Mats Galbe • Guido Zacchi (И)

Dept. of Chemical Engineering, Lund University, P. O. Box 124, 221 00 Lund, Sweden Guido. Zacchi@chemeng. lth. se

1 Introduction……………………………………………………………………………………………….. 42

2 Assessment of Pretreatment…………………………………………………………………………. 44

3 Pretreatment Methods…………………………………………………………………………………. 47

3.1 Physical Methods………………………………………………………………………………………… 48

3.2 Chemical Methods………………………………………………………………………………………. 48

3.3 Physicochemical Methods…………………………………………………………………………….. 49

3.4 Biological Methods………………………………………………………………………………………. 51

4 Results from Pretreatment Studies………………………………………………………………. 51

4.1 Corn Stover…………………………………………………………………………………………………. 52

4.2 Softwood Species………………………………………………………………………………………… 57

4.3 Two-Stage Pretreatment………………………………………………………………………………. 60

5 Conclusions………………………………………………………………………………………………… 62

References……………………………………………………………………………………………………. 62

Abstract Second-generation bioethanol produced from various lignocellulosic materials, such as wood, agricultural or forest residues, has the potential to be a valuable substitute for, or a complement to, gasoline. One of the crucial steps in the ethanol production is the hydrolysis of the hemicellulose and cellulose to monomer sugars. The most promising method for hydrolysis of cellulose to glucose is by use of enzymes, i. e. cellulases. However, in order to make the raw material accessible to the enzymes some kind of pretreatment is necessary. During the last few years a large number of pretreatment methods have been developed, comprising methods working at low pH, i. e. acid based, medium pH (without addition of catalysts), and high pH, i. e. with a base as catalyst. Many methods have been shown to result in high sugar yields, above 90% of theoretical for agricultural residues, especially for corn stover. For more recalcitrant materials, e. g. softwood, acid hydrolysis and steam pretreatment with acid catalyst seem to be the methods that can be used to obtain high sugar and ethanol yields. However, for more accurate comparison of differ­ent pretreatment methods it is necessary to improve the assessment methods under real process conditions. The whole process must be considered when a performance evalua­tion is to be made, as the various pretreatment methods give different types of materials. (Hemicellulose sugars can be obtained either in the liquid as monomer or oligomer sug­ars, or in the solid material to various extents; lignin can be either in the liquid or remain in the solid part; the composition and amount/concentration of possible inhibitory com­pounds also vary.) This will affect how the enzymatic hydrolysis should be performed

(e. g. with or without hemicellulases), how the lignin is recovered and also the use of the lignin co-product.

Keywords Assessment • Enzymatic hydrolysis • Lignocellulose • Pretreatment • Review

1

Introduction

Replacement of gasoline by liquid fuels produced from renewable sources is a high-priority goal in many countries worldwide. It is driven by the aims of a secure and sustainable energy supply and a desire to diminish the green­house effect. The transportation sector in the European Union (EU) is totally dependent on imported fossil fuels, and thus extremely vulnerable to market disturbances. It is also the sector responsible for the main part of the increase in CO2 emissions. The use of biofuels in the EU is encouraged by a Directive that set a target of 2% substitution of gasoline and diesel with biofuels in 2005 on an energy basis, which should have increased to 5.75% by 2010 [1].

Bioethanol is projected to be one of the dominating renewable biofuels in the transportation sector within the coming 20 years, and has already been introduced on a large scale in Brazil, the USA and some European countries. The advantages of bioethanol are that it can be produced from a variety of raw materials, it is non-toxic and is easily introduced into the existing in­frastructure, either as a low blend with gasoline (e. g. E5 and E10) or used in flexi-fuel vehicles at a high concentration (e. g. E85) or as a neat fuel in dedi­cated engines. However, almost all bioethanol today is produced from sugar — or starch-based agricultural crops, using so-called first-generation technolo­gies. Although this ethanol is produced at a competitive cost, the raw material supply will not be sufficient to meet the increasing demand for fuel ethanol, and also the reduction of greenhouse gases resulting from the use of sugar — or starch-based ethanol is not as high as desirable.

One of the most promising options to meet this challenge is the production of bioethanol from lignocellulose feedstocks, such as agricultural residues (e. g. wheat straw, sugar cane bagasse, corn stover) and forest residues (e. g. sawdust, thinning rests), as well as from dedicated crops (salix, switch grass) using second-generation technologies. These raw materials are sufficiently abundant and generate very low net greenhouse gas emissions, reducing en­vironmental impacts.

However, to compete with gasoline the production cost must be substan­tially lowered. Today, raw material and enzyme production are two of the main contributors to the overall production cost [2,3]. Efficient use of the whole crop is required, i. e. high overall yield of ethanol produced by hydro­lysis and fermentation of the carbohydrate fraction (hemicellulose and cel­lulose), as well as a high yield of the main co-product (lignin). However,
producing monomer sugars from cellulose and hemicellulose at high yields is far more difficult than deriving sugars from sugar — or starch-containing crops, e. g. sugar cane or corn.

Ethanol production from lignocellulosic materials comprises the following main steps: hydrolysis of hemicellulose; hydrolysis of cellulose; fermenta­tion; separation of lignin residue; recovery and concentration of ethanol; and wastewater handling. Figure 1 shows a simplified process flowsheet for ethanol production from lignocellulosic materials based on enzymatic hydro­lysis. Some of the most important factors to reduce the production cost are: an efficient utilization of the raw material by having high ethanol yields, high productivity, high ethanol concentration in the distillation feed, and also by employing process integration in order to reduce capital cost and energy de­mand. Part of the lignin can be burnt to provide heat and electricity for the process, and the surplus is sold as a co-product for heat and power appli­cations, which will increase the energy efficiency of the whole system. It is thus necessary to minimize the internal energy demand and to maximize the production of the solid fuel. The two conversion steps can be considered key processes: hydrolysis of the hemicellulose and cellulose to sugars, and fer­mentation of all the sugars.

