7.1

From the Laboratory to the Real World: Strains and Media

Successful expression of XI in S. cerevisiae enabled further engineering for high-yield production of ethanol from D-xylose under anaerobic conditions. D-Xylose fermentation rates reported for S. cerevisiae strains based on the

Piromyces sp. E2 XI were, in principle, sufficiently high for industrial imple­mentation. However, the studies on these strains that have hitherto been cited in this review were all performed under “academic” conditions. These in­volved the use of defined synthetic media controlled at pH 5.0 and, perhaps most importantly, the absence of inhibitors that are characteristic for real-life plant biomass hydrolysates [31,37,49,54].

The S. cerevisiae strains expressing the Piromyces sp. E2 XI are based on the S. cerevisiae CEN. PK platform. Interestingly, preliminary tests showed that the parental strain CEN. PK113-7D demonstrated an almost similar performance in industrial-grade molasses compared with industrial bak­ers’ yeast strains. Moreover, deletion of the GRE3 gene (which encodes a non-specific aldose reductase, [66]) was not detrimental for performance in molasses-based industrial fermentations (W. de Laat, unpublished data). Therefore, trials to test the glucose/xylose fermenting strain S. cerevisiae RWB 218 [44] were initiated in both wheat straw and corn stover hydrolysates. Results from these fermentation trials will be briefly discussed below.

7.2

Evaluation of various pretreatment conditions and the effect on key variables, such as the overall yield of sugars or ethanol, needs to be assessed in an easy way to be able to judge the result. In several studies on pretreatment of biomass the “severity factor” has been used for comparing pretreatment results. Although it does not provide an accurate measure of the severity it can be used for rough estimates [8,9]. The severity correlation describes the severity of the pretreatment as a function of treatment time (minutes) and temperature (°C), Tref = 100 °C.

log(Ro) = log (t exp ^ ^ef) j j. (1)

When pretreatment is performed under acidic conditions (e. g. by impregna­tion with H2SO4), the effect of pH can be taken into consideration by the combined severity [10] defined as:

Combined severity (CS) = log(Ro) — pH (2)

It is well known that more severe conditions during pretreatment will cause greater degradation of hemicellulose sugars [11-13]. A high severity in the pretreatment is nevertheless required to enhance the enzymatic digestibility of cellulose [14]. The ideal pretreatment would hydrolyse the hemicellulose to its monomer sugars without further degradation. It would also cause an increase of the pore size and reduce cellulose crystallinity to enhance the en­zymatic digestibility of the cellulose fibres. However, these two effects are not reached at the same pretreatment severity, at least not using current technologies.

Assessment of pretreatment is usually done by using some of (or a combi­nation of) the following methods:

1. Analysis of the content of sugars liberated during pretreatment to the liquid as a combination of monomer and oligomer sugars, as well as an­alysis of the carbohydrate content of the water-insoluble solids (WIS). This gives the total recovery of carbohydrates in the pretreatment step.

2. Enzymatic hydrolysis (EH) of the WIS, either washed or non-washed.

3. Fermentation of the pretreatment liquid to assess inhibition of the fer­menting microorganism.

4. Simultaneous saccharification and fermentation (SSF) of either the whole slurry or the washed WIS.

The enzymatic hydrolysis (in 1 and 4) is performed using cellulases, i. e. a mixture of various cellobiohydrolases and endoglucanases supplemented with yd-glucosidase. The latter is not a cellulase as it only cleaves cellobiose into two glucose molecules. It has, however, a very important role in hydro­lysis since cellobiose is an end-product inhibitor of many cellulases [15,16]. On the other hand, d-glucosidase is also inhibited by glucose [17]. Since the enzymes are inhibited by the end products, the build-up of any of these products affects the cellulose hydrolysis negatively. The maximum cellulase activity for most fungus-derived cellulases and d-glucosidase occurs at 50 ± 5 °C and a pH of 4.0-5.0. However, the optimal conditions for enzymatic hydrolysis change with the hydrolysis residence time [18] and are also depen­dent on the source of the enzymes.

