Series Editor: T. Scheper

Abstract Industrial biotechnology is the conversion of biomass via biocatalysis, microbial fermentation, or cell culture to produce chemicals, materials, and/or energy. Industrial biotechnology processes aim to be cost-competitive, environmentally favorable, and self­sustaining compared to their petrochemical equivalents. Common to all processes for the production of energy, commodity, added value, or fine chemicals is that raw ma­terials comprise the most significant cost fraction, particularly as operating efficiencies increase through practice and improving technologies. Today, crude petroleum represents the dominant raw material for the energy and chemical sectors worldwide. Within the last 5 years petroleum prices, stability, and supply have increased, decreased, and been threatened, respectively, driving a renewed interest across academic, government, and corporate centers to utilize biomass as an alternative raw material. Specifically, bio-based ethanol as an alternative biofuel has emerged as the single largest biotechnology com­modity, with close to 46 billion L produced worldwide in 2005. Bioethanol is a leading example of how systems biology tools have significantly enhanced metabolic engineer­ing, inverse metabolic engineering, and protein and enzyme engineering strategies. This enhancement stems from method development for measurement, analysis, and data in­tegration of functional genomics, including the transcriptome, proteome, metabolome,

and fluxome. This review will show that future industrial biotechnology process devel­opment will benefit tremendously from the precedent set by bioethanol — that enabling technologies (e. g., systems biology tools) coupled with favorable economic and socio­political driving forces do yield profitable, sustainable, and environmentally responsible processes. Biofuel will continue to be the keystone of any industrial biotechnology-based economy whereby biorefineries leverage common raw materials and unit operations to integrate diverse processes to produce demand-driven product portfolios.

Keywords Bioethanol • Biofuels • Biorefinery • Metabolic engineering • Systems biology

1

Introduction

1.1

The lignin content and type of lignin has a significant effect on the hydro­lysis of various cellulosic substrates as lignin acts as both a physical barrier, restricting access of cellulases to cellulose [68], and as an attractant to cel­lulases, resulting in non-productive binding [69,70]. It has been shown that the chemical and physical structure of lignin plays a significant role in de­termining the magnitude of inhibition it contributes to hydrolysis, and the structure of lignin is heavily dependent on the conditions of the substrate pre­treatment. However, some general observations can be made for substrates treated by specific pretreatments. The main chemical bond linking lignin sub­units is the P-O-4 aryl ether bond [53,54]. As a result, previous work that has examined the structure of lignin in pretreated substrates has mainly observed changes in P-O-4 aryl ethers and the resulting increase in free phenolic groups that occur after P-O-4 cleavage. During SO2-catalyzed steam pretreatment, lignin tends to undergo decreases in both P-5 and P-O-4 aryl ethers [71,72]. Due to the addition of SO2, acid-catalyzed condensation reactions also occur, which are manifested by an increase in the number of aromatics substi­tuted at the C6 [73-75]. It has also been shown that steam pretreatment performed at higher severity results in greater reductions in P-O-4 struc­tures, resulting in more depolymerized lignin and a higher amount of free phenolic groups. Organosolv lignin from a mixed hardwood exhibited sig­nificantly lower amounts of P-O-4 structures than did steam-pretreated lignin from both yellow poplar and aspen [75,76], which is indicative of the greater degree of delignification that occurs during the organosolv process.

Early work showed that exposure of cellulases to soluble lignin obtained from an alkaline organosolv process resulted in reduced hydrolytic activ­ity [77]. Converse et al. [74] reported that the adsorption of cellulases to lignin resulted in decreases in the rate of enzymatic hydrolysis. There have been limited studies investigating the effects of specific lignin functionalities on cellulase activity, however, these studies have concluded that the likelihood of lignin binding cellulases can be linked directly to the presence of specific functional groups. This work is complicated by the fact that subtle changes in pretreatment conditions can result in significant changes in lignin struc­ture [75,76]. Sewalt et al. [78] added powdered lignins to ideal substrates in order to study the effects of lignin structure on cellulose hydrolysis. In the pres­ence of a filter paper substrate and 1.5 mg/mL of lignin, cellulases exhibited reductions in activity of up to 60%. The inhibition by lignin was only moder­ately remedied by increasing the cellulase loading from 5 to 50 FPU/g cellulose, thus indicating that the inhibition resulted from a binding of cellulase to the substrate. It should be noted that these studies added insoluble lignin to the reaction with filter paper, thus the particle sizes of the added lignin should also be considered. The authors concluded that the 6.3% phenolic group content measured in the organosolv lignin compared to 4.3% obtained for the steam- exploded lignin was most likely responsible for the increase in the inhibitory effect of the organosolv lignin. To test this further, the phenolic groups on the added lignin were blocked by hydroxypropylation, which resulted in a virtual elimination of the inhibitory effect of the added lignin.

