UBC Faculty Research and Publications

Cellulose accessibility limits the effectiveness of minimum cellulase loading on the efficient hydrolysis… Arantes, Valdeir; Saddler, Jack N Feb 10, 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13068_2010_Article_79.pdf [ 1.22MB ]
JSON: 52383-1.0223081.json
JSON-LD: 52383-1.0223081-ld.json
RDF/XML (Pretty): 52383-1.0223081-rdf.xml
RDF/JSON: 52383-1.0223081-rdf.json
Turtle: 52383-1.0223081-turtle.txt
N-Triples: 52383-1.0223081-rdf-ntriples.txt
Original Record: 52383-1.0223081-source.json
Full Text

Full Text

Cellulose accessibility limits the effectiveness ofminimum cellulase loading on the efficienthydrolysis of pretreated lignocellulosic substratesArantes and SaddlerArantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3 (10 February 2011)RESEARCH Open AccessCellulose accessibility limits the effectiveness ofminimum cellulase loading on the efficienthydrolysis of pretreated lignocellulosic substratesValdeir Arantes*, Jack N Saddler*AbstractA range of lignocellulosic feedstocks (including agricultural, softwood and hardwood substrates) were pretreatedwith either sulfur dioxide-catalyzed steam or an ethanol organosolv procedure to try to establish a reliableassessment of the factors governing the minimum protein loading that could be used to achieve efficienthydrolysis. A statistical design approach was first used to define what might constitute the minimum proteinloading (cellulases and b-glucosidase) that could be used to achieve efficient saccharification (defined as at least70% glucan conversion) of the pretreated substrates after 72 hours of hydrolysis. The likely substrate factors thatlimit cellulose availability/accessibility were assessed, and then compared with the optimized minimum amounts ofprotein used to obtain effective hydrolysis. The optimized minimum protein loadings to achieve efficient hydrolysisof seven pretreated substrates ranged between 18 and 63 mg protein per gram of glucan. Within the similarlypretreated group of lignocellulosic feedstocks, the agricultural residues (corn stover and corn fiber) requiredsignificantly lower protein loadings to achieve efficient hydrolysis than did the pretreated woody biomass (poplar,douglas fir and lodgepole pine). Regardless of the substantial differences in the source, structure and chemicalcomposition of the feedstocks, and the difference in the pretreatment technology used, the protein loadingrequired to achieve efficient hydrolysis of lignocellulosic substrates was strongly dependent on the accessibility ofthe cellulosic component of each of the substrates. We found that cellulose-rich substrates with highly accessiblecellulose, as assessed by the Simons’ stain method, required a lower protein loading per gram of glucan to obtainefficient hydrolysis compared with substrates containing less accessible cellulose. These results suggest that therate-limiting step during hydrolysis is not the catalytic cleavage of the cellulose chains per se, but rather the limitedaccessibility of the enzymes to the cellulose chains due to the physical structure of the cellulosic substrate.BackgroundBioethanol derived from the bioconversion of lignocellu-losic feedstocks continues to attract global interest as apotentially environmentally compatible alternative tocurrent petroleum-based transportation fuels. However,considerable technical improvements are still neededbefore efficient and economically feasible lignocellulosicbiomass-based bioethanol processes can be commercia-lized. One of the major limitations of this process is theconsistently high cost of the enzymes involved in theconversion of the cellulose component into fermentablesugars [1]. This is primarily due to the comparativelyhigh (compared with amylase loadings required forstarch hydrolysis) protein loadings commonly requiredto overcome the substrate features and enzyme-relatedfactors limiting effective cellulose hydrolysis [2]. Achiev-ing rapid and complete enzymatic hydrolysis of lignocel-lulosic biomass at low protein loadings continues to bea major technical challenge in the commercialization ofcellulose-based processes converting biomass to ethanol.In a typical batch enzyme-based process, celluloseconversion-time experiments are characterized by athree-phase curve (Figure 1A). This usually starts withthe rapid adsorption of the cellulases onto the readilyaccessible cellulose, followed by an initial, fast rate ofhydrolysis. However, the reaction quickly reaches anintermediate phase, characterized by a moderate hydro-lysis reaction rate when about 50-70% of the original* Correspondence: varantes@forestry.ubc.ca; Jack.Saddler@ubc.caForestry Products Biotechnology/Bioenergy Group, Faculty of Forestry,University of British Columbia, 2424 Main Mall, Vancouver BC, V6T 1Z4,CanadaArantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3© 2011 Arantes and Saddler; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.substrate has been hydrolyzed. Thereafter, a very slowphase is characterized by a steady decrease of the reac-tion rate, which results in only a slight increase in theconversion of the remaining (the so-called ‘inaccessible’or recalcitrant) cellulose. Typically, extended hydrolysistimes and/or high protein loadings are required toachieve a near-complete conversion of cellulose (Figure1B). In some cases, depending on the nature of the sub-strate and the pretreatment method used, even at veryhigh protein loadings of the commercially available Figure 1 Typical time course of (A) the enzymatic hydrolysis of cellulose; (B) cellulose hydrolysis with increasing protein loadings.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 2 of 16cellulase mixtures (Figure 1B, curve D) and extensivehydrolysis times, complete cellulose hydrolysis cannotbe achieved (Figure 1B) [3,4].Previous technoeconomic modeling has shown thatthe long hydrolysis time associated with achieving com-plete cellulose saccharification adds significantly to theoperating costs of the enzymatic hydrolysis step and,consequently, to those of the overall biomass to ethanolbioconversion process [5]. Recently, Shen and Agblevor[6] also studied the effect of hydrolysis time and enzymeloading on the hydrolysis of mixtures of cotton ginwaste and recycled paper sludge with the aim of maxi-mizing profit. This work indicated that the use of higherenzyme loadings to achieve >90% cellulose hydrolysislevels was difficult to justify because of the increasedenzyme costs. In the work we describe here, we set atarget of achieving at least 70% glucan hydrolysis of arange of lignocellulosic substrates, using the lowest pos-sible enzyme loading.Various substrate- and enzyme-related factors havebeen suggested to explain the slowdown in the rate ofhydrolysis and, in many cases, the incomplete hydrolysisof cellulosic materials. Although there is still consider-able debate about the contribution of each of these fac-tors, it has been suggested that the accessible surfacearea of the cellulose is one of the most important fac-tors influencing the rate and extent of enzymatic hydro-lysis of lignocellulosic substrates [7-9]. This is notsurprising, as the enzymatic hydrolysis of cellulose is asurface-dominated phenomenon, and direct physicalcontact between the cellulase enzymes and substratemust occur.One of the major barriers faced by cellulase enzymesduring lignocellulose hydrolysis is their limited access tomuch of the cellulose, which is buried within the highlyordered and tightly packed fibrillar architecture of thecellulose microfibrils [10]. Cellulosic materials are typi-cally not smooth but rather heterogeneous porous sub-strates, and their available surface area can generally bedivided into exterior and interior surfaces. The lattercan consist of internal pores, fissures and micro-cracks,which typically arise from ‘discontinuities’ of the mole-cular packing built into the cellulose at the time thesolid substrate is generated [11], or surface openings/internal slits, voids or spaces created by the removal ofnon-cellulosic cell wall components during pretreatment[12-14]. The external surface area of cellulosic-richmaterials is largely determined by the individual overallfiber dimensions [15].Earlier work by Grethlein [16] showed a linear correla-tion between the initial hydrolysis