Arguments in favor of bioethanol (and other biofuels) tend to mix some (or all) of four key points, variously worded, and with different degrees of urgency:

• Fossil energy resources are finite and may become seriously depleted before 2050 (the “peak oil” argument — see chapter 5, section 5.6).

• Biofuels avoid dependency on oil imports (the “energy security” argument).

• Biofuels augment sustainable development across the globe, more efficiently utilizing agricultural wastes (the “long-term environmental” argument).

• Biofuels can reduce greenhouse gas and other harmful emissions (the “acute climatic” argument).

Economists and socioeconomists tend to concentrate the first two cases, whereas biological scientists are far more comfortable with the latter two arguments. So far, the present discussion has touched on topics pertinent to the “energy security” problem, but conflicting data and conclusions have been apparent for nearly three decades on how biofuels (and bioethanol in particular) may — or may not — be solu­tions to the global energy supply and the ecological crisis of global warming caused by greenhouse gas emissions.

3.4.1 "Traditional" Microbial Ethanologens

Given the present indeterminacy regarding “best” choices for producing organism and biomass substrate, it is timely to summarize recent advances in the metabolic and genetic sciences relevant to yeast and bacterial ethanologens, and from them to project forward key areas for their application to industrial ethanol production after 2007. Four themes can readily be identified:

1. A better understanding of stress responses

2. Relating transport phenomena to sugar utilization and ethanol productivity

3. Defining the roles of particular genes in the metabolism of ethanologens

4. Applying whole-genome methodologies and in silico simulations to opti­mize the balance of carbon flows along synergistic or competing pathways

Yeast ethanologens show dose-dependent inhibition by furfural and HMF and are generally more sensitive to furfural; a S. cerevisiae strain converted HMF to 2,5-bis-HMF, a previously postulated alcohol formed from the parent aldehyde, and this suggests that an enzymic pathway for detoxifying these components of lignocel- lulosic hydrolysates should be attainable by gene cloning and/or mutation to improve the activities and substrate preferences of the enzymes.244 That such a reduction reaction as that to form 2,5-bis-HMF might utilize NADPH as cofactor was shown by screening a gene disruption library from S. cerevisiae: one type of mutation was in zwfl, the gene encoding glucose 6-phosphate dehydrogenase, the NADPH-generating entry point to the oxidative pentose pathway; overexpression of zwfl allowed growth in furfural concentrations normally toxic, whereas similar effects with non-redox- associated gene products could be explained as a result of inhibiting the overall activ­ity of the pathway.245 The ADH6 gene of S. cerevisiae encodes a reductase enzyme that reduces HMF and NADPH supports a specific activity of the enzyme some 150­fold higher than with NADH; yeast strains overexpressing this gene had a higher in vivo conversion rate of HMF in both aerobic and anaerobic cultures.246

A key trait for biofuels research is to enable growth in the presence of very high concentrations of both glucose and ethanol, the former in the batch medium (discussed in chapter 4, section 4.3), the latter accumulating as the fermentation proceeds. In yeasts, analysis of the effects of ethanol concentration and tempera­ture point to nonspecific actions on the cell membranes as leading to the variety of cellular responses, including reduced fermentation rate and viability.247 Adapting cells to ethanol causes changes in the cellular membrane composition, in particu­lar the degree of unsaturation of the fatty acids and the amount of sterols present, both indicative of alterations in the fluidity characteristics of the lipid-based mem­branes.248 The low tolerance of Pichia stipitis to ethanol can be correlated with a disruption of protein transport across the cell membrane, an energy-requiring pro­cess involving a membrane-localized ATPase; the system is activated by glucose, but ethanol uncouples ATP hydrolysis from protein transport, thus wasting energy for no purpose, and structural alteration of the ATPase to affect its sensitivity to ethanol could be a quick-win option for yeast ethanologens in general.249

In Z. mobilis, there is evidence for a glucose-sensing system to confer (or increase) the cells’ tolerance to the osmotic stress imposed by high sugar concentrations; the exact functions of the products of the four genes simultaneously transcribed in a glu­cose-regulated operon are unknown, but manipulating this genetic locus would aid adaptation to high concentrations (>100 g/l) of glucose in industrial media.250 The announcement of the complete sequencing of the genome of Z. mobilis by scientists in Seoul, South Korea, in 2005 provided rationalizations for features of the idio­syncratic carbohydrate metabolism in the absence of three key genes for the EMP/ tricarboxylic acid cycle paradigm, that is, 6-phosphofructokinase, 2-oxoglutarate dehydrogenase, and malate dehydrogenase.251 The nonoxidative pentose phosphate pathway is mostly missing, although the genes are present for the biosynthesis of phosphorylated ribose and thence histidine and nucleotides for both DNA and RNA (figure 3.12). The 2-Mbp circular chromosome encodes for 1,998 predicted func­tional genes; of these, nearly 20% showed no similarities to known genes, suggesting a high probability of coding sequences of industrial significance for ethanologens. In comparison to a strain of Z. mobilis with lower tolerance to ethanol and rates of ethanol production, glucose uptake, and specific growth rate, strain ZM4 contains 54 additional genes, including four transport proteins and two oxidoreductases, all potentially mediating the higher ethanol productivity of the strain. Moreover, two genes coding for capsular carbohydrate synthesis may be involved in generating an altered morphology more resistant to osmotic stress. Intriguingly, 25 of the “new” genes showed similarities to bacteriophage genes, indicating a horizontal transfer of genetic material via phages.251

Fructose

6-Phosphogluconate

2-Oxo-3-deoxy-6-phosphogluconate

Glyceraldehyde 3-phosphate

Ribulose 5-phosphate

1,3-Diphosphoglycerate

2-Phosphoglycerate

5-Phosphoribosyl pyrophosphate

Phosphoc^o/pyruvate

Oxaloacetate

Succinate

FIGURE 3.12 Fragmentary carbohydrate metabolism and interconversions in Z. mobilis as confirmed by whole-genome sequencing. (From Seo et al.251)

