The expression of a cofactor-independent, heterologous XI is the solution for bypassing the intrinsic redox constraints of the XR/XDH approach. Suc­cessful implementation, however, requires an in vivo activity of XI similar to that of key glycolytic enzymes such as hexokinase and phosphofructokinase. In practice, this corresponds to an activity, under physiological conditions, of 0.5-1.0 ^mol D-xylose converted per milligram soluble cell protein per minute [68]. The apparent simplicity of this objective turned out to be decep­tive. In fact, studies on the functional expression of heterologous structural genes for XI in S. cerevisiae now spans roughly two decades.

Expression in S. cerevisiae of the E. coli xylA gene (which clusters with the XI genes from other Proteobacteria, Fig. 3), resulted in no [13] or very low in vitro XI activities [59]. Sarthy et al. (1987) showed that, while the E. coli XylA protein was produced in S. cerevisiae, its specific activity was three orders of magnitude below that of XylA protein produced in E. coli [59]. Improper protein folding, sub-optimal intracellular pH, post-translational modification, inter — or intramolecular disulfide bridge formation and a lack of specific cofactors or metal ions in S. cerevisiae were mentioned as possible causes [59]. However, no single factor was identified that could explain the low activity, and attempts to increase E. coli XI expression levels in S. cere­visiae were unsuccessful [59]. Subsequently, attempts were made to express XI-encoding genes from other prokaryotic phyla. Attempts to express XI genes from Clostridium thermosulfurogenes [48], Bacillus subtilis or Actino — planes missouriensis [1], which originate from different prokaryotic phyla (Fig. 3), also failed to result in the production of an active XI enzyme in S. cerevisiae.

The first study that achieved significant activities of a heterologous XI enzyme in S. cerevisiae was based on expression of the XI gene from the thermophile Thermus thermophilus [70]. Indeed, an enzyme activity of up to 1.0 ^mol(mg protein)-1 min-1 was found in cell extracts of the engineered S. cerevisiae strain. However, this activity was assayed at the optimum tem­perature for activity of the T. thermophilus XI of 85 °C, which is not com­patible with yeast growth or survival. At 30 °C, the optimum temperature for growth of S. cerevisiae, activity was only 0.04 ^mol (mg protein)-1 min-1 [70]. Although subsequent random mutagenesis resulted in variants of the T. ther­mophilus XI with improved temperature characteristics [26,47], in vivo en­zyme activities of the T. thermophilus XI in S. cerevisiae strains remained too low to sustain rapid anaerobic growth on D-xylose ( [35], see Sect. 5).

A breakthrough came with the discovery of a XI in an unicellular eu­karyote, the anaerobic fungus Piromyces sp. E2 [28]. Expression of this Piromyces xylA gene in S. cerevisiae resulted in high enzyme activities (up to

1.1 ^mol(mg protein)-1 min-1 at 30 °C [42].

The molecular basis for the high functional expression levels obtained with the Piromyces xylA gene remains unclear. We have recently expressed the XI sequence from Bacteroides thetaiotaomicron into S. cerevisiae. This prokary­otic sequence is 83% identical and 88% similar to the Piromyces xylA gene. S. cerevisiae strains expressing this prokaryotic XI can utilise D-xylose, albeit

at a somewhat lower rate than similar strains expressing the Piromyces xylA gene (A. A. Winkler et al. unpublished). This indicates that its probable evolu­tionary history (horizontal gene transfer followed by evolutionary adaptation to a eukaryotic host) may not be the sole factor in the successful expression of the Piromyces enzyme.

In terms of GC content and codon usage, the Piromyces xylA gene appears to have favourable characteristics for expression in S. cerevisiae. At 45%, its GC content is much closer to that of S. cerevisiae (39%), than that of, for ex­ample, the T. thermophilus gene (little over 64% GC). Also the high codon bias index of the Piromyces gene for expression in S. cerevisiae (0.642 versus — 0.018 for the T. thermophilus gene) may contribute to its efficient expression. Future structure-function studies will likely identify critical factors for high- level functional expression in yeast, in the S. cerevisiae genome as well as in the sequence of heterologous XI genes. However, while of great scientific interest, innovation in D-xylose fermentation is no longer dependent on such research, as the availability of the Piromyces xylA gene has paved the way for metabolic engineering of S. cerevisiae for anaerobic fermentation of D-xylose to ethanol. Recent progress in this area will be discussed in the following paragraphs.

