <10% have been routinely used. When aiming for a final ethanol concentration in the fermentation broth of >4% w/w, which is considered as a prerequisite for feasible large-scale distillation process, the dry-matter content has to exceed 15% for most lignocellulosic substrates [8, 9]. High-solid loadings in the enzymatic hydrolysis have been used in many recent studies, including 30% steam pretreated corn stover [10], 30% hot water-treated or steam-exploded olive tree biomass [11], or 40% hydrothermally pretreated wheat straw [12]. For most technical substrates, these high-solid contents create an environment in which practically no free water exists in the pretreated material, and the slurry becomes difficult to handle. The poor mass-transfer conditions can be improved by means of a partial enzymatic hydrolysis stage (liquefaction), where the structured, porous and water-absorbing high-solid substrate is converted to a more flowable fluid. Recent advances to replace the conventional stirring systems by gravimetric mixing have also helped to overcome this problem and made the liquefaction of pretreated lignocellulose possible at up to 40% initial DM content [8, 12].

Implementation of the liquefaction stage at elevated temperatures, when the material properties and reaction kinetics are more favourable, is expected to bring further improvements to the processes. At higher temperatures, viscosities are inherently lower and reaction rates are higher, potentially reducing processing times and leading to lower investment and energy costs. Integration of a high-temperature pre-hydrolysis step in the process sequence aimed at rapid liquefaction of the high-solid slurry is considered an attractive way to overcome the problems associated with high initial solids contents [13]. Several potential thermostable enzymes acting on lignocellulosic substrates at elevated temperatures have recently been characterized [13, 14].

Rheological studies are difficult to perform in lignocellulose suspensions because of the non-homogenous nature of the substrate, comprising a wide range of particle sizes and shapes, with various interactions occurring between the particles and the surrounding fluid. However, a variety of rheological instruments and tools (for example, vanes, parallel plates, torque rheometer) have been used to study the rheology of concentrated lignocelluloses using mainly pretreated corn stover as the substrate (5% to 30% w/w DM). The few fundamental studies carried out to date were consistent in observing a strongly shear-thinning, viscoelastic behaviour, with a concentration-dependent yield stress, which could be adequately described by various rheological models, including the Bingham, Casson or the Herschel-Buckley models [15–19]. The phenomena of decreasing viscosity during lignocellulose hydrolysis has recently been studied [20, 21], and it has been concluded that the dominant rheological effect of saccharification is caused by material dilution as mass is transferred from a solid to a liquid phase. However, the traditional approach used in these studies (samples collected at specific time points during hydrolysis experiments and subjected to rheological measurements in a separate instrument) is not convenient for observation and/or investigation of rapid changes in the rheological behaviour of the substrate. Recently, a rotational viscometer (Rapid Visco™ Analyzer, RVA) allowing real-time measurement of substrate viscosity was used to follow the initial phase of enzymatic hydrolysis [22]. In this method, the hydrolysis reaction was performed in the viscometer under fully controlled temperatures and agitation rates, and the viscosity of the substrate was monitored continuously.

In the present study, we compared novel thermostable enzymes (three cellobiohydrolases (CBHs), two endoglucanases (EGs), two xylanases and a β-glucosidase (βG)), primarily for their ability to reduce the viscosity of pretreated lignocellulose at high initial solids. Wheat straw, which is an abundant and promising raw material for second-generation ethanol production, was used as the lignocellulose substrate. Viscosity measurements (RVA method) focused on the early stage of the hydrolysis, where changes in the rheological behaviour of the substrate are most pronounced. Tailored mixtures containing an enhanced ratio of the best-performing liquefying enzyme were prepared and evaluated for their hydrolytic performance at different temperatures.

The viscosity of lignocellulosic substrates is known to decrease as a result of cellulolytic activity, with the most probable reason being the collapse of structure and subsequent loss of water-binding capacity upon cellulose degradation [21]. The phenomenon is well documented in the literature, but information on the individual role of the various types of lignocellulose-degrading enzymes in the liquefaction remains limited. Our viscosity trials were carried out to compare the enzymes in short-time hydrolytic reactions.

The RVA viscometer is an approved apparatus for testing the viscous properties of starch, grain and flour, and formulated, extruded or cooked food. Recently, it has also been used for malted grains to mimic the liquefaction process and to identify the mashing properties of grains [33, 34]. Based on these applications, the RVA viscometer was considered to suit to other heterogeneous materials containing soluble and insoluble compounds, such as pretreated lignocellulose, at high consistencies. The studies were aimed to compare the effects of individual enzymes in the improvement of fluidity (reduction of viscosity) of the high-solid substrate at a phenomenal level, and were not meant to discuss the changes in the material properties in terms of absolute viscosity values.