The enzymatic hydrolysis for assessment of pretreatment can be per­formed using various conditions (substrate concentration, enzyme dosage, temperature, stirring speed etc.). A common way is to use washed material at 2 wt % WIS, or alternatively at 1 wt % cellulose, to avoid end-product in­hibition [19]. This could be seen as the maximum achievable digestibility or glucose yield. However, it does not reflect the pretreatment efficiency in terms of avoiding formation of compounds that are inhibitory to the cellulases. In a full-scale process it is crucial to reach high sugar and ethanol concentrations in order to decrease the energy demand in the downstream processes. To in­crease the sugar concentration during large-scale operation, it is assumed that the whole slurry after pretreatment would be used without introducing sepa­ration steps, which would dilute the process stream. Furthermore, the overall substrate loading in enzymatic hydrolysis would probably need to be above 10 wt % WIS to meet the energy requirement for ethanol recovery. To mimic a situation that will be more similar to final process conditions, the enzymatic hydrolysis can be performed using the whole slurry from the pretreatment di­luted to various WIS concentrations, e. g. 10 wt %. In this case also the effect of inhibitors is assessed. However, due to the higher concentration of sugars the enzymes will also suffer from end-product inhibition.

To assess the effect of possible inhibitors acting on the microorganism used for fermentation of the sugars released in the enzymatic hydrolysis, method 2 is most often combined with method 3. The overall ethanol yield depends not only on the sugar yield, but also on the fermentability of the solution. Inhibition is influenced by the concentration of the soluble sub­stances released during pretreatment, present in the original raw material, e. g. acetic acid, or formed in the pretreatment step. Some of the substances present in the slurry are furfural and 5-hydroxymethylfurfural (HMF), which are the result of degradation of pentoses and hexoses, respectively. Furfural may react further to yield formic acid, or it may polymerize. HMF can be converted to formic acid and levulinic acid. In some pretreatments lignin degradation products are also formed. The concentrations of these and all other inhibitory substances in the fermentation step depend on the con­figuration of the preceding process steps. At ethanol concentrations below 4 to 5 wt % the energy demand increases rapidly with decreasing ethanol concentration. It is thus important to evaluate the fermentability of the con­centrated pretreatment hydrolysates. The fermentability test is usually per­formed on the liquid obtained from the pretreatment, either directly or di­luted to a concentration corresponding to what is expected to be suitable in a final process.

Another option for evaluation of the pretreatment step is to perform SSF either on the whole slurry diluted to a suitable WIS concentration or on the washed water-insoluble solid material, in both cases at a WIS around 5% or higher. In this case the glucose produced is immediately consumed by the fermenting microorganism, e. g. Saccharomyces cerevisiae, which removes the end-product inhibition of glucose and cellobiose. SSF adds information about the pretreatment efficiency, since SSF usually gives a higher overall ethanol yield than separate enzymatic hydrolysis and fermentation (SHF) due to con­version by the microorganism of some compounds that are inhibitory to the enzymes to less inhibitory compounds [20]. Also in the assessment by SSF the conditions may vary, e. g. substrate concentration, enzyme dosage, concentra­tion of microorganism etc.

It has to be added that variations between different laboratories in con­figurations and conditions used for assessment of the pretreatment make it very difficult to compare various pretreatment methods unless they are assessed in exactly the same way. Even so, the conclusions may be incor­rect as the conditions used may be unfavourable to a specific method. For instance, the use of hemicellulases in the enzymatic hydrolysis, instead of only cellulases, will be beneficial to pretreatment methods that result in large amounts of oligomer hemicellulose sugars, as will be discussed in the results section.

It is our opinion that the assessment of pretreatment has to be performed in a more rigorous way. The standard enzymatic hydrolysis at low substrate concentration may well be used to assess the maximum digestibility. However, in this case both cellulases and hemicellulases are needed. The “real” assess­ment should be performed by optimizing the conditions for all subsequent process steps under more realistic process conditions, taking into account the special features of the pretreated material, and then comparing the produc­tion cost for the various alternatives.

3

The primary factor in the high cost of enzymes for biomass hydrolysis is sim­ply the amount of enzyme that must be used. Compared to starch hydrolysis, 40- to 100-fold more enzyme protein is required to produce an equivalent amount of ethanol (Novozymes data). It was recognized very early on that efficient cellulose hydrolysis requires a complex, interacting collection of en­zymes during initial characterization of the T. reesei cellulase system [35]. To significantly reduce the amount of these enzymes requires that either more efficient component enzymes are identified or that additional enzymes can be added that reduce the total enzyme loading. Synergy, the ability of two or more enzymes to work simultaneously more effectively than in succession, was first described in cellulases more than 30 years ago when describing the action of CBH I and EG activities [36]. In this case, the synergy can be mech­anistically explained by the production of new cellulose ends by the action of the endoglucanase, creating new sites of exoglucanase attack by the CBH. Similarly, studies of the observed synergism between CBH I and CBH II from Humicola insolens, revealed that this CBH II, although capable of acting pro — cessively from non-reducing chain ends, does also cleave the cellulose chains in an endo fashion [37]. To drive enzyme loading down, we needed to search for similar synergistic enzyme pairs that could complement the preferred T. reesei cellulase system.