Analysis of mutants of the yeast Yarrowia lipolytica hypersensitive to acetic acid and ethanol identified a single gene that could direct multiple phenotypic changes, including colony morphology, the morphology of intracellular membranes, and the appearance of vacuoles and mitochondria, and induce early death on glucose even with­out the presence of acetate or ethanol; this GPR1 gene may be a target for manipulation
and mutation to increase solute tolerance.252 A more direct approach in S. cerevisiae involved multiple mutagenesis of a transcription factor and selection of dominant muta­tions that conferred increased ethanol tolerance and more efficient glucose conversion to ethanol: 20% for growth in the presence of an initial glucose concentration of 100 g/l, 69% for space-time yield (grams of ethanol per liter per hour), 41% for specific pro­ductivity, and 15% for ethanol yield from glucose.253 It is likely that the expression of an ensemble of hundreds of genes was increased by the mutations; of the 12 most over­expressed genes, deletions mutations abolished the increased glucose tolerance in all but one gene, and selective overexpression of each of the top three up-regulated genes could replicate the tolerance of the best mutant. Because this work was, by necessity, carried out on a laboratory haploid strain, it is possible that similar changes could have occurred gradually during many generations in industrial strains. The effect on gene expression was, however, greatly diminished at 120 g/l of glucose.

Comparable analyses of gene expression and function are now being introduced into commercial potable alcohol research. A common driver between breweries and industrial bioethanol plants is (or will be) the use of concentrated, osmotically stressful fermentation media. In breweries, the capability to use high-gravity brewing in which the wort is more concentrated than in traditional practice saves energy, time, and space; the ethanol-containing spent wort can be diluted either with lower alcohol-content wort or with water. Variants of brewer’s yeast (S. pastorianus) were obtained subjecting UV- mutagenized cells to consecutive rounds of fermentation in very-high-gravity wort; variants showing faster fermentation times and/or more complete sugar utilization were identified for investigations of gene expression with microarray technologies.254 Of the 13 genes either overexpressed or underexpressed in comparison with a control strain, a hexokinase gene was potentially significant as signaling a reduced carbon catabolite repression of maltose and maltotriose uptake; more speculatively, two ampli­fied genes in amino acid biosynthesis pathways could point to increased requirements for amino acid production for growth in the highly concentrated medium. Although deletion mutants are usually difficult to isolate with polyploid industrial strains, the dose-variable amplification of target genes is more feasible.

Strains of S. cerevisiae genetically engineered for xylose utilization depend, unlike naturally xylose-consuming yeasts, on xylose uptake via the endogenous hexose transporters; with a high XR activity, xylose uptake by this ad hoc arrange­ment limits the pentose utilization rate at low xylose concentrations, suggesting that expression of a yeast xylose/pentose transporter in S. cerevisiae would be beneficial for the kinetics and extent of xylose consumption from hemicellulose hydrolysates.255 A high-affinity xylose transporter is present in the cell membranes of the efficient xylose fermenting yeast Candida succiphila when grown on xylose and in Kluyvero — myces marxianus when grown on glucose under fully aerobic conditions.256

In ethanologenic bacteria, there are also grounds for postulating that radically chang­ing the mechanism of xylose uptake would improve both anaerobic growth and xylose utilization. Mutants of E. coli lacking PFL or acetate kinase activity fail to grow anaero­bically on xylose, that is, the cells are energy-limited without the means to generate ATP from acetyl phosphate; in contrast, the mutants can grow anaerobically on arabinose — and the difference resides in the sugar transport mechanisms: energy-expending active transport for xylose but an energy-conserving arabinose.257 Reconfiguring xylose uptake

is, therefore, an attractive option for removing energetic constraints on xylose metabo­lism. In yeast also, devising means of conserving energy during xylose metabolism would reduce the bioenergetic problems posed by xylose catabolism — gene expres­sion profiles during xylose consumption resemble those of the starvation response, and xylose appears to induce metabolic behavioral responses that mix features of those elic­ited by genuinely fermentable and respirative carbon sources.258

Looking beyond purely monomeric sugar substrates, the soil bacterium Geobacil­lus stearothermophilus is highly efficient at degrading hemicelluloses, possessing a 30- gene array for this purpose, organized into nine transcriptional units; in the presence of a xylan, endo-1,4-P-xylanase is secreted, and the resulting xylooligosaccharides enter the cell by a specialized transport system, now known to be a three-gene unit encod­ing an ATP-binding cassette transport system, whose transcription is repressed by xylose but activated by a response regulator.259 This regulation allows the cell to rapidly amplify the expression of the transport system when substrates are available. Inside the cell, the oligosaccharides are hydrolyzed to monomeric sugars by the actions of a fam­ily of xylanases, P-xylosidase, xylan acetylesterases, arabinofuranosidases, and gluc­uronidases. Such a complete biochemical toolkit for hemicellulose degradation should be considered for heterologous expression in bacterial bioethanol producers.

Control of the gene encoding the major PDC in wild-type S. cerevisiae was shown by analysis of gene transcription to be the result of ethanol repression rather than glucose induction; an ethanol-repression ERA sequence was identified in the yeast DNA; the Era protein binds to the PDC1 promoter sequence under all growth conditions, and an autoregulatory (A) factor stimulates PDC1 transcription but also binds to the pdc1 enzyme — the more enzyme is formed, the more A factor is bound, thus making the transcription of the FDC1 gene self-regulating.260 This may be a useful target for strain improvement; industrial strains may, however, have already been “blindly” selected at this locus.

Focusing on single-gene traits, a research group at Tianjin University (China) has recently demonstrated three approaches to reduce glycerol, acetate, and pyruvate accumulation during ethanol production by S. cerevisiae:

1. Simply deleting the gene encoding glycerol formation from glycolysis, that is, glycerol 3-phosphate dehydrogenase reduced the growth rate compared with the wild type but reduced acetate and pyruvate accumulation. Simul­taneously overexpressing the GLTl gene for glutamate synthase (catalyzing the reductive formation of glutamate from glutamine and 2-oxo-glutarate) restored the growth characteristics while reducing glycerol accumulation and increasing ethanol formation.261

2. Deleting the gene (FSP1) encoding the transporter protein for glycerol also improved ethanol production while decreasing the accumulation of glycerol, acetate, and pyruvate—presumably increasing the intracellular concentration of glycerol (by blocking its export) accelerates its metabolism by a reversal of the glycerol kinase and glycerol 3-phosphate dehydrogenase reactions.262