rate of pretreated bio-mass and the pore size accessible to a molecule with adiameter of 5.1 nm, which is about the diameter of a‘representative’ cellulase. More recent work by Thygesenet al. [17], using fluorescent-labeled enzymes combinedwith confocal fluorescence microscopy, showed that cel-lulases were able to penetrate into the porous regions ofthe cellulose before any significant cellulose depolymeri-zation was observed. Indeed, it has been suggested thatthe enzymatic hydrolysis of cellulose could occur bothon the external surface by a sequential ‘shaving’ or‘planing’ of the cellulose fibrils, or by key components ofthe cellulase mixture entering pores/fissures largeenough to accommodate enzymes and then initiatingthe actual cellulose depolymerization process after aswelling action to increase substrate availability [10]. Ineither case, the cellulose topology/porosity can beexpected to be an important factor that would influencethe amount of protein adsorbed onto the substrate.Although previous studies have highlighted the impor-tance of cellulose accessibility during enzymatic hydroly-sis [16-20], the majority of the studies have employedonly a small number of samples and, in many cases,made use of a highly digestible ‘model’ or pure cellulosicsubstrates, which are not really indicative of how therealistic, natural heterogeneous, lignocellulosic feed-stocks might behave. At the same time, the relationshipbetween the substrate surface area and cellulose digest-ibility is sometimes contradictory and in many casesinconclusive. This is also, at least in part, probably dueto the dependency of the accessible cellulose surfacearea on the nature of the substrate (for example, itssource, pretreatment and storage) and the enzyme pre-paration used (complexity, type, composition, concentra-tion), and on difference in the methods employed toassess changes. It is recognized that, some methodsused to measure the available surface area of cellulosicmaterials are not particularly accurate (for example,water retention value, mercury porosimetry), and othersinvolve drying the samples (for example, nitrogenadsorption technique). In the latter case, the pores ofthe wet cellulose fibers have been shown to shrink suc-cessively as the moisture content is decreased (an irre-versible phenomenon termed ‘hornification’ [21]),resulting in smaller pore sizes and narrowed pore sizedistribution [22], which make the material less suscepti-ble to enzymatic hydrolysis [20]. In addition to the diffi-culties experienced in measuring substrate changesoccurring during hydrolysis, comparatively high enzymedosages have been employed in many of these past stu-dies, possibly masking any differences that might havebeen observed in the substrate characteristics. Thus, theinfluence that the specific surface area of the cellulosemight have on our ability to achieve fast and completeenzymatic hydrolysis of pretreated lignocellulosicfeedstocks at low protein loadings remains ambiguous.Therefore, further work, using a broad range of lignocel-lulosic substrates, moderate protein loadings, andArantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 3 of 16substrate characterization methods that do not requiresample drying, is warranted.In the present study, a range of lignocellulosic feed-stocks (including agricultural, softwood and hardwoodsubstrates) were subjected to sulfur dioxide (SO2)-catalyzed steam and ethanol organosolv (EO) pretreat-ment at previously determined conditions [23-27] thatwere deemed optimal for both good hemicelluloserecovery and subsequent hydrolysis of the cellulose-richstream. These pretreated substrates were used to deter-mine the key factors that might help achieve efficienthydrolysis at low enzyme loadings. With this goal inmind, a statistical design approach was first used todefine what might constitute a minimum protein load-ing for the efficient hydrolysis of a range of pretreatedsubstrates. In parallel, the substrate factors (that is, theexternal and internal surface area of cellulose-rich sub-strates) that might limit the accessibility of the cellulasecomplex to the cellulose and the maximum proteinadsorption capacity were measured for each substrate,in an attempt to correlate cellulose accessibility with theminimum protein-loading requirement for efficienthydrolysis. We also evaluated the influence of increasinghydrolysis times and solids loadings on the minimumprotein loading (cellulase and b-glucosidase) required toachieve efficient hydrolysis.The aim of this work was that, by defining the mini-mum protein loading required to achieve efficienthydrolysis, it would help us to better understand whichkey factors limit the fast and near-complete hydrolysisof cellulosic substrates at moderate protein loadings. Asindicated in the paper, the results provided us withsome insights into how we could improve accessibilityto the cellulose fibers/microfibrils, consequently improv-ing the enzymatic digestibility of lignocellulosicmaterials.Materials and methodsEnzyme preparationsTwo commercial preparations (both Novozymes, Frank-linton, NC, USA) - a cellulase cocktail (Celluclast 1.5 L;protein content 129.8 mg/mL) derived from Tricho-derma reesei and a b-glucosidase preparation (Novozym188; protein content 233 mg/mL) derived from Aspergil-lus niger - were used in the enzymatic hydrolysis experi-ments. Protein concentrations were determined usingthe modified ninhydrin method [28]. Bovine serum albu-min was used as the protein standard.Lignocellulosic feedstocks and pretreatment technologiesRepresentatives of agricultural residues (corn stover andcorn fiber), softwood (douglas fir and beetle-killed lod-gepole pine) and hardwood (hybrid poplar) were used inthis study.Lignocellulosic feedstocks were pretreated by SO2-catalyzed steam and/or EO pretreatment as describedpreviously [29]. Most of the pretreatments were per-formed at near-optimal pretreatment conditions (Table1), which have previously been determined in ourlaboratories (steam-pretreated corn stover (SPCS) [25],corn fiber (SPCF) [23], douglas fir (SPDF) and lodgepolepine (SPLP) [27], and EO-pretreated lodgepole pine(OPLP) [26] and poplar (OPP) [24]) to obtain goodoverall carbohydrate recovery (that is, hemicellulosesand cellulose) while producing cellulose-rich substratesamenable to enzymatic hydrolysis. After pretreatment,all substrates (solid fractions) were thoroughly washed,filtered, and kept in refrigerated storage until they wereused for analysis and hydrolysis.Chemical analysis of pretreated feedstocksThe chemical composition of the pretreated materialswas determined according to a standard method (T222om-88; Technical Association of the Pulp and PaperIndustry), as previously described [30]. Monosaccharideswere analyzed by high-performance liquid chromatogra-phy with fucose as the internal standard, as previouslydescribed [31]. All analyses were performed in triplicate.Carbohydrate and lignin contents are shown in Table 1.Defining minimum protein loadings for efficienthydrolysisOptimization of minimum protein loadings required forefficient glucose release from a broad range of pre-treated substrates was performed according to a centralcomposite design in the form of a 24 full factorial designexperiment with three central points. The dependentvariable was glucan conversion, expressed as percentage,and the independent variables were the cellulase (Cellu-clast 1.5 L) and b-glucosidase (Novozym 188) loadings,hydrolysis time, and solids loading. The range and thelevels of these variables are given in Table 2.To describe and predict the effect precisely and quan-titatively, the hydrolysis data for each of the pretreatedmaterials was fitted using a second-order polynomialmodel and Statistica software (version 6.0; Statsoft Inc.,Tulsa, OK, USA).Enzymatic hydrolysisBatch hydrolysis of pretreated substrates was carried outin sodium acetate buffer 50 mmol/L pH 4.8, supplemen-ted with 0.02% w/v tetracycline and 0.015% w/v cyclohex-amide, to prevent microbial contamination. The reactionmixtures (1 mL) were mechanically shaken in an orbitalshaker incubator (Combi-D24 hybridization incubator,FINEPCR®, Yang-Chung, Seoul, Korea) at 50°C. The con-ditions for cellulase and b-glucosidase loadings, hydroly-sis time, and solids loadings were determined accordingArantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 4 of 16to the statistical design of experiments (Table 3). Glucoseconcentration was determined using a microscale enzy-matic assay involving glucose oxidase and horseradishperoxidase as adapted by Berlin et al. [32]. Hydrolysisyields (%) of the pretreated substrates were calculatedfrom the cellulose content as a percentage of thetheoretically available cellulose in the pretreated sub-strate. Enzymatic digestibility of the pretreated materialsrefers to the enzymatic digestibility of cellulose only,unless otherwise stated.Available surface areaProtein adsorptionThe maximum extent of protein (cellulase and b-glucosidase) adsorption was used as an indication of thesurface area of a particular substrate available for proteinbinding. Protein adsorption isotherms were established byvarying the amounts of protein (cellulase + b-glucosidase)added to the different pretreated substrates (2 mg/mL) insodium acetate buffer (50 mmol/L, pH 4.8). The cellulase:b-glucosidase ratios were obtained by assessing the mini-mum protein loading required for efficient hydrolysis. Freeprotein was determined by measuring the amount of pro-tein in the supernatant after incubation at 4°C and 150rpm for 1 hour to reach equilibrium. Bound protein wascalculated as the difference between free protein and thetotal protein initially added to the reaction medium. Theprotein content was determined using the ninhydrin assay[28]. The experimental data was fitted to the Langmuiradsorption isotherm using the following linearized form ofthe equation:1 P 1 P K 1 P Pads max p max/ / ( / ) , in which P is the concentration of unadsorbed protein(mg of protein/mL), Pads is the concentration ofadsorbed protein (mg of protein/mg of substrate), Pmaxis the maximal adsorbed protein (mg of protein/mg ofsubstrate) and Kp is the equilibrium constant (mL/mg ofprotein).Table 1 Pretreatment conditions and chemical composition of pretreated lignocellulosic substratesSubstrate Pretreatment conditions Composition of pretreated feedstocks AbbreviationSO2-steam pretreatmenta Arab Galc Glud Xyle Manf AILgCorn stover 190°C, 5 minutes, 3% SO2 0.8 0.2 55.1 12.0 1.9 18.9 SPCSCorn fiber 190°C, 5 minutes, 4% SO2 6.9 2.8 38.2 15.3 2.2 12.6 SPCSDouglas fir 200°C, 5 minutes, 4% SO2 BDLg BDL 50.6 0.4 1.0 47.0 SPDFLodgepole pine 200°C, 5 minutes, 4% SO2 BDL BDL 52.4 0.6 1.0 45.9 SPLPEthanol-organosolv pretreatmentCorn fiber 170°C, 30 minutes; 65% EtOH, 0.75% H2SO4 2.1 1.6 57.9 11.5 3.0 15.7 OPCFPoplar 195°C, 40 minutes; 70% EtOH, 1.0% H2SO4 BDL BDL 77.0 6.0 2.4 16.0 OPPLodgepole pine 170°C, 60 minutes; 65% EtOH, 1.1% H2SO4 0.1 0.1 74.8 1.6 1.8 17.3 OPLPaSulfur dioxide.bArabinan.cXylan.dGlucan.eGalactan.fMannan.gAcid-insoluble lignin.hBelow detectable level.Table 2 Coded and actual levels of variables chosen forthe statistical design of experimentFactors Level PretreatmentSO2 steama EthanolorganosolvCS,b DF,cLPdCFe CF LP, PfSolids loading, % -1 2 2 2 20 6 6 6 61 10 10 10 10Hydrolysis time, hours -1 24 24 24 240 48 48 48 481 72 72 72 72Cellulase,g mg protein/g glucan -1 25 5 13 200 50 15 26 451 75 25 39 70b-glucosidase,h mg protein/gglucan-1 0 0 0 00 15 15 10 101 30 30 20 20aSO = sulfur dioxide.bCS = corn stover.cDF = douglas fir.dLP = lodgepole pine.eCF = corn fiber.fP = poplar.gCelluclast 1.5L.hNovozym 188.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 5 of 16Fiber lengthThe external surface area of the cellulosic-rich substratesmeasured as the average fiber length of the pretreatedsubstrates was determined using a high-resolution fiberquality analyzer (FQA) (LDA02; OpTest Equipment, Inc.,Hawkesbury, ON, Canada) in accordance with the proce-dure described by Robertson et al. [33]. Briefly, a dilutesuspension of fibers with a fiber frequency of 25 to 40events per second was transported through a sheath flowcell where the fibers were oriented and positioned. Theimages of the fibers were detected by a built-in charge-coupled device (CCD) camera, and the length of thefibers was measured by circular polarized light. All sam-ples were run in duplicate.Simons’ stainSimons’ stain (SS), a staining technique used in the pulpand paper industry to examine changes in the physicalstructure of pulp fibers under the microscope, andadapted for evaluating the pore structure (internal sur-face area) of cellulosic materials [34], was performedaccording to the modified procedure by Chandra et al.[35]. Pontamine fast orange 6RN (direct orange; DO)and Pontamine fast sky blue 6BX (direct blue; DB) dyeswere used (Pylam Products Co. Inc., Garden City, NY,Table 3 Matrix and results of a 24 full factorial design with centered face and three repetitions at the center point forsteam- and organosolv-pretreated lignocellulosic substratesRun Factors Glucan to glucose, %Number Time Solids Cell.a BGb SPCFc SPCSd SPLPe SPDFf OPPg OPCFh OPLPi1 -1 -1 -1 -1 16 46 8 18 12 27 52 1 -1 -1 -1 20 59 22 28 12 48 223 -1 -1 -1 1 74 62 40 43 27 82 394 1 -1 -1 1 79 68 44 46 33 95 575 -1 -1 1 -1 31 57 37 44 39 48 276 1 -1 1 -1 29 71 60 42 58 82 857 -1 -1 1 1 53 73 90 64 88 98 788 1 -1 1 1 54 61 72 63 82 76 1019 -1 1 -1 -1 5 32 16 15 13 14 1210 1 1 -1 -1 16 42 25 24 26 28 2411 -1 1 -1 1 52 41 33 33 34 43 3412 1 1 -1 1 54 46 45 44 45 61 5213 -1 1 1 -1 24 43 34 35 34 33 3014 1 1 1 -1 37 53 52 52 56 62 5715 -1 1 1 1 55 52 53 60 61 69 6316 1 1 1 1 56 67 63 74 82 90 9217 0 -1 0 0 78 73 72 70 65 98 8418 0 1 0 0 59 53 47 52 63 70 6819 0 0 -1 0 51 61 45 35 42 68 1020 0 0 1 0 63 64 74 75 82 78 8521 0 0 0 -1 31 55 40 42 37 43 3822 0 0 0 1 61 65 71 68 63 91 7923 -1 0 0 0 63 62 60 61 52 74 6224 1 0 0 0 63 70 57 72 89 97 9125 0 0 0 0 58 63 61 67 72 73 6726 0 0 0 0 68 64 63 62 66 66 7727 0 0 0 0 60 61 69 65 73 67 74aCellulase.bBG = b-glucosidase.cSPCF = steam-pretreated corn fiber.dSPCS = steam-pretreated corn stover.eSPLP = steam-pretreated lodgepole pine.fSPDF = steam-pretreated douglas fir.gOPP = organosolv-pretreated poplar.hOPCF = organosolv-pretreated corn fiber.iOPLP = organosolv-pretreated lodgepole pine.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 6 of 16USA). Fractionation of DO was performed according toEsteghlalian et al. [20].Results and DiscussionIt has been predicted that a diverse range of plant bio-mass will be needed to satisfy the projected demands forsecond-generation bioethanol [36]. As would beexpected, different feedstocks (for example, agriculturalresidues vs. forest biomass) have significant qualitativeand quantitative differences in their component andstructural arrangements. Additionally, further differencessuch as composition/distribution and arrangement ofcomponents are introduced during the pretreatmentstep, and are heavily influenced by the pretreatmentprocess employed. This variability is known to have asignificant effect on the enzymatic hydrolysis step [37].Therefore, our initial approach was to select a broadrange of lignocellulosic feedstocks, including representa-tives of agricultural and forest biomass, and to pretreatthese materials under conditions that allowed maximumhemicellulose recovery and good enzymatic hydrolysis ofthe cellulosic component. Subsequently, a statisticalexperimental design was used to define the minimumamounts of protein required for efficient hydrolysis ofthe pretreated substrates, in order to establish a reliableassessment of the factors governing the minimum pro-tein loading required for efficient hydrolysis of each ofthe pretreated substrates.The minimum protein loading for efficient hydrolysiswas initially optimized before any correlation was made,to account for the possibility that previous predictionsof the hydrolyzability of pretreated lignocellulosics basedon either low or high protein levels might not be asmeaningful or as accurate as predicted. For example,predictions based on low protein loadings might onlyinclude saccharification of the so-called ‘easy/accessible’cellulose, and thus factors that control the digestibilityof cellulose at high levels of conversion might not havebeen assessed. By contrast, experiments carried out athigh protein loadings might, by saturating the substratewith enzymes, mask important factors limiting efficienthydrolysis.The enzymatic digestibility of the seven pretreatedsamples using varying protein and solids loadings andhydrolysis times was