4

The use of polylactic acid (PLA) as a biodegradable carbohydrate-based plas­tic is rapidly expanding in many areas such as food packaging, drug deliv­ery, textiles, medical implants, and cosmetics [111-114]. Both the physical properties and the rate of biodegradation can be controlled by adjusting the ratio of the blended enantiomers, D(-)-lactate and L(+)-lactate [115]. For decades, lactic acid bacteria have been used to produce optically pure d(-)- and L(+)-lactate. However, high costs due to the need for complex media and the inability to ferment a broad range of sugars have con­strained the use of PLA to the manufacture of medical grade sutures and implants. Alternative biocatalysts from a variety of organisms are currently being investigated for efficient and inexpensive production of optically pure isomers [116-121].

Biocatalysts derived from E. coli K-12 had been previously engineered to produce D(-)-lactate but these were not able to metabolize 10% glucose or sucrose to completion in rich or minimal media [122-124]. The success and robustness of E. coli W derived-ethanologenic KO11 prompted the redirection of metabolism in this organism from ethanol to D(-)-lactate production [39]. Elimination of adhE, ackA, and the Z. mobilis homoethanol pathway from KO11 yielded strain SZ110 [39]. SZ110 was subjected to metabolic evolution in LB 100 g L-1 glucose [39], mineral salts medium with 100 g L-1 sucrose, and mineral salts medium with 100 gL-1 glucose, along with genetic manipula­tions to reduce co-product formation and remove foreign genes, to ultimately generate D(-)-lactate-producer SZ194 [125]. Replacement of the native SZ194

IdhA gene with the Pediococcus acidilactici IdhL gene and further metabolic evolution in mineral salts medium with glucose resulted in L(+)-lactate pro­ducer strain TG103 [121]. Both SZ194 and TG103 produced 1.2 M lactate from 12% glucose in mineral salts medium supplemented with 1 mM betaine. How­ever, lactate optical purity decreased from 99.5% to 95% in the presence of betaine [42,121]. This chiral impurity was associated with high glycolytic flux rates; spillover of carbon to lactic acid through the methylglyoxal pathway was the source of the contamination [126-128]. Elimination of the first com­mitted enzyme of the methylglyoxal pathway (mgsA) restored the product optimal purity to close to 100% [121]. The resulting E. coli strains, TG114 and TG108, consistently produced high titers of greater than 99.9% chirally pure d(-)- and L(+)-lactate, respectively, from 12% glucose at greater than 95% of the theoretical yield [121]. The lactate titer, yield, and optical purity attained by TG114 and TG108 are the highest compared to other lactate-producing organisms and were achieved in simpler fermentation medium and condi­tion, making these microbial biocatalysts some of the most efficient lactate producers.

5.2

Process simulations cannot replace experiments, but constitute a useful tool in the planning and evaluation of experiments. Furthermore, they highlight factors that are sometimes neglected in experimental studies, for example, the amounts of chemicals needed in the process (catalyst in pretreatment, acid/base for pH adjustment, nutrients and not least enzymes and yeast), which constitute a significant contribution to the production cost. The over­all demand of steam, process water and cooling water are other important factors.

Optimization of ethanol production from lignocellulosic feedstock requires a model that includes all the major process steps, since changing the condi­tions in one process step is likely to affect other parts of the process. Although no full-scale plant based on enzymatic hydrolysis has yet been built, most of the process steps (e. g. distillation, evaporation, drying and incineration) are considered to be technically mature, i. e. their operational performance is well known. Of course, the application of these unit operations in a lignocellulose — to-ethanol plant still requires to be verified on pilot scale before a full-scale plant can be constructed. However, the ethanol process includes other process steps, which are associated with greater uncertainties regarding design and performance on full scale. This is definitely true for the pretreatment step, ir­respective of the pretreatment method chosen, or how it is configured (Galbe, in this volume). It also applies to enzymatic hydrolysis or SSF at high solids concentrations, as well as solid-liquid separation of the stillage.