Sewalt et al. [78] also incubated cellulases with lignin in the absence of substrate, which resulted in a 10-30% decrease in the protein content in the liquid phase, indicating a precipitation with lignin. The authors surmised that the enzyme was bound to lignin, resulting in its deactivation. However, the binding was strictly due to the presence of phenolic groups that mediated the addition of the enzyme to quinone methide groups on lignin. In a recent study, Berlin et al. [18] compared organosolv lignin isolated in the ethanol — soluble stream to the pulp residual lignin isolated by digest by cellulases. Both lignin preparations contained low amounts of P-O-4 and P-5 linkages, in­dicative of their extensive delignification. The dissolved lignin contained 19% more phenolic hydroxyl groups than the isolated residual lignin; however, the residual lignin contained 46% more aliphatic hydroxyl groups and 67% more carboxylic groups. Not surprisingly, the residual lignin was also found to be more condensed than the ethanol-soluble lignin. The incubation of cellulases with the ethanol-soluble lignin, with its higher phenolic content, resulted in a decrease in hydrolytic activity to a greater degree than the enzymatically liberated residual lignin sample. Unlike Sewalt et al, Berlin et al. attributed the difference in the inhibitory effects between the two lignin preparations to the lower amount of carboxylic groups and aliphatic hydroxyl groups of the ethanol-soluble lignin. This may have resulted in a more hydrophobic lignin preparation that was more amenable to hydrophobic interactions with cellu- lases. Unlike previous studies, the particle size of the lignin preparations was considered; however, the amount of cellulases that may have bound to the lignin preparations was not measured. The most likely explanation for the re­sults is that a combination of increased lignin phenolic groups and increased hydrophobicity was responsible for the inhibition of cellulases by the various lignin preparations.

There has also been strong evidence [70, 79] supporting the role of hy­drophobic interactions in the non-productive binding of cellulases to lignin. Multiple studies [70,80,81] have shown that the addition of the surfactant Tween, to cellulolytic hydrolysis improved hydrolysis yields. Similarly, the addition of agents such as BSA (bovine serum albumin) [69,78], gelatin, and PEG (polyethylene glycol) [78] have also reduced the inhibition of cellulases by lignin. It seems safe to assume that during the hydrolysis of lignocellu — losic substrates, the addition of a hydrophobic compound or surfactant to compete with the cellulase proteins for adsorption sites on lignin would re­suit in a reduction in non-productive binding and an increase in hydrolysis performance [79,80]. The surfactant may also facilitate the desorption of cel — lulases that have bound to lignin, similar to the enhancement in cellulase desorption observed during the hydrolysis of pure cellulose substrates in the presence of non-ionic surfactants. Overall, it is apparent that the surfactant, added protein or compound possessing both a hydrophobic and hydrophilic component, aids in reducing the adsorption of cellulases to lignin thereby improving the hydrolysis performance.

Considering the detrimental effect of lignin-enzyme interactions on hydrolysis performance, Berlin et al. [82] introduced a novel approach to enzyme improvement involving a reduction in the affinity of enzymes for lignin rather than an alteration of the substrate. It was shown that natu­rally occurring enzymes with similar catalytic activities tested on “model” substrates such as Avicel and Sigmacell may differ in their interaction with lignin, which may therefore affect performance on the native substrate [82]. As mentioned earlier, Berlin et al. [18] investigated enzyme-lignin interac­tions, and isolated and characterized two lignin preparations from softwood using ethanol organosolv pretreatment. After testing seven different cellulase preparations, three different xylanase preparations and one P-glucosidase preparation, it was shown that the various cellulases differed by up to 3.5- fold in their inhibition by lignin, while the xylanases showed less variability. Moreover, в-glucosidase was least affected by lignin. The authors concluded that the selection or engineering of so-called “weak-lignin-binding enzymes” in the future will offer an alternative means of enzyme improvement [82]. Overall, it has been demonstrated that the presence of lignin presents a sig­nificant obstacle during hydrolysis. However, early work [83] has also shown that hemicellulose removal during pretreatment also results in significant improvements in hydrolysis performance. Hemicelluloses differ significantly from lignin, since their recovery is quite sensitive to changes in processing conditions, and their hydrolysis can potentially be used to fortify recovered sugars to increase ethanol yields in subsequent fermentation [59].

