Natural S. cerevisiae strains reduce xylose to xylitol with an endogenous xylose (aldose) reductase encoded by the GRE3 gene [28]. Xylitol strongly inhibits the activity of XI [94], and therefore deleting the GRE3 gene im­proves efficient xylose utilization in XI-expressing S. cerevisiae strains [95]. Improved ethanol yields at the expense of reduced xylitol yields were in­deed observed for XI-carrying strains [96,97] (strain RWB217, Table 1). Furthermore, the GRE3 deletion decreased xylitol formation also in strains carrying XR and XDH, albeit only under continuous fermentation [98] (strain TMB3120, Tables 2 and 3). However, the aldose reductase endoded by GRE3 belongs to a group of generally stress-induced proteins [28] and the deletion of it reduces the growth by 30% [96]. This limits the usefulness of GRE3 deletion in strains aimed at industrial applications.

4.3

Hemicellulose refers to a number of heterogeneous structures, such as (ara — bino)xylan, galacto(gluco)mannan, and xyloglucan [107,108]. These chem-

REF SFI EG1 CEL5 CEL5

PASC

ically diverse polymers are linked together through covalent and hydrogen bonds, as well as being intertwined. Although many pretreatments remove variable amounts of hemicellulosics, it remains imperative from an economic perspective that sugars contained in the hemicellulose fraction of lignocellu- lose are also converted to ethanol [9,36]. The hydrolysis of xylans, the second most abundant sugar polymer in nature, and utilization of xylose, its main constituent, are therefore crucial in a viable CBP configuration.

The cross-linked and partially crystalline nature of the matrix offer great resistance to enzymatic hydrolysis. Furthermore, as the structure of xylan

is variable, involving not only linear P-1,4-linked chains of xylose, but also branched heteropolysaccharides; its degradation requires the synergistic ac­tion of a range of different enzymes [109-111]. To date, hydrolytic enzymes for the cleavage of almost all chemical bonds found in plant structures have been identified in microbial sources (Fig. 2). Hemicellulases, such as в- xylanases and в-mannanases, have drawn attention as they can help facilitate industrial processes such as bleaching in the pulp and paper industry.

There have been several reports of the successful expression of hemi — cellulases in S. cerevisiae (Table 3). Xylan hydrolyzing enzymes such as в- xylanase, в-xylosidase, and auxiliary enzymes such as a-glucuronidase and a-arabinofuranosidase have all been produced successfully in yeast [114,120, 121,128]. The heterologous production of mannanase and a-xylosidase, ac­tive against mannan and xyloglucan, respectively, was also reported [107,124, 127].

Degradation of the e-1,4-xylan backbone requires the action of endo-в — 1,4-xylanases (e-1,4-D-xylan xylanohydrolase EC 3.2.1.8) and в-xylosidases (e-1,4-D-xylan xylohydrolase EC 3.2.1.37) (Fig. 2a). cDNA copies of в- xylanase encoding genes cloned from the yeasts Cryptococcus albidus and Aureobasidium pullulans and the filamentous fungi A. niger, A. kawachii, and

T. reesei have been expressed in S. cerevisiae under transcriptional control of glycolytic promoters [112-114,116,129]. Secreted active enzyme could be as­sayed in all cases. The cDNA copy of the T. reesei в-xylanase II (xyn2) was expressed in S. cerevisiae under the transcriptional control of the PGK1 and ADH2 promoters [114]. Efficient secretion of the heterologous в-xylanase was achieved by the native T. reesei xyn2 secretion signal and the recombinant в-xylanase was 27 kDa in size. The molecular mass of the mature protein in T. reesei was found to be 21 kDa, with virtually no glycosylation. The extra molecular weight of the heterologous Xyn2 protein secreted by S. cerevisiae was shown to be the result of hyperglycosylation of the protein; however, the extra sugar moieties did not influence the activity of the enzyme.

The в-xylosidase encoding gene of Bacillus pumilus (xynB) was cloned from a genomic DNA library and expressed in S. cerevisiae under tran­scriptional control of the S. cerevisiae ADH2 promoter [118]. To promote secretion of the enzyme, the gene was fused in reading frame with the S. cere­visiae mating pheromone a-factor (MFa1) secretion signal. Biologically ac­tive в-xylosidase was obtained, but remained mostly cell associated. When this fusion gene and T. reesei xyn2 were coexpressed in S. cerevisiae under transcriptional control of the S. cerevisiae ADH2 promoter, a 25% increase in the amount of reducing sugars released from birchwood xylan was obtained, compared to strains expressing в-xylanase alone. However, no xylose was produced from birchwood xylan, presumably due to very low в-xylosidase activity. A cDNA copy of the A. niger в-xylosidase encoding gene was sub­sequently cloned [115]. The mature protein encoding region was fused in reading frame with the S. cerevisiae MFa1 secretion signal to ensure secretion

U = micromole substrate released/min, DCW = dry cell weight, NR = not reported; sub­strate used for activity determination is given in parentheses; italics indicate calculation based on assumptions (0.45 g DCW/g glucose, 0.45 g protein/g DCW, 1.3 x 107 cells/g DCW, 1 OD(600) = 0.57 g DCW/L).