3. Improved production of ethanol was also achieved by deleting FPS1 and overexpressing GLTl, accompanied by reduced accumulation of glycerol, acetate, and pyruvate.263

A more subtle metabolic conundrum may limit ethanol accumulation in Z. mobilis: although the cells are internally “unstructured” prokaryotes, lacking membrane — limited intracellular organelles, different redox microenvironments may occur in which two forms of ADH (rather than one) could simultaneously catalyze ethanol synthesis and oxidation, setting up a futile cycle; this has only been demonstrated under aerobic growth conditions, but the functions of the two ADH enzymes (one Zn-dependent, the other Fe-dependent) inside putative redox microcompartments within the cell warrant further investigation.264

S. cerevisiae, along with E. coli, B. subtilis, and a few other organisms, formed the initial “Rosetta Stone” for genome research, providing a well-characterized group of genes for comparative sequence analysis; coding sequences of other organisms lacking detailed knowledge of their biochemistry could be assigned protein func­tionalities based on their degree of homology to known genes. Genome-scale meta­bolic networks can now be reconstructed and their properties examined in computer simulations of biotechnological processes. Such models can certainly generate good fits to experimental data for the rate of growth, glucose utilization, and (under micro — aerobiosis) ethanol formation but have limited predictive value.265 One limitation is that the biosynthetic genes comprise too conservative a functional array, lacking the sensing mechanisms and signaling pathways that sense and respond to nutri­tional status and other environmental factors.266 These elements function to estab­lish developmental and morphological programs in the wild-type ecology and have unpredictable effects on growth patterns in the fluctuating nutritional topologies of large imperfectly mixed industrial fermentors. If computer-aided design for metabo­lism is to be feasible for industrially relevant microbes, added sophistications must be incorporated — equally, however, radical restructuring is necessary to accommo­date the gross differences in metabolic patterns between bacteria such as E. coli and, for example, the Entner-Doudoroff pathway organisms or when P. stipitis metabo­lism is compared in detail with the far better characterized pathway dynamics in S. cerevisiae.267,268 Comparing metabolic profiles in 14 yeast species revealed many important similarities with S. cerevisiae, but Pichia augusta may possess the ability to radically rebalance the growth demand for NADPH from the pentose phosphate pathway (figure 3.2); this suggests a transhydrogenase activity that could be used to rebalance NADH/NADPH redox cofactors for xylose metabolism (section 3.2.1).269

In the immediate and short-term future, therefore, applied genomics can suc­cessfully define the impacts of more focused changes in quantitative gene expression. Developing a kinetic model of the pentose phosphate and Entner-Doudoroff pathways in Z. mobilis, and comparing the predictions with experimental results, showed that the activities of xylose isomerase and transaldolase were of the greatest significance for ethanol production but that overexpressing the enzymic link between the two pathways (phosphoglucose isomerase) was unnecessary.270 Such systematic analysis can help not only in maximizing fermentation efficiencies but also in minimizing the degree of expression of heterologous enzymes to reduce “metabolic burdens” in recombinant strains. Although in silico predictions failed to account for contin­ued xylitol accumulation by a recombinant, xylose-utilizing S. cerevisiae strain on xylose (suggesting continued problems in balancing cofactor requirement in reduc — tase/dehydrogenase reactions and/or still unidentified limitations to later catabolic steps in xylose consumption), computer simulations did accurately highlight glycerol formation as a key target for process improvement: expressing a nonphosphorylating NADPH-dependent glyceraldehyde 3-phosphate dehydrogenase increased carbon flow to pyruvate (rather than being diverted to glycerol formation), reduced glycerol accumulation, and improved ethanol production by up to 25%.271272

Three serious drawbacks have been noted with cellulases — especially of fungal origin — for the efficient saccharification of cellulose and lignocellulosic materials on the industrial and semi-industrial scales. First, cellulases have often been described as being catalytically inferior to other glycosidases. This statement is certainly true when crystalline cellulose is the substrate for cellulase action.84 When more accessible forms of cellulose are hydrolyzed, the catalytic efficiency increases, but comparison with other glycosidases shows how relatively poor are cellulases even with low-molecular-weight substrates (table 2.6).

Second, cellulases from organisms not normally grown at elevated temperatures have poor stability at incubation temperatures higher than 50°C typically used for cellulose digestion: fungal cellulases show half-life times at 65°C as low as 10 min­utes, whereas thermophilic clostridial enzymes may be stable for periods longer than ten days.84 This is one of the much-explored causes for the rapid decline of hydro­lysis rate when cellulase is mixed with cellulose as a substrate.60 Another possible explanation for this third problem encountered with cellulase-mediated sacchari­fication is the inhibition of cellulases by cellobiose, a major immediate product of cellulase action; a strong product inhibition would be a major drawback, limiting the amounts of soluble sugars that can be produced unless they are rapidly removed; in other words, batch hydrolysis would be inferior to a continuous process with with­drawal of the fermentable sugars or a simultaneous release and utilization of cello — biose would be required. Although such a product inhibition is readily demonstrable

with low-molecular-weight soluble substrates for cellulase, macromolecular cellu­lose does not show a marked sensitivity to cellobiose.85 The hydrolysis of cellobiose by P-glucosidase to yield free glucose is itself inhibited by glucose.86

The failure to maintain the initial rate of degradation of macromolecular cellulose, however, necessitates an explanation involving the interaction between cellulose and cellulase, and various formulations of a hypothesis have been made in which the “reactivity” of the cellulose decreases during cellulase digestion.60 A more intuitive line of reasoning is simply that, cellulose being macroscopic and inevitably heterogeneous on an enzyme protein scale, cellulases would attack most rapidly any sites on the cellulose substrate that are, by their very nature, most vulnerable (e. g., with directly accessible regions of the glucan polymer chain); once these sites are cleaved, the active sites remain as intrinsically active as they were initially but most then are restricted to operating at less accessible regions of the available surface. The experimental evidence is contradictory: although scanning electron microscopy has revealed changes in the conformations and packing of cel­lulose microfibers during prolonged cellulase-catalyzed hydrolysis, “restart” exper­iments (where the cellulase is removed and fresh enzyme added) show that, at least with crystalline cellulose, no rate-limiting decrease in substrate site accessibility occurs while enzyme action continues.8788 Detailed kinetic models have postulated physical hindrance to the movement of exoglucanases along the cellulose and a time variation in the fraction of the P-glucosidic bonds accessible by cellulases bound to the macromolecular cellulose surfaces as factors leading to the failure to maintain hydrolysis rates.89,90

