Available online at www.sciencedirect.com

Bioresource Technology 99 (2008) 5216–5225

Enzyme characterization for hydrolysis of AFEX and liquidhot-water pretreated distillers’ grains and their conversion to ethanol

Bruce S. Dien a,, Eduardo A. Ximenes b, Patricia J. O’Bryan a, Mohammed Moniruzzaman c,Xin-Liang Li a, Venkatesh Balan d, Bruce Dale d, Michael A. Cotta a

a National Center for Agricultural Utilization Research, USDA, Agricultural Research Service1, 1815 North University Street, Peoria, IL 61604,

United Statesb Microbiology Department, 204 Biological Sciences, University of Georgia, Athens, GA 30602-2605, USA

c Bioenergy International, LLC, 99 Longwater Circle, Norwell, MN 02061, United Statesd Biomass Conversion Research Laboratory, Department of Chemical Engineering and Material Science, Michigan State University, E. Lansing, MI 48823,

United States

Available online 8 November 2007

Abstract

Dried distillers’ grains with solubles (DDGS), a co-product of corn ethanol production, was investigated as a feedstock for additionalethanol production. DDGS was pretreated with liquid hot-water (LHW) and ammonia fiber explosion (AFEX) processes. Cellulose wasreadily converted to glucose from both LHW and AFEX treated DDGS using a mixture of commercial cellulase and b-glucosidase; how-ever, these enzymes were ineffective at saccharifying the xylan present in the pretreated DDGS. Several commercial enzyme preparationswere evaluated in combination with cellulase to saccharify pretreated DDGS xylan and it was found that adding commercial grade (e.g.impure) pectinase and feruloyl esterase (FAE) preparations were effective at releasing arabinose and xylose. The response of sugar yieldsfor pretreated AFEX and LHW DDGS (6 wt%/solids) were determined for different enzyme loadings of FAE and pectinase and modeledas a response surfaces. Arabinose and xylose yields rose with increasing FAE and pectinase enzyme dosages for both pretreated mate-rials. When hydrolyzed at 20 wt%/solids with the same blend of commercial enzymes, the yields were 278 and 261 g sugars (i.e. total ofarabinose, xylose, and glucose) per kg of DDGS (dry basis, db) for AFEX and LHW pretreated DDGS, respectively. The pretreatedDDGS’s were also evaluated for fermentation using Saccharomyces cerevisiae at 15 wt%/solids. Pretreated DDGS were readily fer-mented and were converted to ethanol at 89–90% efficiency based upon total glucans; S. cerevisiae does not ferment arabinose or xylose.Published by Elsevier Ltd.

Keywords: Distillers’ grains with solubles; Maize; Bioethanol; Arabinoxylan

1. Introduction

Dried distillers’ grains with solubles (DDGS) is themajor, and often only, co-product of ethanol manufacturedfrom whole ground corn, which is used to produce 75% of

0960-8524/$ - see front matter Published by Elsevier Ltd.

doi:10.1016/j.biortech.2007.09.030

Corresponding author.E-mail address: Bruce.Dien@ars.usda.gov (B.S. Dien).

1 Names are necessary to report factually on available data; however,the USDA neither guarantees nor warrants the standard of the product,and the use of the name by USDA implies no approval of the product tothe exclusion of others that may also be suitable.

the 4 billion gallons made each year in the US (RFA,2005). Annually, 9.0 million metric tons of DDGS are pro-duced; a bushel of corn (55 lb) yields 2.8 gal of ethanol and17 lb of DDGS (Dien et al., 2002). As ethanol productioncontinues to grow, e.g. forecasted to be 7.5 billion US gal-lons/year by 2010, the supply of DDGS is expected toreach 12–14 million metric tons per year (RFA, 2005).While demand for fuel ethanol appears limitless, 140 billiongallons of gasoline is used in the US each year; the marketfor DDGS has not been as resilient to expanded produc-tion. DDGS is sold as a feed, primarily for ruminant live-stock (e.g. cattle), and there are concerns that its selling

B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225 5217

price could erode further in response to increasing supplies.The high fiber content of DDGS, limits it use in non-rumi-nant (e.g. swine and poultry) diets.

In this special journal issue, DDGS, specifically the fiberfraction, is investigated as an alternate feed source for pro-ducing ethanol and butanol. Converting DDGS fiber intoethanol would directly lower the amount of DDGS onthe market and increase the overall ethanol yield from abushel of corn by up to 12%, which is based upon 31%wt carbohydrates, 90% dryness, and 85% ethanol conver-sion efficiency. Furthermore, the resulting modified low-fiber DDGS that is produced should have greater utilitybecause it could be more suitable for inclusion in dietsfor monogastrics livestock (i.e. swine and poultry) and inaquaculture.

DDGS can be converted to ethanol or butanol by pre-treating the biomass, saccharifying the carbohydrates,and fermenting the sugars followed by product recovery(see Kim et al., 2008a,b and Ezeji and Blaschek, 2008). Pre-treatment of DDGS with ammonia fiber explosion (AFEX)and liquid hot-water (LHW) was described in another arti-cle in this series (Kim et al., 2008b). While these pretreat-ments are chemically and physically dissimilar, bothremove the xylan portion of the cell wall and open up thehighly crystalline cellulose structure. Subsequently, fer-mentable monosaccharides are produced by saccharifyingthe carbohydrates with enzymes, including cellulase andhemicellulases. The subject of this paper is the productionof monosaccharides from pretreated biomass using com-mercial enzyme preparations.