BG-xylan = birchwood glucuronoxylan, PNP-|3-X = p-nitrophenyl-|3-D-xylopyranoside, AZCL-XG = azurine-dyed cross-linked xyloglucan, ABIU = aldobiouronic acid, ATRU = aldotriouronic acid, ATEU = aldotetraouronic acid, PNPA = p-nitrophenyl-a-L-arabino — furanoside, LBG = locust bean gum, INM = ivory nut mannan, PGGM = pinewood galactoglucomannan, PNPaGal = p-nitrophenyl-a-D-galactopyranoside from S. cerevisiae and secreted в-xylosidase activity was obtained. When this fusion gene and T. reesei xyn2 were coexpressed in S. cerevisiae, high levels of в-xylanase and в-xylosidase activity were obtained in autoselective strains grown in rich medium. Coproduction of these two enzymes allowed this re­combinant S. cerevisiae strain to convert up to 46% of birchwood xylan to xylose [115].

Using a cell surface engineering system based on a-agglutinin, S. cere­visiae strains displaying the в-xylanase II separately or in combination with the в-xylosidase from Aspergillus oryzae on the cell surface were con­structed [119,130]. When xylan was incubated with high cell concentrations of these strains, HPLC analysis showed that xylose was the main product of the yeast strain codisplaying the в-xylanase and в-xylosidase, while xylobiose and xylotriose were detected as the main products of the yeast strain dis­playing the в-xylanase. Subsequently, a xylan utilizing S. cerevisiae strain was constructed by introducing genes for xylose utilization, specifically, those en­coding xylose reductase and xylitol dehydrogenase from Pichia stipitis and xylulokinase from S. cerevisiae into the strain codisplaying the P-xylanase and в-xylosidase. Ethanol was directly produced from birchwood xylan, and the yield in terms of grams of ethanol per gram of carbohydrate consumed was 0.30 g/g. This strain, though not able to completely degrade xylan and still suffering from the redox imbalance problem during xylose utilization, supports the potential of using S. cerevisiae in a CBP configuration for con­verting xylan to ethanol.

In order to achieve complete degradation of complex substituted xylans, a series of accessory or debranching enzymes are also needed, namely a-D — glucuronidases (EC 3.2.1), a-L-arabinofuranosidases (a-L-arabinofuranoside arabinofuranosidase EC 3.2.1.55), and acetylesterases or acetyl xylan es­terases (EC 3.1.1.6) [131]. Successful expression of an a-glucuronidase in S. cerevisiae was recently reported [120]. The secreted enzyme was active on aldouronic acids from aldobiuronic to aldopentauronic acid. The T. ree — sei a-arabinofuranosidase encoding gene, abfl, was expressed in S. cere­visiae and the resulting enzyme released L-arabinose from p-nitrophenyl-a — L-arabinofuranoside and arabinoxylans [117]. Successful expression of the gene encoding a-L-arabinofuranosidase B (abfB) from A. niger was also shown [121,122].

The major hemicelluloses in softwoods are acetylated galactoglucoman — nans [107]. These consist of a backbone of P-1,4-linked mannose and glucose residues substituted with a-1,6-linked galactosyl side groups. Mannanase (endo-1,4-P-mannanase; mannan endo-1,4-|3-mannosidase; EC 3.2.1.78) ran­domly hydrolyzes the 1,4-^-mannosidic bonds of the main chain of gluco — mannan and galactomannan (Fig. 2c). Endomannanases of T. reesei (man1), Aspergillus aculeatus (man1), and Orpinomyces PC-2 (manA) have all been expressed and secreted in S. cerevisiae [107,108,124]. The secreted en­zymes showed activity toward locust bean gum and ivory nut mannan with the A. aculeatus enzyme exhibiting the highest titer and activity. The en­zyme a-galactosidase (a-D-galactoside galactohydrolase) catalyzes hydrolysis of a-1,6-linked galactosyl residues from galacto(gluco)mannans and simple oligosaccharides such as raffinose and is required for the complete hydro­lysis of galactomannan [125]. Three a-galactosidases, agl1, agl2, and agl3, from T. reesei were cloned and expressed in S. cerevisiae. The recombinant enzymes were able to hydrolyze raffinose, melibiose, and p-nitrophenyl-a-D — galactopyranoside and release galactose from galacto(gluco)mannan.