The enzymatic process is regarded as the most attractive way to degrade cellulose to glucose [4-6]. However, enzyme-catalysed conversion of cellulose to glucose is very slow unless the biomass has been subjected to some form of pretreatment, as native cellulose is well protected by a matrix of hemicel­lulose and lignin. Pretreatment of the raw material is perhaps the single most crucial step as it has a large impact on all the other steps in the process, e. g. enzymatic hydrolysis, fermentation, downstream processing and wastewater handling, in terms of digestibility of the cellulose, fermentation toxicity, stir­ring power requirements, energy demand in the downstream processes and wastewater treatment demands.

An effective pretreatment should have a number of features. It has to:

• Result in high recovery of all carbohydrates.

• Result in high digestibility of the cellulose in the subsequent enzymatic hydrolysis.

• Produce no or very limited amounts of sugar and lignin degradation prod­ucts [7]. The pretreatment liquid should be possible to ferment without detoxification.

• Result in high solids concentration as well as high concentrations of liber­ated sugars in the liquid fraction.

• Have a low energy demand or be performed in a way so that the energy can be reused in other process steps as secondary heat.

• Have a low capital and operational cost.

Additional positive features would be if hemicellulose sugars were obtained in the liquid as monomer sugars, as this would help to avoid the use of hemi — cellulases, and/or if the lignin—without being oxidized—was separated from the cellulose, as this would alleviate the unproductive binding of cellulases on lignin in the enzymatic hydrolysis step.

This chapter will focus on pretreatment of lignocellulosic raw materials. Some of the most common methods that have been investigated will be pre­sented and to some extent compared for various raw materials.

2

The most widespread, commercial enzyme products currently available for biomass hydrolysis are produced by submerged fermentation of the sapro­phytic mesophilic fungus T. reesei [30]. This organism, first isolated over 60 years ago from decaying cotton tents during World War II [31] is a prolific producer of secreted cellulases. Since its initial isolation, numerous mutants have been isolated that increase the productivity of the strain by over 20­fold [28,32,33]. Three enzymes classes form the core of the T. reesei cellulase system: exoglucanases comprised of two primary cellobiohydrolases, a num­ber of endoglucanases, and в-glucosidases (Fig. 5). There are two types of

cellobiohydrolases, CBH I and CBH II, that constitute roughly 60% and 20% of the secreted protein mix and are critical to the efficient hydrolysis of cel­lulose [34]. The CBH I and II hydrolyze the cellulose chain processively from the reducing and non-reducing ends of cellulose chains, respectively, releas­ing the glucose disaccharide cellobiose. Endoglucanases (EG I-IV) constitute roughly 15% of the secreted protein and hydrolyze в-1,4 linkages within the cellulose chains, creating new reducing and non-reducing ends that can then be attacked by the CBHs. в-Glucosidases (BGL I and II), constituting roughly 0.5% of the secreted protein mix, and hydrolyze cellobiose and some other short-chain cellodextrins into glucose.

4.2

In contrast to the energy-conserving function of glycolysis, the main metabolic function of the PPP is to provide anabolic intermediates such as ribulose 5-phosphate, erythrose 4-phosphate, and NADPH for biosynthesis and cell growth. The flux through the nonoxidative PPP in S. cerevisiae was found to be much lower than in other yeasts [105], which was later con­firmed by metabolome analysis [106]. The low PPP activity in S. cerevisiae is sometimes interpreted to be a result of the domestication of S. cerevisiae by prolonged selection for carbon dioxide and ethanol production from hexose sugars. However, PPP activity is a crucial part of pentose metabolism, since it is virtually the only way to introduce xylulose into the central metabolism. It was early pointed out that the PPP activity may limit xylose metabolism in S. cerevisiae [47,107], which was further supported when excretion of PPP intermediates was observed in xylulose — and xylose-metabolizing S. cere­visiae [43,108].

The insufficient flux through the nonoxidative PPP in S. cerevisiae has been indirectly confirmed in several genome-scale and enzymatic analyses of mutant strains with improved xylose metabolism, where invariably ei­ther the transaldolase (TAL1) or the transketolase (TKL1) genes, or both, have been found to be upregulated [57,71,91,109-111]. Directly, the im­portance of the flux through the PPP has been confirmed by the superior pentose utilization and ethanolic fermentation by strains in which the en­zymes of the nonoxidative PPP have been overexpressed. An early attempt to overexpress P. stipitis transketolase in xylose-metabolizing S. cerevisiae was not successful [112], whereas overexpression of the endogenous S. cere­visiae transaldolase (TAL1) resulted in improved growth on xylose [78]. Later, the overexpression of all four nonoxidative PPP genes, including not only TAL1 and TKL1 but also ribulose-5-phosphate 4-epimerase (RPE1) and ribu — lokinase (RKI1), was shown to improve xylulose consumption by S. cere­visiae [90,113] (strain TMB3026, Table 3). Moreover, the improvement result­ing from the overexpression of the four genes was higher than when each gene was overexpressed alone [90]. The simultaneous overexpression of the whole nonoxidative PPP, together with GRE3 deletion, allowed growth on xylose in a strain carrying a bacterial XI [42] (strain TMB3050, Table 2). The usefulness of this combination of modifications was confirmed when it allowed aerobic and anaerobic growth on xylose in a strain carrying the Piromyces XI [97] (strain RWB217, Table 1). PPP overexpression also al­lowed superior xylose fermentation rates in combination with high levels of XR and XDH [42,54] (cf. strains TMB3057, TMB3056, and TMB3062, Table 1 and Fig. 5).