The xylan from DDGS originates from the pericarp andgerm fractions of the corn kernel. The pericarp covers thekernel and has been intensely studied, in part, because ofits importance in protecting the starchy endosperm fromfungal infection. This xylan is notable for it chemical com-plexity – over 80% of the beta-1-4 xylose units forming itsbackbone are substituted with a wide variety of side-groups, and (appropriately) its resistance to fungal hydro-lytic enzymes (Grabber et al., 1995; Grabber et al., 1998a,b; Saha, 2003; Saulnier et al., 1995; Saulnier and Thibault,1999). These two traits are related because the side-groupsprotect the fibers from xylanases (ibid). AFEX and LHWtreatment break apart the xylan strands, allowing the lowerdegree of polymerization xylan to dissolve, but these pre-treatments only partially remove the side-groups. In thispaper, mixtures of commercial enzymes were tested at dif-ferent loadings and in different combinations for complet-ing the hydrolysis begun by the pretreatments, and toenhance hydrolysis of the hemicelulose (xylan) fractionthat current commercial enzyme preparations do not tar-get. Experiments were first carried out at lower biomassconcentrations, thereby avoiding mass transfer and endproduct inhibition effects, and later at higher biomass con-centrations, such as those that are more realistic for fer-mentation. Finally, the sugars produced were evaluatedfor their fermentability to ethanol using Saccharomyces

cerevisiae.

2. Methods

2.1. Materials

DDGS was collected from Big River Resources, LLC(W. Burlington, IA) and stored at room temperature; a sin-gle lot was used for all experiments. DDGS was finelyground using a coffee mill prior to use in experiments toensure a representative sample. DDGS was treated byeither AFEX or liquid hot-water (LHW) as described inthe previous two papers in this special edition. Enzymeswere provided by Genencor International, Inc. (Rochester,NY), with the exceptions of Novo188 (Novozyme A/S,purchased from Sigma Chemical Co., St. Louis, MO) andDepol 740L (Biocatalysts Inc., Wales, UK). Reagents usedfor enzyme activity assays were purchased from SigmaChemicals and all other chemicals from Fisher ScientificInternational, Inc. (Hampton, NH). All reagents used forthis study were of analytical quality.

2.2. Measuring enzyme activities

Enzyme activities in the presence of dissolved oat speltxylan, Whatman #1 filter paper, or 2.5 mM q-nitrophenylconjugated substrates were determined at 50 �C and pH4.8 using published methods (Adney and Baker, 1996;Hespell, 1992; Hespell and Whitehead, 1990). The releaseof reducing sugars in cellulase and xylanase assays wasdetermined as described in (Miller, 1959). One unit of cel-lulase and xylanase activities was defined as the release ofone lmol of either glucose or xylose equivalents per min-ute. For q-nitrophenyl conjugated substrates, one unit ofactivity was defined as one lmol of q-nitrophenyl (q-NP)released per minute. Feruloyl esterase (FAE) activity wasassayed measuring the conversion of methyl ferulate (pre-pared as 100 mM in 50% DMSO v/v, and added to a finalconcentration of 2 mM) to ferulate. Ferulic acid andmethyl ferulate were measured by reverse phase HPLC aspreviously described (Dien et al., 2006).

2.3. Low-solids digestion assay

Pretreated hydrolysates were diluted to 6% w/w solidsby adding dH2O, sodium citrate buffer (pH 4.8, final con-centration 50 mM), and thymol (500 mg/l). Suspendedsolutions were distributed (4 ml each) to scintillation vialsand enzymes added at loadings indicated in the text. Thevials were rolled (The Mini Tube Roller Bellco Biotech-nology, Inc, Vineland, NJ) for 72 h at 50 �C. Monosac-charide concentrations were measured using HPLC andconverted to % yields by dividing by the maximum sugarconcentrations attainable based upon the followingDDGS composition: 20.7% w/w glucan, 10.2% xylan +galactan, and 5.6% arabinan, after correcting for massgained from hydrolysis. Xylose and galactose were ana-lyzed together because they co-elute on the HPLC columnused for sugar analysis.

5218 B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225

2.4. High-solids digestion assay

Except as noted, high-solid digestions were conducted in asimilar manner as low-solids digestions. AFEX pretreatedDDGS was prepared and treated exactly as noted above,except diluted to 20% w/w solids. For LHW, DDGS was pre-treated at 20% w/w solids and (following cooling) sodium cit-rate buffer, thymol, and enzymes added directly to thepretreatment reactor, a rubber seal inserted into the cap,and the DDGS treated under the same conditions describedabove; reactors were rolled using the same apparatus usedfor the glass vials. A more complete description of the pre-treatments are given by Kim et al., 2008b.

2.5. Saccharomyces cerevisiae fermentations

AFEX treated DDGS solutions (15% w/w) were placed inPyrex Bottles (50 ml), adjusted to pH 4.8 with 1 M HCl andsodium citrate to 50 mM (pH 4.8), capped, and autoclavedfor 15 min. Upon cooling, filter sterilized solutions ofGC220 cellulase (15 FPU/g cellulose) and beta-glucosidaseNovo188 (40 U/g cellulose) were added to the bottles.LHW DDGS was prepared as follows: 3 large reactors wereloaded with 15% w/w DDGS and heated at 160 �C, for20 min. The contents were combined. While stirring, 30 mlaliquots were removed to 50 ml Pyrex bottles. Sodium citratewas added to 50 mM, pH 4.8. Bottles were capped and auto-claved for 15 min. Enzymes were added as noted above forAFEX treated DDGS. Both sets of bottles were allowed tosaccharify for 18 h at 50 �C in a shaker/incubator set to100 rpm. To the bottles containing LHW DDGS, ammo-nium sulfate was added to 0.05% w/v. After cooling, the bot-tles were inoculated with S. cerevisiae to an O.D. @600 nm of1.0 and the solid caps replaced with caps containing siliconesepta, though which 22 g needles had been pierced to exhaustCO2 released by fermentation. The temperature of the sha-ker/incubator was reset to 32 �C and the bottles returned.Bottles were sampled periodically for the next 72 h and sam-ples were stored at �20 �C.