4

In the post-genomic era, increasing efforts have been made to quantitatively describe the relationship between the genome and phenotype of cells. At the interface between the environment and DNA-encoded processes, metabolite levels are quantitative phenotypic indicators that provide an important com­plement to the measurements of mRNA and proteins when studying cellular function. In the same way as for proteomics, where mRNA expression is often assumed to correlate linearly with protein expression and further correlate with protein activity, the false pretence of a one-to-one relationship between all gene expression and metabolite formation exists. In fact, metabolite lev­els may be viewed as the final result of a complex integration of gene ex­pression, RNA translation, post-translational modification, enzyme activity, and pathway regulation [117]. Metabolomics is a burgeoning field produc­ing volumes of data that, like other x-omic data, brings together analytical technology, genomics, bioinformatics, and model construction, and lies at the core of the systems biology agenda [118]. The general idea of metabolomics was first defined several years ago in the field of microbiology [119], and its importance in plant science was pointed out soon thereafter [120]. To­day, metabolomics is also a powerful tool in drug discovery and develop­ment, especially for the identification of drug metabolites and biomarkers for organ-specific toxicities [121,122]. Industrial biotechnology has also be­gun to benefit from integration of metabolomics into the systems biology framework. In metabolic engineering, quantitative metabolomics, by assign­ing function and confirming in silico pathways, could provide a measure of changes in regulatory driving forces and elucidate the impact of changes in enzyme activities on fluxes [123].

Panagiotou and colleagues performed a thorough examination of the metabolome (amino and non-amino acids of the pyruvate, glycine, serine,

threonine, phenylalanine, tyrosine, tryptophan, histidine, glutamine, gluta­mate and dibasic acid metabolism, and the TCA cycle) of Fusarium oxys — porum, a promising microorganism for bioethanol production, in different physiological states [124-127]. They demonstrated that in spite of the di­versity of mechanisms in fungi that regulate the flux of intracellular amino acids, the production of amino acids was closely correlated with the oxy­gen supply and growth medium, while less so to the cultivation phase [126]. By investigating the profile of several intracellular metabolites during culti­vations on glucose and cellulose, metabolic limitations in F. oxysporum that determine the reduced growth rate of this organism compared to other fila­mentous fungi could be pinpointed [125,127]. For example, one of the major drawbacks on the glucose-to-ethanol conversion by F. oxysporum is the for­mation of significant amounts of acetic acid as a by-product. A systematic metabolome analysis suggested that the y-aminobutyric acid (GABA) shunt is active under anaerobic conditions [125]. This led to the suggestion that the high production of acetic acid, which indicates NAD(P)H regeneration, may be associated with a GABA shunt activation since such pathways act as sinks for excess NAD(P)H, e. g., when the TCA cycle is inhibited [128]. Also, a determination of the sugar phosphorylated profiles under aerobic and anaerobic cultivations in order to improve the understanding of slow arabi- nose fermentation by F. oxysporum [126] was performed. The identification of key metabolites in F. oxysporum cultivations uncovered the activation of novel pathways and possible bottlenecks of others, offering specific genetic targets for improved fermentation performance (overexpression of phospho — glucomutase, transaldolase/transketolase).

Metabolomics has not only been used as a tool for identification of targets for metabolic engineering as described above, but also as an all — encompassing approach to understanding total, yet fundamental, changes occurring after subtle genetic perturbations. For example, key intermediates in the pentose phosphate pathway (PPP) and the Entner-Doudoroff pathway (EMP) pathway were analyzed to gain further insight into the metabolism of laboratory and industrial S. cerevisiae strains [129]. The results verified that the profiles of metabolites are indicative of the reference genetic background of the strains and engineered genotype. Devantier et al. (2005) have investi­gated the influence of very high gravity simultaneous saccharification and fer­mentation process conditions on yeast cellular metabolism [130]. Laboratory and industrial S. cerevisiae strains were cultured mimicking fermentation conditions commonly found in the fuel ethanol industry. Concurrently, GC — MS metabolite profiling was performed to determine if there was a metabolic stress response under defined conditions. Metabolite profiling and multivari­ate data analysis was demonstrated by the ability to distinguish strains and fermentation conditions based on intra — and extracellular metabolites. Fur­thermore, the increased energy consumption of stressed cells was reflected in increased intracellular concentrations of pyruvate and related metabolites.

Consequently Villas-Boas and coworkers (2005) used the metabolite profile of S. cerevisiae during very high gravity ethanol fermentation [130] to elucidate un-described metabolic pathways [131]. They proposed a de novo pathway for glycine catabolism and glyoxylate biosynthesis in recombinant S. cerevisiae strains, demonstrating the great potential of coupling metabolomics and iso­tope labeling analysis for pathway reconstructions.

A recent literature review of the applications of metabolome data from mi­croorganisms was summarized by Wang et al. (2006), and included compara­tive metabolite studies, fermentation control, metabolic control analysis and flux studies, and integration of metabolomics for strain improvement [132]. Clearly, metabolomics will have a strong impact on industrial biotechnology in the coming years as one of the corner stones of the systems biology toolbox being applied to metabolic engineering for bioethanol strain improvement.