The modeling of a lignocellulose-to-ethanol process poses a number of unique challenges. In contrast to well-defined systems, such as pure ethanol — water systems, it involves not only vapor and liquid phases but also a solid phase, including atypical compounds like cellulose, lignin and yeast. There­fore, when simulating such a process it is necessary to use a flowsheeting program that is able to handle solid components. In most techno-economic evaluations of the lignocellulose-to-ethanol process that have been performed during the past 10 years, Aspen Plus from Aspen Technologies has been used [16—20]. In Aspen Plus, a separate solid stream is used that does not interact with the liquid phase and never ends up in the vapor phase. There is thus no need to estimate vapor phase data such as heat of vaporization or vapor pressure for components treated as solids (lignin, glucan, yeast, etc.).

3

4.2.1

Cofactor Dependence

The production of xylitol and L-arabitol during pentose consumption by natural as well as recombinant pentose-utilizing yeasts has been rational­ized with the difference in cofactor preferences between the enzymes in the initial pentose utilization pathways. XR from P stipitis preferentially uses NADPH, but can also use NADH as a cofactor [23], whereas XDH exclu­sively uses NAD+ [24,46]. This may result in excess NADH formation and lack of NAD+, since yeasts do not harbor a transhydrogenase enzyme that would allow direct conversion of NADP+ to NAD+ [32]. Numerous investiga­tions have supported this metabolic model; external electron acceptors, which are reduced by NADH-dependent enzymes in S. cerevisiae, reduce xylitol for­mation [32,74-76]. Xylitol formation in recombinant S. cerevisiae was also reduced by changing the kinetic properties of the enzymes involved by ex­pressing a fusion protein of XR and XDH [87], or by expressing mutated XR with altered cofactor affinity [88,89] (strains TMB3270, TMB3271, R267H, Ta­bles 1-3). However, significantly increased ethanolic xylose fermentation as a result of such engineering strategies has been less frequently reported [89].

4.2.2

The composition of plant biomass can vary substantially but all plant biomass is composed of four major polymeric compounds: cellulose (~ 33-51%), hemicellulose (~ 19-34%), pectin (~ 2-20%), and lignin (~ 20-30%) [16, 31]. Upon hydrolysis, plant biomass yields a variety of fermentable hexoses (glucose, 36-50%; mannose, 0.3-12%; galactose, 0.1-2.4%) and pentoses (xy­lose, 3.4-23%; arabinose, 1.1-4.5%). S. cerevisiae can ferment all the hexoses to ethanol, but not the pentoses, which can be a significant portion (25%) of, for example, sugarcane bagasse, a preferred feedstock for bioethanol pro­duction [32]. More than three decades of research have been devoted to the development of yeast for efficient xylose fermentation, initiating with the search for alternative yeasts, such as Pachysolen tannophilus, Pichia stipitis, and Candida shehatae, and in the last two decades focusing on the genetic en­gineering of S. cerevisiae to utilize xylose and arabinose [33]. Please refer to the chapters by Hahn-Hagerdal et al. and van Maris et al. (in this volume) for detailed reviews of this topic [34,35].

After several unsuccessful attempts to produce a functional bacterial xy­lose isomerase in S. cerevisiae, many groups focused for the last decade on efficient expression of fungal xylose utilizing genes and manipulating the pentose phosphate pathway to enhance xylose utilization and fermentation in S. cerevisiae [36]. These research efforts ensured steady but slow progress toward the development of xylose utilizing S. cerevisiae strains, and it was recent successes with the production of a functional Piromyces sp. xylose iso — merase in recombinant S. cerevisiae that opened the way for efficient xylose fermentation by S. cerevisiae at low oxygen levels [37,38]. The main advan­tage of this approach is the circumvention of the redox imbalance problem created by expressing xylose reductase (XYL1) and xylitol dehydrogenase (XYL2) fungal genes in S. cerevisiae. Recognizing the need for S. cerevisiae to ferment all hexoses and pentoses produced during enzymatic hydrolysis of wood, both laboratory and industrial S. cerevisiae platform strains have been developed that can utilize xylose [39] and later co-utilize xylose and arabinose [40].