The S. cerevisiae inoculum was prepared by growingstrain D5A on solid YPD, containing per L: 10 g yeastextract, 20 g Protease Peptone, and 20 g/l dextrose supple-mented with 15 g Bacto Agar. The solid culture was incu-bated at 32 �C for 2 days. A single colony was transferred to a50 ml Erlenmeyer flask containing 5 ml of YPD and grown at32 �C with agitation (200 rpm) for 18 h. This culture was usedto inoculate the seed culture, which consisted of an Erlen-meyer flask (250 ml) containing 25 ml of YPD, but with50 g/l dextrose and grown under similar conditions also for 18 h.For inoculation, the cells were concentrated to an O.D.@600 nm of 50, using a phosphorus saline buffered solution.

2.6. Analytical methods

Moisture contents of DDGS and pretreated materialswere determined from weight loss after drying at 105 �Cfor 24 h.

Total glucan (cellulose and starch) was measured using atwo-stage H2SO4 protocol (Ruiz and Ehrman, 1996).Xylan sugars (arabinose and xylose and galactose) weremeasured by digesting with 2N trifluoroacetic acid for 1 hat 100 �C (Dien et al., 1999). The later method was favoredfor analysis of xylan sugars because the final solution wasmore concentrated for HPLC analysis.

Monosaccharide and ethanol concentrations were mea-sured using a using a SpectraSYSTEM liquid chromatog-raphy system (Thermo Electron Corporation, CA)equipped with an automatic sampler, column heater, iso-cratic pump, refractive index detector, and computer basedintegrator running Chromquest ver. 2.5 (Thermo ElectronCorporation, CA). Samples were injected (20 ll) onto anorganic acid column (Aminex HPX-87H Column,300 · 7.8 mm, Bio Rad Laboratories, Inc., Hercules, CA)and eluted with 5 mM H2SO4 at 0.6 ml/min and 65 �C.

2.7. Statistics

Response surfaces were generated using data from cen-tral composite experimental design where the center pointwas replicated 5 times and all other points twice. Datawas analyzed using Design-Expert (ver. 6.0.11, Stat-Ease,Minneapolis, MN).

3. Results

3.1. Spectra of activities for commercial enzyme

preparations

Enzyme preparations supplied by Genencor Interna-tional and b-glucosidase (Novo188, Novozyme A/S) wereevaluated for their range of hydrolytic enzyme activities(Table 1). While the most significant activity for each prep-aration corresponded to its marketed activity, e.g. GC220cellulase had the highest cellulase activity of the testedblends; all of the supplied preparations also containednumerous additional activities. In particular, they all con-tained significant xylanase and carbohydrate debranchingactivities. The most balanced enzyme profile, in terms ofdebranching activities, was the Multifect Pectinase PE.Also of note, only Novo188 and Multifect Pectinase PEcontained measurable feruloyl esterase. The commercialferuloyl esterase (Depol 740L) contained 15.6 U/ml ofFAE and 13.8 U/ml of para-coumaroyl esterase activities(pH 4.8), as well as numerous other activities (data notpresented).

3.2. Optimizing xylan digestion at low-biomass

concentrations

Earlier it was noted that treating LHW or AFEXDDGS with GC220 cellulase (15 FPU/g cellulose) blendedwith Novo188 b-glucosidase (40 U/g cellulose) released86% and 93% of the available glucose, respectively.However, this combination of enzymes was ineffective at

Table 1Enzyme activities detected in commercial enzyme preparations

Activity GC220 Spezyme CP Novo188 Multifect xylanase Multifect pectinase PE

Activities units/ml

Cellulase (FPUa) 92.8 58.2 8.5 0.77 4.18b-Glucosidase 99.7 128 665 35.9 345.8Xylanase (OSXb) 2782 2622 123 25203 1664a-Arbinofuransoidase 3.06 22.6 29.3 9.44 1862b-Xylosidase 23.3 7.3 16.6 22.6 186.2a-Galactosidase 3.9 0.39 116 2.39 31.9Feruloyl esterase 0.0 0.0 0.6 0.0 9.67p-Coumaroyl esterase nmc nm nm 1.3 21.7

a Filter paper units.b Oat spelt xylan.c Not measured.

B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225 5219

digesting xylan and only 18% and 24% of the availablexylose were released (Table 2). To develop a more effectiveblend of enzymes, a test tube scale assay was developedusing pretreated ground DDGS diluted in a solution ofsodium citrate buffer (pH 4.8, 50 mM) and thymol, whichwas added to prevent microbial contamination; AFEX pre-treated material required additional pH adjustment withHCl to bring it to the target pH.

Possible deficiency in xylan hydrolyzing activity presentin the cellulase preparation was alleviated by adding addi-tional xylanase activity. The following xylanases were eval-uated: Multifect 720, GC260, Multifect XL, MultifectXylanase, Multifect A40, Optimash BG, Optimash XL.The enzymes were evaluated on ground LHW DDGS.Each was blended at one-quarter the cellulase loading rateon a weight basis with the previously described combina-tion of cellulase and b-glucosidase. While none dramati-cally increased the liberation of arabinose or xylose,Multifect Xylanase appeared to be the most effective (datanot shown). Supplementing the cellulase mixture with fur-ther additions of Multifect Xylanase led to no additionalimprovements in any of the sugar yields (Figs. 1a and b)

Table 2Summary of enzyme digestion results for HW- and AFEX-treated DDGS at

Multifect xylanase Multifect pectinase FE DEPOL 740L Gluco

Yield

Liquid hot-water pretreated DDGS

– – – 197+ – – 193– + – 204+ + – 196– + + 209

AFEX pretreated DDGS

– – – 214+ – – 203– + – 236+ + – 224– + + 232

a Maximum enzyme loadings are listed in Figs. 1a, 1b, 2a, 2b, 3a, 3b or in thebased upon the following DDGS composition (mg of anhydrous sugar/g, db)

and, especially, xylose yields were only improved a coupleof percent compared to digestions in which it was notincluded (Table 2).