4.5

Significant advances related to recombinant enzyme expression support the potential for S. cerevisiae as a CBP host. However, the challenge of integrating all the different aspects of enzymatic hydrolysis of cellulose and hemicellu — lose and subsequent fermentation of the sugars released to ethanol in a single reactor with a CBP should not be underestimated. A pertinent question of­ten asked by critics is, “Would S. cerevisiae be able to simultaneously express multiple genes, while producing and secreting the different cellulases, hemi- cellulases, and pentose utilizing enzymes required?” Several studies demon­strate coexpression of multiple genes in S. cerevisiae, for example in the case of the expression of tethered cellulolytic and xylanolytic enzymes [59,119], xylose and arabinose utilizing enzymes [40,162], as well as xylose and cel — looligosaccharide utilizing enzymes [45]. The expression and secretion of a variety of cellulases, amylases, and pectinase has also been demonstrated without adversely affecting yeast growth [51, 52].

However, the number of genes expressed is probably not as important a challenge as the need for high-level expression as well as the stress re­sponses that may accompany such high-level expression. Factors that may impose unnecessary stress on the cell are (1) sequestering of transcription factors at highly expressed promoters used for heterologous gene expression,

(2) impact of unfavorable codon bias on the translation of heterologous pro­tein (can be overcome by the use of codon-optimized synthetic genes), and

(3) improper folding of foreign proteins that can evoke the (4) unfolded pro­tein response (UPR) and consequently the endoplasmic reticulum-associated protein degradation (ERAD) response [163]. Some of these effects may be exacerbated by (5) interrupted transport of foreign proteins through the se­cretion pathway, or (6) accumulation of larger proteins at the cell wall due to low permeability [164]. The answer would thus not be simply overexpression of all the required genes to ensure a functional CBP yeast with the desirable enzymatic activities, but much more attention should also be devoted to the careful manipulation of the required enzyme activities and producing them at the right concentration to provide functionality without exerting too much unnecessary stress on the CBP yeast.

Essentially all work carried out thus far involving heterologous expres­sion of saccharolytic enzymes in yeast has involved laboratory strains. Much of this work has to be transferred to industrial strains that provide the fer­mentation capacity and robustness desired for industrial processes. Different strategies have been used for the overexpression of multiple genes in in­dustrial S. cerevisiae strains. High copy-number episomal YEp vectors, often using the two-micron autonomous replicating sequence (ARS), have been very helpful in demonstrating proof of concept in laboratory strains of S. cere­visiae [43,51,102,115]. However, these vectors are usually mitotically unsta­ble and require selection for the episomal plasmid, which often means using a defined medium that is not applicable to industrial uses [164]. The preferred route taken for industrial strains has been the use of integrative YIp vectors that facilitate direct integration of foreign expression cassettes into a target gene on the yeast genome [165,166] or recycling dominant selectable markers for multiple integration [167-170]. Although these methods provide stable expression from the yeast genome and are amendable to industrial strains, the major drawback has been low expression levels and often not delivering high enough quantities of the required gene product.

Different approaches have been pursued in an attempted to combine the advantages of overexpression from multicopy plasmids with the stability of chromosomal integration, which is also applicable to industrial strains when dominant selectable markers are used. These include the use of repetitive chromosomal DNA sequences such as rDNA [171] and 8-sequences [172]. There are approximately 140-200 copies of rDNA existing in the haploid yeast genome; however, rDNA is located in the nucleolus, which may af­fect the accessibility to RNA polymerase II transcription. Also, the size of pMIRY (multiple integration into ribosomal DNA in yeast) vectors could determine the mitotic stability of these multiple integrations [173]. The 8- sequences are the long terminal repeats of S cerevisiae retrotransposon Ty. More than 400 copies of 8-sequences can exist either Ty associated or as sole sites in the haploid yeast genome [174]. 8-Integration thus makes it possible to integrate more copies of a gene of interest into the yeast genome than the conventional integration systems [78,175]. Host strains and integrated gene size can significantly affect the transformation efficiency at 8-sequences; however, the transformation frequency can be 10- to 100-fold those ob­tained when transforming with vectors that target a single gene on the yeast genome [176].

Although the necessary tools exist for multiple and repeated integration of genes of interest into the genome of industrial strains to complement the required features for CBP (Table 1), a more strategic approach would be re­quired to design a yeast that produces the required enzyme activities, yet re­tains the competence to still perform well under industrial conditions for the economic conversion of plant biomass to ethanol. Such an approach will most probably start by building on a platform industrial yeast that cometabolises hexoses and pentoses, and subsequently finding the right combination and level of expression for saccharolytic enzymes. This approach will inevitably use reiterated metabolic engineering and flux analysis, selection and mutage­nesis strategies, and even strain breeding to allow the microorganism itself to overcome rate-limiting hurdles toward developing an efficient CBP yeast. Ex­amples of such approaches in the past have been performed to enhance xylose fermentation in laboratory and industrial strains [33,37,39,177].