assessed by monitoring the amountof glucose released (Table 3), and the effect of each ofthe variables and their interactions during hydrolysiswas assessed by direct analysis of their statistical signifi-cance with a reliability of 95% (Table 4). This approachwas chosen because the significance of the interactionsbetween the variables would have been lost if the experi-ments were carried out using the classic methods ofvarying the level of one parameter at a time over acertain range, while holding constant the rest of thetested variables.Regression analyses (ANOVA) were carried out toobtain mathematical models (Table 5) that betterdescribe the relation between the independent variables(cellulase loading, b-glucosidase loading, hydrolysis time,and solids loading) and the studied response (glucosereleased). To prepare the adjusted models and their sur-faces (Figure 2), only terms found to be significant atP ≤ 0.05, or values near to this, were included in themodels. The validity of the models was evaluated as afunction of their respective coefficients of determination(R2). The value of the correlation coefficient provides ameasure of variability in the observed response valuesthat can be explained by the experimental factors andtheir interactions (the closer the R2 value to 1.0, the bet-ter the fit of the model to the experimental data). Themodels computed for the R2 value ranged between 0.91and 0.96 (Table 5), indicating that the models wereappropriate and could be used for quantitative predic-tion of the minimum protein loadings (cellulase andb-glucosidase) required to attain efficient cellulose con-version, and for assessment of the effect of time andsolids loading during hydrolysis.Determining the minimum cellulase and b-glucosidaserequirement for efficient hydrolysisThe commercial cellulase cocktail (Celluclast 1.5 L)derived from the filamentous fungus T. reesei, consistsmainly of cellobiohydrolases and endoglucanases[38-40]. Owing to the low level of in situ b-glucosidaseactivity, this T. reesei cellulase system is commonly sup-plemented with an excess of b-glucosidase to avoid anyend-product inhibition caused by the accumulation ofcellobiose, which would mask the actual minimum cel-lulase requirement at higher levels. It has also beenreported that synergism between cellulase enzymesdecreases at high cellulase concentrations (aroundsaturation levels) [41]. Thus, to avoid the use ofexcess protein and to take maximum advantage of thesynergistic effect between the cellulases, the minimumb-glucosidase supplementation required for efficienthydrolysis of various pretreated lignocellulosic substrateswas also determined.In this work, we defined effective hydrolysis as at least70% of the original cellulose in the pretreated lignocellu-losic materials being hydrolyzed to glucose. With thispercentage conversion as a target, a meaningful assess-ment of cellulose saccharification could be made beforethe typical, significant slowdown in hydrolysis rate tookplace (Figure 1A). As mentioned earlier [6], a recenteconomic assessment of the influence of protein loadingon the maximum profit rate for ethanol productionArantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 7 of 16Table 4 Estimated effects (P-value at 95% confidence level) for glucan conversion during hydrolysis of variouspretreated lignocellulosic substratesFactor SPCFa SPCSb SPLPc SPDFd OPPe OPCFf OPLPgMean/Interc. <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.00011Time (Lh) 0.1209 <0.0001** 0.0037* 0.0120* 0.0002** <0.0001** <0.0001**Time (Qi) 0.7302 0.4838 0.2711 0.6937 0.4510 0.3613 0.22552Consistency (L) 0.0056* <0.0001** 0.0049* 0.1038 0.4281 <0.0001** 0.0295*Consistency (Q) 0.2199 0.2965 0.4479 0.2823 0.4004 0.4919 0.24973Cellulase (L) 0.1387 <0.0001** 0.0000** 0.0000** <0.0001** <0.0001** <0.0001**Cellulase (Q) 0.0420* 0.1679 0.4215 0.0163* 0.1568 0.1228 0.0013*4BGj (L) <0.0001** <0.0001** <0.0001** <0.0001** <0.0001** <0.0001** <0.0001**BG (Q) 0.0001** 0.0145* 0.0783* 0.0171* 0.0005 0.0109* 0.0563*1L by 2L 0.5554 0.7365 0.7954 0.3060 0.2724 0.6278 0.14001L by 3L 0.8592 0.2955 0.3451 0.8947 0.1263 0.2986 0.0122*1L by 4L 0.5278 0.2785 0.2945 0.7946 0.7840 0.3451 0.86092L by 3L 0.0179* 0.8421 0.0087* 0.6421 0.0059* 0.9680 0.0537*2L by 4L 0.1884 0.1729 0.0197* 0.6285 0.1901 0.0372* 0.22983L by 4L 0.0006* 0.2711 0.2314 0.7789 0.0293* 0.5689 0.1829R2 0.9692 0.9692 0.9549 0.9529 0.9704 0.9538 0.9589aSPCF = Steam-pretreated corn fiberbSPCS = steam-pretreated corn stover.cSPLP = steam-pretreated lodgepole pine.dSPDF = steam-pretreated douglas fir.eOPP = organosolv-pretreated poplar.fOPCF = organosolv-pretreated corn fiber.gOPLP = organosolv-pretreated lodgepole pine.hL = linear.iQ = quadratic.jBG = b-glucosidase.*Significant model terms; ** highly significant model terms.Table 5 Predictive models describing the relationship between hydrolysis yields of various pretreated lignocellulosicsubstrates and the significant variablesSubstrate Modela R2SPCFb H = 0.1349 + 0.00085T - 0.0257S + 0.0214C - 0.0006C2 + 0.0412B - 0.00075B2 + 0.00096SC - 0.00043CB 0.9563SPCSc H = 0.4346 + 0.002T - 0.0229S + 0.0023C + 0.0131B - 0.00032B2 0.9406SPLPd H = -0.2426 + 0.0023T + 0.0234C + 0.0095C + 0.0327B - 0.0006B2 - 0.0005CB - 0.0007SC 0.9163SPDFe H = -0.2953 + 0.0018T - 0.0065S + 0.0221C - 0.000171C2 + 0.02152B - 0.00047B2 0.9402OPPf H = -0.2409 + 0.0032T + 0.0207S + 0.0094C + 0.05006B - 0.00217B2 - 0.0005C + 0.00015CB 0.9434OPCFg H = 0.1178 + 0.00438T - 0.0214S + 0.00876C + 0.0552B - 0.0014B2 - 0.0012SB 0.9295OPLPh H = -0.5703 + 0.00166T + 0.061S + 0.0312C - 0.00028B2 + 0.0301B - 0.00064B2 + 0.00009TC - 0.00037SC 0.9294aH = hydrolysis yield; C = cellulase loading, mg protein/g glucan; B = b-glucosidase loading, mg protein/g glucan; T = hydrolysis time, hours; S = solids loading,% w/v.bSPCF = steam-pretreated corn fiber.cSPCS = steam-pretreated corn stover.dSPLP = steam-pretreated lodgepole pine.eSPDP = steam-pretreated douglas fir.fOPP = organosolv-pretreated poplar.gOPCF = organosolv-pretreated corn fiber.hOPLP = organosolv-pretreated lodgepole pine.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 8 of 162550750153030%40%50%60%70%80%90% 80 % 70% 60%Hydrolysis yieldSPCSCellulase (mg/g cellulose)B-glucosidase (mg/g cellulose)2550750153020%30%40%50%60%70%80%90% 70% 60% 50% 40% 30%SPDFB-Glucosidase (mg/g cellulose)Cellulase (mg/g cellulose)Hydrolysis Yield510152025 0153010%20%30%40%50%60%70%80%90%SPCFHydrolysis YieldCellulase (mg/g cellulose) B-Glucosidase (mg/g cellulose) 70% 60% 50% 40% 30%2045700102020%40%60%80%100%OPPHydrolysis YieldB-Glucosidase (mg/g cellulose) Cellulase (mg/g cellulose) 100% 80% 60% 40% 20%2550750153020%40%60%80%100%SPLPHydrolysis YieldB-Glucosidase (mg/g cellulose) Cellulase (mg/g cellulose) 80% 60% 40% 20%2045700102025%50%75%100%OPLPHydrolysis YieldB-Glucosidase (mg/g cellulose)Cellulase (mg/g cellulose) 100% 75% 50% 25% 1326390102025%50%75%100% 100% 75%OPCFHydrolysis YieldB-Glucosidase (mg/g cellulose)Cellulase (mg/g cellulose) Figure 2 Response surface fitted to the experimental data corresponding to the hydrolysis of a broad range of pretreated substrates.Hydrolysis times and solids loadings were kept constant at 72 hours and 2% (w/v), respectively.