Next, it was speculated that adding pectinase mightincrease saccharification of xylan because these typesof preparations also contained hemicellulase relatedactivities. Even though DDGS does not contain pectin,commercial pectinase preparations contain multiple side-activities that may aide the cellulase in hydrolyzingDDGS. As such, the pectinase preparation was usedsolely for its hemicellulose relevant activities. Therefore,the pectinase mixture was added based upon its xylanaseactivity as opposed to pectinase activity. Adding MultifectPectinase PE significantly improve the yields of bothxylose and arabinose. The amount of Multifect PectinasePE selected for further work was 50 U of xylanase per gDDGS (Figs. 2a and b) because this amount appearedto give good sugar yields while minimizing the enzymeloading. At this loading (50 U/g), xylose yields wereimproved 172% for LHW DDGS and 248% for AFEXDDGS compared to digestions without added MultifectPectinase PE (Table 2).

maximum loadingsa

se Xylose Arabinose Glucose Xylose Arabinose

s (mg per g DDGS, db) Efficiencies (% sugar released)

33 23 86 29 3744 25 84 39 4081 41 89 72 6576 39 85 67 6393 44 91 82 70

16 12 93 14 2025 17 88 22 2875 49 103 66 7970 47 97 61 7592 61 101 81 98

text. Symbols: ‘+’ = added and ‘–’ = not added to reaction. Efficiencies are: Arabinose (56), Glucose (207), and Xylose (102).

Added Xylanase (xylanase IU/g DDGS, db)

0 50 100 150 200

Mon

osac

char

ide

Rel

ease

d (m

g / g

DD

GS

, db)

0

20

40

60

80

100

120

140

160

180

200

220

GlucoseXylose Arabinose

Fig. 1a. Digestion of LHW DDGS with GC220 cellulase (15 FPU/gcellulose) + Novo188 b-glucosidase (40 U/g cellulose) supplemented withvarious loadings of multifect xylanase.

Added Xylanase (xylanase IU/g DDGS, db)

0 50 100 150 200

Mon

osac

char

ide

Rel

ease

d (m

g / g

DD

GS

, db)

0

50

100

150

200

250

GlucoseXylose Arabinose

Fig. 1b. Digestion of AFEX DDGS with GC220 cellulase (15 FPU/gcellulose) + Novo188 b-glucosidase (40 U/g cellulose) supplemented withvarious loadings of multifect xylanase.

Pectinase LHW

Added Pectinase (xylanase IU/g DDGS, db)

0 50 100 150 200

Mon

osac

char

ide

Rel

ease

d (m

g / g

DD

GS

, db)

0

50

100

150

200

250

GlucoseXylose Arabinose

Fig. 2a. Digestion of LHW DDGS with GC220 cellulase (15 FPU/gcellulose) + Novo188 b-glucosidase (40 U/g cellulose) supplemented withvarious loadings of Multifect Pectinase PE. Pectinase preparation wasspecified in units of xylanase activity because DDGS contains arabinoxy-lan and not pectin.

AFEX-DDGS

Added Pectinase (xylanase IU/g DDGS, db)

0 50 100 150 200

Mon

osac

char

ide

Rel

ease

d (m

g / g

DD

GS

, db)

0

50

100

150

200

250

GlucoseXylose Arabinose

Fig. 2b. Digestion of AFEX DDGS with GC220 cellulase (15 FPU/gcellulose) + Novo188 b-glucosidase (40 U/g cellulose) supplemented withvarious loadings of multifect pectinase FE.

5220 B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225

Pectinase and xylanase were blended on an equal xylan-ase basis to test if a mixture of these two preparationsmight further improve sugar yields. The mixture wasblended with cellulase (supplemented with Novo188) overa range of 0–250 U xylanase per g DDGS (db) (data notshown), but no further improvements were observed com-pared to just adding pectinase alone (Table 2).

A final attempt was made to improve the yields of arab-inose by supplementing the tri-enzyme blend (cellulase, b-glucosidase, and pectinase) with a commercial feruloylesterase (Depol 740L). FAE was added at a 0–5 U FAE/g DDGS (db) (Figs. 3a and b). Adding in FAE improvedsugar yields for both AFEX and LHW DDGS and espe-cially the former. Yields were found to plateau off afterapproximately 1 U FAE/g DDGS for LHW and 2 UFAE/g DDGS for AFEX. The overall xylose yields were81–82% for LHW and AFEX DDGS. Arabinose yields

were only 70% for LHW, but as high as 98% for AFEXtreated DDGS (Table 2).

3.3. Response surfaces

In as far as adding pectinase and FAE togetherincreased xylan sugar yields from pretreated DDGS, itwas of interest to determine their interaction and the min-imum amount of each needed to produce viable sugaryields. Response surfaces were constructed for pectinasevs. FAE for both AFEX and LHW DDGS using a centralcomposite face-centered experimental design. Pectinase

Added FAE (FAE IU/g DDGS, db)

0 1 2 3 4 5

Mon

osac

char

ide

Rel

ease

d (m

g / g

DD

GS

, db)

0

50

100

150

200

250

Glucose XyloseArabinose

Fig. 3a. Digestion of LHW DDGS with GC220 cellulase (15 FPU/gcellulose) + Novo188 b-glucosidase (40 U/g cellulose) + Multifect Pectin-ase (50 U xylanase/g DDGS db) supplemented with various loadings ofDepol 740L FAE. Sugar yields are shown by solids lines and maximumsugar yields are shown by dashed lines.

0 1 2 3 4 5

Mon

osac

char

ide

Rel

ease

d (m

g / g

DD

GS

, db)

0

50

100

150

200

250

Glucose Xylose Arabinose

Added FAE (FAE IU/g DDGS, db)

Fig. 3b. Digestion of AFEX DDGS with GC220 cellulase (15 FPU/gcellulose) + Novo188 b-glucosidase (40 U/g cellulose) + Multifect Pectin-ase (50 U xylanase/g DDGS db) supplemented with various loadings ofDepol 740 FAE. Sugar yields are shown by solids lines and maximumsugar yields are shown by dashed lines.