R. P. Chandra1 • R. Bura2 • W. E. Mabee1 • A. Berlin1 • X. Pan3 •

J. N. Saddler1 (И)

faculty of Forestry, University of British Columbia, 2424 Main Mall,

Vancouver, British Columbia V6T 1Z4, Canada jack. saddler@ubc. ca

2College of Forest Resources, University of Washington, Box 352100,

Seattle, WA 98195-2100, USA

3Biological Systems Engineering, University of Wisconsin-Madison,

460 Henry Mall, Madison, WI 53706, USA

1 Background………………………………………………………………………………………………… 68

1.1 Summary of Pretreatment Processes…………………………………………………………….. 69

1.2 Steam Pretreatment of Biomass…………………………………………………………………… 71

2 Substrate Characteristics of Steam-Pretreated Wood…………………………………….. 73

3 Substrate Lignin………………………………………………………………………………………….. 75

3.1 The Effects of Pretreatment on Lignin Content……………………………………………….. 76

3.2 The Effects of Substrate Lignin on Enzymatic Hydrolysis……………………………….. 78

4 Substrate Hemicelluloses……………………………………………………………………………… 80

4.1 The Effect of Pretreatment on Hemicellulose Content……………………………………… 81

4.2 The Effect of Substrate Hemicellulose Content on Hydrolysis…………………………. 83

5 Physical Properties Affecting the Hydrolysis of Substrates by Cellulases. 84

5.1 Specific Surface Area…………………………………………………………………………………… 85

5.2 Cellulose Crystallinity and Degree of Polymerization…………………………………….. 87

6 Conclusions………………………………………………………………………………………………… 88

References……………………………………………………………………………………………………. 90

Abstract Although the structure and function of cellulase systems continue to be the subject of intense research, it is widely acknowledged that the rate and extent of the cellulolytic hydrolysis of lignocellulosic substrates is influenced not only by the effective­ness of the enzymes but also by the chemical, physical and morphological characteristics of the heterogeneous lignocellulosic substrates. Although strategies such as site-directed mutagenesis or directed evolution have been successfully employed to improve cellulase properties such as binding affinity, catalytic activity and thermostability, complemen­tary goals that we and other groups have studied have been the determination of which substrate characteristics are responsible for limiting hydrolysis and the development of pretreatment methods that maximize substrate accessibility to the cellulase complex. Over the last few years we have looked at the various lignocellulosic substrate characteristics

at the fiber, fibril and microfibril level that have been modified during pretreatment and subsequent hydrolysis. The initial characteristics of the woody biomass and the effect of subsequent pretreatment play a significant role on the development of substrate prop­erties, which in turn govern the efficacy of enzymatic hydrolysis. Focusing particularly on steam pretreatment, this review examines the influence that pretreatment conditions have on substrate characteristics such as lignin and hemicellulose content, crystallinity, degree of polymerization and specific surface, and the resulting implications for effective hydrolysis by cellulases.

Keywords Biomass • Cellulose • Cellulases • Hemicellulose • Hydrolysis • Lignin •

Steam pretreatment

1

Background

There have been several recent reviews [1-7] that have considered the various enzymatic factors that influence the efficiency of hydrolysis of lignocellu — losic substrates. However, it is apparent that the physical and chemical nature of lignocellulosic substrates imparted by different pretreatment procedures are just as complex and influential as the enzyme systems used to break­down the various components that comprise a lignocellulosic substrate into fermentable monosaccharides and other industrially relevant chemical com­pounds. Despite intensive research over the last 30 years or so, obtaining the rapid, complete and efficient conversion of cellulosic substrates by enzy­matic hydrolysis remains a challenging goal. Up until about 5 or 6 years ago, various technoeconomic models had indicated that the enzyme production step of the overall biomass-to-ethanol process was one of the most expensive. Recent efforts by some of the world’s leading industrial enzyme manufac­turers have resulted in an approximate 20- to 30-fold reduction in the cost of cellulases utilized for the hydrolysis of pretreated corn stover [8]. How­ever, it is acknowledged that the nature of the substrate and pretreatment method used continue to influence the effectiveness of the enzyme mix em­ployed [9]. The significant decreases in the cost of the enzyme hydrolysis step have highlighted how the cost and nature of the biomass feedstock and the pretreatment method used to enhance both overall product recovery and enzymatic hydrolysis of the cellulosic and hemicellulosic components are sig­nificant technical and economic considerations.

Typically, after an initial rapid phase, the hydrolysis rate decreases rapidly during the saccharification process, resulting in lower glucose yields and longer processing times and, in most cases, the accumulation of a recalcitrant residue due to incomplete hydrolysis of the substrate. When a typical progress curve for enzymatic hydrolysis of cellulose is plotted, the reaction rate usually remains constant during the first few hours. However, the reaction rate even­tually slows down and it has been suggested that the decrease in reaction rate can be attributed to both enzyme — and substrate-related factors [2-4,6]. Var­ious substrate-related factors that affect hydrolysis include: how the presence of extraneous materials such as lignin and hemicellulose impede the action of cellulases, the influence of cellulose crystallinity and degree of polymeriza­tion (DP), and the amount of accessible surface area available to react with cellulases [2]. Enzyme-related factors that have been studied include: shear or thermally induced deactivation [10] occurring during mixing or exposure to high temperatures, the separation of enzyme components by the physical characteristics of the substrate resulting in a loss of synergism [11], as well as product inhibition due to an accumulation of cellobiose and glucose in the re­action medium. It is known that both enzyme — and substrate-related factors influence the efficiency of enzymatic hydrolysis [2]. However, depending on the nature of the substrate and pretreatment used, one factor could be more influential than another.

As mentioned earlier, an effective pretreatment method should be cheap (both capital and operating costs), effective on a wide range of lignocellulosic materials, require minimum preparation/handling steps prior to pretreat­ment, ensure recovery of all of the lignocellulosic components in a useable form, and provide a cellulosic stream that can be efficiently hydrolyzed with low concentrations of enzymes. With regard to the latter requirement, it would be beneficial if the pretreatment process could degrade the cell-wall structure by reducing the cellulose crystallinity, DP and particle size, while removing hemicellulose and lignin and increasing pore volume such that the cellulosic and hemicellulosic surface area available to the enzymes is greatly increased. However, as will be discussed in more detail, no one currently available pretreatment process can provide all of these desired outcomes on all lignocellulosic materials and it is the nature of the compromised condi­tions that will be described in this review.