4.2.1

Reducing xylitol formation has been a major challenge in xylose fermentation by recombinant S. cerevisiae carrying the P. stipitis xylose pathway enzymes XR and XDH. Xylitol formation has primarily been ascribed to the differ­ence in cofactor requirements of the two enzymes, so that the intracellular concentration of NAD+ controls the amount of xylitol being converted to xy­lulose [21,32,47,74-76]. However, xylitol formation during ethanolic xylose fermentation also depends on the strain background, i. e., the metabolism of the host cell, since for example some strains of P. stipitis do not produce xyl­itol [47,49,50]. Thus, engineering the redox metabolism of the S. cerevisiae host has been given great attention where the aim primarily has been to ma­nipulate the intracellular concentrations and fluxes of cofactors to minimize xylitol formation.

4.5.1

L. R. Jarboe1,2 (И) • T. B. Grabar1 • L. P. Yomano1 • K. T. Shanmugan1 •

L. O. Ingram1

department of Microbiology and Cell Science, University of Florida,

Gainesville, FL 32611, USA Jarboe@UFL. edu

department of Chemical and Biological Engineering, Iowa State University,

Ames, IA 50011, USA

1 Introduction……………………………………………………………………………………………… 238

2 Engineering and Performance of Ethanologenic E. coli……………………………………. 240

2.1 Ethanologenic Biocatalysts KO11 and LY01………………………………………………… 240

2.1.1 Engineering Scheme………………………………………………………………………………….. 240

2.1.2 Utilized Substrates…………………………………………………………………………………… 242

2.1.3 Limitations and Challenges………………………………………………………………………. 243

2.2 Ethanologenic Biocatalyst, Strain LY168 …………………………………………………….. 243

2.2.1 Conversion of SZ110 to LY168 ………………………………………………………………….. 244

2.2.2 Ethanol Production by LY168 …………………………………………………………………… 244

2.3 Other Recombinant Ethanologenic E. coli Strains…………………………………………… 245

2.4 Non-recombinant Ethanologenic E. coli…………………………………………………………. 246

2.5 Ethanol Production in Organisms Other than E. coli………………………………………. 246

3 Metabolic and Transcriptomic Changes Accompanying Ethanologenicity. 247

3.1 Physiological Differences Conferring Ethanol Resistance to LY01…………………… 248

4 Challenges for Ethanol Production……………………………………………………………… 248

4.1 Cost Effective Growth Media……………………………………………………………………… 248

4.2 Osmolyte Stress Limits Performance in Mineral Salts Media………………………… 249

4.3 Hemicellulose Hydrolysate Contains Inhibitors……………………………………………. 250

4.4 Reducing the Requirement for Fungal Cellulases………………………………………….. 251

5 Application of Ethanol Design Scheme to Other Commodity Products. . 252

5.1 Optically Pure d(-)-and L(+)-Lactic Acid……………………………………………………. 252

5.2 Acetate and Pyruvate………………………………………………………………………………… 253

5.3 Xylitol……………………………………………………………………………………………………… 254

5.4 Succinate………………………………………………………………………………………………….. 255

5.5 L-Alanine…………………………………………………………………………………………………. 255

6 Summary…………………………………………………………………………………………………. 256

References……………………………………………………………………………………………………. 257

Abstract The utilization of lignocellulosic biomass as a petroleum alternative faces many challenges. This work reviews recent progress in the engineering of Escherichia coli and Klebsiella oxytoca to produce ethanol from biomass with minimal nutritional supplemen-

tation. A combination of directed engineering and metabolic evolution has resulted in microbial biocatalysts that produce up to 45 g L-1 ethanol in 48 h in a simple mineral salts medium, and convert various lignocellulosic materials to ethanol. Mutations contributing to ethanologenesis are discussed. The ethanologenic biocatalyst design approach was ap­plied to other commodity chemicals, including optically pure d(-)- and L(+)-lactic acid, succinate and L-alanine with similar success. This review also describes recent progress in growth medium development, the reduction of hemicellulose hydrolysate toxicity and reduction of the demand for fungal cellulases.