AFEXXylose Released (g/kg DDGS db)

Pectinase (U xylanase / g DDGS db)

FA

E (

U/g

DD

GS

db)

12.5 21.9 31.3 40.6 50.0

0.50

0.88

1.25

1.63

2.00

69.25

72.3275.40

78.48

81.56

Fig. 4a. Contour plot showing effect of varying feruloyl esterase andpectinase loadings on xylose yield for AFEX treated DDGS. The data wasfitted with the Xylose Yield = 60.0 + 0.321 * Pec. + 4.38 * FAE. Eachfactor was significance at the 99% level and the model fitted the data with ar2 = 0.91.

AFEX Design

Arabinose Released (g/kg DDGS db)

Pectinase (U xylanase/ g DDGS db)

FA

E (

U/g

DD

GS

db)

12.5 21.9 31.3 40.6 50.0

0.50

0.88

1.25

1.63

2.00

50.5

51.853.1

54.4

55.7

Fig. 4b. Contour plot showing effect of varying feruloyl esterase andpectinase loadings on arabinose yield for AFEX treated DDGS. The datawas fitted with the Arabinose Yield = 46.54 + 0.134 * Pec. +1.90 * FAE.Each factor was significance at the 99% level and the model fitted the datawith an r2 = 0.80.

B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225 5221

was varied from 12.5 to 50 xylanase U/g DDGS, db andDepol 740L from 0.5 to 2.0 FAE U/g DDGS, db.Response surfaces showing arabinose and xylose yieldsfor both pretreatments are displayed in Figs. 4a–d; fittedequations and r2 values are included in the figure captions.

The AFEX DDGS data were fitted by a linear modeland no significant interaction was detected between theactions of Depol 740L and Multifect Pectinase PE withinthe sampling space. Furthermore, arabinose and xyloseyields climbed with increasing amounts of enzymesthroughout the sampling space. At the center point, yieldswere arabinose (85%), glucose (95%), and xylose (67%).The yields were lower than those reported in the previous

experiment because of lower enzyme loadings used for thistreatment level.

For the LHW DDGS, arabinose and xylose yields werefitted best with a quadratic model. All model terms weresignificant at the 99% level, except for the interaction ofFAE and pectinase, which was not significant for eitherAFEX or LHW DDGS. At the center point, the yields

LHW Pretreatment

Arabinose Released (g/kg DDGS db)

Pectinase (U/g DDGS db)

FA

E (

U/g

DD

GS

db)

12.5 21.9 31.3 40.6 50.0

0.50

0.88

1.25

1.63

2.00

45.7 46.4 47.2 47.9 48.7

Fig. 4d. Contour plot showing effect of varying feruloyl esterase andpectinase loadings on arabinose yield for LHW treated DDGS. The datawas fitted with Arabinose Yield = 42.1 + 0.235 * Pectinase + 0.427 * FAE– 2.10E-003 * Pectinase2. Each factor was significance at the 99% level andthe model fitted the data with an r2 = 0.96.

Table 3Higher solids enzymatic saccharification

AFEX DDGS LHW DDGS

Final sugar concentrations (g/l)Glucose 42.2 36.2Xylose 15.1 14.9Arabinose 10.6 7.1Total 68.0 58.2

Yields (g/kg DDGS, db)Glucose 173 160Xylose 62 66Arabinose 43 33Total 278 261

LHW Pretreatment

Xylose Released (g/kg DDGS, db)

Pectinase (U/g DDGS db)

FA

E (

U/g

DD

GS

db)

12.5 21.9 31.3 40.6 50.0

0.50

0.88

1.25

1.63

2.00

87.4

89.9 92.5 95.0

97.6

Fig. 4c. Contour plot showing effect of varying feruloyl esterase andpectinase loadings on xylose yield for LHW treated DDGS. The datawas fitted with Xylose Yield = 75.5 + 0.731 * Pectinase + 2.45 * FAE –6.74E-003 * Pectinase2. Each factor was significance at the 99% level andthe model fitted the data with an r2 = 0.97.

5222 B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225

for LHW DDGS were arabinose (77%), glucose (99%), andxylose (84%). That the factors were fitted with differentmodels for LHW and AFEX DDGS is not surprisinglybecause both pretreatments depend upon different chemis-tries for their effects. Other evidence of the differences inthese two pretreated materials is the observation that the

% of xylose released was greater than for arabinose forLHW DDGS and the converse for AFEX DDGS.

It was important at this point to examine the possibleeffect of grinding the samples prior to saccharification onthe release of sugars. It is possible that grinding might leadto increased yields by removing steric hindrances or afford-ing greater surface area. To test this, ground and ungroundsamples of DDGS were treated with LHW and each treat-ment hydrolyzed with enzymes in a side-by-side compari-son. Likewise, AFEX pretreated samples were evaluatedin a similar manner, except that the test samples wereground after pretreatment. Each comparison was run intriplicate. The results showed that grinding had no signifi-cant effect (compared to sample to sample variation) onyields. For the LHW, the yields [mg/g] were (ungroundvs. ground): glucose (207 ± 4 vs. 199 ± 2), xylose (86 ± 1vs. 84 ± 2), and arabinose (43 ± 1 vs. 40 ± 3). For AFEXtreated DDGS, the yields [mg/g] were (unground vs.ground): glucose (200 ± 7 vs. 196 ± 1), xylose (78 ± 3 vs.75 ± 0), and arabinose (52 ± 2 vs. 50 ± 0). Interestingly,the unground samples all gave higher yields than theground samples, however, differences between treatmentswere always within one standard deviation and thereforenot significant.