1.1

The thermostable enzyme preparations were kindly provided by ROAL, Fin­land. The genes encoding three thermostable enzymes: cellobiohydrolase (CBH/Cel7A) from Thermoascus aurantiacus, fused with the T. reesei CBHI cel­lulose binding domain (CBM), endoglucanase (EG/Cel45A) from Acremonium thermophilum and a xylanase and в-glucosidase (BG/Cel3A) from T. auran­tiacus were inserted under the control of a strong T. reesei cbhl promoter and transformed into a host strain where all the major cellulase genes were deleted (phenotype CBHI/Cel7A — CBHII — Cel6A — EGI/Cel7B — EGII/Cel5A). Fermenter supernatants produced at pilot scale were used to produce vari­ous mixtures of the thermostable enzymes. The background activities in the deletion strains were measured. The composition of the mixture of the three thermostable enzymes was optimised based on the average cellulase compo­sition of T. reesei. The enzyme components were mixed in different ratios and the total cellulase activity of the mixtures was measured at 50 °C(as FPU mL-1) and used as the basis of enzyme dosing. In addition, a family 10 thermostable xylanase from T. aurantiacus, cloned and expressed in the T. reesei deletion strain, was added to some preparations to ensure complete hydrolysis.

Celluclast 1.5 L FG (Novozymes, Denmark) and Econase CE (ABEnzymes, Finland), eventually supplemented with BG from Novozym 188 (Novozymes, Denmark) were used as reference enzymes. The standard enzyme dosage was 10 FPU g-1 cellulose for Celluclast 1.5 L FG, supplemented with additional BG (100-500 nkatg-1 cellulose). Assuming an average 50% cellulose content of the lignocellulose substrates, the enzyme dosage was thus 5 FPU g-1 substrate. For the hydrolysis experiments at elevated temperatures, higher dosage of Celluclast (22 FPU g-1 cellulose) was used.

The total cellulolytic activity used as a basis for dosing of the enzyme mixtures was evaluated using the FPU activity, measured against Whatman no. 1 filter paper [36]. The EG activity was assayed using hydroxyl ethyl cel­lulose as substrate [36]. The CBH activity was determined by using 4-methyl umbelliferyl-b-D-lactoside as substrate, estimating the effect of EGs by carry­ing out the assay with or without 20 mM cellobiose in the reaction [6]. The xy- lanase activity was assayed using birchwood glucuronoxylan as substrate [4] and that of BG using p-nitrophenyl-b-D-glucopyranosidase as substrate [5]. Protein was assayed according Lowry et al. [43]. All the enzyme activity assays were carried out at pH 5.

Substrates

The substrates used were steam pretreated, washed spruce solid fraction kindly provided by Guido Zacchi at the Lund University, Sweden, and steam pretreated corn stover kindly provided by Francesco Zimbardi at ENEA, Italy. The solid fraction of spruce substrate after the pretreatment was sep­arated from the liquid fraction by filtration, washed and used in the hydro­lysis experiments. The composition of the fibre fractions of the substrates is presented in Table 2. In addition to the insoluble fibre fraction, the corn stover substrate contained significant amounts, about 17% (d. w.) of solu­bilised mono — and oligosaccharides, solubilised mainly from hemicelluloses. Based on secondary analytical enzymatic hydrolysis and HPLC analysis, the carbohydrates in the soluble fraction consisted of xylose (74%), arabinose (15%), galactose (5%) and glucose (6%). Comparative hydrolysis experiments were carried out using crystalline cellulose (Avicel, Sigma).

Antonius J. A. van Maris1 • Aaron A. Winkler2 • Marko Kuyper2 •

Wim T. A. M. de Laat3,4 • Johannes P. van Dijken1,2 • Jack T. Pronk1 (И)

department of Biotechnology, Delft University of Technology, Julianalaan 67,

2628 BC Delft, The Netherlands J. T. Pronk@TUDelft. NL

2Bird Engineering B. V., Westfrankelandsedijk 1, 3115 HG Schiedam, The Netherlands