3.5

The principal components of biomass are: cellulose (~ 30-50%), hemicellu­lose (~ 20-30%) and lignin (~ 20-30%); with minor components of starch, protein and oils. The exact composition of each biomass varies depending both on the plant species and the plant tissue utilized. Table 1 shows a var­iety of substrates in an effort to illustrate the variability of the composition of different substrates. In addition to the variability seen between plant species, work at the US National Renewable Energy Laboratory has demonstrated that even within a single plant species there is considerable variability in compo­sition [6]. Using near infrared spectroscopy, they showed that the total sugar content contributed by cellulose and hemicellulose varied from 45 to 68% of dry mass between 1061 samples of corn stover. Lignin content, which has a direct impact on enzymatic digestibility, varied between 12 and 20%. These differences can be attributed to the genetic background of the corn variety, environmental factors such as weather, location, and pest invasion, and dif­ferences in farming practices.

The substrate characteristics that have been shown to impact the rate of hydrolysis include accessibility, degree of cellulose crystallinity, and the type and distribution of lignin [8]. The presence of lignin in a cellulose-cellulase

reaction is hypothesized to decrease the quantity of the enzyme associated with the cellulose due to nonspecific adsorption of the enzyme to lignin [9] and steric hindrance [10]. Steric hindrance occurs when lignin encapsulates the cellulose and makes it less accessible to enzyme attack [11]. Each of the factors summarized above are known to effect enzyme action and no sin­gle parameter correlates absolutely with enzymatic digestibility. The variation in composition of a given biomass requires some tailoring in the conversion method.

3.2

There is a vast range of materials suitable for the production of ethanol. They can, somewhat arbitrarily, be put into three categories: agricultural, hard­wood and softwood materials. It must be emphasized that it is not always possible to transfer the results from one type of material to another. During the last three decades, many types of materials using various pretreatment methods have been studied. Some hardwood materials, such as poplar, salix or aspen, have been frequently used in various investigations [48-52]. Other

materials examined have been straw [53-58], sugar cane bagasse [59-61] and olive tree wood [62], to mention a few.

In this summary some lignocellulosic materials were chosen for a more in­depth discussion. The materials that are discussed are an agricultural residue (corn stover) for which the hemicellulose is mainly composed of the pen­tose sugar xylose (about 22% xylose, 5.5% arabinose and 3% galactose; all as anhydro-sugars) and a softwood (spruce) where the hemicellulose mainly consists of the hexose sugar mannose (about 12-13% mannose, 5% xylose, 2% galactose and 2% arabinose; all as anhydro-sugars). Table 1 shows the composition of these materials as well as the theoretical amount of ethanol that can be produced from the hexose and the pentose fractions. It is clear that in some cases it is very important to utilize not only the hexose fraction, but also the pentose part of the material.

4.1

Liisa Viikari1 (И) • Marika Alapuranen2 • Terhi Puranen2 •

Jari Vehmaanpera2 • Matti Siika-aho3

University of Helsinki, P. O. Box 27, 00014 Helsinki, Finland liisa. viikari@helsinki. fi

2ROAL, Valta-akseli, 05200 Rajamaki, Finland

3VTT Technical Research Centre of Finland, P. O. Box 1000, 02044 Espoo, Finland

1 Introduction…………………………………………………………………………………… 122

2 Enzymatic Hydrolysis of Cellulose………………………………………………….. 122

3 Thermostable Cellulases…………………………………………………………………. 123

4 Process Concepts……………………………………………………………………………. 127

5 Evaluation of Novel Thermophilic Enzymes; Materials and Methods… 129

6 Composition of the Thermophilic Enzyme Mixtures………………………….. 131

7 Performance of Commercial Fungal Preparations at Elevated Temperatures 132

8 Evaluation of New Thermostable Enzyme Mixtures………………………….. 133

9 Performance of the Thermostable Enzymes at Lower Temperatures…. 137

10 Discussion…………………………………………………………………………………….. 138

References ……………………………………………………………………………………………………. 141

Abstract Thermostable enzymes offer potential benefits in the hydrolysis of lignocellulosic substrates; higher specific activity decreasing the amount of enzymes, enhanced stability allowing improved hydrolysis performance and increased flexibility with respect to pro­cess configurations, all leading to improvement of the overall economy of the process. New thermostable cellulase mixtures were composed of cloned fungal enzymes for hydrolysis ex­periments. Three thermostable cellulases, identified as the most promising enzymes in their categories (cellobiohydrolase, endoglucanase and в-glucosidase), were cloned and produced in Trichoderma reesei and mixed to compose a novel mixture of thermostable cellulases. Thermostable xylanase was added to enzyme preparations used on substrates containing residual hemicellulose. The new optimised thermostable enzyme mixtures were evaluated in high temperature hydrolysis experiments on technical steam pretreated raw materials: spruce and corn stover. The hydrolysis temperature could be increased by about 10-15 °C, as compared with present commercial Trichoderma enzymes. The same degree of hydro­lysis, about 90% of theoretical, measured as individual sugars, could be obtained with the thermostable enzymes at 60 ° C as with the commercial enzymes at 45 ° C. Clearly more effi­cient hydrolysis per assayed FPU unit or per amount of cellobiohydrolase I protein used was