Xyloglucan is the main hemicellulosic polysaccharide present in the pri­mary cell walls of dicotyledonous plants [127]. It consists of a linear 1,4- в-linked D-glucan backbone that carries a-D-xylosyl, P-D-galactosyl-1,2-a — D-xylosyl, and a-L-fucosyl-1,2-P-D-galactosyl-1,2-a-D-xylosyl side chains at­tached to the OH-6 of в-glucosyl residues. a-Xylosidase releases the unsubsti­tuted side chain xylosyl residue attached to the backbone glucosyl residue sit­uated farthest from the reducing end of the molecule (Fig. 2d). When a gene encoding the Arabidopsis thaliana a-xylosidase (AtXYLl) was expressed in S. cerevisiae, activity could be detected inside the cell [127]. A xyloglucan — specific endo-P-1,4-glucanase from Aspergillus aculeatus was isolated and ex­pressed in yeast [126]. The recombinant enzyme was active in yeast, showing clearing zone formation on azurine-dyed cross-linked xyloglucan containing plates.

As the technologies of pentose sugar utilization and hemicellulase produc­tion in S. cerevisiae mature, integration of these processes and subsequent single-step processing of biomass hemicellulose to commodity products such as ethanol becomes ever more easily envisioned.

6

Numerous pretreatment investigations have been carried out using agricul­tural residues and hardwoods. Softwood, on the other hand, has not been as thoroughly investigated. Table 4 presents a list of studies using softwood

as raw material. As with any type of lignocellulosic starting material it is very difficult to compare the yields from the different investigations. The pretreatment step is usually evaluated using subsequent enzymatic hydro­lysis; different substrate and enzyme concentrations in this step result in overall yields that are difficult to compare. Yields are often reported for a sin­gle step and occasionally no description is given for the yield calculations, which makes comparisons even more difficult. However, investigations on steam pretreatment and dilute-acid pretreatment by Clark et al. [82], Sten — berg et al. [83], Tengborg et al. [84] and Soderstrom et al. [85-87] were all performed in a fairly similar fashion.

One of the most extensive investigations on the softwood Pinus radiata has been performed by Clark and Mackie [82]. They used steam pretreatment covering a temperature range of 148-248 °C, residence times of 0.5-18 min and catalyst concentrations of 0.5-12 wt % SO2 (w/w dry wood). The pre­treatment was assessed by enzymatic hydrolysis of washed solid material at 2 wt % WIS and 20 FPU/g WIS. The sugar yield increased with the impreg­nation concentration of SO2, up to a concentration of 3 wt %. The optimal temperature and residence time, 215 °C and 3 min, were the same for the different concentrations of SO2. The highest sugar yield after steam pretreat­ment and enzymatic hydrolysis was 57-60 g/100 g original dry wood (ODW), corresponding to 80-84% of the theoretically obtainable yield. Enzymatic hydrolysis improved with more severe pretreatment conditions, which also decreased the amount of carbohydrates in the solids. No investigations of the fermentability and the formation of by-products were reported.

Stenberg et al. and Tengborg et al. investigated steam pretreatment of softwood including impregnation with either SO2 or H2SO4 in two different studies [83,84]. In the SO2 study [83] mixed softwood (Picea abies, Pinus sylvestris) was used. The temperature range studied was 190-230 °C, with res­idence times of 2-15 min and SO2 concentrations of 1-6% (w/w) WIS. The pretreatment was assessed by enzymatic hydrolysis on 2 wt % WIS of washed material with a cellulase activity of 15 FPU/g WIS and a ^-glucosidase ac­tivity of 22 IU/gWIS. Fermentation of the liquid after pretreatment was performed with a yeast cell concentration of 9 g DM/L. Increasing severity re­sulted in the release of more sugars during pretreatment as well as a lower fibre yield. The yield in the enzymatic hydrolysis showed an optimum at medium severity, while the overall sugar yield increased with severity in the range studied. The fermentability decreased with increasing temperature. The optimal conditions were at a severity factor around 4.0 (215 °C, 3 min) with an SO2 concentration of 3.5 wt %.

In the study with H2SO4-impregnated spruce [84], pretreatment was per­formed in a temperature range of 180-240 °C, with residence times of 1 -20 min and H2SO4 concentrations of 0.5-4.4 wt %. Evaluation of the pre­treatment was performed in the same way as in the SO2 study. The yield of hexoses indicated that the optimal combined severity (2.3-2.7) for mannose was lower than that for glucose (2.9-3.4). Degradation of sugar increased with harsher pretreatment conditions. Enzymatic hydrolysis showed an optimum glucose yield in the same range of combined severity as that for pretreatment (2.9-3.4). Fermentation of the liquids after pretreatment showed that mate­rial pretreated at a combined severity higher than 3.4 was not fermentable. The highest overall yield of fermentable sugars, 35 g/100 gDM (70%), was obtained at 225 °C, 5 min and 0.5 wt % H2SO4.