Keywords Escherichia coli ■ Ethanol ■ Hemicellulose hydrolysate ■ Lactic acid

1

Introduction

Increasing petroleum costs, together with our increasing dependency on crude oil imports, have provided an opportunity for bio-based fuels and chemicals to become economically competitive. With the development of new technologies, replacement of the current petroleum-based automotive fuels and petrochemicals and supplementation of the national energy supply with sustainable resources, such as plants and plant-derived materials, is now feas­ible. Currently, 65% of the oil consumed in the USA is imported. More than 211 billion gallons, or roughly half of the total US energy consumption, were burned as automotive fuel in 2005 [1]. Therefore, development of an alterna­tive renewable transportation fuel, such as ethanol, will significantly reduce US imported oil dependency, contribute to preservation of finite natural re­sources, and improve the environment.

The use of sugar-derived ethanol as the chief component of automotive fuel was successfully implemented in Brazil nearly three decades ago. While the USA already has automobiles capable of utilizing ethanol blended with gasoline and the infrastructure required to distribute ethanol across the na­tion, ethanol production lags significantly behind the 168 billion gallon do­mestic fuel demand. In 2006, the USA produced approximately 4.9 billion gallons of fuel ethanol [2]. Lignocellulosic materials provide the opportunity to further expand ethanol production.

Lignocellulose is a complex substance that accounts for approximately 90% of the dry weight of plant material. It represents the most abundant renewable energy source in the world and is comprised of cell wall structural polymers (cellulose, hemicellulose, pectin, and lignin) (Fig. 1). Due to the complexity of lignocellulose and the biological limitations of existing biocatalysts, the cur­rent conceptual process designs for lignocellulose-based ethanol production are more complex than starch-based processes. The development of a micro­bial biocatalyst that is capable of metabolizing lignocellulose and all of the constitutive sugars will simplify the process and reduce the cost of ethanol production

The common bacterial ethanol production pathway, shown in Eq. 1 and Fig. 2, does not allow complete, balanced conversion of glucose to ethanol. In contrast, the homoethanol pathway, comprised of pyruvate decarboxylase (PDC) and alcohol dehydrogenase (ADH), allows balanced production of two ethanol molecules per glucose. The homoethanol pathway is present in yeast, plants, and fungi, but is rare in prokaryotes and animals. Bacterial PDCs have a low pyruvate Km relative to other pyruvate-utilizing enzymes, resulting in effective competition for the pyruvate pool [3]; Km values are indicated for pyruvate-consuming reactions in Fig. 2.

Native E. coli Glucose ^ Ethanol + Acetate + 2 Formate (1)

Homoethanol Glucose ^ 2 Ethanol + 2CO2 (2)

Recombinant expression of the Zymomonas mobilis homoethanol pathway in E. coli was first described nearly 20 years ago and has been previously re­viewed [4-8]; this review will focus on progress made in the past 10 years. Additionally, this review will discuss advances in hemicellulose utilization and the application of the ethanologenic microbial biocatalyst design scheme to successful production of other commodity chemicals.

Pretreatment strategies have generally been categorized into biological, phys­ical and chemical processes, or a combination of these approaches.

Biological pretreatments typically utilize wood degrading fungi (soft, brown and white rot) to modify the chemical composition of the lignocel — lulosic feedstock. Generally, soft and brown rot fungi primarily degrade the hemicellulose while imparting minor modifications to lignin. White-rot fungi can more actively attack the lignin component [12]. Although there has been a fair amount of work done in this area, the primary application has been as a biopulping option for the pulp and paper industry rather than as a pretreat­ment for bioenergy applications. In addition to the requirements for careful control of growth conditions and for large amounts of space to perform bio­logical treatments, a major disadvantage of biological/fungal treatments is the typical residence time of 10-14 days. For these reasons, biological pretreat­ments are considered to be less attractive commercially.

Physical pretreatments involve the breakdown of the biomass feedstock into smaller particles that are more amenable to subsequent enzymatic hydrolysis. Physical treatments such as hammer — and ball-milling [13-16] have been shown to improve hydrolysis yields by disrupting cellulose crys­tallinity and by increasing the specific surface area of the feedstock, rendering them more accessible to attack by cellulases. One of the advantages of physi­cal pretreatment is that it is relatively insensitive to the physical and chemical characteristics of the biomass employed. However, the physical pretreatment processes are energetically demanding and do not result in lignin removal. Lignin has been shown to restrict access and inhibit cellulases [17,18]. Fur­thermore, physical pretreatments have yet to be shown to be economically viable at a commercial scale.