obtained. The maximum FPU activity of the novel enzyme mixture was about 25% higher at the optimum temperature at 65 ° C, as compared with the highest activity of the com­mercial reference enzyme at 60 °C. The results provide a promising basis to produce and formulate improved enzyme products. These products can have high temperature stability in process conditions in the range of 55-60 ° C (with present industrial products at 45-50 ° C) and clearly improved specific activity, essentially decreasing the protein dosage required for an efficient hydrolysis of lignocellulosic substrates. New types of process configurations based on thermostable enzymes are discussed.

Keywords Thermostable • Cellulases • Cellobiohydrolase • Endoglucanase •

P-Glucosidase • Lignocellulose • Hydrolysis

1

Introduction

The present challenge is to substantially increase the production and use of biofuels for the transport sector. In order to reach the future goals of sub­stituting fossil based fuels, it will be necessary to promote the transition towards second generation biofuels. These can be produced from a wider range of feedstock, including lignocellulosic raw materials. Biomass resources can be broadly categorised as agricultural or forestry-based, including secondary sources derived from agro — and wood industries, waste sources and municipal solid wastes. Fuels from lignocellulosic biomass have a higher potential to re­duce greenhouse gas emissions, and hence are an important means to fulfil the CO2 emissions targets, as compared with first generation biofuels. Lignocellu- losic raw materials comprise an abundant source of carbohydrates (cellulose and hemicellulose) for a variety of biofuels, including bioethanol. The conver­sion technologies of lignocellulosic raw materials are, however, more complex and need novel enzyme systems and advanced fermentation technologies. The rate-limiting step in the conversion of cellulose to fuels is hydrolysis, especially the initial attack on the highly ordered, insoluble structure of crystalline cellu­lose. In spite of recent achievements, further developments are still needed to improve the overall economy of the lignocellulose-to-ethanol process. These novel conversion techniques would also be applicable for the production of other sugar platform-based chemicals.

2

The term “industrial biotechnology” first widely appeared in the literature in the early 1980s when genetic engineering, propelled by recombinant DNA technology, was searching for applications beyond health care and medi­cal biotechnology [1, 2]. Today, industrial biotechnology represents a well — defined field with strong academic, government, and corporate representa­tion. Formally, industrial biotechnology is the bioconversion, either via mi­crobial fermentation, cell culture, or biocatalysis, of organic raw materials extracted from biomass or their derivatives to chemicals, materials, and/or energy. Biomass is the result of photosynthetic carbon fixation by plants to form organic polymers that may be enzymatically or chemically digested to carbohydrate, protein, and lipid monomers. Industrial biotechnology, of­ten referred to as white biotechnology [3], aims to provide cost-competitive, environmentally friendly, self-sufficient alternatives to existing or newly pro­posed petrochemical processes. Processes that exploit industrial biotechnol­ogy have recently garnered increasing global attention with traditional petro­chemical processing under scrutiny due to increasing raw material costs, environmental constraints, and decreasing self-sufficiency.

Industrial biotechnology has experienced unprecedented growth with bio-based production processes representing 5% of the total chemical pro­duction sales volume. By 2010, several studies have estimated that the total fraction will increase to 20%, representing $310 billion of a projected total sales volume of $1600 billion. Industrial biotechnology will continue to capture significant sales volume percentages in the arenas of basic chem­icals and commodities (2 to 15%), specialty or added-value chemicals (2 to 20%), and polymers (1 to 15%). However, the greatest percentage gain is likely to occur in the fine chemical market (16 to 60%), where indus­trial biotechnology platforms enable complex chemistries that are presently produced via synthetic or combinatorial routes [4]. Furthermore, indus­trial biotechnology is enabling new products, particularly novel therapeutic agents such as polyketides and specialty chemicals not previously identified, such as the diverse polyunsaturated fatty acids and biopolymers produced by microalgae [5].

The significant increases in fundamental research and development, and commercialization at industrial scales of biotechnological processes may be attributed to several key observations. These observations may be classified and discussed in the context of four broader themes:

1. Petroleum economics in terms of raw material price, stability, and avail­ability

2. Significant technical and scientific achievement within the fields of en — zyme/protein engineering, metabolic engineering, systems biology, pro­cess life-cycle analysis, and process integration

3. Environmental awareness and preservation

4. National energy self-sufficiency and security

Within each of these categories there have been several identifiable and quan­titative drivers fueling the application of industrial biotechnology to pro­cesses previously exclusive to the petrochemical industry.