Nguyen et al. studied the pretreatment of Douglas fir and Ponderosa pine [88]. Impregnation with 0.4 wt % H2SO4 was used and pretreatment was performed at temperatures from 201 to 231 °C and residence times of 125-305 s. The pretreatment was assayed with enzymatic hydrolysis, SSF and determination of the fermentability. Enzymatic hydrolysis was performed at a solids concentration corresponding to 1 wt % cellulose, and at an enzyme activity of 60 FPU/g cellulose. The overall glucose yield could only be calcu­lated in one case (corresponding to pretreatment conditions of 212 °C and 105 s), resulting in a yield of 80% of the theoretical. The fermentability test showed that samples treated at 230 °C did not ferment, while some of those treated at 215 °C fermented poorly.

Schwald et al. [89] investigated SO2 impregnation prior to steam pre­treatment of Black spruce. Pretreatment was performed at temperatures of 190-220 °C, residence times of 50-250 s and an SO2 concentration of 0.5­5 wt % (dry wood basis). Alkali treatment and H2O2 treatment after steam pretreatment were included in the study. The effects of pretreatment and post-treatment were evaluated with enzymatic hydrolysis ofwashed solid ma­terial at 2 wt % DM and 15 FPU/g substrate. The highest overall sugar yield was 50 g/100 g ODW, but no theoretical yield can be calculated, as the com­position of the raw material was not given. Oligomers, which may have been present in the liquid, were not included in the sugar yield. The alkali extrac­tion decreased the efficiency of enzymatic hydrolysis, while treatment with H2O2 improved it. Sulphur dioxide had a positive effect on enzymatic hydro­lysis up to a concentration of 3.5 wt %. However, the yield in the enzymatic hydrolysis step alone is not a good measure of the overall yield, since loss of sugars in the pretreatment step must be taken into account. By-products were determined, but no fermentation was performed.

In the Lignol organosolv process softwood has successfully been treated, yielding a material that is susceptible to enzymatic hydrolysis and simultan­eous saccharification and fermentation [28]. Ethanol (40-60%) was used with H2SO4 as catalyst at elevated temperatures (around 200 °C) to extract most of the lignin, which was recovered as a precipitate. The enzymatic hydro­lysis yield was higher than 90%. However, the concentration of solid material (spruce/pine/fir) during hydrolysis was only 2%, which may be too low to see the effects of possible inhibitors. The overall sugar yield was not presented.

4.3

Thermostable enzymes are gaining wide industrial and biotechnical inter­est due to the fact that they are more stable and thus generally better suited for harsh process conditions. The concept of thermostability is, however, not very clear, and the thermostability is a relative term. The enzymatic activity is known to increase with increasing temperature up to the tem­perature where inactivation starts to occur [25]. Thermostability is usually defined as the retention of activity after heating at a chosen temperature for a prolonged period. The drawback is that it only measures how well an en­zyme tolerates high temperature and does not take into consideration the number of variables affecting this measurement. The most appropriate way to express thermostability is to measure the half-life of enzyme activity at elevated temperatures. Thermostable enzymes are produced both by ther­mophilic and mesophilic organisms. Although thermophilic microorganisms are a potential source for thermostable enzymes, the majority of industrial thermostable enzymes originate from mesophilic organisms. Thermophilic bacteria have, however, received considerable attention as sources of highly active and thermostable enzymes.

Thermostable enzymes in the hydrolysis of lignocellulosic materials have several potential advantages: higher specific activity (decreasing the amount of enzyme needed), higher stability (allowing elongated hydrolysis times) and increased flexibility for the process configurations. The two first character­istics would expectedly improve the overall performance of the enzymatic hydrolysis even at the range of conventional enzymes active at around 50 °C. Thus, carrying out the hydrolysis at higher temperature would ultimately lead to improved performance, i. e. decreased enzyme dosage and reduced hydro­lysis time and, thus, potentially decreased hydrolysis costs. Thermostable enzymes would expectedly also allow hydrolysis at higher consistency due to lower viscosity at elevated temperatures and allow more flexibility in the process configurations. The characteristics of thermostable cellulases are re­viewed in Table 1. The enzymes are categorised as endo — or exoglucanases, based on the information available.