Of the studied enzymes, the EGs were clearly the most efficient in rapidly reducing the viscosity of the biomass suspension. EGs are known to act preferentially on the amorphous water-binding parts of the cellulose fibres [35] and, therefore, their distinguished liquefaction potential can be explained by the loss of water-binding capacity and eventual collapse of the substrate upon EG action. By contrast, the CBHs, xylanases and βG had no substantial effect on the fluidity of the material. These results are consistent with our previous observations using purified enzymes from the industrially relevant fungus T. reesei in similar viscosity trials [22]. No clear difference between the EGII/Cel5A and EGV/Cel45A could be observed at 50°C or 60°C. The relative molar dosage of EGV/Cel45A was about 30% higher than that of EGII/Cel5A, thus reflecting a slightly better performance of the latter when used at the same protein dosage. Based on its temperature profile showing a clear decrease in viscosity even at 75°C, Ta EGII/Cel5A was chosen as the enzyme with the highest potential for efficient liquefaction at high temperatures. Primarily, the high thermostability of this EG type of enzyme can make a liquefaction stage possible at a notably elevated temperature. There is no direct comparison available in the literature on the hydrolytic performance of the two enzymes, but At EGV/Cel45A has recently been compared with EGII enzymes from other thermophilic sources [36]. At EGV/Cel45A was found to solubilize at least twice as much sugar from phosphoric acid-swollen cellulose, Avicel or bacterial cellulose as the Myceliphtora thermophila EGII/Cel5A after 24 hours of hydrolysis, but was approximately half as efficient as the Thielavia terrestris EGII/Cel5A in similar hydrolysis studies based on protein content.

The efficient hydrolytic performance of the enzyme(s) used in the pre-hydrolysis step is important. In an ideal case, the liquefying component(s) potentially already present in the pre-hydrolysis mixture would contribute to a significant extent to the efficient saccharification of the substrate, and thus reduce the overall enzyme requirement. Owing to the inherent synergistic behaviour of the various types of hydrolytic enzymes [37–39], it may be beneficial to use a dedicated mixture during the pre-hydrolysis step, thus allowing maximized exploitation of complementing hydrolytic activities. With this in mind, the combined hydrolytic performance of thermostable enzymes was studied in mixtures containing an enhanced proportion of the key liquefying enzyme, Ta EGII/Cel5A, in various ratios with the other components. Based on the protein content, the mixtures contained the same amount of EG as used in the RVA studies, to ensure at least as efficient liquefaction as obtained previously. The EG:CBH ratio was kept constant, and the mixtures were supplemented with the XYL and βG preparations.

The saccharification properties of mixtures containing the different CBH enzymes differed significantly, and the order of their performance also varied depending on the temperature. The results were not fully consistent with the optimum temperatures of the CBH enzymes determined by Voutilainen et al.[14] using the initial hydrolysis rate of single enzymes at various temperatures. In that study, using a short time activity assay (10 minutes) against the soluble substrate methylumbelliferyl lactoside (MUL), the Ct CBHI was always superior to the other thermostable CBH enzymes at temperatures from 30°C to 75°C. However, we found At CBHI/Cel7A to be clearly the best-performing CBH on a solid substrate up to a temperature of 55°C, above which the activity decreased.

The addition of xylanase improved the saccharification performance of mixtures in all combinations, with better performance achieved using Ta XYL/Xyn10A. The mechanism most probably involved an increase in free cellulose surface by removal of the residual hemicellulose from the pretreated fibres, thus helping the action of the cellulases synergistically, as suggested by the results of GC analysis.

The mixtures used in this study contained an enhanced ratio of the key liquefying enzyme, EGII/Cel5A, to achieve efficient reduction of substrate viscosity in the high-solid process. The mixtures were not optimized for efficient saccharification, thus the relatively low CBH content and the lack of CBHII and other minor activities (accessory enzymes) were expected to limit the hydrolytic performance considerably. Nevertheless, some of the more efficient mixtures outperformed the commercial reference enzymes at higher reaction temperatures.

In this study, heterologously expressed enzymes from various thermophilic organisms were compared in the liquefaction of hydrothermally pretreated wheat straw using a real-time viscometric technique. Of the studied enzymes, EGs gave the most pronounced reduction in viscosity of the substrate, even during the first hour of the hydrolysis. EGII/Cel5A from T. aurantiacus was efficient up to 75°C. The performance of various thermostable CBHI/Cel7A enzymes was compared in mixtures containing a high proportion of the key liquefying enzyme. The most efficient CBH enzyme was At CBHI/Cel7A. The optimal combination of primary liquefying and secondary saccharifying enzymes can be expected to result in a reduced total enzyme use and correspondingly decreased costs in the production of bioethanol from high-solid lignocelluloses.