Figure 1 presents four categories that any industrial biotechnology pro­cess must consider and evaluate prior to development. These areas include process economics, environmental impact, public perception and policy sup­port, and sustainability and self-sufficiency. Figure 2 provides a more focused schematic overview of how modern industrial biotechnology process devel­opment has evolved with the integration of x-ome data.

This review aims to support the hypothesis that industrial biotechnology has benefited from bio-based ethanol production, and that fundamental tools developed previously, but applied in bioethanol development will be applied to future processes. Bioethanol in many cases has served as an industrial proof-of-concept for many biotechnology approaches. In particular, the tools and analysis developed are enabling the vision of a future biorefinery — an integrated process platform that converts biomass-derived feed streams to a diversified portfolio of product streams, adjusted according to market de­mands — to become a reality. Similar to the existing model of a petrochemical refinery, biorefineries will allocate renewable and sustainable raw materials to a diverse array of products, produced by environmentally favorable and cost-effective bioconversions.

1.1.1

Lignocellulosic biomass feedstocks are composed of various types of hemicel­luloses. A recent review by Saha [84] detailed hemicellulose structure, enzy­matic saccharification, fermentation and the production of potentially valu­able products from a hemicellulose hydrolyzate. Hemicelluloses are hetero­geneous, branched polymers of pentoses (xylose, arabinose), hexoses (man­nose, glucose, galactose) and acetylated sugars. Hardwood hemicelluloses are composed mainly of xylan [54]. Xylans from many plant materials are heteropolysaccharides with homopolymeric backbone chains of 1,4-linked в-D-glucopyranose units. Besides xylose, xylan may contain arabinose, glu­curonic acid or its 4-O-methyl ether, and acetic, ferulic, and p-coumaric acids. The frequency and composition of branches depends on the source of xylan [54]. Softwood hemicelluloses consist of mostly galactoglucoman — nans, with a linear or branched chain of 1,4-linked glucose or mannose units. Other softwood hemicelluloses include arabinoglucuronoxylan, ara — binogalactan and others [54]. Within the plant cell wall architecture, hemi­celluloses are thought to coat the cellulose-fibrils resulting in a reduced ac­cessibility of the cellulose-fibrils [83]. Therefore, enzymatic hydrolysis of the hemicelluloses is essential to facilitate complete cellulose degradation. Given the diversity of xylan and mannan structures, a variety of hemicellulases such as endo/exoxylanases, arabinosidases, acetylesterase, glucoronisidases, and mannanases should be required to degrade hemicellulose. Therefore, a pretreatment such as SO2-catalyzed steam pretreatment, which degrades a significant amount of hemicellulose to monomeric sugars, would be invalu­able for their potential utilization in fuel and/or bioproduct production.

4.1

A metabolic flux is defined as a quantitative measurement of the rate of con­version of reactants to products, where rate may be defined as the mass or concentration per unit time of reactant consumption and product formation. For metabolic engineers, flux analysis represents a critical determinant of whether a given strategy has succeeded in re-directing flux from undesired to target products. In the classic textbook, Metabolic engineering: principles and methodologies, the authors note: “The combination of analytical methods to quantify fluxes and their control with molecular biology techniques to implement suggested genetic modifications is the essence of metabolic en­gineering” [133]. There is a significant body of literature describing methods for metabolic flux estimation and measurement, from single, clearly defined, and experimentally determined stoichiometric reactions or sets of reactions for extracellular metabolites, to the more recent in silico estimation of in­tracellular metabolites based upon feeding of isotope-labeled substrates. For more in-depth reviews of methods employed, including their advantages and limitations, please refer to the following references: [133-137]. The focus of this review will be to explore how recent fluxomic data has contributed to improved metabolic engineering strategies for bioethanol production. Fur­thermore, it will reveal that the fluxomic methods developed and data pre­sented thus far have directly contributed to improvements in other industrial biotechnology process development efforts.

Prior to discussion, it should be noted that although the flux data is grow­ing to include measurement and estimation of many fluxes, at present, the re­constructed metabolic network used in flux estimation represents only a frac­tion (less than 5%) of the genome-scale metabolic network. For example, the genome-scale metabolic model for S. cerevisiae contains a total of 708 open reading frames, 584 metabolites, and 1175 reactions [14], while recent work describing the flux network of xylose fermenting S. cerevisiae strains was based on a total of 40 measured fluxes (17 measured fluxes and 23 determined by sum fractional labeling using [1-13C] glucose with 99% abundance) that were then applied to a reconstructed metabolic network model consisting of 37 irreversible reactions, seven reversible reactions, and 24 balanced metabo­lites [138]. Although flux analysis has rapidly been upgraded to fluxome, and continuing expansion of intracellular and extracellular metabolome measure­ment capabilities will enable more flux determinations, the large majority of fluxes have yet to be experimentally determined.