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 9 of 16from biomass substrates suggested that the costsinvolved in achieving complete hydrolysis are prohibi-tive, and that fast but incomplete hydrolysis, leavingabout 10-20% of the original substrate unhydrolyzed,might be a more effective strategy.The effect of cellulase and b-glucosidase loading oncellulose saccharification yields for most pretreatedsubstrates was significantly affected by cellulase andb-glucosidase loadings in the linear term, and less signif-icantly affected by their interaction with each other, andby their interaction with solids loading and hydrolysistime (Table 4). Unexpectedly, the cellulase loading (lin-ear term) was not significant for SPCF. The likely causefor this lack of significance is the heterogeneity of thispretreated substrate (observed visually) making accuraterepresentative sampling unfeasible and resulting in therelatively high standard deviation observed with thehydrolysis yields (Table 3, runs 25-27). The other pre-treated substrates were more homogeneous, exhibiting amudlike consistency, and were thus easier to samplerepresentatively.The significance of the quadratic coefficients ofcellulase loading for SPCF, SPDF and OPLPP, and ofb-glucosidase loading for SPCS, SPCF, SPLP, SPDF,OPCF, OPLPP, indicate that cellulose saccharificationyields increase with protein loading up to a certain level(Table 4). Beyond that, the entire variable has an inhibi-tory effect on cellulose conversion. It was apparent that,within the range of pretreated substrates and cellulaseand b-glucosidase loadings used in the present study,hydrolysis yields were influenced more by high b-gluco-sidase than by high cellulase loadings.The mathematical models (Table 5) obtained afterregression of the results shown in Table 3 were used toquantitatively predict the minimum protein requirementfor efficient hydrolysis (70% glucan conversion) (Table6). Considering the difference in the degrees of hydro-lyzability of the pretreated substrates (Table 3), thehydrolysis time was kept at 72 hours to ensure that cel-lulose conversion yields reached, or were near to, the‘plateau phase’ of hydrolysis. Solids loadings were keptat 2% (w/v), in an attempt to generate data that couldbe further correlated with the protein adsorption dataobtained from experiments that were also carried out ata 2% (w/v) solids loading. The influence of hydrolysistime and solids loading on the hydrolysis yields of thepretreated materials and on the minimum proteinrequired to achieve efficient hydrolysis were nextassessed.The reliability of the equations was also assessed bycomparing the experimental values of the responses atthe centre point conditions, an average of three inde-pendent experiments (Table 3, runs 25-27), with thevalues calculated using the equations shown in Table 5.The results (data not shown) indicated that the pre-dicted values agreed well with the observed values forthe hydrolysis yields (Table 6). All of the predicted mini-mum protein loadings fell within the range accuratelypredicted by the empirical models.It was apparent that the minimum protein require-ment ranged between 18 and 63 mg protein per gram ofglucan. The minimum protein requirement increased asfollows: OPCF < SPCF < OPLP < OPP < SPCS < SPDF< SPLP. It was observed that for the same feedstock (forexample, corn fiber and lodgepole pine), EO pretreat-ment generally resulted in substrates that required lessprotein to achieve efficient hydrolysis than did steampretreatment. Within the group of feedstocks pretreatedby the same process, the pretreated agricultural residues(corn stover and corn fiber) required lower protein load-ing per gram of glucan to achieve high glucan conver-sion than did the pretreated forest biomass (poplar,douglas fir and lodgepole pine). This confirmed that thenature of the lignocellulosic feedstock plays an impor-tant role in determining the amount of protein requiredfor effective hydrolysis. This was not unexpected, as theplant cell-wall architecture and molecular structure,which are the primary lignocellulosic factors contribut-ing to biomass recalcitrance, are likely to be different inwoody biomass and herbaceous plant-derived biomass.For instance, softwoods have a more rigid structure anda higher lignin content, and are therefore expected todisplay more resistance towards deconstruction (bemore recalcitrant) than the less structurally recalcitrantbiomass derived from herbaceous plants.Table 6 Minimum cellulase and b-glucosidase loadingsrequired for efficient hydrolysis (70% glucan conversion)of a broad range of pretreated lignocellulosic substratesas predicted by the equations shown in Table (5) for 2%(w/v) solids loading and 72 hoursSubstrate Cell.a BGb Cell./BG TotalSPCFc 5 18 0.3 23SPCSd 30 24 1.3 54SPLPe 42 21 2.0 63SPDFf 45 16 2.8 61OPPg 38 10 3.8 48OPCFh 14 4 3.5 18OPLPi 32 11 2.9 43aCell = cellulase (Celluclast 1.5), mg protein/g glucan.bBG = b-glucosidase (Novozym 188) mg protein/g glucan.cSPCF = steam-pretreated corn fiber.dSPCS = steam-pretreated corn stover.eSPLP = steam-pretreated lodgepole pine.fSPDF = steam-pretreated douglas fir.gOPP = organosolv-pretreated poplar.hOPCF = organosolv-pretreated corn fiber.iOPLP = organosolv-pretreated lodgepole pine.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 10 of 16Solids loading and hydrolysis timeTo achieve efficient bioconversion of cellulose to etha-nol, it is desirable that the hydrolyzate obtained afterenzymatic hydrolysis contains a sufficiently high concen-tration of fermentable sugars to result in a high ethanolconcentration. To obtain this high sugar concentration,hydrolysis should be carried out at high solids loading.Raising the solids loading in the enzymatic hydrolysisstep is crucial to minimizing subsequent distillationcosts, and it is also expected to decrease process cost bylowering the reactor size and minimizing water require-ments [42,43]. It has been shown that the time requiredto achieve high conversion rates also contributes to thepoor economics of the hydrolysis step [5]. Therefore, aswell as determining the minimum protein loadingrequired for efficient cellulose hydrolysis, the influenceof time and solids loading on hydrolysis yields (Table 4)were also assessed for all of the pretreated substrates(Figure 3).Regardless of the pretreatment process used, the effectof solids loading was highly significant for all of the pre-treated agricultural residues (SPCF, SPCS and OPCF)(Table 4). This negative effect was probably due to thehigher xylan content in these materials, which at highsolids loading would be likely to result in the release ofhigh concentrations of xylooligomers, which have beenshown to inhibit the action of cellulases [44]. When thepretreated woods were assessed, the solids loading wasonly significant for the SPLP substrate. It was apparentthat the interactive effect of solids loading with cellulaseloading (2L by 3L) was more significant than the inter-active effect with b-glucosidase (2L by 4L) (Table 4), butno correlation was observed between different feed-stocks or pretreatments.The linear effect of hydrolysis time was significant forall of the pretreated substrates, with the exception ofthe SPCF sample (Table 4). This lack of significanceindicated that increasing the hydrolysis time from 24 hto 48 h or 72 hours does not necessarily result in statis-tically higher hydrolysis yields for the SPCF substrate,suggesting that the ‘plateau phase’ was reached withinthe first 24 hours of hydrolysis within the range of pro-tein loadings used in this work. Previous results haveshown that the SPCF substrate can be effectively hydro-lyzed within 24 hours when moderate protein loadingsare used [45].The hydrolysis yields obtained with minimum proteinloading for the steam pretreated wood substrates (SPDFand SPLP) did not seem to be affected by increasing thehydrolysis time from 24 hours to 72 hours, and thehydrolysis yields for the steam-pretreated agriculturalresidues (SPCF and SPCS) were only slightly affected(Figure 3). The greatest influence of hydrolysis time onthe hydrolysis yields at minimum protein loading wasobserved with the EO pretreated samples (OPP, OPCFand OPLP) (Figure 3).When the effect of solids loading on the hydrolysisefficiency of the pretreated materials at the optimizedminimum protein loading was assessed (Figure 3), it wasapparent that increasing the substrate concentrationfrom 2% to 10% (w/v) decreased the hydrolysis yieldsfor the SPCS, OPCF, SPCF and OPP substrates, whereasit had no effect on the yields for the SPDF, OPLP andSPLP substrates. The latter group of substrates had verylow or undetectable levels of xylan, whereas the formergroup of samples had a relatively high xylan content.Again, this negative effect of solids loading on thehydrolysis yields was probably a result of inhibition ofcellulase enzymes by high concentrations of xylooligo-mers at these higher substrate concentrations. Althoughthe cellulase:b-glucosidase ratio was optimized for mini-mum protein loading at a 2% solids loading, it is possiblethat at higher substrate concentrations, cellooligomersmight be produced. As cellooligomers inhibit cellulasesas potently as do xylooligomers, this might also contri-bute to cellulase inhibition, thereby lowering hydrolysisyields at high solids loading, as a result of limitedb-glucosidase levels.Protein