3.4. Digestion studies at higher biomass concentrations

AFEX and LHW pretreated DDGS were treated at 20%w/w solids to test if higher sugar concentrations could beobtained by hydrolysis. Enzyme loadings were chosenbased upon the response surfaces in order to give highsugar yields with the least amount of added enzymes. Thechosen enzyme loadings were: cellulase (15 FPU/g cellu-lose), b-glucosidase (40 U/g cellulose), pectinase (40 Uxylanase/g DDGS (db), and FAE (1.2 U/g DDGS (db)).Each digestion was carried out in triplicate for 72 h (Table3). The higher solids concentrations were successful in sub-stantially increasing the final sugar concentrations to 58 g/lfor LHW DDGS and 68 g/l for AFEX DDGS. However,the sugar recovery efficiencies were significantly lower thanobserved at dilute biomass concentrations for both pre-treated DDGS. Efficiencies for LHW DDGS were: arabi-nose (59%), glucose (70%), and xylose (54%) and for

B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225 5223

AFEX DDGS: arabinose (70%), glucose (75%), and xylose(55%). Possible reasons for the lower yield at higher bio-mass concentrations might be end product inhibition ormass transfer limitations.

3.5. Fermentation of pretreated DDGS with S. cerevisiae

The fermentabilities of the pretreated materials wereevaluated using S. cerevisiae. AFEX and LHW DDGS at15% solids loading were mixed with cellulase and b-gluco-sidase and a simultaneous saccharification and fermenta-tion (SSF) carried out for 72 h. Pectinase and FAE werenot included because S. cerevisiae does not ferment arabi-nose or xylose. As a control, SSF of untreated DDGSwas also included in the experiment. Ethanol yields were88% and 90% of theoretical for AFEX and LHW DDGS,indicating the materials were readily fermented. Theuntreated DDGS ethanol yield was 69% of theoretical, pre-sumably with much of the fermented glucose coming fromthe residual starch. It should be noted that samples wereground for this experiment, in keeping with the establishedprotocol, and this might explain the relatively high yield.An extra fermentation flask was included for each DDGSmaterial to allow for repeated sampling; the flasks usedfor yields were only sampled at the end of the experimentin order to ensure a good material balance. These flasksindicated the fermentations were completed within 27 to33 h, which is further evidence of the fermentability ofthe pretreated DDGS.

4. Discussion

Earlier studies have reported on converting DDGS toethanol by pretreating the material with dilute sulfuric acid(Dien et al., 2004; Tucker et al., 2004). In this study, twoalternative pretreatments were used: AFEX and LHW(Mosier et al., 2005, and Kim et al., 2008b). These pretreat-ments have the advantage of being more benign and (wesuspect) easier to integrate into existing dry grind corn eth-anol processes because of the absence of H2SO4. Yet, xylanrepresents 40% of the carbohydrates present in DDGS andthese pretreatments do not convert xylan into monosaccha-rides, the form needed for ethanol fermentation. Therefore,hemicellulases are required for completing hydrolysis of thexylan and converting it into fermentable arabinose, galact-ose, and xylose monomers. An alternate approach that uti-lizes a fixed bed for ion exchange catalyst to carry outhydrolysis is reported by Bootsma et al. (2007), later in thisvolume. However, comparison of enzyme hydrolysis to afixed acid catalyst shows that enzyme hydrolysis gives rea-sonable results, and is preferred, given the current state ofcatalyst development.

Treating AFEX or LHW DDGS with a commercial cel-lulase preparation supplemented with extra b-glucosidase(Novo188) was sufficient to release nearly all of the avail-able glucose. Some of the glucan present in DDGS is inthe form of starch, which requires amylase to saccharify

to glucose. Novo188 was the source of amylase in theseexperiments (data not shown). Both of these enzyme prep-arations were found to have significant hemicellulase activ-ities and it was hoped at first that they would be equallyeffective at digesting the xylan fraction, although they werenot developed for this type of application. Hence, it wasnot surprising that they were largely ineffective at releasingarabinose and xylose from either AFEX or LHW DDGS.Yields were limited to 22–39% of maximum for xyloseand 28–40% for arabinose.

There are few studies that have examined treatingDDGS with enzymes to produce monosaccharides (Xim-enes et al., 2007). A number of prior studies have, however,examined the chemical structure and enzymatic treatmentof corn pericarp arabinoxylan, the type that predominatesin DDGS. Pericarp xylan consists of xylose (45%), arabi-nose (30%), galactose (7.4%), acetate (4.2%), glucuronicacid (4.8%), para-courmaric acid (0.5%), ferulic acid(4.2%), and diferulic acid (0.9%) (Saulnier et al., 1995;Saulnier and Thibault, 1999). As expected from the heter-ogeneous composition, over 80% of the xylan backboneis substituted at the O-2 and/or O-3 position with mono-saccharides and short oligomers of the above mentionedsugars and organic acids. The phenolic acids are boundto the arabinose, with the diferulates and triferulates (Bun-zel et al., 2005) forming cross bridges between separatexylan strands. This chemically complex structure and feru-lic cross bridges may be responsible for much of the xylan’sresistance to enzymes because the side-groups protect thexylan backbone from the action of endo-xylanase (Grabberet al., 1998a; Saulnier et al., 2001).

A large number of commercial enzyme preparations,including labeled xylanases, hemicellulases, a pectinase,and even feruloyl esterase, were blended with cellulase inan attempt to release arabinose and xylose. No benefitwas derived from adding any of these commercial xylanaseor hemicellulase preparations for either LHW or AFEXDDGS. In contrast, blending cellulase with pectinase andFAE did increase the amount of arabinose and xylosereleased from the xylan. Therefore, the final selected blendconsisted of 4 enzyme preparations: cellulase, b-glucosi-dase, pectinase, and feruloyl esterase. For LHW andAFEX DDGS, the arabinose yields were 70%, 98% andthe xylose yields 82%, 81%, respectively (Table 2).