3DSM Anti-Infectives, A. Fleminglaan 1, 2613 AX Delft, The Netherlands

4Royal Nedalco, Van Konijnenburgweg 100, 4612 PL Bergen op Zoom, The Netherlands

1 Introduction……………………………………………………………………………………………… 180

1.1 Saccharomyces cerevisiae and Fermentation of Lignocellulosic Hydrolysates 180

1.2 Introduction of Heterologous Genes Encoding Xylose Reductase

and Xylitol Dehydrogenase: Redox Restrictions………………………………………. 182

1.3 Native D-Xylose-Metabolising Enzymes in S. cerevisiae………………………………….. 185

1.4 One-Step Conversion of D-Xylose into D-Xylulose via Xylose Isomerase. . 186

2 Xylose Isomerase: Properties and Occurrence………………………………………………. 186

3 Expression of Xylose Isomerases in S. cerevisiae:

a Long and Winding Road…………………………………………………………………….. 187

4 Characterisation of Yeast Strains

with High-Level Functional Expression of a Fungal Xylose Isomerase… 190

5 Metabolic Engineering

for Improved Xylose-Isomerase Based D-Xylose Utilisation……………………… 192

6 Evolutionary Engineering

for Improved Xylose-Isomerase-Based D-Xylose Utilisation……………………… 194

6.1 Evolutionary Engineering of D-Xylose-Consuming S. cerevisiae

for Improved Mixed Substrate Utilisation………………………………………………. 194

6.2 Evolutionary Engineering of S. cerevisiae

only Containing Fungal Xylose Isomerase……………………………………………………. 197

7 Towards Industrial Application:

Fermentation Trials with Xylose-Isomerase-Expressing S. cerevisiae. . . . 198

7.1 From the Laboratory to the Real World: Strains and Media…………………………… 198

7.2 Batch Fermentation of Wheat Straw Hydrolysate…………………………………………. 199

7.3 Fed-Batch Fermentation of Corn Stover Hydrolysate…………………………………….. 200

8 Outlook…………………………………………………………………………………………………… 201

References

Abstract Metabolic engineering of Saccharomyces cerevisiae for ethanol production from D-xylose, an abundant sugar in plant biomass hydrolysates, has been pursued vigorously for the past 15 years. Whereas wild-type S. cerevisiae cannot ferment D-xylose, the keto — isomer D-xylulose can be metabolised slowly. Conversion of D-xylose into D-xylulose is therefore crucial in metabolic engineering of xylose fermentation by S. cerevisiae. Ex­pression of heterologous xylose reductase and xylitol dehydrogenase does enable D-xylose utilisation, but intrinsic redox constraints of this pathway result in undesirable byprod­uct formation in the absence of oxygen. In contrast, expression of xylose isomerase (XI, EC 5.3.1.5), which directly interconverts D-xylose and D-xylulose, does not have these constraints. However, several problems with the functional expression of various bacterial and Archaeal XI genes have precluded successful use of XI in yeast metabolic engineer­ing. This changed with the discovery of a fungal XI gene in Piromyces sp. E2, expression of which led to high XI activities in S. cerevisiae. When combined with over-expression of the genes of the non-oxidative pentose phosphate pathway of S. cerevisiae, the result­ing strain grew anaerobically on D-xylose with a doubling time of ca. 8 h, with the same ethanol yield as on glucose. Additional evolutionary engineering was used to improve the fermentation kinetics of mixed-substrate utilisation, resulting in efficient D-xylose utilisa­tion in synthetic media. Although industrial pilot experiments have already demonstrated high ethanol yields from the D-xylose present in plant biomass hydrolysates, strain ro­bustness, especially with respect to tolerance to inhibitors present in hydrolysates, can still be further improved.

1

Introduction

1.1

Bioethanol was introduced into the transportation fuel supply chain as early as the 1970s with the introduction of the PROALCOOL program by the Brazil­ian government in an original effort to stabilize the international price of sugarcane, which was highly sensitive to subsidies by other domestic produc­ers. In 1979, the Brazilian government strengthened the program by spon­soring development of a fleet of ethanol-fueled vehicles [42]. Although the history of bioethanol in Brazil is quite tumultuous with significant govern­ment sponsorship, tax incentives, and subsidies, Brazil has emerged as the second largest producer of bioethanol (4.3 billion gallons/year in 2005) re­quiring 25% ethanol blends in transportation fuel, and has become energy self-sufficient by supplementing internal petroleum supplies and refining cap­acity with bioethanol production [43]. In 2005, total Brazilian petroleum production was estimated at 2 million bbl/day with consumption estimated at 1.6 million bbl/day, in contrast to the USA which produced 7.6 billion bbl/day, yet consumed 20 billion bbl/day [35].

Bioethanol may serve both as an additive or complete replacement for petroleum-derived transportation fuels, particularly gasoline in spark igni­tion (SI) engines. The volumetric energy fraction of ethanol is approximately 66% that of gasoline, suggesting a one-third reduction in the total kilometers per volume of ethanol consumed. However, review of the comparative phys­ical chemistry data provides insight into why ethanol combustion results in a 15% higher efficiency [44]. Ethanol (C2H5OH, 34.7 wt % oxygen) is a par­tially oxidized fuel compared to gasoline (C4-C12, 0 wt % oxygen), resulting in a lower stoichiometric air-to-fuel ratio. Therefore, a larger mass or volume of ethanol compared to gasoline is required to yield the same caloric output from combustion. However, ethanol also has a higher octane number, per­mitting the fuel to be burned at a higher compression ratio (defined as the ratio of the volume between the piston and cylinder head before and after a compression stroke). A higher compression ratio results in higher power output, efficiency, and consequently favorable fuel economy [45]. Compared to gasoline, there is only a 20-25% reduction in kilometer efficiency [44]. Fur­thermore, as a result of the significantly higher latent heat of vaporization for ethanol (1177 kJ/kg compared to 348 kJ/kg at 60 °C) there is an effective engine cooling and leaner operation. This leads to significant reductions in CO(g) and NOx,(g) emissions, with 85% ethanol blends of gasoline (referred to as E-85) yielding NOx,(g) emission reductions of 20% compared to pure gasoline. However, the emission of reactive aldehydes, including acetalde­hyde and formaldehyde, is increased [46,47]. Several studies on the effect of ethanol-gasoline blends (up to 60% ethanol) on engine performance and exhaust emissions have suggested that proper fine tuning of engine parame­ters can lead to excellent performance with significantly reduced hydrocarbon and CO(g) emissions [46-48].

In 1990 the USA enacted the Clear Air Act Amendments, mandating that oxygenated additives (methyl-tertiary-butyl ether, MTBE; ethyl-tertiary-butyl ester, ETBE; or ethanol) be included at 2 wt % oxygen to decrease hazardous emissions. In 1999, 21 million tons of MTBE were produced globally, primar­ily in the USA, although Europe produced approximately 3.3 million tons.

In the USA, it is among the most frequently found groundwater contami­nants with over 400 000 underground storage tanks identified to be leaking by the US Environmental Protection Agency (EPA) since 1988 [49]. Although there is still debate in the public health community as to the toxicity level and health risk that MTBE human consumption poses, a number of US states have banned the use of MTBE as a fuel additive. Furthermore, many European nations, including the UK, Germany, and Switzerland have identified MTBE — contaminated sites and are transitioning to ethanol enrichment [50,51]. As a result, ethanol has been the favored fuel additive for increasing oxygenation.