adsorptionThe binding of cellulase enzymes onto insoluble andheterogeneous lignocellulosic biomass has beenreported to have a strong role in governing the ratesand yields of hydrolysis of cellulose [3,46], and also tobe influenced by the available surface area of cellulose[46,47]. Therefore, the maximum cellulase adsorptionwas used as a parameter to measure the accessibility ofthe seven pretreated lignocellulosic substrates. Mix-tures of cellulase and b-glucosidase over a range ofconcentrations were incubated with 2% (w/v) of pre-treated material. The ratios of cellulase to b-glucosi-dase were based on the optimized minimum cellulaseand b-glucosidase loadings required for efficient hydro-lysis (Table 6). When the maximum amount of proteinadsorbed onto the substrates was determined by fittingthe experimental data to the Langmuir adsorption iso-therm model, a good correlation (R2 > 0.9790) wasobtained. An assessment of the Langmuir adsorptionisotherm revealed significant differences for proteinadsorption onto the different pretreated lignocellulosicmaterials. The maximum adsorption capacity (Pmax) ofproteins onto pretreated materials ranged from 11.0 to89.3 mg/g substrate and increased as follows: steam-pretreated samples: SPLP < SPDF < SPCS < SPCF;then EO-pretreated samples: OPP < OPLP < OPCF(Figure 4).When the optimized minimum protein loading forefficient hydrolysis was plotted against the accessibilityArantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 11 of 16of the substrate, determined as the maximum amount ofprotein that was adsorbed (Pmax), two distinct curveswere observed (Figure 4). Each fitted curve corre-sponded to substrates pretreated by the same process. Itappears that different pretreatment technologies havedifferent effects on the adsorption of proteins onto thesubstrates, and that different feedstocks undergo similarmodifications during the same pretreatment. The differ-ent protein adsorption patterns observed for the EO-and steam-pretreated samples may be, at least in part, aresult of differences in the content and structure of thelignin, which is also known to bind proteins [48]. It hasbeen reported that steam-pretreated substrates containmore binding sites on the lignin-particle surface due to4824487220%40%60%80%100% 80% 70% 60% 50% 40% 30%Solids Loading (% w/w)Hydrolysis yieldHydrolysis TimeOPCF4824487230%40%50%60%70%80%90%100%Solids Loading (% w/w) 70%  65%   60% 58% Hydrolysis Time (h)Hydrolysis yieldSPDF261 02 44 87 220 %40 %60 %80 %10 0%O PPSo lid s  L o a d in g  (w /w )Hyd ro lys is  T im e  (h )Hydrolysis yield 7 0% 6 0%4824487220%40%60%80%100% 74% 70% 60%OPLPHydrolysis yieldSolids Loading (% w/w) Hydrolysis Time (h)4824487230%40%50%60%70%80% 70% 60% 50%SPCSSolids Loading (% w/w)Hydrolysis YieldHydrolysis Time (h) 261 0 2 44 87 220 %40 %60 %80 %10 0%SPCFHyd ro lys is  tim e  (h )Hydrolysis YieldSo lid s  L o a d in g  (%  w /w ) 7 5% 7 0%   6 0%   2610 2448720. Loading (% w/w)Hydrolysis YieldHydrolysis Time (h) 70% 60%  50% Figure 3 Effect of hydrolysis time and solids loading on the minimum protein requirement for efficient hydrolysis of a variety oflignocellulosic substrates. Cellulase and b-glucosidase were kept constant according to the protein level shown in Table 6.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 12 of 16the preservation of functional groups (phenolic hydroxyland benzyl) and lignin branches during this type of pre-treatment than do substrates produced by organosolvpretreatment, which is a delignifying process thatdecreases the total lignin content [49,50]. This suggeststhat protein adsorption patterns can be comparedbetween different feedstocks pretreated by the same pre-treatment technology, but not when pretreated by differ-ent pretreatment technologies.The high correlation values, R2 = 0.993 for the steam-pretreated and R2 = 0.999 for EO-pretreated samples,indicate the strong dependency of the minimum proteinloading required to achieve efficient hydrolysis on themaximum capacity (Pmax) of the substrates to bind protein.For feedstocks pretreated by the same pretreatmenttechnology, it was observed that the higher the capacityof the substrate to adsorb proteins, the lower theamount of protein required to attain efficient hydrolysis.This suggests that the more available the surface area ofthe cellulose-rich material for the proteins to bind to,the lower the protein-loading requirement for efficientcellulose saccharification. We next wanted to confirmthe role that the available surface area of cellulose mighthave on the minimum protein loading required for effi-cient hydrolysis of lignocellulosic substrates.External and internal surface area versus minimumprotein loadingIt has been suggested that the cellulose surface areaaccessible to the cellulase enzymes is one of the mostimportant factors determining the ease of hydrolysis ofcellulosic materials, and it is also affected by several sub-strate characteristics. These features include distributionof particle size, pore volume, degree of crystallinity anddegree of polymerization (DP) [9,35,37,46,51], amongothers. Although previous work has tried to correlateDP and crystallinity with enzymatic digestibility of cellu-losic materials, using a comparison between the hydroly-sis of a fully bleached eucalyptus Kraft pulp and that ofSO2-catalyzed steam-pretreated eucalyptus chips, thesubstrate accessibility to the cellulases could not bereadily predicted from the differences in their celluloseDP or crystallinity, but these substrate characteristicsdid indicate the likely mode of action of the enzymes[52]. From this and other work, it has been shown thatthe specific surface area of a mixture of particles isinversely proportional to the average diameter of theparticles. Therefore, a smaller average particle sizeresults in an increased surface area. Thus, it could beanticipated that a relationship between particle size andcellulose hydrolysis would occur [9].In this study, we assessed the influence of the exteriorsurface area of the cellulosic-rich materials, determinedby fiber dimension/length, on the minimum proteinrequirement for effective enzymatic digestibility of pre-treated lignocellulosic substrates, using a FQA, which isan automated particle size analyzer. We found that theminimum protein requirement for efficient hydrolysishad no correlation with the average initial particle size(Figure 5). Several factors could explain this lack of cor-relation, including the fact that the FQA analysis pro-vides only a gross estimation, as it assumes that thefiber particles are smooth and it does not consider thesurface topology and porosity (cracks and fissures) ofthe particles. Additionally, the size of cellulosic particlescan be difficult to measure because of the presence ofdifferent types of particles and their agglomerates [53].Another possibility is that the minimum proteinFigure 4 Relationship between maximum protein adsorptioncapacity of a range of pretreated lignocellulosic biomass andthe optimized minimum protein loading for efficienthydrolysis.Figure 5 Relationship between minimum protein loading forefficient hydrolysis and external surface area determined asaverage initial fiber length.