Apart from monosaccharides, S. cerevisiae can utilize the disaccharides su­crose and maltose, and some Saccharomyces strains can also utilize melibiose and the trisaccharides maltotriose and raffinose [41,42]. However, the major end products of cellulose hydrolysis are cellobiose and cellooligosaccharides, which cannot be utilized by S. cerevisiae. The heterologous expression of four different в-glucosidases in S. cerevisiae was evaluated and the P-glucosidase (BGL1) of Saccharomycopsis fibuligera was found to be produced at the high­est activity levels [43]. Expression of the в-glucosidases encoding genes of Candida wickerhamii, Aspergillus kawachii, and T. reesei yielded activities at least one order of magnitude lower than that of Saccharomycopsis fibuligera. It was shown that multicopy expression of the S. fibuligera BGL1 gene could enable growth on cellobiose as sole carbon source at a rate equivalent to that found on glucose [43,44]. Recently, a S. cerevisiae strain was developed that could utilize both xylose and cellobiose [45].

Even with the introduction of pentose and cellobiose utilizing genes, S. cerevisiae strains preferentially utilize glucose before the other mono — and disaccharides. Deregulation of the strong glucose repression effect in S. cerevisiae would be required to allow cometabolism of sugars derived from plant biomass for high ethanol productivity. Disrupting both the MIG1 and MIG2 genes allowed cometabolism of glucose and sucrose [46], and simi­lar strategies could be used to allow co-utilization of sugars released from plant biomass. Furthermore, simultaneous co-transport of glucose and xylose must be facilitated as the delayed utilization of xylose (in a recombinant xy­lose utilizing strain) is in part an effect of competition for the same glucose transporters in the absence of a xylose-specific transporter [37].

4

On the basis of earlier kinetic modelling for the conversion of glucose to ethanol by wild-type Z. mobilis [59], a further model has been developed for the fermentation of glucose/xylose mixtures by ZM4 (pZB5) [60]. A two — substrate model was constructed based on Monod kinetics for substrate lim­itation, as well as functions for product (ethanol) inhibition and substrate inhibition at the higher glucose and xylose concentrations. The model simu­lation data for various glucose/xylose concentrations were compared with the experimental results using a Microsoft Excel-based program and statistical analysis for error minimization. Using this approach, it was established that the model (with relevant values of the constants) provided good agreement with the experimental batch culture data for 25/25, 50/50 and 65/65 g L-1 glu — cose/xylose media. It should be noted that the model did not include any repression of xylose uptake by glucose as experimentally both glucose and xylose are taken up simultaneously even at the high initial glucose concen­trations. However, this does not preclude the possibility that some glucose repression of xylose might be occurring. The results indicate that ethanol inhibition of xylose utilization is likely to be the more dominant factor in influencing its kinetics.

2.6

Another common policy instrument to promote biofuel use and consump­tion is exemption from excise taxes or mineral spirits taxes. Excise taxes are commonly used in the transportation sector and are designed to fill the gap between property and income taxes. These types of taxes can be imposed on the sale or use of certain articles, including fuels, and on certain transac­tions and occupations. In many cases, these taxes are not itemized in sales receipts and cannot be easily detected, and thus result in a hidden cost to the consumer [81,82]. As shown previously in Table 1, excise tax rates for the countries under consideration range considerably.

In North America, excise taxes have been used as a tool to support renew­able biofuels for some time. The federal governments of both Canada and the USA offer an exemption on bioethanol, which results in a slightly reduced tax rate for E10 blends. In addition, some state and provincial governments also offer exemptions. The largest North American exemption on excise taxes is currently offered in Manitoba, although that status is dependent upon the value of Canadian and American currency.