In as the pectinase (Table 1) and feruloyl esterase (datanot shown) each have a wide-spectrum of activities, it is notapparent which were directly associated with the increasein yields. However, there is ample evidence in the literaturethat combining T. reesei cellulase preparations with eitherA. niger or H. insolens enzyme preparations substantiallyimproves the sugar yields from grain (e.g. barley, wheat,as well as maize) associated arabinoxylan. Synergy betweenT. reesei and A. niger enzyme preparations have beenreported previously by us and other groups (Dien et al.,2006; Ximenes et al., 2007). In this regard, it is also signif-icant that T. reesei does not appear to produce FAE (Dienet al., 2006; Ximenes et al., 2007) unlike A. niger and

5224 B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225

H. insolens. Commercial preparations from H. inoslens

have long been reported to complement the activities ofT. reesei cellulase (Akin et al., 2006; Grabber et al.,1998a; Sorensen et al., 2003; Sorensen et al., 2005; Sorensenet al., 2006). Current research suggests that H. insolens

preparations contains highly active FAE activities forremoving feruloyl and diferuloyl groups (Faulds et al.,2004; Faulds et al., 2006a,b), high levels of a familyGH11 xylanase, and (most recently) an arabinoxylan arabi-nofuranohydrolyase activity (Sorensen et al., 2005).

Another important observation was derived from theresponse surfaces of arabinose and xylose yields for varyingadditions of pectinase and FAE. Their effects were inde-pendent when applied to either LHW or AFEX treatedDDGS. This is important because it suggests that it mightbe possible to isolate the active components from eachpreparation independent of each other. Eventually it mightbe possible to only produce (xenogenetically) the significantproteins to add to the cellulase blend.

The response surface experiment also indicated that yieldscontinued to increase as more and more pectinase and FAEwere added. Unfortunately, it was deemed impractical toadd the optimum amount of pectinase and FAE for yieldsof arabinose and xylose and a compromise enzyme loadingwas settled upon (1.2 FAE U/g DDGS of FAE and 40 xylan-ase U/g DDGS of pectinase). When the DDGS solids load-ing was increased to 20% w/w for each AFEX and LHWDDGS, with the same enzyme loading per g DDGS, thesugar yields per g DDGS fell. Lower efficiencies at higherbiomass solids loading are to be expected from the literatureand are indicative of end product inhibition, mass transferlimits, and/or xylan solubility/association effects. For thehigher solids, total monosaccharide yields were 78% and75% and final sugar concentrations were 68 and 58 g/l,respectively, for AFEX and LHW DDGS (Table 3).

On the positive side, the sugar hydrolysates appear to befermentable, which indicates the absence of inhibitory side-products. Both AFEX and LHW pretreated DDGS wasdigested with cellulase and b-glucosidase and simultaneousfermentation with S. cerevisiae. As a control, untreatedDDGS was also treated in a similar manner. Xylan relatedenzymes were not added because S. cerevisiae does not fer-ment arabinose or xylose; collaborators at Purdue Univer-sity are currently carrying out fermentations using a xylosefermenting recombinant Saccharomyces strain. Ethanolfermentations were concluded in 27–33 h and yields forthe pretreated material were 89% and 90% of theoretical.If these results could be replicated at a dry grind plant,the increase in ethanol yield would be 0.22 gal/bu of cornand greater if the xylan sugars was also converted.

Acknowledgements

The authors thank Mr. Loren Iten for excellent techni-cal assistance and the following companies for their giftsof research material: Genencor International, Inc. for en-zymes, Biocatalysts Ltd for DEPOL 740L, and Big River

Resources, LLC for DDGS. We would also like to thankDr. Ladisch and Dr. Mosier of Purdue University for use-ful discussions, especially regarding liquid hot-water pre-treatment of DDGS. This work was funded in partthrough the Midwest Consortium for Biobased Productsand Bioenergy, USDOE Award Number DE-FG36-04GO14220.

References

Adney, B., Baker, J., 1996. Measurement of Cellulase Activities. LAP-006NREL Analytical Procedure. National Renewable Energy Laboratory,Golden, CO.

Akin, D.E., Morrison, W.H., Rigsby, L.L., Barton, F.E., Himmelsbach,D.S., Hicks, K.B., 2006. Corn stover fractions and bioenergy. Appl.Biochem. Biotechnol. 129–132, 104–116.

Bootsma, J.A., Entorf, M., Eder, J., Shanks, B.H., 2007. Hydrolysis ofoligosaccharides from distillers grains using organic–inorganic hybridmesoporous silica catalysts. Biotechnology Resources.

Bunzel, M., Ralph, J., Funk, C., Steinhart, H., 2005. Structural elucida-tion of new ferulic acid-containing phenolic dimers and trimersisolated from maize bran. Tetrahedron Lett. 46, 5845–5850.

Dien, B.S., Iten, L.B., Bothast, R.J., 1999. Conversion of corn fiber toethanol by recombinant E-coli strain FBR3. J. Ind. Microbiol.Biotechnol. 22, 575–581.

Dien, B.S., Bothast, R.J., Nichols, N.N., Cotta, M.A., 2002. The US cornethanol industry: An overview of current technology and futureprospects. Int. Sugar J. 104, 204.

Dien, B.S., Nichols, N.N., O’Bryan, P.J., Iten, L.B., Bothast, R.J. 2004.Enhancement of ethanol yield from corn dry grind process byconversion of kernel fiber fraction. In Agricultural Applications inGreen Chemistry. ACS, Washington DC.

Dien, B.S., Li, X.L., Iten, L.B., Jordan, D.B., Nichols, N.N., O’Bryan,P.J., Cotta, M.A., 2006. Enzymatic saccharification of hot-waterpretreated corn fiber for production of monosaccharides. EnzymeMicrob. Technol. 39, 1137–1144.

Ezeji, I., Blaschek, H., 2008. Fermentation of dried distiller 0s grainsand solubles (DDGS) hydrolysates to solvents and value-addedproducts by solventogenic Clostridia. Bioresour. Technol. 99, 5232–5242.