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 13 of 16requirement for efficient hydrolysis is independent ofthe overall external surface area of the lignocellulosicsubstrate. This is not unexpected, as fiber dimensions,although a good indicator of the external surface area ofcellulosic materials, do not necessarily reflect the overallcellulose surface area available to the cellulase enzymesin the pretreated lignocellulosic materials.It is known that cellulose microfibrils are porous sub-strates, and their overall accessible surface area isexpected to be a combination of exterior and interiorsurface area (for example, substrate porosity and topol-ogy). One method that we have adapted is the SS tech-nique, which measures a combination of both theinterior and exterior surface area of the exposed/accessi-ble cellulose [20,35]. SS is a two-color differential stainthat is sensitive to variations in the accessibility of theinterior structure of fibers [54]. When the cellulosic sub-strates are treated with a mixture of DO and DB dyes,the DB molecules initially populate the pores of thefibers, then the DO molecules gain access to the largerpores and displace the DB molecules because of thehigher molecular size and higher affinity of the DO dye[9,43]. In the present study, in addition to the exteriorarea estimated by measuring particle size, the ratios ofadsorbed DO and DB onto the pretreated lignocellulosicmaterials were used to assess the overall accessible sur-face area of cellulose to cellulases (Figure 6). It has beenshown previously that the molecular diameter of theDO dye molecules is in the range of 5 to 36 nm [54],which is close to the molecular diameter of a ‘typical’fungal cellulase.The overall available surface area of celluloseincreased with the increasing glucan content of the pre-treated substrates, with the exception of SPCF, as evi-denced by the linear correlation observed between theglucan content and the DO:DB ratio (Figure 6). Thisconfirmed the previous suggestion [35] that the use ofSS dyes, more specifically the DO:DB ratio, as molecularprobes is a good indicator of the total (external andinternal) surface area of cellulose available to theenzymes. It was also evident that the higher the DO:DBratio, the lower the protein loading required for efficienthydrolysis (Figure 7). This indicated the strong depen-dency of the minimum protein requirement on theaccessibility of the available cellulose in the pretreatedlignocellulosic materials, and the importance of both theexternal and the internal surface areas (for example,pore volume, fissures and micro-cracks).It has been reported that the internal surface area ofcellulose is much larger than the external surface area[15]. Therefore, it seems logical that the porous struc-ture of cellulose has a major influence on the diffusionof reactants such as cellulase enzymes into the cellulosenetwork. This is in good agreement with the proposedmechanism of enzymatic hydrolysis of cellulose by cellu-lases, which suggests that, rather than cellulose fibrilsbeing slowly eroded by surface ‘shaving’ or ‘planing’, thecellulase enzymes enter through pores large enough toaccommodate them, facilitating the disaggregation andfragmentation of the cellulose. Therefore, the topology/porosity of the available cellulose is an important factorthat may play a key role in limiting the amount of pro-tein that can penetrate into the microfibril defects/poresof the cellulose. Previous work by Thygesen et al. [17]supports this suggested mechanism; they showed thatcellulases first penetrated into the porous regions of cel-lulose, precipitating the subsequent depolymerization. Ithas also been shown that enzymatic degradation doesnot necessarily promote cleavage in the fiber axialFigure 6 Relationship between glucan content and distributionof large and small pores (combination interior/exterior surfacearea) determined by the Simons’ staining technique.Figure 7 Relationship between distribution of large and smallpores (combination interior/exterior surface area) andminimum protein loading for efficient hydrolysis.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 14 of 16direction, as evidenced by significant decrease in fiberlength, but not fiber width [17,55].ConclusionsPrevious work has suggested that the limited available sur-face area of cellulose is a key factor that necessitates theneed for relatively high enzyme dosages to attain effectivecellulose hydrolysis. However, the majority of these studiesused highly digestible, purified cellulosic substrates such asfilter paper, Solka floc or Avicel. In the present study, weused a broad range of more realistic heterogeneous, ligno-cellulosic feedstocks pretreated by promising technologiesunder more representative conditions. Regardless of signif-icant differences in the origin, structure and chemicalcomposition of the feedstocks and the pretreatment pro-cess used, it appears that the minimum protein loadingrequired for efficient hydrolysis of pretreated lignocellulo-sic substrates has no direct relationship with only theexternal surface area of the cellulose-rich materials. How-ever, protein loading did appear to be strongly influencedby the overall enzyme accessibility, as determined by theSS technique, which as well as measuring the external cel-lulose surface area, also takes into account the porosity/topology of the available cellulose.A strong linear relationship between cellulose accessi-bility and the minimum amount of protein required toachieve effective hydrolysis was apparent, at least withthe enzyme cocktail used in this study. As regards theenzymatic mechanism, these results suggest that someof the cellulase components may initially penetrate intoareas of the cellulose, particularly the amorphousregions that are large enough to accommodate cellulaseenzymes, disrupting/fragmenting the cellulose fibersbefore significant hydrolysis of cellulose takes place.The fact that the more available/exposed cellulose inthe pretreated lignocellulosic structure required lowerprotein levels per gram of glucan to attain high diges-tion rates suggests that the rate-limiting step duringhydrolysis may not be the actual catalytic cleavage ofthe cellulose chains per se but rather the limited accessi-bility of the enzymes to the cellulose chains within thesubstrate matrix.AcknowledgementsThe Natural Sciences and Engineering Research Council of Canada (NSERC),Natural Resources Canada (NRCan) and Genome BC are gratefullyacknowledged for the support of this work. We thank our colleagues LinojKumar, Luis Del Rio and Richard Chandra for providing or helping with thepreparations of the pretreated samples. We also thank Mr Brian Chan for histechnical support during the protein adsorption assay. Finally, we thank ourcolleagues at Novozymes for the donations of enzymes and many fruitfuldiscussions.Authors’ contributionsVA planned and carried out the experiments, analyzed the results and wrotethe paper. JNS participated in the design of the study, helped analyzing theresults and contribute to the draft of the manuscript. All authors read andapproved the final manuscript.Competing interestsThe authors declare that they have no competing interests.Received: 9 August 2010 Accepted: 10 February 2011Published: 10 February 2011References1. Lynd LR, Laser MS, Bransby D, Dale BE, Davison B, Hamilton R, Himmel M,Keller M, McMillan JD, Sheehan J, Wyman CE: How biotech can transformbiofuels. Nature Biotechnol 2008, 26:169-172.2. Merino ST, Cherry J: Progress and challenges in enzyme development forbiomass utilization. Adv Biochem Eng/Biotechnol 2007, 108:95-120.3. Hogan CH, Mes-Hartree M, Saddler JN, Kushner D: Assessment of methodsto determine minimal cellulase concentrations for efficient hydrolysis ofcellulose. Appl Microbiol Biotechnol 1990, 32:614-620.4. Mooney CA, Mansfield SD, Tuohy MG, Saddler JN: The effect of initial porevolume and lignin content on the enzymatic hydrolysis of softwoods.Biores Technol 1998, 64:113-119.5. Gregg D, Saddler JN: Factor affecting cellulose hydrolysis and thepotential of enzyme recycle to enhance the efficiency of an integratedwood to ethanol process. Biotechnol Bioeng 1996, 51:375-383.6. Shen J, Agblevor FA: Optimization of enzyme loading and hydrolytictime in the hydrolysis of mixtures of cotton gin waste and recycledpaper sludge for the maximum profit rate. Biochem Eng J 2008,41:241-250.7. Sinitsyn AP, Gusakov AV, Vlasen EY: Effect of structural and physico-chemical features of cellulosic substrates on the efficiency of enzymatichydrolysis. App Biochem Biotechnol 1991, 30:43-59.8. Thompson DN, Chen HC, Grethlein HE: Comparison of pretreatmentmethods on the basis of available surface area. Bioresour Technol 1992,39:155-163.9. Chandra PR, Esteghlalian AR, Saddler JN: Assessing substrate accessibilityto enzymatic hydrolysis by cellulases. In Characteristics of lignocellulosicmaterials. Edited by: Hu TQ. Oxford: Blackwell; 2008:60-80.10. Arantes V, Saddler JN: Access to cellulose limits the efficiency ofenzymatic hydrolysis: the role of amorphogenesis. Biotechnol Biofuels2010, 3:4.11. Rowland SP: Cellulose: pores, internal surfaces, and the water interface.In Textile and paper chemistry and technology, ACS Symposium Series Editedby: Arthur JC 1977, 49:20-45.12. Stone JE, Scallan AM: Effect of component removal upon the porousstructure of the cell wall of wood. J Polym Sci Part C 1965, 