Formulating cost-effective media for the recombinant microorganisms developed for broad-spectrum pentose and hexose utilization (chapter 3, sections 3.2 and 3.3) commenced in the 1990s. For pentose-utilizing E. coli, for example, the benchmark was a nutrient-rich laboratory medium suitable for the generation of high-cell-density cultures.111 Media were then assessed using the criteria that the final ethanol concen­tration should be at least 25 g/l, the xylose-to-ethanol conversion efficiency would be high (90%), and a volumetric productivity of 0.52 g/l/hr was to be attained; in a defined minimal salts medium, growth was poor, only 15% of that observed in the laboratory medium; supplementation with vitamins and amino acids improved growth but could only match approximately half of the volumetric productivity. The use of corn steep liquor as a complex nitrogen source was (as predicted from its wide industrial use in fermentations) the best compromise between the provision of a complete nutritional package with plausible cost implications for a large-scale process. As an example of the different class of compromise inherent in the use of lignocellulosic substrates, the requirement to have a carbon source with a high content of monomeric xylose and low hemicellulose polymers implied the formation of high concentrations of acetic acid as a breakdown product of acetylated sugar residues; to minimize the associated growth inhibition, one straightforward strategy was that of operating the fermentation at a relatively high pH (7.0) to reduce the uptake of the weak acid inhibitor.

In a study conducted by the National Center for Agricultural Utilization Research, Peoria, Illinois, some surprising interactions were discovered between nitrogen nutrition and ethanol production by the yeast P. stipitis.112 When the cells had ceased active growth in a chemically defined medium, they were unable to ferment either xylose or glucose to ethanol unless a nitrogen source was also provided. Ethanol pro­duction was increased by the amino acids alanine, arginine, aspartic acid, glutamic acid, glycine, histidine, leucine, and tyrosine (although isoleucine was inhibitory); a more practical nitrogen supply for industrial fermentations consisted of a mixture of urea (up to 80% of the nitrogen) and hydrolyzed milk protein supplemented with tryptophan and cysteine (up to 60%); the use of either urea or the protein hydrolysate was less effective than the combination of both. Adding small amounts of minerals, in particular, iron, manganese, magnesium, calcium, and zinc salts as well as amino acids could more than double the final ethanol concentration to 54 g/l.

Returning to recombinant E. coli, attempts to define the minimum salts concen­tration (to avoid stress imposed by osmotically active solutes) resulted in the formu­lation of a medium with low levels of sodium and other alkali metal ions (4.5 mM) and total salts (4.2 g/l).113 Although this medium was devised during optimization of lactic acid production, it proved equally effective for ethanol production from xylose. Because many bacteria biosynthesize and accumulate internally high concentrations of osmoprotective solutes when challenged with high exogenous levels of salts, sugars, and others, modulating known osmoprotectants was tested and shown to improve the growth of E. coli in the presence of high concentrations of glucose, lactate, sodium lactate, and sodium chloride.114 The minimum inhibitory concentrations of these sol­utes was increased by either adding the well-known osmoprotectant betaine, increas­ing the synthesis of the disaccharide trehalose (a dimer of glucose), or both, and the combination of the two was more effective than either alone. Although the cells’ tol­erance to ethanol was not enhanced, the use of the combination strategy would be expected to improve growth in the presence of the high sugar concentrations that are becoming ever more frequently encountered in media for ethanol fermentations.

Accurately measuring the potential for ethanol formation represented by a cel — lulosic biomass substrate for fermentation (or fraction derived from such a material) is complex because any individual fermentable sugar (glucose, xylose, arabinose, galactose, mannose, etc.) may be present in a large array of different chemical forms: monomers, disaccharides, oligosaccharides, even residual polysaccharides. Precise chemical assays may require considerable time and analytical effort. Bioassay of the material using ethanologens in a set medium and under defined, reproducible condi­tions is preferable and more cost effective — and broadly analogous to the use of shake flask tests to assess potency of new strains and isolates and the suitability of batches of protein and other “complex” nutrients in conventional fermentation laboratories.115

Genetic Manipulation with Genes for Xylose Isomerization

Historically, the earliest attempts to engineer xylose metabolic capabilities into S. cerevisiae involved the single gene for xylose isomerase (XI, catalyzing the intercon­version of xylose and xylulose) from bacteria (E. coli and B. subtilis), but these failed because the heterologous proteins produced in the yeast cells were enzymically inac- tive.4,69 A greater degree of success was achieved using the XI gene (xylA) from the bacterium Thermus thermophilus; the transformants could exhibit ethanol formation in O2-limited xylose fermentations.95

A crucial breakthrough, however, was made in 2003, when the gene encoding XI in the fungus Piromyces sp. strain E2 (capable of anaerobic growth on xylose) was recognized as part of the known “bacterial” pathway for xylose catabolism and, for the first time, was revealed to be functional in a eukaryote.96 The same research group at the Delft Technical University, The Netherlands, soon demonstrated that xylA gene expressed in S. cerevisiae gave high XI activity but could not by itself induce ethanol production with xylose as the carbon source.97 The additional genetic manipulations required for the construction of ethanologenic strains on xylose were the overexpression of XK, transketolase, transaldolase, ribulose 5-phosphate epim — erase, and ribulose 5-phosphate isomerase and the deletion of nonspecific AR, fol­lowed by selection of “spontaneous” mutants in xylose-limited continuous cultures and anaerobic cultivation in automated sequencing-batch reactors on glucose-xylose media.98-100 The outcome was a strain with negligible accumulation of xylitol (or xylulose) and a specific ethanol production three — to fivefold higher than previously publicized strains.4 Mixtures of glucose and xylose were sequentially but completely consumed by anaerobic cultures of the engineered strain in anaerobic batch culture, with glucose still being preferred as the carbon source.99

A side-by-side comparison of XR/XDH — and XI-based xylose utilizations in two isogenic strains of S. cerevisiae with genetic modifications to improve xylose metab­olism (overexpressed XK and nonoxidative pentose phosphate pathway enzymes and deleted AR) arrived, however, at widely different conclusions for the separately opti­mal parameters of ethanol production:101

• In chemically defined medium, the Xl-containing variant showed the high­est ethanol yield (i. e., conversion efficiency) from xylose.