In Fig. 4, the excise tax exemptions are shown for the USA, and are re­lated to bioethanol production capacity. The federal and state exemptions are illustrated by the shading on the map, with blue indicating the base fed­eral exemption, and shades of red from light to dark indicating increasingly higher state-level exemptions. Bioethanol production capacity in 2005 is in­dicated by the size of the yellow circles, increasing on a logarithmic scale as shown in the legend. Expected additional bioethanol capacity for 2007 is shown by the dark orange circles.

In the USA, Idaho offers the largest combined exemption on E10 fuels at US $ 0.021 L-1, but has no active production of bioethanol. Of the largest bioethanol-producing states, South Dakota and Iowa are the only two produc­ing states that offer an additional exemption on state excise taxes. It may be inferred that excise tax exemptions provide a benefit for producers, but are

■ >$0 20 SO 16-SO 20 $0 14-SO 16

SO 14 (base US Federal exemption)

■ SO 073 (base Canadian Federal exemption)

Fig. 4 Geographic distribution of North American federal and state/provincial-level ex­cise tax exemptions (2005), existing bioethanol production capacity (2005), and projected bioethanol production capacity (2007) [15,17,21,22,69,81] not the deciding factor in determining where to install capacity for produc­tion.

Similarly, exemptions on excise taxes cannot be simply related to bio­ethanol production in Canada or Europe. In Canada, Manitoba offers com­bined exemptions that are higher than any offered in the USA. Combined federal and provincial excise tax exemptions on E10 reach as high as US$ 0.0256 L-1 in Manitoba, as compared to US$ 0.0181 L-1 in Ontario. At the current time, however, Ontario continues to lead Canada in the amount of bioethanol produced, while Manitoba currently lags behind jurisdictions such as Saskatchewan (which has individual incentives) and Quebec (where exemptions are limited to the federal level). In Europe, high excise taxes mean that exemptions for bioethanol (and other biofuels) are very significant and orders of magnitude larger than those found in North America. France offers the largest incentive in the form of tax exemptions, but has focused pro­duction of biofuels on ETBE, while Spain produces a significant amount of bioethanol under a significantly lower excise exemption regime, as indicated in Table 1 [35,43].

In Fig. 5, the level of excise tax exemptions are plotted against bioethanol production capacity and the correlation between the two is examined, using bioethanol production capacities for 2003 and 2005. The two years of data are differentiated by the shaded and white circles. In 2003, no correlation was found between state-level excise tax exemptions and bioethanol pro­duction capacity (r2 = 0.01). This may be evidence that the federal level exemption, which applies to all states, is a sufficient incentive for produc­ers, and that additional incentives are not required to spur development of bioethanol capacity. It could thus be concluded that this is a less effective policy tool for state-level planners. By 2005, the changes in production cap­acity has slightly changed this correlation, but not to any significant extent (r2 = 0.04). It may be postulated that excise tax exemptions have far less in­fluence over the development of bioethanol capacity than does the amount of funding available to capital projects, feedstock availability, or other market influences. Follow-up analyses compared excise tax exemptions to estimates of bioethanol consumption [83,84] and indicate that in the same period, no correlation (r2 < 0.01) could be found between the use of bioethanol in gaso­line blends and state-level excise tax exemptions. This indicates that excise tax exemptions do not serve as a particularly effective tool in enforcing the use of renewable fuels, and that there is no clear cause-and-effect relationship between the level of these exemptions and the establishment of the industry within individual jurisdictions. While excise tax exemptions are undoubtedly an important economic component of a bioethanol producer’s business plan,

they would seem to have less effectiveness as a policy tool to create biofuel capacity or increase its consumption.