Several hyperthermostable cellulolytic enzymes have been isolated from various thermophilic bacteria including the anaerobic Thermotoga [11,14,21], Anaerocellum thermophilum [82] and Rhodothermus strains [34]. Signifi­cant research efforts have been invested in the thermophilic bacterial cel — lulosome systems of Clostridia (reviewed by [17]). Concepts for the direct conversion of lignocellulose into ethanol using clostridial co-culture pro­cess have been studied [33]. In addition, thermostable ascomycete cellu — lases have been characterised [30,37,57]. Several mesophilic or moderately thermophilic fungal strains are also known to produce thermostable en­zymes. These enzymes are stable and active at temperatures that are essen­tially higher that the optimum temperatures for the growth of the micro­organism [65]. Some filamentous fungi produce cellulases that retain rela­tively high cellulose-degrading activity at elevated temperatures, particularly those from the species Talaromyces emersonii [27,50,78], Thermoascus au — rantiacus [26,59,70], Chaetomium thermophilum [48], Myceliophthora ther-

mophila [63], Thielavia terrestris and Corynascus thermophilus [45]. Ther­mophilic в-glucosidases have been obtained from e. g. Aureobasidium sp. [66], Chaetomium thermophila [79], Talaromyces emersonii [15], Thermoascus au — rantiacus [23,26,59,70] and Thermomyces lanuginosa [40]. The literature data shows that a number of enzymes are stable at temperatures around 70 °C for elongated periods, but the data does not allow comparison of the proper­ties under similar conditions.

4

Reports on xylose fermentation by recombinant strains in industrial sub­strates are relatively few [6,10,12-14]. Laboratory strains are usually not viable in toxic lignocellulose hydrolysates, and strains with industrial back­ground must be used. In general, xylose fermentation in hydrolysates occurs more slowly than in laboratory media even by industrial strains. For example, the rate of xylose fermentation of strain TMB3400 in dilute-acid spruce hy­drolysate was an order of magnitude lower than that in mineral medium [54]. Another general observation is that very little xylitol is produced when xy­lose in lignocellulose hydrolysates is fermented by XR — and XDH-carrying industrial strains. This is most likely due to the presence of external electron acceptors in industrial media [74-76,144], paradoxically removing the prob­lem of xylitol formation, which has been considered the main drawback of the XR — and XDH-based metabolic engineering strategy.

6

Functional genomics is the quantitative collection, analysis, and integration of whole genome scale data sets that enable biologically relevant and often predictive mathematical models to be constructed. With genome sequences becoming readily available for production organisms, bioethanol process de­velopment has been a benefactor of the scientific achievements in functional genomics, particularly in the areas of transcriptome analysis, proteomics, and fluxomics. Such developments today encompass a systems biology toolbox that may be further exploited for bioethanol and other industrial biotechnol­ogy processes.

This volume will present a diverse collection of technical contributions that aim to provide insight and an update to the state of the art in bioethanol biotechnology. This introductory chapter and the subsequent chapters will be focused on the upstream bioprocess developments and challenges. Topics will include pretreatment of lignocelluloses, enzymatic hydrolysis, enzyme engin­eering, and metabolic engineering of production hosts including S. cerevisiae, Pichia stipitis, and Zymomonas mobilis. The examples cited in this chapter of systems biology tools will draw examples from numerous fermentation or­ganisms; however, with focus on S. cerevisiae.

S. cerevisiae today is the preferred bioethanol production host primarily as a result of proven industrial process robustness and exceptional physi­ological and x-omics characterization [6-9]. The S. cerevisiae genome se­quence, consisting of 6604 total open reading frames (4437 verified; 1343 uncharacterized; 834 dubious) [10], was first made publicly available in 1996 largely through Andre Goffeau’s coordination of the European yeast research community [11]. Soon thereafter, in 1997 and 1998, the first cDNA spotted microarray exploring metabolic gene regulation, and the first commercial

Fig.2 Evolution of Industrial Biotechnology Process Development. A modern industrial ► biotechnology process is composed of five major unit operations: raw material treat­ment, biocatalysis, fermentation, downstream engineering, and process integration. Raw material sources must first be selected based on a set of criteria that includes overall en­ergy balance, availability, abundance, transportation costs, sustainability, self-sufficiency, and associated agricultural costs. Raw materials are then fed through primary and/or secondary treatment unit operations. Primary treatment may include chemical, heat, and/or mechanical force for hydrolysis of lignocelluloses and other relevant biomass ma­terials. Secondary treatment will often include biocatalysis and enzyme treatment to further hydrolyze to monosaccharides. Monosaccharides are then converted to an end-of — fermentation product that is further converted and purified via downstream processing. Throughout this process there are numerous opportunities for energy, water, and in­termediate recycling, along with waste management. These various operations are often lumped together and referred to as process integration. The traditional biotechnology platform, prior to the availability of genome sequences for many production organisms, was based on traditional random mutagenesis, selection, and conventional biochemical engineering solutions. This typically included selection of a high-producing host and then phenomenological optimization of fermentation processes. However, x-ome scale characterization and subsequent bioinformatics has enabled predictive solutions to be en­gineered not only to production organisms, but has also impacted biomass/biofuel crop engineering and enzyme and biocatalyst engineering. This improved approach enables a higher experimental probability of success through in silico prediction. The result is a modern platform for industrial biotechnology process development

platform (Affymetrix) microarray data exploring mitotic cell regulation were reported, respectively [12,13]. The genome sequence coupled with exten­sive annotation based on fundamental biochemistry, peer-review literature, and available transcription data enabled publication of the first genome-scale metabolic model for S. cerevisiae in 2003 [14]. The genome-scale metabolic model represents an integration of extensive amounts of data into an an­notated, defined, and uniform format permitting simulations of engineered genotypes to elicit desired phenotypes [14,15].