• The XR/XDH transformant had the higher rate of xylose consumption, spe­cific ethanol production, and final ethanol concentration, despite accumu­lating xylitol.

• In a lignocellulose hydrolysate, neither transformant accumulated xylitol, but both were severely affected by toxic impurities in the industrially rel­evant medium, producing little or no ethanol, xylitol, or glycerol and con­suming little or no xylose, glucose, or mannose.

The bacterial XI gene from T. thermophilus was revisited in 2005 when this path­way for xylose utilization was expressed in S. cerevisiae along with overexpressed XK and nonoxidative pentose phosphate pathway genes and deleted AR; the engineered strain, despite its low measured XI activity, exhibited for the first time aerobic growth on xylose as sole carbon source and anaerobic ethanol production at 30°C.102

Even before the end of the 1970s, serious scientific discussion had commenced regarding the thermodynamics of fuel ethanol production from biological sources. The first in-depth study that compiled energy expenditures for ethanol production from sugarcane, sorghum, and cassava under Brazilian conditions concluded that the net energy balance (NEB), that is, the ratio between the energy produced (as ethanol) and the total energy consumed (in growing the plants, processing the harvest, and all the various stages of the ethanol production process) was positive.60 This issue must not be confused with economic price but is one of thermodynamics: does the production of ethanol require a net input when a summation of the inputs is made, including such obviously energetic factors as the heat employed to distill the ethanol from the aqueous fermentation broth and also more subtle energy costs such as those involved in the manufacture of fertilizers and pesticides?

Thermodynamically, energy can be neither created nor destroyed but merely converted from one form into another. Energy is unavoidably expended in the production of ethanol, and the summation can be made across the entire produc­tion process including human labor (that must be replenished, i. e., energy goes into supplying the food for the work force), machinery, fuel, seeds, irrigation, and all agro­chemicals (many of which are derived from petrochemicals); these can be described as “direct” or “mechanical” as distinct from essential environmental energy inputs such as sunlight.[8] The output (i. e., ethanol) has a measurable energy yield in internal combustion engines; the energy inputs I have termed “energy conversion debits” that can be — although they are not unavoidably — fossil fuel consumptions (as diesel, coal, natural gas, etc.). The ratio between energy in the ethanol produced and the energy consumed in the conversion debits is the net energy yield or energy balance.

Four distinct cases can be distinguished with far-reaching implications for mac­roeconomics and energy policy:

1. Ethanol production has a net yield of energy and has no absolute depen­dency on nonrenewable energy inputs.

2. Ethanol production has a net yield of energy but has a dependency on non­renewable energy inputs.

3. Ethanol production has a net loss of energy (a yield of <1) but has no depen­dency on nonrenewable energy inputs.

4. Ethanol production has a net loss of energy as well as being dependent on nonrenewable energy inputs.

Viewed broadly, the most acceptable conclusion economically as well as to policy makers is case 1. Case 2 would pose problems to the short-term adoption of biofu­els, but eventually, there could be no insuperable requirement to either “fund” bio­ethanol by expending fossil-fuel energy as alternative energy sources become more widely available or to support the required agricultural base with agrochemicals derived exclusively from petrochemicals or other nonrenewable sources (especially if fermentation-derived bioprocesses supplanted purely synthetic routes). Moreover, cases 3 and 4 are not in themselves intrinsically unacceptable — as Sama pointed out nearly 20 years ago, power stations burning oil or coal all show a net energy loss but that is not used as an argument to shut them down.61 Domestic and industrial devices that are electrically driven require power stations as an interface for energy conversion, with the associated economic and thermodynamic costs. Filling gaso­line tanks with sugar, corncobs, or wood chips cannot fuel automobiles; converting these biological substrates to ethanol can — but again at economic and energetic costs. Case 3 and especially case 4 are equivalent to wartime scenarios where fuel is produced “synthetically”; case 3 is one of sustainable, long-term development if nonfossil energy sources such as hydroelectric, geothermal, wind, wave, and solar can be used to meet the net energy costs inherent in bioethanol production.

A review of a broad portfolio of renewable energy technologies in the mid-1970s that included methane production from algae and livestock waste as well as sugar­cane, cassava, timber, and straw concluded that only sugarcane-derived ethanol was capable of yielding a net energy gain.62 Data from early Brazilian and U. S. studies of sugarcane-derived ethanol are summarized in table 1.7. It was evident even in these early studies that the NEB was highly influenced by the utilization of bagasse, that is, the combustion of the waste product to supply steam generation and other energy requirements on-site or (although not considered at that time) as a saleable power commodity to the broader community. The energy balance was, however, much lower than the comparative quoted case of gasoline produced from oil extracted in the Gulf of Mexico oil fields, that is, 6:1.63

Several reports of the technical feasibility of ethanol production from corn had been published by 1978. Researchers from the University of Illinois, Urbana, collected the data and developed a mathematical model that extended the analysis to include mileage estimates and measurements for 10% ethanol blends with gasoline.64 From the data, the overall energy balance could be computed to be small (approximately 1.01); even attaining the energy break-even point required the efficient utilization of wastes for on-site energy generation and the sale and agricultural use of by-products counted as energy outputs (see figures 1.21 and 1.22). Few data were available at that time for mileages achieved with gasohol blends; if positive advantages could be gained, this would have translated into significantly increased net energy gains in what was, in effect, the first testing of a “well-to-wheel” model for fuel ethanol.64