Most of the chemical pretreatments that have been assessed to date (typ­ically acid and alkali based) have had the primary goal of enhancing enzyme accessibility to the cellulose by solubilizing the hemicellulose and lignin, and to a lesser degree decreasing the DP and crystallinity of the cellulosic component. Pretreatments that reduce cellulose crystallinity include mild swelling agents such as NaOH, hydrazine and anhydrous ammonia, and ex­treme swelling agents such as sulfuric acid, hydrochloric acids, cupram, cuen, and cadoxen [19]. Treatments that reduce the lignin content of the substrate include organosolv pulping with various solvents including ethanol, glycerol and ethylene glycol.

Typically, all chemical pulping processes in commercial use today involve the removal of lignin to produce pulp for various paper products. Although these processes could be considered as potential pretreatment methods, they are optimized to maintain the fiber/strength integrity of the pulp, not to in­crease accessibility to the cellulose. The relatively high value of pulp (at the time of writing, approximately US$730 per tonne of northern bleached soft­wood Kraft pulp in Europe according to the PIX Pulp Benchmark Index) can justify the high capital and operating costs of chemical pulping, while lower-value biofuels must seek cheaper pretreatment alternatives. Despite these apparent drawbacks, various groups have looked at modified pulping processes as potential pretreatment methods, most likely since these pulp­ing processes produce readily hydrolyzable substrates. For example, in a Kraft pulping process NaOH and Na2S are combined in an aqueous liquor to cook wood chips under elevated pressures, followed by a pressure-release defi — bration step. The resulting Kraft pulps have been shown to be receptive to subsequent hydrolysis by cellulases [16], most likely because of the combina­tion of chemical dissolution of lignin and a decrease in average particle size that occurs during physical defibration.

Pretreatments that combine both chemical and physical processes are re­ferred to as physiochemical processes. These pretreatment methods have received the most attention in recent years and are the major focus of this review. In particular, steam pretreatment has received significant attention for its suitability in generating easily hydrolyzable substrates from lignocellu — losic biomass. However, several aspects that affect the viability of the overall process will be discussed in more detail later in this review, including the handling and preparation of the feedstock prior to the pretreatment step, the need to minimize processing costs, and the need to maximize the value of co­products derived from the hemicellulose and lignin streams. For example, if a pretreatment method has a requirement for very fine, uniform feedstock with a particle size of less than 10 mm, this will have a significant impact on the overall economic viability of the overall process because of the en­ergy requirements to produce this fine material [20,21]. Similarly, although acid-based pretreatment processes have been shown to be effective on a range of lignocellulosic substrates, downstream costs including the need for alka­line neutralization chemicals such as CaOH2 [22], must be considered. At the same time alkaline-based pretreatment methods such as lime, ammonia freeze explosion (AFEX), and ammonia recycle percolation (ARP) processes can effectively reduce the lignin content of agricultural crops such as wheat straw and corn stover, but have a much more difficult time processing recal­citrant substrates such as softwoods.

To summarize this general introduction, it is unlikely that one pretreat­ment process will be declared a “winner” as each method has its inherent advantages/disadvantages. However, as discussed in more detail below, steam pretreatment is one method that is effective on a range of lignocellulosic sub­strates and, through companies such as Masonite, has been shown to work effectively at a commercial scale.

1.2

The hydrolysis experiments were carried out at a substrate consistency of 10 gL-1 in 50 mM sodium acetate, pH 5, in a volume of 100 mL, and incubated in shake flasks with shaking (200 rpm) at different temperatures from 55 °C to 70 °C. Duplicate shake flasks were sampled (5 mL sample) at 2 h, 4 h, 6 h, 24 h, 48 h and 72 h from the start of the hydrolysis. Possible evaporation was checked by weighing and corrected when necessary by adding a correspond­ing amount of water. The release of hydrolysis products was followed during the hydrolysis.

The worldwide annual ethanol production via microbial fermentation amounted to ca. 40 Mt in 2005 (according to the Renewable Fuel Associa­tion; www. ethanolrfa. org) and is rapidly growing. Although bacteria such as Zymomonas mobilis and engineered Escherichia coli strains are capable of homoethanolic fermentation of sugars [17], the yeast Saccharomyces cere­visiae remains the organism of choice for large-scale industrial production of ethanol. Factors contributing to the popularity of S. cerevisiae as an industrial ethanol producer include its high ethanol tolerance, its ability to grow under strictly anaerobic conditions and — an important characteristic distinguishing it from prokaryotic organisms — its insensitivity to bacteriophage contamina­tions. Moreover, S. cerevisiae grows well at low pH, reducing problems with contamination of industrial processes with, for example, lactic acid bacteria.