Strain development has classically been dominated by random mutagene­sis, largely by chemical mutagens and radiation, of a production host followed by screening and selection in controlled environments for a desired pheno­type. Although this methodology has endured tremendous success, particu­larly in the areas of amino acid production (ь-glutamate, L-lysine) [16-19], antibiotics (penicillin) [20,21], and vitamins (ь-ascorbic acid) [22], it has largely been end-product driven with minimal mechanistic understand­ing. Today, with the exponential increase in genome sequences of exist­ing and future production hosts, coupled with tools from bioinformatics that enable integration and interrogation of x-omic data sets, it is pos­sible to identify high-probability, targeted, genetic strategies to increase yield, titer, and/or productivity [23-25]. It is also now possible to per­form inverse metabolic engineering, where previously successful production systems may be x-omically characterized to elucidate key metabolic path-

ways and control points for future rounds of targeted metabolic engineer­ing [26]. In both forward and reverse metabolic engineering, systems level models and simulations are accelerating bio-based process development, re­sulting in reduced time to commercialization with significantly less resource commitment.

1.1.3

Compared to studies on lignin content, studies evaluating the effect of hemi — cellulose content on the rate and extent of hydrolysis of lignocellulosics have been far less frequent. The lack of studies in this area is most likely due to the sensitivity of the hemicellulose component to the pretreatment conditions. However, it has been widely accepted that, similarly to lignin, hemicellulose acts as a barrier within the lignocellulosic matrix restricting access of cel — lulases to cellulose [11,91,92]. In the case of steam-pretreated substrates, strong evidence supporting the role of hemicellulose in cellulose hydrolysis has been shown on numerous occasions, as increasing solubilization of hemi — cellulose during pretreatment has facilitated subsequent hydrolysis by cellu — lases [26,34,93]. As discussed earlier, the pretreatment process is a balance between maximizing recovery and removal of hemicelluloses from the solid fraction, while minimizing degradation of hemicellulose sugars to fermenta­tion inhibitors. However, especially in the case of softwood substrates, it is challenging to obtain the conditions that minimize hemicellulose degrada­tion while solubilizing sufficient hemicellulose sugars in the solid component to promote the complete degradation of cellulose in subsequent hydrolysis. This point was studied in depth by Boussaid et al. and Wu et al. [34,94] who concluded that during steam pretreatment of Douglas-fir wood chips, higher severities resulted in a near complete solubilization of the hemicellulose with a concomitant increase in enzyme digestibility of the cellulose component; however, this was accompanied by the production of fermentation inhibitors. Due to these factors, researchers utilizing SO2-catalyzed steam explosion pretreatments have focused on employing a medium severity pretreatment, which improves the enzymatic hydrolyzability of the solids and results in the recovery of the majority of hemicellulose in a monomeric form within the water-soluble stream [23,95]. Although pretreatment and hydrolysis studies have strongly suggested that hemicellulose hinders the hydrolysis of lignocel — lulosic substrates, further evidence has been presented in studies that have employed cellulases in combination with hemicellulases to hydrolyze pre­treated substrates.

In nature, woody lignocellulosics are degraded by microorganisms such as white-rot fungi, which are capable of degrading the entire wood struc­ture, utilizing the combined activities of a host of enzyme such as cellulases, hemicellulases, pectinases, lignin peroxidases, manganese peroxidases, and laccases [12]. Therefore, it is to be expected that the presence of periph­eral materials such as hemicellulose and lignin would be an obstruction when applying cellulases exclusively for the hydrolysis of lignocellulosic sub­strates. Indeed, recent work by Berlin et al. compared the hydrolysis per­formance of seven fungal cellulase preparations from both Trichoderma and Penicillium on steam-pretreated softwoods and organosolv-pretreated soft­wood and hardwood substrates, only to find that the differences in perform­ance among the enzyme preparations was heavily dependent on the level of в-glucosidase supplementation [96,97]. Consequently the P-glucosidase added to the hydrolysis reaction also contained xylanases, which according to the authors, most likely facilitated the cellulose hydrolysis by removing xylan, thus improving accessibility of the substrate to cellulases. Similar results have been shown when combining cellulase with xylanase for treating pulps used in papermaking, as xylanases have been shown to act synergistically with cellulases to improve paper properties. In the case of Douglas-fir Kraft pulp fibers, cellulase-xylanase combinations were shown to improve fiber/paper strength properties [98], most likely by improving accessibility to cellulases by the removal of xylan and its substituents as well as any associated low molecular weight lignin fragments by the xylanases [99]. Likewise, cellulases have been shown to improve the accessibility of xylanases to softwood Kraft pulps as mild cellulase pretreatments increased the apparent median pore size of the pulp to facilitate subsequent prebleaching by xylanases [100].