Subsequently, both the energetics of corn — and sucrose-derived bioethanol pro­cesses have been the subjects of scrutiny and increasingly elaborate data acquisition and modeling exercises. Economic analyses agree that the production of ethanol from sugarcane is energetically favorable with net energy gains that rival or equal those for the production of gasoline from crude oil in subterranean deposits.65,66 For corn-derived fuel alcohol, however, the energy position is highly ambiguous. By 2002, that is, before the steep rise in world oil prices (see figures 1.3 and 1.11) — a factor that might easily have exercised undue weight on the issue — conflicting estimates had appeared, but the published accounts during two decades displayed very different methodological approaches and quantitative assumptions. Even as crucial a parameter as the energy content of ethanol used in the computations varied in the range 74,680 to 84,100 Btu/gallon — the choice of lower or higher heat values (measured by reference to water or steam, respectively) is also required for the energy inputs, and if the choice is used consistently in the calculations, there will be no effect on the net energy gained or lost. Ten of the studies were discussed in a 2002 review.67-76 Table 1.8 collects the data, adding to the comparison an extra publication from 2001.77 The obvious spread of energy balance values has generated a sustained argument over the choice of relevant input parameters, but a more per­tinent conclusion is that none of the balances greatly exceeded 1.00 (the arithmeti­cal average is 1.08); it could have been concluded before 2002, therefore, that only a highly efficient production process (including the maximal utilization of what­ever by-products were generated) could deliver a net energy gain, although changes in agricultural practices, higher crop yields, increased fermentation productivity, and others might all be anticipated to contribute to a gradual trend of piecemeal improvement.

Since 2002, the polemics have continued, if anything, with an increased impres­sion of advocacy and counteradvocacy. Among the peer-reviewed journal articles, the following considered NEB values (on a similar mathematical basis to that used in tables 1.7 and 1.8), but with various assumptions and calculations for energy input data:

• Patzek78 — 0.92 (with no by-product energy credits allowed).

• Pimentel and Patzek79 — 0.85 (after adjustment for the energy in by-products).

• Dias de Oliveira et al.80 — 1.10 (compared with 3.7 for sugar-derived ethanol in Brazil).

• Farrell et al.81 — 1.20.

• Hill et al.58 — 1.25, although this depended mostly on counting the energy represented by the DDGS by-product (Figure 1.21) into the calculations for NEB and NEB ratio.

The continued failure to demonstrate overall NEBs much more than 1.00 is again striking. This has provided an impetus to define better metrics, including decreased fossil fuel usage and greenhouse gas emissions in the assessments.81 Put into the perspective of the historical measures of energy balance from the oil industry corn — derived ethanol remains relatively inefficient.82-84 Assuming an NEB of 1.2, the notional expenditure of 5 units of bioethanol would be required to generate each net unit. This argument against ethanol as a biofuel is, of course, far less persuasive if all the energy needs for bioethanol production are met by renewable sources — this was highly speculative in the 1970s, and energy inputs to biofuels have (implicitly or explicitly) assumed a fossil fuel basis. With an energy input defined as “nonrenewable,” Hammerschlag normalized data from six studies to compute a range of “energy return on investment,” that is, total product energy divided by the nonrenewable energy to its manufacture, of 0.84-1.65.86 For an economy like Brazil’s, where hydroelectricity is the single largest source of power generation (64% in 2002), renewable energy is a practical option, but for the world at large (where hydroelectricity and all other renewable energy sources account for less than 10% of total power generation), the dependency on coal, oil, and natural gas is likely to remain for some decades.87

In comparison with corn ethanol, biomass-derived ethanol has received less attention. Estimates of the NEB range from the “pessimistic” (0.69 from switchgrass [Panicum virgatum], 0.64 from wood) to the “optimistic” (2.0).79,81 Hammerschlag suggested a range for cellulosic ethanol of 4.40-6.61 but ignored one much lower value.86 The Greenhouse gases, Regulated Emissions and Energy use in Transportation (GREET) model developed by the Argonne National Laboratory (Argonne, Illinois) also predicts large savings in petroleum and fossil fuels for ethanol produced from switchgrass as the favored candidate lignocellulosic feedstock.88 Technologies for bioethanol manufactured from nonfood crops appear, therefore, to warrant further attention as energy-yielding processes capable of narrowing the considerable gap in energy gain and expenditure that exists between corn-derived ethanol production and the oil industry.

Why the large discrepancy between corn-derived and lignocellulosic etha­nol? Intuitively, lignocellulosic substrates are more intractable (and, therefore, more costly to process) than are starches. As all practitioners of mathemati­cal modeling appreciate, the answer lies in the assumptions used to generate the modeling. Specifically, modeling methodologies for cellulosic ethanol processes assume that electricity is generated on-site from the combustion of components of biomass feedstocks not used for fermentation (or that are unused after the fer­mentation step) in combined heat and power plants; this was built into the earli­est detailed models of biomass ethanol production in studies undertaken by the Argonne National Laboratory and by the National Renewable Energy, Oak Ridge National, and the Pacific Northwest Laboratories, and the results are pivotal not only for energy balances but also for reductions in pollutant emissions.81,88,89 This methodology approximates energy use in biomass (cellulose) ethanol plants to that in sugar ethanol plants in Brazil (see table 1.7).

In Europe, where bioethanol production is much less developed than in North America and Brazil, attention has focused on wheat starch and beet sugar. Studies of energy balances with wheat have consistently shown negative energy balances (averaging 0.74), whereas values for sugarbeet range from 0.71 to 1.36 but average 1.02, although projections made by the International Energy Agency suggest that both feedstocks will show positive energy balances as combined fertilizer and pesti­cide usage drops and biotechnological conversion efficiencies improve.90

The question of deriving metrics adequate to accurately compare fuel alcohol and gasoline production processes also continues to exercise the imagination and inventiveness, particularly of European analysts. Portuguese analysts defined energy renewability efficiency (ErenEf) as:

ErenEf = (FEC — Ein, ^ц, prim) x 100 / FEC

where FEC is the fuel energy content and Ein, fossil, prim represents the unavoidable fossil energy input required for the production of the biofuel. Under French conditions, sugarbeet-derived ethanol was renewable even without taking into account any coproduct credits but was maximal with an allocation of the energy inputs based on the mass of the ethanol and coproducts (ErenEf = 37%), whereas wheat grain-derived ethanol was entirely dependent on this allocation calculated for the DDGS (ErenEf = 48%).91 These values were equivalent to positive NEBs of 1.59 (sugarbeet) and 1.92 (wheat). Conversion of the ethanol from either source to ethyl tertiary butyl ether by reaction with petroleum-derived isobutylene results in a product superior as an oxygenate to MTBE, but the energy gains calculated for the fuel ethanol were almost entirely lost if this extra synthetic step was included.