Global concern about carbon dioxide emissions and climate change, deple­tion of oil reserves and geopolitical issues all contribute to a drive to increase the production of ethanol as a renewable transport fuel (see the contribution of Otero et al. in this volume). Presently, ethanol is exclusively produced from the starch or the sucrose fraction of a small number of (edible) agricultural crops such as corn, sugar cane, sugar beet and grain. To expand the feed­stock range for large-scale ethanol production and to improve productivity, it is of vital importance to enable efficient ethanol production from agri­cultural residues and other low-value sources of carbohydrates. Feedstocks such as corn stover, bagasse, wheat straw, non-recyclable paper or dedicated crops such as switchgrass represent an enormous potential in terms of avail­able carbohydrates. However, instead of starch and sucrose, the carbohydrates in these feedstocks consist of a complex matrix of cellulose, hemicellulose, pectin and lignin [69].

The use of lignocellulosic raw materials for ethanol production poses a number of major challenges compared to the use of conventional starch — or sucrose-based feedstocks:

(i) Release of monomeric sugars from lignocellulosic biomass requires a mix of physicochemical (extreme pH, high temperature, high pressure) and enzymic polysaccharide (hydrolases) treatments [19,37].

(ii) The resulting lignocellulose hydrolysates contain a wide variety of com­pounds that may inhibit the fermentation process. These compounds are either formed during the pretreatment process (e. g. furfural and hydroxymethylfurfural) or are biomass constituents that are released during hydrolysis (e. g. acetate, formate) [31,37,49,54].

(iii) Whereas starch — and sucrose-based feedstocks yield hexoses upon hydro­lysis, lignocellulosic biomass, and in particular its hemicellulose frac­tion, also contains large amounts of the pentose sugars D-xylose and L-arabinose. D-Xylose, generally the most abundant pentose, comprises up to 25% of the total sugar content in some hydrolysates [24,46,69].

Whereas S. cerevisiae spp. can rapidly ferment hexose sugars such as glucose, fructose, mannose and galactose, they cannot grow on nor ferment D-xylose or L-arabinose [7,69]. Given the importance of xylose fermentation for the efficient production of ethanol from lignocellulosic biomass [24,46,69], it is not surprising that introduction and optimisation of heterologous path­ways for xylose fermentation into S. cerevisiae has long been a hot topic in metabolic engineering of yeast.

Interestingly, it has long been known that S. cerevisiae is able to slowly metabolise the pentose sugar D-xylulose [30,71]. This keto-isomer of xylose is phosphorylated to D-xylulose-5-phosphate by xylulokinase (XKS1, [57]) and subsequently metabolised via the non-oxidative part of the pentose phos­phate pathway and glycolysis. It is therefore logical that strategies for convert­ing D-xylose into D-xylulose are an exhaustively studied topic in the quest for alcoholic fermentation of D-xylose by S. cerevisiae. These strategies will be briefly discussed in Sects. 1.2-1.4.

1.2

During the same time period that the cost of crude petroleum rose 150%, from January 2001 to 2005, the total number of bioethanol refineries in the USA increased from 56 to 81, with total production capacity increasing from 6.6 bil­lion L/year to 13.8 billion L/year. Within the last year, from January 2005 to 2006, the total number of refineries increased to 95 and output further increased to 14.3 billion L/year, a 1500% increase since 2001. Total world production in 2005 was 46 billion L, with the USA and Brazil representing a combined 70% of the world’s production. It should be further noted that by the end of 2005, 29 ethanol refineries and nine expansions of existing refineries were under con­struction, with a combined annual capacity of 5.7 billion L. If you consider all of the US ethanol production capacity currently on-line, under expansion, and under construction, then the projected capacity is approximately 24 billion L — approximately 85% of that required by the RFS by 2012 [43].

In the USA, the raw material of choice for bioethanol production is corn. Approximately 13% of the US corn crop is dedicated to ethanol production,

third only to livestock feed and exports [43]. In Brazil, however, the raw ma­terial of choice is sugarcane. With over 100 countries producing sugarcane, no one has yet to match Brazil’s cost structure and supply chain. In mid-2005, the sugar production costs in the three lowest countries were estimated to be $145/metric ton in Brazil, $185/metric ton in Australia, and $195/metric ton in Thailand. About 25% of worldwide sugar production is at $200-250/metric ton, above which the figure escalates to $400/metric ton and higher. Sugar­cane is a highly efficient crop for producing biomass, representing the highest biomass per growing area of any major commercial crop, including corn. This is a result of sugarcane’s ability to incorporate C3 and C4 compounds in its photosynthetic pathway, while most plants only incorporate C3 compounds. Brazilian ethanol is most likely the cheapest in the world, with an estimated production cost of between $0.19 and $0.21/L in 2005. For this reason Brazil is not only looking to expand its ethanol production capacity, but to further expand into biorefineries [54].