There have been several examples where flux measurements and analysis has significantly contributed to bioethanol strain development, particularly with respect to engineering xylose — and pentose-consuming fermentations. Native S. cerevisiae is incapable of metabolizing xylose and has therefore been an area of very active research and metabolic engineering [114]. For example, Grotkj^r et al. (2005) compared the flux profile of two recombinant S. cere­visiae strains, TMB3001 and CPB. CR4, both expressing xylose reductase (XR) and xylitol dehydrogenase (XDH) from P. stipitis, and the native xylulokinase (XK). CPB. CR4 included a GDH1 deletion and GDH2 being put under a PGK promoter [138]. Expression of XR, XDH, and XK led to highly inefficient xy­lose utilization due to a cofactor imbalance, where excess NADH must be regenerated via xylitol production resulting in reduced ethanol yield. There­fore, metabolic engineering of the ammonium assimilation through deletion of the NADPH-dependent glutamate dehydrogenase (GDH1) and overexpres­sion of the NADH-dependent glutamate dehydrogenase (GDH2) resulted in a 16% higher ethanol yield due to a 44% xylitol reduction [138,139]. Using a reverse metabolic engineering approach, metabolic flux analysis was used to characterize the intracellular fluxes for both strains based on experimen­tal data of anaerobic continuous cultivations using a growth-limited feed of 13C-labeled glucose, confirming that XR activity shifted from being mostly NADPH to partly NADH dependent in the CPB. CR4 strain. Furthermore, the analysis revealed, unexpectedly, activation of the glyoxylate cycle in CPB. CR4, generating the question of whether glyxoylate cycle activation may be pre­ferred for ethanol yield. It was only through flux measurements and analysis that the distribution of carbon believed to have been altered via targeted ge­netics could be confirmed.

In a separate example, again addressing the issue of redox balance result­ing from xylose fermentation, metabolic flux analysis was used to predict a priori that activation of the phosphoketolase pathway (PKP), which leads to the net re-oxidation of one NADH per xylose converted to ethanol, would be preferred [140]. The PKP converts xylose-5-P to acetyl-P and glyceraldehyde — 3-P, enabling the maximum theoretical yield of 0.51 g-ethanol/g-xylose with­out affecting the NADPH/NADH consumption ratio of the XR reaction. A functional PKP was reconstructed in strain TMB3001c and the ethanol yield was increased by 25% due to minimization of xylitol formation; how­ever, metabolic flux analysis predicted that only about 30% of the optimum flux required to eliminate xylitol and glycerol accumulation was present. Further overexpression of PKP; however, lead to increased acetate and a re­duced xylose consumption rate, prompting the investigators to overexpress the acetaldehyde dehydrogenase, ald6. This reduced acetate formation, and produced a strain with 20% higher ethanol yield and a 40% higher xylose consumption rate compared to the reference strain [141]. Metabolic flux an­alysis served two purposes: (i) determination a priori of preferred metabolic engineering targets, and (ii) experimental confirmation of carbon flux distri­butions, neither possible based on visual inspection of biochemical pathways. For a more in-depth review of the recent advances in metabolic engineering of xylose fermentation in S. cerevisiae, specifically focusing on the introduc­tion of a xylose isomerase from Piromyces sp. as a critical milestone in xylose substrate utilization for ethanol production, refer to chapters co-authored by Antonius J. A. van Maris and Barbel Hahn-Hagerdal in this volume.

Fluxome analysis is developing at an accelerated pace, particularly in two areas that will have direct impact on strain development for industrial biotechnology applications. First, significant effort is being dedicated towards developing bioinformatic tools that enable integration of experimentally or in silico-determined flux values with other x-ome data. For example, experimen­tally determined flux values have recently been used as a quality control check of previously generated E. coli genome-scale metabolic models, whereby re­action constituents, direction, or stoichiometry have been revised to reflect in vivo observations [141]. In addition to bridging fluxomics with genomics, integration of transcriptome and fluxome data was also previously discussed under transcriptomics.