Accepting that the fossil fuel requirement of corn-derived ethanol approximates the net energy value of the product, Nielsen and Wenzel then advanced the argu­ment that gasoline requires an equal amount of energy as fossil fuels in its pro­duction — the net fossil fuel saving when counting in the gasoline saved when corn ethanol is combusted in an automobile therefore generates a fossil energy saving 90% of that of the fuel alcohol.92 No supporting data were quoted for this assertion, that is that gasoline required such a high outlay of fossil fuel, and other analysts markedly disagree.82-84 A commentator from the Netherlands argued that the opportunity costs for crop production must be taken into account, that is, that the energy costs for corn- and biomass-derived ethanol cannot include those implicit in the generation of the fermentation substrate unless the land for their production is otherwise left idle.93 This is again contentious, as no arable land must be gainfully farmed; as a citizen of a European Union state, the Dutch author would have been aware that land deliberately left idle even has a monetary value: the so-called “set-aside” provisions of the Common Agricultural Policy aim to financially encourage farmers not to overproduce agricultural surpluses, which otherwise would require a large financial outlay for their storage and disposal (see chapter 5, section 5.2.3). Even with this economic adjustment, the production of bioethanol was not a net energy process with either switchgrass or wood as the biomass input.

When analysts tacitly assume that all (or most) of the energy required for biofuel production is inevitably derived from fossil fuels, this is equivalent to the

International Energy Agency’s Reference Scenario for future energy demands, that is, that up until 2030, coal, gas, and oil will be required for 75% of the world’s power generation, with nuclear, hydro, biomass, and other renewables accounting for the remaining 25%.94 The agency’s alternative scenario for 2030 postulates fossil fuel usage for power generation decreasing to 65% of the total. In the future, there­fore, the fossil fuel inputs and requirements for corn — and biomass-derived ethanol may significantly decrease, thus affecting the quantitative assessment of (at least) the fossil fuel energy balance. As an interim measure, however, defining the crucial net energy parameter as the ratio between the energy retrieved from ethanol and the fossil fuel energy inputs involved in its production, the “fossil energy ratio” or “bioenergy ratio”66 gives:

(Enet + Ecoproduct)/(EA + EB + EC +ED)

where Ea is the fossil fuel energy required for the production of the plant inputs, EB is the fossil fuel energy required during crop growth and harvesting, Ec is the fossil fuel required during transport of the harvested crop, and Ed is that expended during the conversion process.

On this basis, different industrial processes could result in widely divergent bioenergy ratios (figure 1.26). The most obvious inconsistency is that between two molasses-derived ethanol processes, one (in India) was derived from the case of a distillery fully integrated into a sugar mill, where excess low-pressure steam was used for ethanol distillation, whereas a South African example was for a distillery distant from sugar mills and reliant on coal and grid electricity for its energy needs.66 In terms of energy, corn and corn stover and wheat and wheat straw were all inferior to Brazilian sugarcane, whereas Indian bagasse was a biomass source that gave a high result.

FIGURE 1.26 Ratio of ethanol energy content to fossil fuel energy input for ethanol production systems. (Data from von Blottnitz and Curran.66)

After more than two decades of intensive molecular genetic research, S. cerevisiae remains the ethanologen of choice. Although industrial strains are genetically largely undefined, it is easily demonstrated that laboratory strains can function perfectly adequately for ethanol production, even for the complex process of manufacturing potable spirits with defined requirements for sensory parameters in the finished prod­uct such as volatiles, higher alcohols, and glycerol content.273 Nevertheless, it is more likely that applying the knowledge gained to the genomic improvement of hardy industrial Saccharomyces (or other Crabtree-positive yeast) strains with proven track records of fermenting very concentrated media to high volumetric yields of ethanol would generate highly suitable biocatalysts for the demanding tasks of growing and metabolizing sugars and oligosaccharides in lignocellulosic hydrolysates.

Even better would be the importing of a methodology developed in the world of industrial biotechnology, that is, “genome breeding,” as outlined by Kyowa Hakko Kogyo,[30] Tokyo, Japan. By comparing the whole genome sequence from the wild — type Corynebacterium glutamicum, the major producing organism for L-lysine and L-glutamate, with gene sequence from an evolved highly productive strain, it was possible to identify multiple changes; with these data, transforming a “clean” wild type to a hyperproducer of lysine was accomplished with only three specific and known mutations in the biosynthetic genes.274 This minimal mutation strain had dis­tinct productivity advantages over the industrial strains that had been developed by chance mutation over decades — a reflection of how many unwanted changes may (and did) occur over prolonged periods of random mutagenesis and selection in the twentieth century.275 Repeating such an exercise with any of the multitude of com­mercial alcohol-producing Saccharomyces yeasts would rapidly identify specific genomic traits for robustness and high productivity on which to construct pentose­utilizing and other capabilities for bioethanol processes.

A radically different option has been outlined in a patent application at the end of May 2007.276 Work undertaken at the J. Craig Venter Institute, Rockville, Maryland, defined a minimal set of 381 protein-encoding genes from Mycoplasma genitalium, including pathways for carbohydrate metabolism, nucleotide biosynthesis, phospho­lipid biosynthesis, and a cellular set of uptake mechanisms for nutrients, that would suffice to generate a free living organism in a nutritionally rich culture medium. Adding in genes for pathways of ethanol and/or hydrogen formation would result in a biofuels producer with maximum biochemical and biotechnological simplicity. The timeline required for practical application and demonstration of such a synthetic organism is presently unclear, although the research is funded by the DOE, under the genome projects of the Department’s Office of Science, with the target of develop­ing a novel recombinant cyanobacterial system for hydrogen production from water and a cellulosome system for the production of ethanol and/or butanol in suitable clostridial cells.277 The drawbacks to such an approach are that it would be highly dependent on the correct balance of supplied nutrients to the organism (with its lim­ited capabilities), requiring highly precise nutrient feeding mechanisms and a likely protracted optimization of the pathways to rival rates of product formation already attainable with older patented or freely available ethanologens.