11:13-25.13. Kerr AJ, Goring DA: I. The role of hemicellulose in the delignification ofwood. Can J Chem 1975, 53:952-959.14. Ishizawa CI, Davis MF, Schell DF, Johnson DK: Porosity and its effect onthe digestibility of dilute sulfuric acid pretreated corn stover. J AgricFood Chem 2007, 55:2575-2581.15. Chang MM, Chou TYC, Tsao GT: Structure, pretreatment and hydrolysis ofcellulose. In Advances in Biochemical Engineering. Volume 20. Edited by:Fiecher A. Bioenergy Springer-Verlag Berlin; 1981:15-42.16. Greithlein HE: The effect of pore size distribution on the rate ofenzymatic hydrolysis of cellulosic substrates. Biotechnol 1985, 2:155.17. Thygesen L, Hidayat BJ, Johansen KS, Felby C: The significance ofsupramolecular structures of cellulose for the enzymatic hydrolysis ofplant cell walls. Proceeding of the 32nd Symposium on Biotechnology forFuels and Chemicals April 19-22 2010; Clearwater Beach 2010, 43.18. Stone JE, Scallan AM, Donefer E, Ahlgren A: Digestibility as a simplefunction of a molecule of similar size to a cellulase enzyme. In Cellulasesand their applications. Volume 95. Edited by: Hajny GJ, Reese ET. AmericanChemical Society, Washington DC; 1969:219-241.19. Tanaka M, Ikesaka M, Matsuno R: Effect of pore size in substrate anddiffusion of enzyme on hydrolysis of cellulosic materials with cellulases.Biotechnol Bioeng 1988, 32:698-706.20. Esteghlalian AR, Bilodeau M, Mansfield SD, Saddler JN: Do enzymaticdigestibility and Simons’ stain reflect the differences in the availablesurface area of lignocellulosic substrates. Biotechnol Prog 2001, 17:1049-1054.21. Laivins GV, Scallan AM: The influence of drying and beating on theswelling of fines. J Pulp Paper Sci 1996, 22:J178-J184.Arantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 15 of 1622. Häggkvist M, Li TQ, Ödberg L: Effects of drying and pressing on the porestructure in the cellulose fibre wall studied by 1H and 2H NMRrelaxation. Cellulose 1998, 5:33-49.23. Bura R, Bothast RJ, Mansfield SD, Saddler JN: Optimization of SO2-catalyzed steam pretreatment of corn fiber for ethanol production. ApplBiochem Biotechnol 2003, 105-108:319-335.24. Pan X, Gilkes N, Kadla J, Pye K, Saka S, Gregg D, Ehara K, Xie D, Lam D,Saddler J: Bioconversion of hybrid poplar to ethanol and co-productsusing an organosolv fractionation process: optimization of processyields. Biotechnol Bioeng 2006, 94:851-861.25. Öhgren K, Bura R, Saddler JN, Zacchi G: Effect of hemicellulose and ligninremoval on enzymatic hydrolysis of steam pretreated corn stover. BioresTechnol 2007, 98:2503-2510.26. Del Rio L, Saddler JN: The effect of varying organosolv pretreatmentchemicals on the physicochemical properties and cellulolytic hydrolysisof mountain pine beetle - killed lodgepole. Appl Biochem Biotechnol 2009,161:1-21.27. Kumar L, Chandra R, Chung PA, Saddler J: Can the same steampretreatment conditions be used for most softwoods to achieve good,enzymatic hydrolysis and sugar yields? Bioresour Technol 2010,101:7827-7833.28. Starcher B: A ninhydrin-based assay to quantitate the total proteincontent of tissue samples. Anal Biochem 2001, 292:125-129.29. Tu MB, Chandra RP, Saddler JN: Recycling cellulases during the hydrolysisof steam exploded and ethanol pretreated lodgepole pine. BiotechnolProg 2007, 23:1130-1137.30. Bura R, Mansfield SD, Saddler JN, Bothast RJ: SO2-catalyzed steamexplosion of corn fibre for ethanol production. Appl Biochem Biotechnol2002, 98-100:59-72.31. Boussaid A, Robinson J, Cai Y, Gregg DJ, Saddler JN: Fermentability of thehemicellulose-derived sugars from steam-exploded softwood (douglas-fir). Biotechnol Bioeng 1999, 64:284-289.32. Berlin A, Maximenko V, Bura R, Kang KY, Gilkes N, Saddler J: A rapidmicroassay to evaluate enzymatic hydrolysis of lignocellulosic substrates.Biotechnol Bioeng 2005, 93:880-886.33. Robertson G, Olson J, Allen P, Chan B, Seth R: Measurement of fiberlength, coarseness, and shape with the fiber quality analyzer. Tappi J1999, 82:93-98.34. Yu X, Atalla RH: A staining technique for evaluating the pore structurevariations of microcrystalline cellulose powders. Powder Technol 1998,98:135-138.35. Chandra R, Ewanick S, Hsieh C, Saddler JN: The characterization ofpretreated lignocellulosic substrates prior to enzymatic hydrolysis, Part1: Modified Simons’ staining technique. Biotechnol Prog 2008,24:1178-1185.36. Perlack RD, Wright LL: Biomass as feedstock for a bioenergy andbioproducts industry: The technical feasibility of a billion-ton annualsupply. Oak Ridge TN: Oak Ridge National Laboratory/U.S. Department ofEnergy; 2005.37. Chandra RP, Bura R, Mabee WE, Berlin A, Pan X, Saddler JN: Substratepretreatment: the key to effective enzymatic hydrolysis oflignocellulosics? Adv Biochem Engin/Biotechnol 2007, 108:67-93.38. Duff SJB, Murray WD: Bioconversion of forest products industry wastecellulosics to fuel ethanol: a review. Biores Technol 1996, 55:1-33.39. Nieves RA, Ehrman CI, Adney WS, Elander RT, Himmel ME: Technicalcommunication: survey and analysis of commercial cellulasepreparations suitable for biomass conversion to ethanol. World JMicrobiol Biotechnol 1998, 14:301-304.40. Berlin A, Maximenko V, Gilkes N, Saddler J: Optimization of enzymecomplexes for lignocellulose hydrolysis. Biotechnol Bioeng 2007,97:287-296.41. Kim E, Irwin DC, Walker LP, Wilson DB: Factorial optimization of a six-cellulase mixture. Biotechnol Bioeng 1998, 58:494-501.42. Wingren A, Galbe M, Zacchi G: Techno-economic evaluation of producingethanol from softwood: comparison of SSF and SHF and identification ofbottlenecks. Biotehnol Prog 2003, 19:1109-1117.43. Roche CM, Dibble CJ, Stickel JJ: Laboratory-scale method for enzymaticsaccharification of lignocellulosic biomass at high-solids loadings.Biotechnol Biofuels 2009, 2:28.44. Kumar R, Wyman CE: Effect of enzyme supplementation at moderatecellulase loadings on initial glucose and xylose release from corn stoversolids pretreated by leading technologies. Biotechnol Bioeng 2009,102:457-467.45. Mabee WE, Gregg DJ, Arato C, Berlin A, Bura R, Gilkes N, Mirochnik O,Pan X, Pyg EK, Saddler JN: Updates on softwood-to-ethanol processdevelopment. Appl Biochem Biotechnol 2006, 55:129-132.46. Mansfield SD, Mooney C, Saddler JN: Substrate and enzyme characteristicsthat limit cellulose hydrolysis. Biotechnol Prog 1999, 15:804-816.47. Jeoh T, Ishizawa CI, David MF, Himmel ME, Adney WS, Johnson DK:Cellulase digestibility of pretreated biomass is limited by celluloseaccessibility. Biotechnol Bioeng 2007, 98:112-122.48. Berlin A, Gilkes N, Kurabi A, Bura R, Tu M, Kilburn D, Saddler J: Weak lignin-binding enzymes: a novel approach to improve activity of cellulases forhydrolysis of lignocellulosics. Appl Biochem Biotechnol 2005, 121-124:163-70.49. Tu M, Pan X, Saddler JN: Adsorption of cellulase on cellulolytic enzymelignin from lodgepole pine. J Agric Food Chem 2009, 57:7771-7778.50. Shevchenko SM, Beatson RP, Saddler JN: The nature of lignin from steamexplosion/enzymatic hydrolysis of softwood: structural features andpossible uses. App Biochem Biotechnol 1999.51. Zhang P, Lynd LR: Toward an aggregated understanding of enzymatichydrolysis of cellulose: Non-complexed cellulase systems. BiotechnolBioeng 2004, 88:797-824.52. Ramos LP, Breuil C, Saddler JN: Effect of enzymatic hydrolysis on themorphology and fine structure of pretreated cellulosic residues. EnzymeMicrob Technol 1993, 15:821-831.53. Ragnar E, Alderborn G, Nyström C: Particle analysis of microcrystallinecellulose: differentiation between individual particles and theiragglomerates. Int J Pharm 1994, 111:43-50.54. Yu X, Minor JL, Atalla RH: Mechanism of action of Simons’ stain. Tappi J1995, 78:175-180.55. Park S, Venditti RA, Abrecht DG, Jameel H, Pawlak JJ, Lee JM: Surface andpore structure modification of cellulose fibers through cellulasetreatment. J App Polymer Sci 2007, 103:3833-3839.doi:10.1186/1754-6834-4-3Cite this article as: Arantes and Saddler: Cellulose accessibility limits theeffectiveness of minimum cellulase loading on the efficient hydrolysisof pretreated lignocellulosic substrates. Biotechnology for Biofuels 20114:3.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitArantes and Saddler Biotechnology for Biofuels 2011, 4:3http://www.biotechnologyforbiofuels.com/content/4/1/3Page 16 of 16


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items