In August 2005, the Energy Policy Act (EPACT) was enacted into US law creating the national Renewable Fuels Standard (RFS). The RFS calls for

15.1 billion L of renewable fuels (primarily ethanol but may include alterna­tive fuels such as biodiesel) to be used by 2006, increasing by 2.6 billion L/year until 2011 when a final volume of 28.4 billion L will be called for in 2012 [43].

The USA is neither alone nor first with actively passing legislation that requires and promotes the integration of biofuels into the transportation economy. As previously discussed, Brazil presently requires a 25% ethanol blend of all gasoline, and continues to provide preferential tax treatment for ethanol producers and consumers. Argentina is requiring a 5% ethanol blend over the next 5 years. Thailand requires that all gasoline sold in Bangkok must be composed of 10% ethanol. India is requiring 5% ethanol gasoline blends. Canada has provided tax benefits for ethanol producers and consumers since 1992 [43].

The European Union (EU) has also taken an aggressive stance in reshap­ing its transportation fuel and energy supply chain, in addition to promoting industrial biotechnology as a sustainable and cost-effective alternative to petrochemical processes. In December 2005, the EU adopted the Biomass and Biofuels Action Plan. This plan encompasses more then 20 specific ac­tion items, including creation of the Biofuels Technology Platform with the explicit purpose of advancing research into the use of forestry, agricultural, and woody crops for energy purposes. In February 2006, the EU adopted the Strategy for Biofuels, which set out three objectives: (1) to promote biofuels in both the EU and developing countries, (2) to prepare for large-scale use of biofuels by improving their cost-competitiveness and increasing research into second-generation fuels, and (3) to support developing countries where biofuel production could stimulate sustainable economic growth [52].

Furthermore, the EU has established quantitative targets for incorporation of biofuels into a broader and emerging bio-based economy. The EU trans­port sector accounts for more than 30% of the total energy consumption, with a 98% dependency on fossil fuels. There is also significant pressure for the EU to comply with the Kyoto Protocol, an agreement under the United Nations Framework Convention on Climate Change, ratified by 160 countries to sig­nificantly reduce greenhouse gas emissions, specifically CO2,(g). The EU has failed to meet the Kyoto targets with 90% of the increases in CO2,(g) emissions between 1990 and 2010 attributable to transportation fuel usage. Therefore, significant reform in transportation fuel usage is required. There are three specific legislative actions in place [53]:

• Promotion of renewable energy-based electricity generation from 14% in 1997 to 21% by 2010 for the EU 25 (22.1% for EU 15) (Directive 2001/77/EC)

• Promotion of biofuels for transport applications by replacing diesel and petrol to the level of 5.75% by 2010 (Directive 2003/30 EC) accompanied by detaxation of biofuels

• Promotion of cogeneration of heat and electricity (Directive 2004/8/EC)

It is clear, however, that the EU is not meeting the objectives set forth. Spe­cifically, the current production of liquid biofuels in the EU is 2 million tons of oil equivalent (Mtoe), less than 1% of the market. The EU policy target for 2010 was 18 Mtoe in the transportation sector alone. Although it is un­likely the target will be met, it should be noted that between 4 and 13% of the total agricultural land in the EU would be required to meet the above target. Therefore, through the creation of the various plans and platforms highlighted before, the EU has established, “An ambitious and realistic vi­sion for 2030 is that up to one-fourth of the EU’s transport fuel needs could be meet by clean and CO2-efficient biofuels” [53]. Although it remains to be seen whether the appropriate resources will be allocated to meet this goal, it is certainly clear that industrial biotechnology, in particular the concept of a bio-based economy with biorefineries at its core, has taken center stage in the EU meeting its energy needs and environmental targets.

2.3

As has been demonstrated previously with pretreated substrates [30], the rates and extents of hydrolysis of pulp fibers have also been directly corre­lated to their initial specific surface area. This is not surprising, since the very existence of a substrate pretreatment step stems from the necessity to increase the accessibility of reaction sites on substrates to cellulases, as lig — nocellulosic substrates such as wood possess limited reactive surface area available to cellulases prior to pretreatment. Coincidentally, both the chem­ical and mechanical pulping processes that have been applied to produce pulp for the formation of paper also result in an increase in accessibility to cellulases compared to the starting lignocellulosic furnish. During pulping, wood chips are subjected to either physical or physiochemical action, re­sulting in the breakdown of the lignocellulosic matrix into fiber cells [48]. Consequently, the breakdown of the wood yields fibers with various phys­ical attributes such as length, coarseness, width, kink and curl [102,103]. Surface area of pulp fibers can be divided into exterior surface area affected mainly by fiber length and width, or interior surface area, which is gov­erned by the size of the lumen and the number of fiber pores and cracks. The varying fiber lengths and widths produced during pulping can be viewed in a similar manner as the array of particle sizes produced during the pre­treatment oflignocellulosic substrates for bioconversion. The specific surface area of a mixture of particles is inversely proportional to their average diam­eter, therefore, a smaller average particle size results in an increase in surface area. Indeed, cellulases have been shown to act on the surface of pulp fibers, resulting in a “peeling effect” [104]. Therefore, smaller particle sizes with a greater amount of specific surface area would be expected to hydrolyze at a faster rate.