Faulds, C.B., Mandalari, G., Lo Curto, R., Bisignano, G., Waldron,K.W., 2004. Arabinoxylan and mono- and dimeric-ferulic acid releasefrom brewer’s grain and wheat bran by feruloyl esterases and glycosylhydrolases from Humicola insolens. Appl. Microbiol. Biotechnol. 64,644–650.

Faulds, C.B., Mandalari, G., Lo Curto, R.B., Bisignano, G., Christak-opoulos, P., Waldron, K.W., 2006a. Synergy between xylanases fromglycoside hydrolase family 10 and family 11 and a feruloyl esterase inthe release of phenolic acids from cereal arabinoxylan. Appl. Micro-biol. Biotechnol. 71, 622–629.

Faulds, C.B., Mandalari, G., Lo Curto, R.B., Bisignano, G., Waldron,K.W., 2006b. Influence of the arabinoxylan composition on thesusceptibility of mono- and dimeric-ferulic acid release by Humicola

insolens feruloyl esterases. J. Sci. Food Agric. 86, 1623–1630.Grabber, J.H., Hatfield, R.D., Ralph, J., Zon, J., Amrhein, N., 1995.

Ferulate cross-linking in cell walls isolated from maize cell suspensions.Phytochemistry 40, 1077–1082.

Grabber, J.H., Hatfield, R.D., Ralph, J., 1998a. Diferulate cross-linksimpede the enzymatic degradation of non-lignified maize walls. J. Sci.Food Agric. 77, 193–200.

Grabber, J.H., Ralph, J., Hatfield, R.D., 1998b. Ferulate cross-links limitthe enzymatic degradation of synthetically lignified primary walls ofmaize. J. Agric. Food Chem. 46, 2609–2614.

Hespell, R.B., 1992. Fermentation of xylans by Butyrivibrio fibrisolvens

and Thermoanaerobacter strain B6A: Utilization of uronic acids andxylanolytic activities. Curr. Microbiol. 25, 189–195.

B.S. Dien et al. / Bioresource Technology 99 (2008) 5216–5225 5225

Hespell, R.B., Whitehead, T.R., 1990. Physiology and genetics of xylandegradation by gastrointestinal tract bacteria. J. Dairy Sci. 73, 3013–3022.

Kim, Y., Mosier, N., Ladisch, M., 2008a. Process simulation of modifieddry grind ethanol plant with recycle of ptreated and enzymaticallyhydrolyzed distillers’ grains. Bioresour. Technol. 99, 5177–5192.

Kim, Y., Hendrickson, R., Mosier, N., Bals, B., Balan, V., Dale, B.E.,Ladisch, M.R., 2008b. Enzyme hydrolysis and ethanol fermentation ofliquid hot-water and AFEX pretreated distillers’ grains at high solidsloadings. Bioresour. Technol. 99, 5206–5215.

Miller, G.L., 1959. Use of dinitrosalylic acid reagent for determination ofreducing sugars. Anal. Chem. 31, 127–132.

Mosier, N., Wyman, C., Dale, B., Elander, R., Lee, Y.Y., Holtzapple, M.,Ladisch, M., 2005. Features of promising technologies for pretreat-ment of lignocellulosic biomass. Bioresour. Technol. 96, 673–686.

Ruiz, R., Ehrman, T., 1996. Determination of carbohydrates in biomassby HPLC. LAP-002 NREL. Analytical Procedure. National Renew-able Energy Laboratory, Golden, CO.

Saha, B.C., 2003. Hemicellulose bioconversion. J. Ind. Microbiol.Biotechnol. 30, 279–291.

Saulnier, L., Marot, C., Chanliaud, E., Thibault, J.F., 1995. Cell wallpolysaccharide interactions in maize bran. Carbohydrate Polym. 26,279–287.

Saulnier, L., Marot, C., Elgorriaga, M., Bonnin, E., Thibault, J.F., 2001.Thermal and enzymatic treatments for the release of free ferulic acidfrom maize bran. Carbohydrate Polym. 45, 269–275.

Saulnier, L., Thibault, J.F., 1999. Ferulic acid and diferulic acids ascomponents of sugar-beet pectins and maize bran heteroxylans. J. Sci.Food Agric. 79, 396–402.

Sorensen, H.R., Meyer, A.S., Pedersen, S., 2003. Enzymatic hydrolysis ofwater-soluble wheat arabinoxylan. 1. Synergy between L-arabinofur-anosidases, endo-1,4-b-xylanases, and b-xylosidase activities. Biotech-nol. Bioeng. 81, 726–731.

Sorensen, H.R., Pedersen, S., Meyer, A.S., 2006. Optimization of reactionconditions for enzymatic viscosity reduction and hydrolysis of wheatarabinoxylan in an industrial ethanol fermentation residue. Biotech-nol. Progress 22, 505–513.

Sorensen, H.R., Pedersen, S., Vikso-Nielsen, A., Sorensen, H.R., Meyer,A.S., 2005. Efficiencies of designed enzyme combinations in releasingarabinose and xylose from wheat arabinoxylan in an industrialethanol fermentation residue. Enzyme Microb. Technol. 36, 773–784.

Tucker, M.P., Nagle, N.J., Jennings, E.W., Ibsen, K.N., Aden, A.,Nguyen, Q.A., Kim, K.H., Noll, S.L., 2004. Conversion of distiller’sgrain into fuel alcohol and a higher-value animal feed by dilute-acid pretreatment. Appl. Biochem. Biotechnol. 113–116, 1139–1159.

Ximenes, E.A., Dien, B.S., Ladisch, M.R., Mosier, N., Cotta, M.A., Li,X.L. 2007. Enzyme production by industrially relevant fungi culturedon feed co-product from corn dry grind ethanol plants. Appl.Biochem. Biotechnol., in press, doi:10.1016/j.biortech.2007.09.028.