Brazil’s sugarcane production is unique, and not representative of the gen­eral challenge almost all other nations face when determining which raw ma­terial source is preferred. Raw material utilization for bioethanol and biotech­nology processes in general represents a significant challenge and opportu­nity for research and development. The US Department of Agriculture and Department of Energy estimate that the resources exist to produce over 1 bil­lion tonnes of biomass annually, representing approximately 30% displace­ment of current fossil fuel usage (302 billion L) [55]. Biomass is composed of cellulose (40-50%), hemicellulose (25-35%) and lignin (15-20%) [56]. Significant effort in the fields of non-food agricultural engineering, enzyme catalysis of cellulose and hemicellulose, and hexose and pentose fermentation will be required to extrapolate the full energetic value of lignocellulose.

Figure 2 schematically shows how research in the aforementioned areas is integrated into bioethanol process development, specifically focusing on the secondary pretreatment of feedstocks and microbial metabolic engineer­ing. In both examples, the application of systems biology to the metabolic engineering framework can yield improved products, either in the form of enzymes or microbial platforms. We will further explore how scientific and technical achievements in the fields of metabolic engineering and systems bi­ology as applied to the afore mentioned areas and others, driven by industrial biotechnology and demand for bioethanol, will improve bioethanol process development.

3

There have been only limited studies assessing the significance of initial cellulose crystallinity and DP of lignocellulosic substrates with regard to subsequent substrate hydrolysis by cellulases; however, the importance of these factors has been the subject of considerable debate [113,114]. It has been suggested that amorphous cellulose is hydrolyzed, initially resulting in an accumulation of crystalline cellulose rendering the substrate increasingly recalcitrant as the hydrolysis progresses [113,114]. Most studies that have established a correlation between crystallinity and hydrolysis have utilized substrates of relatively pure cellulose, which most likely do not represent the heterogeneous lignocellulosic substrate encountered during the hydrolysis of substrates pretreated for bioconversion [115-117]. Furthermore, to demon­strate the effect of crystallinity on hydrolysis, these studies frequently utilize physical treatments such as ball milling [116] or gamma irradiation [118] to alter the initial substrate crystallinity, which can also result in increases in specific surface area. As a result, in previous work both crystallinity and spe­cific surface area of pure cellulose substrates have been combined into models predicting the rate and extent of hydrolysis [119]. As for crystallinity, it is dif­ficult to assess the effects of DP exclusively, since altered DP can be associated with crystallinity or accessible surface area. Nevertheless, there have been a few studies investigating the effects of the crystallinity and DP of chemical pulps on their hydrolysis by cellulases.

Employing unbeaten, beaten and recycled softwood pulps as substrates to assess various substrate characteristics that influence the enzymatic hydro­lysis of cellulose, Nahzad et al. [105] showed that although the pulps pos­sessed similar crystallinity, the beaten pulp hydrolyzed more readily than the unbeaten pulp without any appreciable changes in crystallinity occur­ring during the hydrolysis. Similar results have been found by Ramos et al. during the treatment of fully bleached eucalyptus Kraft pulp [104] and by Mansfield et al. during combined cellulase-xylanase treatment of Douglas — fir Kraft pulps [99]. Nahzad et al. [105] also showed that the initial DP of the pulps did not play a role in affecting subsequent hydrolysis; however, the DP was decreased by 2/3 during the hydrolysis period and the polydispersi — ties of all the hydrolyzed pulps were quite similar. Mansfield et al. [99] did not observe appreciable changes in cellulose DP during hydrolysis, however, it should be noted that they employed low cellulase loadings to impart sub­tle modifications to the pulp fiber. It is evident from the literature presented here that attributing the ease of enzymatic digestibility of a given substrate to initial crystallinity or DP is a dubious task, compared to studies that have tied the ease of hydrolysis of substrates to their initial surface area. However, pore volume determinations require a significant investment in time to ob­tain reproducible results. Also, it is likely that the pores in a lignocellulosic substrate will have irregular shapes, thus affecting the accuracy and preci­sion of the measurement [120]. Another drawback is that the method does not measure the areas in pores that are larger than the size of the probe, which would provide the easiest access for cellulases. Investigations into the sub­strate physical factors that affect hydrolysis should be aided by the continual evolution of analytical techniques such as thermoporosimetry [121] and high resolution fiber quality analysis [122], which may be capable of dealing with the diversity of pretreated lignocellulosic substrates produced for subsequent hydrolysis and fermentation in the bioconversion process.