In an investigation assessing the hydrolysis of Douglas-fir Kraft and me­chanical pulps, Mooney et al. showed that at equal lignin contents, the “fines” (small particles) of a delignified mechanical pulp were hydrolyzed faster than the longer fibers (large particles) of the Kraft pulp [17]. When each fiber length fraction was hydrolyzed separately, it was shown that the isolated long fiber fraction hydrolyzed slower and consequently adsorbed a lower amount of cellulases than the whole pulp [17]. The increased hydrolysis rate of the whole pulp was attributed to the greater amount of specific surface available for the adsorption of cellulases provided by the pulp fines and short fibers. Although it is apparent that particle size has a significant effect on cellulose hydrolyzability, it has also been shown that the fiber delamination and en­hanced swelling that results from mechanical treatment of Kraft and recycled pulp fibers has a greater effect on hydrolysis by cellulases than does a de­crease in particle size [105]. Since recycled pulps originate from fiber sources that undergo irreversible changes in their structure upon drying [106], their swelling properties must be regenerated by employing a mechanical treat­ment referred to as “refining” or “beating”. The swelling properties of pulps can be measured using the water retention value measurement [107]. After beating, the pulp sample usually drains at an inadequate rate to be used on a high-speed paper machine. Consequently, cellulases have been shown to improve the drainage of recycled pulps. Oksanen et al. [108] applied sepa­rate EG1, EG2 and CBH1 cellulase components to pulps during each recycling round. As each pulp was beaten after recycling, the water retention value (WRV) increased and the pulp became more responsive to cellulases, espe­cially EG1 and EG2. Although the particle size and swelling properties of pulps have been shown to be related to the ease of hydrolysis of lignocellulosic substrates, it has been shown that a greater amount of information related to the action of cellulases can be obtained from measurements of the pores or “interior” surface area of pulp fibers available for penetration by cellulases.

Direct correlations have been found between the initial pore volume or interior surface area of lignocellulosic substrates and their extent of hydro­lysis [30,83,90]. It has been proposed that the efficacy of cellulose hydrolysis is enhanced when the pores of the substrate are large enough to accommo­date both large and small enzyme components to maintain the synergistic action of the cellulase enzyme system [11]. From extensive studies, Grethlein et al. [83] and others [109-111] have found that the rate-limiting pore size for the hydrolysis of lignocellulosic substrates was 5.1 nm, thus the solute exclu­sion technique utilizing molecular probes in this size range has been shown to be effective for assessing the pore volume of substrates. Mooney et al. [68] utilized dextran molecular probes in the solute exclusion method to measure the pore volume of refiner mechanical pulp (RMP), sulfonated RMP, sodium chlorite delignified RMP and Kraft pulp from Douglas-fir to assess the ease of these pulps to subsequent hydrolysis by cellulases. As mentioned earlier, the delignification of the RMP resulted in a greater rate and extent of hydrolysis than the Kraft pulp sample, which may be attributed to the smaller particle size of the RMP. The sulfonation of the RMP dramatically increased swelling. Unlike delignification, however, this did not translate into either enhanced access to the 5.1 nm probe or hydrolysis performance. The most feasible ex­planation for these results is that the lignin content of the sulfonated pulp (30.9%) inhibited hydrolysis, regardless of the greater swelling of the pores, thus demonstrating the detrimental effect of substrate lignin on hydrolysis as mentioned earlier.

Since it is well known that the pore volumes of pulps undergo significant reductions upon drying [106], Esteghlalian et al. [112] innovatively applied the Simons’ stain technique to measure changes in pore volume imparted by air, oven and freeze drying prior to enzymatic hydrolysis. As expected, drying significantly reduced the number of larger pores in the pulp sample, which most likely restricted the access of the fibers to cellulases and thus de­creased hydrolysis rate over 12 h [112]. Although the specific surface area of the substrate provided by decreased particle size and increased swelling and pore volume plays a significant role in facilitating hydrolysis by cellulases, the interconnecting role of other substrate factors such as crystallinity and DP should also be considered.

5.2

The observation that most xylose-utilizing fungi produce considerable amounts of xylitol from xylose, and that only species containing also the

NADH-dependent XR activity are capable of producing ethanol from it, sug­gested that the different cofactor preferences of XR and XDH limit ethanolic xylose fermentation by yeast [21,32]. Since S. cerevisiae ferments xylu­lose [1,2], it was suggested that xylose fermentation could be easily achieved by heterologous expression of an XI [32,33]. Indeed, xylose was fermented to ethanol when extracellular XI was added to the medium [33]. This enzyme, with activity not only for xylose but also for glucose, is industrially used for the production of high-fructose corn syrup (HFCS) [18] to convert starch — derived glucose into the sweeter sugar fructose to reduce the sugar demand in the food industry. Heterologous expression of bacterial XI genes in S. cere­visiae proved to be challenging, and for many years no actively expressed enzyme was reported [34-39]. The first functionally expressed XI in S. cere­visiae [40] originated from the bacterium Thermus thermophilus [41]. It was later shown that the low activity of the bacterial XIs in yeast could be par­tially related to intracellular precipitation [39], and it was suggested that the rigid nature of the thermotolerant T. thermophilus XI aided correct folding of the protein in S. cerevisiae. However, the activity of this enzyme at 30 °C was too low to allow xylose fermentation. Still, when combined with other genetic modifications, aerobic growth on xylose was demonstrated by S. cerevisiae carrying the T. thermophilus XI [42] (strain TMB3050, Table 2).

More recently, an XI from the obligate anaerobe rumen fungus Piro — myces [20] was expressed in S. cerevisiae with an activity of about 1 U/mg protein at 30 °C [43] (strain RWB202, Tables 3 and 4). Later, bacterial XIs with high sequence similarity to the Piromyces XI, such as those from Bac- teroides thetaiotaomicron [44] and Xanthomonas campestris [45], were also expressed in S. cerevisiae, but the activity of these enzymes in S. cerevisiae was lower than that of the Piromyces XI. Despite the relatively high activity of Piromyces XI in S. cerevisiae, the expression of this enzyme alone did only al­low very slow growth on xylose [43], suggesting that the conversion of xylose to xylulose does not alone control the xylose metabolism in S. cerevisiae [42]. This observation may also set in a new light the failures of early trials for het­erologous XI expression where, in many cases, functional XI expression was only assayed as growth on xylose [35,37].

2.3