5

Many compounds that result from the pretreatment and heating of the lig — nocellulosic material, e. g., furfural and hydroxymethylfurfural, are severely inhibitory to most microorganisms [117,137]. Detoxification procedures for lignocellulose hydrolysates are under development [134]; however, the prac­tical large-scale use of detoxification is technically complex and adds cost to the fermentation process [138]. Similarly to inhibitors, low pH is required in the industrial context both for the function of cellulases and for avoiding bacterial infections, which is a frequently occurring problem in fermentation plants [139].

S. cerevisiae is the most robust microorganism among those with po­tential for efficient pentose fermentation, but differences between different S. cerevisiae strains are considerable. Laboratory yeast strains have been se­lected with regard to properties such as biomass yield and stability under well-defined conditions [140], whereas industrial isolates have been nat­urally selected for tolerance to industrial conditions. Although laboratory yeast strains are useful to evaluate metabolic engineering strategies and to compare cellular physiology, these strains do not possess the robust­ness that is required in the industrial context. Several investigations have shown that laboratory S. cerevisiae strains are generally less tolerant to lig — nocellulose hydrolysates than more robust industrial strains [7,141,142]. In addition, tolerance and robustness varies between different industrial strains [7]. Whereas some industrial strains require detoxification of the hydrolysate for efficient fermentation [6], others are able to ferment un­detoxified hydrolysates [12,13,143]. However, it is important to note that hydrolysates prepared with different methods and from different raw materi­als contain significantly different concentrations of inhibitors as well as of the fermentable sugars, as detailed elsewhere in this volume (pretreatment and hydrolysis). Therefore, the hydrolysate to be used has to be taken into account when selecting a strain for a fermentation process [144].

5.2

LY168 produced 0.5 g ethanol per gram of xylose during growth in min­eral salts medium with betaine, a value close to the theoretical maximum of

LB yeast extract + tryptone

Min minerals + 1 mM betaine

YE yeast extract supplemented with phosphate

YEP supplemented with yeast extract and peptone

Synth minerals supplemented with a mixture of vitamins

5 G 8 g of glucose added per liter

a Yomano et al. 2007

0.51 (Yomano et al., submitted). As described in detail below, osmolyte stress in mineral salts media limits biocatalyst performance and betaine supple­mentation combats this stress [42]. The amount of ethanol produced by LY168 during 24-h fermentations in mineral salts medium with betaine or LB were equivalent. Thus, strain LY168 fulfills the goal of constructing a microbial biocatalyst that produces ethanol without dependence on costly nutritional supplements. As presented in Table 2, the LY168 ethanol yield from xylose is higher than any previously reported ethanologenic biocatalyst.

2.3

In open anaerobic ecosystems where biomass is not sterilized, the degra­dation carried out by ubiquitous microorganisms normally follows a rather well-defined pathway as shown in Fig. 1. If no inorganic electron acceptors such as sulfate or nitrate are present, methane is the inevitable terminal biofuel product since all intermediates from the fermentative bacteria can be degraded to methane, carbon dioxide, and water. The natural end prod­ucts of the fermentative bacteria in such open systems are short-chained volatile fatty acids, hydrogen, and carbon dioxide. Alcohols are only formed in small amounts. Approximately 90% of the energy of the converted biomass is conserved in the end products, and only 10% is used by the fermentative bacteria [1]. In the terminal formation of methane from the fermentation

products, the biomass carbon is sequestered completely to the most oxi­dized (CO2) and the most reduced (CH4) states. Only 4% of the original biomass energy is utilized by the terminal link, leaving 86% of the ori­ginal energy content in the formed methane (Fig. 1), which constitutes the sound energetic rationale for the extensive exploitation of the biogas pro­cess. The obligate biology leading to methane formation has an intrinsic stability governed by thermodynamics, which ensures that methanogenesis proceeds within a wide spectrum of physical and chemical conditions. In most methanogenic fermentations the methane yield lies close to the theor­etical maximum of 3 mol of methane per mol of glucose, calculated from the Buswell equation [12,13]:

CaHbOc + (a — b/4 — c/2)H2O ^

(a/2 — b/8 + c/4)CO2 + (a/2 + b/8 — c/4)CH

4