6

Conclusions

In this review we suggested that, although the properties of the cellulase en­zyme complex has a significant effect on how effectively a lignocellulosic material will be hydrolyzed, it is the biomass pretreatment and the intrin­sic structure/composition of the substrate itself that are primarily responsible for its subsequent hydrolysis by cellulases. It is apparent that in sequential series of events, the conditions employed in the chosen pretreatment will af­fect various substrate characteristics, which in turn govern the susceptibility of the substrate to hydrolysis by cellulases and the subsequent fermentation of the released sugars. Choosing the appropriate pretreatment for a particu­lar biomass feedstock is frequently a compromise between minimizing the degradation of the hemicellulose and cellulose components while maximizing the ease of hydrolysis of the cellulosic substrate. The digestibility of pretreated lignocellulosic substrates is further complicated by the lignin-hemicellulose matrix in which cellulose is tightly embedded. Pretreatment conditions can be tailored to create either solid or solid/liquid substrates with varying levels of cellulose, hemicellulose and lignin. It is apparent that lignin affects enzy­matic hydrolysis by blocking cellulose and by chemical interactions facilitated by its hydrophobic surface properties and various functional groups. The role of hemicellulose is less obvious although there is good evidence to support the action of hemicellulose as a barrier restricting access to cellulases. In the past, many investigators have attributed the enhanced enzymatic hydrolysis performance of a particular pretreatment to changes in the proportion of the lignin, hemicellulose and cellulose in the substrate. However, it is important to advance this conclusion one step further as it is likely that decreases in lignin and hemicellulose content that occur as a result of pretreatment also affect the physical properties of the cellulosic component, such as its crys­tallinity, the degree of polymerization and the surface area of the substrate accessible to cellulases.

Various studies conducted with different cellulase systems and a range of cellulosic substrates all indicate that it is ultimately the “accessibility” of the cellulose fraction to the enzyme system that determines how fast (reaction rate) and how far (% conversion) the hydrolysis reaction can proceed [112]. In work conducted with either wood pulps or substrates pretreated for bio­conversion we and other groups have shown that accessibility is a property that describes the static environment encountered by the cellulase complex when it is combined with the substrate, and its action is governed by the in­trinsic pore size distribution, degree of swelling and other gross and detailed substrate characteristics. As enzymatic hydrolysis commences, the situation becomes dynamic, as substrate attributes begin to change due to cellulose hydrolysis and the hydrolysis rate decreases. Some workers have reported de­creases in accessible surface area as hydrolysis proceeds without appreciable changes in crystallinity [99,105,123], while others have reported decreases in crystallinity and increases in accessible surface area during hydrolysis [124]. The discrepancy in results regarding the decrease in the hydrolysis rate of pretreated substrates can most likely be attributed to variations in the sub­strates being studied and the techniques used for measurement of substrate properties. Furthermore, as cellulose is hydrolyzed, the lignin and hemicellu — lose that accumulate in the hydrolysis residue can potentially restrict access to cellulases and decrease the hydrolysis rate. Therefore, pretreatments should aim to produce a readily hydrolyzable substrate by increasing accessibility to cellulases and limiting the negative effects of hemicellulose and lignin on hydrolysis, while maximizing the total carbohydrate recovery.

However, it is important to recognize that studies which try to optimize pretreatment (as assessed by product recovery) need to be performed in parallel with measurements of key substrate characteristics in order to as­sociate specific aspects of pretreatment to substrate attributes that facilitate subsequent hydrolysis by cellulases. This emphasizes the significance of the pretreatment since the effectiveness of pretreatment affects both the up­stream selection of biomass, the efficiency of recovery of the overall cellulose, hemicellulose and lignin components, and the chemical and morphological characteristics of the resulting cellulosic component, which governs down­stream hydrolysis and fermentation.

Adv Biochem Engin/Biotechnol (2007) 108: 95-120 DOI 10.1007/10_2007_066 © Springer-Verlag Berlin Heidelberg Published online: 27 June 2007

Progress and Challenges