The second area of rapid progress is the experimental determination of fluxes. Presently, most isotope labeling experiments are performed in well — controlled stirred tank bioreactors, often at working volumes ranging from

0. 1 to 1.0 L, and many times in a continuous culture mode to ensure both growth and isotopic steady state (i. e., 1-[13C]-glucose feed). These systems, while reliable, are low-throughput and costly to sustain, particularly the large volumes of isotope-labeled substrate required to reach isotopic steady state (generally five residence volumes). Numerous groups are working on en­abling high-throughput flux measurements in small-volume (i. e., 1-100 mL) culture systems. For example, Fischer et al. (2003) reported the development of a novel methodology for 13C-constrained flux balancing applied to data of E. coli cultures fed [U-13C]-glucose and [1-13C]-glucose in shake flasks and 1-mL deep-well microtiter plates [142]. There was excellent agreement of flux values with conventional and comprehensive isotopmer metabolic models [142,143]. In another example, Yang et al. (2006) developed a novel method for metabolic flux studies of central carbon metabolism based ex­clusively on online measurement of carbon dioxide evolved from shake flask cultures of Corynebacterium glutamicum for improving lysine production.

This method, referred to as respirometric 13C flux analysis, was experimen­tally validated in cultures supplemented with [1-13C]-, [6-13C]-, and [1,6- 13C]-glucose, and successfully resolved the major fluxes of central carbon metabolism [144,145]. These examples are on the forefront of enabling high — throughput flux analysis and measurement to become common place among bioprocess development efforts.

3.6

For an industrial process to be economically viable, enzymatic breakdown of lignocellulose to fermentable sugars must occur as quickly as possible, preferably in hours or days. No known enzyme or mix of enzymes are able to accomplish this feat on crude biomass. To make lignocellulose more amenable to breakdown, a wide array of thermochemical pretreatments have been devised [12]. Pretreatment has been described as the second most ex­pensive unit cost in the conversion of lignocellulose to ethanol using enzy­matic hydrolysis [13], after feedstock cost. A wide variety of pretreatments have been extensively described, including comminution [14], steam explo­sion [15], ammonia fiber explosion [16], and acid [17] or alkaline treat­ments [18] with different chemicals [19]. It is not our intent to review these sundry pretreatments, but only to indicate how they differ in terms of their impact on downstream enzymatic hydrolysis.

Pretreatments vary from extremely acidic to quite alkaline, modifying the composition of the biomass and making it more accessible to the enzymes. For example, acidic pretreatments will hydrolyze the majority of the hemi — cellulose while largely leaving the cellulose and lignin intact [20,21]. Dilute acid (0.5-1.0% sulfuric) at moderate temperatures (140-190 °C) hydrolyzes most of the hemicellulose to soluble pentose sugars (both monomers and oligomers), with a concomitant increase in the efficiency of enzymatic cel­lulose digestion [22]. Although very little lignin is solubilized, the lignin is disrupted or redistributed in such a way that enzymatic digestion is en­hanced [23]. Alkaline pretreatments typically solubilize less of the hemicel — lulose and lignin than acidic pretreatments, but modify or redistribute the lignin [24, 25]. Alkaline pretreatments therefore require enzymes that hy­drolyze both cellulose and hemicellulose. Pretreatments differ not only in the degree of hemicellulose depolymerization, but also in the formation of com­pounds such as furfurals, acetate, and other chemicals that may be deleterious to the fermentative organism [26]. The effect of inhibitors released during pretreatment can also inhibit enzyme activity [27,28]. An ideal pretreatment would be inexpensive both in capital and operating costs, create cellulose and hemicellulose substrates that require low enzyme dosages to release the monomeric sugars, generate no hazardous waste, and produce a sugar stream that is fermentable without detoxification. Alterations in the type and severity of pretreatment can have a profound impact on enzyme dosage required for hydrolysis, and therefore on the cost of enzymatic hydrolysis.

As an example, a comparison of enzyme digestibility was made at Novozymes between alkaline pretreated and acid pretreated corn stover using the same enzyme mix composed of T. reesei cellulase (Novozymes’ Cellu — clast 1.5 L) supplemented with cellobiase (Novozym 188) (Fig. 3). For the alkaline pretreatment, two conditions of severity were supplied for analysis, while for the acid pretreated material, one sample was washed exhaustively with water to remove solubles, while the other was simply adjusted to pH 5 with base. Although the samples contained the same cellulose content, large differences were seen in the enzyme loading required to hydrolyze the cel­lulose, with only the cellulose in the washed acid pretreated material being hydrolyzed to completion. The unwashed acid pretreated material was more

resistant to hydrolysis, likely due to the presence of inhibitors that block en­zyme action. While the cellulose in the unwashed stover was hydrolyzed to a greater extent with increasing enzyme dose, both of the alkaline pretreat­ments show a plateau in cellulose hydrolysis, likely due to steric hindrance by unremoved hemicellulose or lignin. Addition of hemicellulase activities can improve the cellulose digestion in these cases, but at an increased cost for the additional activities. Our enzyme mix was optimized for an acid pre­treatment, and better enzyme mixes for both alkaline and acid pretreatments could and should be possible.

3.3