From the discussion thus far it is apparent that the pretreatment methods currently being advocated for their potential application in bioconversion processes have a significant effect on the proportions of lignin cellulose and hemicellulose within the substrate, while the presence of hemicellulose and lignin plays a significant role in affecting the ease of enzymatic digestibil­ity of lignocellulosics. However, the effects of lignin and hemicellulose are only part of the “hydrolysis puzzle”, as it is likely that changes in the primary components of lignocellulose also have significant effects on the physical and chemical characteristics of the substrate.

5

Since S. cerevisiae cannot utilize xylose, but does utilize and ferment its iso­mer D-xylulose [1,2], the obvious first step to allow xylose metabolism is to introduce a heterologous pathway converting xylose to xylulose. Over the years, several approaches have been explored to express a pentose utilization pathway from naturally pentose-utilizing bacteria and fungi in S. cerevisiae. Figure 1 summarizes the initial pathways for D-xylose utilization in bacteria and fungi.

Ul

О

NR: not reported

a Batch culture by pulsing a chemostat culture b Calculated from reference c Ethanol yield on xylose d Calculated after glucose depletion

NR: not reported a Calculated from reference

NR: not reported

a Measured and corrected by closing the DR balance b Calculated from reference

2.1

6.1

Evolutionary Engineering of D-Xylose-Consuming S. cerevisiae for Improved Mixed Substrate Utilisation

The sub-optimal kinetics of mixed-substrate utilisation by the genetically engineered XylA-expressing strain RWB 217 [43] suggested a low affinity (qmax/Ks) for D-xylose. Soon after the invention of the chemostat it was al­ready established that prolonged cultivation in nutrient-limited chemostats leads to selection of spontaneous mutants with an improved affinity for the growth-limiting nutrient [52,53]. This principle, which has since been demonstrated for many micro-organisms and nutrients [40,58,72,73] was applied to improve the affinity of S. cerevisiae RWB 217 for D-xylose [44].

Indeed, during prolonged anaerobic D-xylose-limited chemostat cultivation at a dilution rate of 0.06 h-1, the residual D-xylose concentration decreased threefold, indicating that cells with improved affinity for D-xylose were se­lected for [44]. After 1000 h (85 generations) of this directed evolution in chemostat cultures, single-colony isolates were tested for batch growth on a mixture of glucose and D-xylose. Although the fermentation kinetics of some of these single-cell lines, as evaluated by carbon dioxide production profiles, were already drastically improved relative to the parental strain (Fig. 6), the D-xylose phase remained slower than anticipated based on batch cultivation on D-xylose alone. A further 85 generations of chemostat cultivation resulted in only marginal improvement of the D-xylose consumption characteristics.

To select for further improvement of D-xylose fermentation kinetics, an additional evolutionary engineering strategy was applied, which involved sequential anaerobic batch cultivation on glucose-xylose mixtures [44]. To maximise the number of generations that the cells grow on D-xylose, the D-xylose concentration in the cultures was raised to 90 gL-1, with a glucose concentration of 20gL-1. After 20 cycles, the evolved culture was capable of complete anaerobic conversion of a mixture of 20 g L-1 glucose and 20 g L-1 D-xylose in about 20 h, with an inoculum size of 5% (v/v) [44].

Characterisation of the resulting strain RWB 218 (derived from single colony isolate) showed that D-xylose consumption followed the consump­tion of glucose rapidly (Fig. 7). The D-xylose consumption rate observed in these cultures was 0.9 g D-xylose (gdryweight)-1 h-1. This evolved Xl-based strain, in contrast to strains based on xylose reductase and xylitol dehydroge­nase, produced only 0.45 mM of xylitol, indicating that redox imbalance does

not occur during alcoholic fermentation of D-xylose. The ethanol yield on total sugar in batch cultures co-fermenting glucose and D-xylose was typic­ally 0.40 g g-1, which is identical to the ethanol yield that would be achieved in glucose-grown cultures in a similar set-up. Even when tested in more concen­trated sugar mixtures (100 g L-1 glucose and 25 g L-1 D-xylose), resembling an industrial situation, this strain consumed both sugars within 24 h, starting with 1.1 gL-1 yeast dry weight as the inoculum [44].

With evolutionary engineering as a proven tool for obtaining (yeast) strains with improved properties, a full understanding of the underlying molecular changes becomes the next challenge. In an attempt to unravel the changes between the original metabolically engineered and the subsequently evolved Piromyces XI-based strains, anaerobic chemostat cultivations on D-xylose as the sole carbon source were used as the basis for transcriptome analysis with Affymetrix DNA arrays (J. T. Pronk, unpublished data). The most striking observation amongst the genes with a changed transcript level was the repre­sentation of various members of the hexose transport family, including HXT1, HXT2 and HXT4. Interestingly, HXT1 and HXT4 have been associated with D-xylose transport in previous studies [27,62]. To investigate whether the improved fermentation characteristics were indeed due to changes in sugar transport, zero trans-influx assays were performed using both the strain that was only metabolically engineered and the subsequently evolved strain [44]. The D-xylose uptake kinetics obtained for the metabolically engineered strain (Km 132 mM, Vmax 15.8 mmol (gdryweight)-1 h-1) were in agreement with other studies [22,39]. Strikingly, the D-xylose uptake kinetics of the evolved strain had changed drastically, with a 25% reduction in the Km (to 99 mM) and a twofold increase of Vmax to 32 mmol (g dryweight)-1 h-1.

6.2

The focus of this review has been twofold. First, to present a summary of the economic and socio-political landscape that has fueled the resurgence in bioethanol as a biofuel, and consequently, the general adoption of industrial biotechnology as a cost-effective, sustainable, and preferred alternative to traditional petrochemical processing. Second, to offer the hypothesis that sig­nificant scientific achievements in metabolic engineering and systems biology have been applied to bioethanol and other chemical products for successful commercialization, suggesting a graduation of the field to industrial systems biology. If we revert back to Fig. 1, we have focused most of our attention on process economics, with some indirect discussion of environmental im­pact and sustainability/self-sufficiency. Within the discussion of industrial systems biology we have focused only on the upstream bioprocessing steps schematically shown in Fig. 2, namely, systems biology used for enhance­ment of metabolic engineering. An area that we have not discussed, but is addressed in the chapter co-authored by Warren E. Mabee in this volume, and suggested in Fig. 1, is public perception and policy.

In the July 2006 issue of Nature Biotechnology there were a series of edito­rials and commentaries written exploring bioethanol as a biofuel [146-150]. Amongst these articles was a discussion of the various public perception is­sues facing bioethanol, ranging from statements of support, such as, “At least three major factors — rapidly increasing atmospheric CO2 levels, dwindling fossil fuel reserves and their rising costs — suggest that we now need to ac­celerate research plans to make greater use of plant-based biomass for energy production and as a chemical feedstock as part of a sustainable energy econ­omy” [149], to highly critical statements, such as, “At present, ethanol is not price competitive by any stretch of the imagination — even with the absurd and decidedly anti-free-market 54-cent per gallon tariff Washington imposes upon Brazilian ethanol” [147]. Both in the scientific peer-review and main­stream literature, there is still debate as to whether bioethanol for biofuel makes sense. This debate has prompted the development of new methods for analysis of whether bioethanol is economically feasible, and more impor­tantly, sustainable over the long-term. A general classification often used to evaluate process feasibility is life-cycle analysis.

Life-cycle energy analysis is a methodology used to answer the bottom­line question: is more energy contained in the fuel than is used in the production of that fuel? Life-cycle energy analysis, much like the tools em­ployed in functional genomics, is a systems approach to evaluate all aspects of the production process from feedstock processing, availability, and trans­portation, to opportunities for recycling of energy and mass pre-, during, and post-production [151-153]. Life-cycle energy analysis, unlike earlier ap­proaches, has suggested that process integration, energy recycling, and care­ful selection of raw materials and unit operations can yield bioethanol pro­cesses that are energetically favorable [151-153]. Consequently, biorefineries are viewed as a natural extension of bioethanol production facilities given the opportunities for integration, recycling, and production of higher value chemicals.

Holistic approaches that take a systems level approach must be refined, im­proved, and presented to policy makers and key stakeholders to finally put an end to the question — does bioethanol make sense? The question should be revised to ask under which conditions does bioethanol makes sense, and what is required to commercialize those conditions. Bioethanol, biorefineries, and industrial biotechnology will not be successful and expand to novel areas if we do not focus on engineering public perception and policy. Bioethanol has become a commercial vehicle for industrial systems biology applied to in­dustrial biotechnology. Now, it needs to become the commercial vehicle for reaching public, government, and corporate support, and again, it will take systems level data to get there.

Acknowledgements J. M. Otero is a Merck Doctoral Fellow and acknowledges financial support from the Division of Bioprocess Research & Development, Merck Research Labs, Merck & Co., Inc.

G. Panagiotou, PhD, acknowledges financial support from the Villum Kann Rasmussen foundation.