The consequence of these limitations of cellulases as catalysts for the degradation of cellulose is that large amounts of cellulase have been considered necessary to rapidly process pretreated lignocellulosic substrates, for example, 1.5-3% by weight
of the cellulose, thus imposing a high economic cost on cellulose-based ethanol production.91 From the standpoint of the established enzyme manufacturers, this might have been a welcome and major expansion of their market. The multiplicity of other uses of cellulases meant, however, that competition resulted in massive increases in cellulase fermentation productivity — in the 1980s, space-time yields for cellulases increased by nearly tenfold; between 1972 and 1984, total cellulase production doubled every two years by the selection of strains and the development of fed-batch fermentation systems.70 Some trends in Hypocrea jecorina cellulase productivity and costs were published by the Institut Frangais du Petrole from their pilot-plant scale-up work (figure 2.8).92 By 1995, the productivities in industrial fermentations were probably 400% higher, and process intensification has undoubt­edly continued.93 The development of some of the H. jecorina strains from one originally isolated at the U. S. Army Laboratory (Natick, Massachusetts) by physical and chemical mutagenesis treatments has also been described (figure 2.9).94 From a baseline price of $15/kg in 1990, cellulases have decreased markedly, but they still average three to five times the cost of the much more readily produced (and more enzymically active) amylases.95

Major enzyme manufacturers including Genencor (Palo Alto, California), Novozymes (Denmark), Iogen Corporation (Canada), and Rohm (Finland) have long produced cellulases from organisms including H. jecorina, Aspergillus ory — zae, A. niger, Humicola insolens, and Penicillium spp. Different producing organ­isms yield cellulases with different profiles of cellulase components (figure 2.10), and nonbiofuel markets recognize premium products from particular biological sources with optimum properties for particular applications.70

5-і

-C >

и

5

i

For all microbial producers of cellulase, however, a key area of knowledge is that of the genetic regulation of cellulase synthesis in the life cycle of the organism and — most acutely — when the producing cells are functioning inside the fermentor. For a fungus such as H. jecorina, the evidence points to the low-level basal production
of cellulase components generating an inducer of rapid cellulase synthesis when cellulose enters the nutritional environment:

• Deletion of the genes coding for discrete cellulase components prevents the expression of other cellulase genes.96,97

• The general carbon catabolite repressor protein CRE1 represses the transcription of cellulase genes, and a hyperproducing mutant has a crel mutation rendering cellulase production insensitive to glucose.98,99

• Production of cellulases is regulated at the transcriptional level, and two genes encoding transcription factors have been identified.100-102

• The disaccharide sophorose (P-1,2-glucopyranosyl-D-glucose) is a strong inducer of cellulases in H. jecorina.103-105

• Cellulases can also catalyze transglycosylation reactions.106

Low activities of cellulase are theorized to at least partially degrade cellulose, liberating cellobiose, which is then transglycosylated to sophorose; the inducer stimulates the transcription of cellulase genes, but this is inhibited if glucose accumulates in the environment, that is, there is a triple level of regulation.107-109 This overall strategy is a typical response of microbes to prevent the “unneces­sary” and energy-dependent synthesis and secretion of degradative enzymes if readily utilizable carbon or nitrogen is already present. Augmenting this elabo­rate mechanism, sophorose also represses P-glycosidase; because sophorose is hydrolyzed by P-glycosidase, this repression acts to maintain sophorose concentra­tions and thus maximally stimulate cellulase formation.105 Another disaccharide, gentiobiose (P-1,6-glucopyranosyl-D-glucose), also induces cellulases in H. jeco­rina.105 In anaerobic cellulolytic bacteria, cellobiose may be the inducer of cellu­lase, but whether induction occurs in anaerobic bacteria is unclear.60 In at least one clostridial species, cell wall-attached cellulosomes are formed during growth on cellulose but not on cellobiose; although the cellulase synthesis is determined by carbon catabolite repression, cellulose (or a breakdown product) is a “signal” that can be readily recognized by the cells.69,110

Can sophorose be used to increase fungal cellulase expression in fermentations to manufacture the enzyme on a large scale? As a fine chemical, sophorose is orders of magnitude more expensive than glucose,[15] and its use (even at low concentrations) would be economically unfeasible in the large fermentors (>50,000 l) mandated for commercial enzyme manufacture. Scientists at Genencor discovered, however, that simply treating glucose solutions with H. jecorina cellulase could generate sophorose — taking advantage of the transglycosylase activity of P-glucosidase — to augment cellulase expression and production in H. jecorina cultures.111 Lactose is used indus­trially as a carbon source for cellulase fermentations to bypass the catabolite repres­sion imposed by glucose but adding cellulase-treated glucose increased both cellulase production and the yield of enzyme per unit of sugar consumed (figure 2.11).

The improving prospects for cellulase usage in lignocellulosic ethanol produc­tion has engendered an intense interest in novel sources of cellulases and in cellu — lase-degrading enzymes with properties better matched to high-intensity cellulose saccharification processes. Both Genencor and Novozymes have demonstrated tangible improvements in the catalytic properties of cellulases, in particular, thermal stability; such enzyme engineering involved site-directed mutagenesis and DNA shuffling (table 2.7). Other discoveries patented by Genencor derived from an extensive and detailed study of gene expression in H. jecorina that revealed 12 previously unrecognized enzymes or proteins involved in polysaccharide degradation; some of these novel proteins may not function directly in cellulose hydrolysis but could be involved in the production and secretion of the cellulase complex or be relevant when other polysaccharides serve as growth substrates.112 The “traditional” and long-established four major components of H. jecorina cellulase — two cellobiohydrolases and two endoglucanases — together constitute more than 50% of the total cellular protein produced by the cells under inducing conditions and can reach more than 40 g/l in contemporary industrial strains that are the products of many years of strain development and selection.112,113 The poor performance of H. jecorina as a cellulase producer — once described as the result of nature opting for an organism secreting very large amounts of enzymically incompetent protein rather than choosing an organism elaborating small amounts of highly active enzymes84 — have engendered many innovative and speculative

TABLE 2.7

Post-2000 Patents and Patent Applications in Cellulase Enzymology and Related Areas

studies on radical alternatives to the “Trichoderma cellulase” paradigm. Proxi­mal to commercially realizable applications are cellulases immobilized on inert carriers that can offer significant cost savings by the repeated use of batches of stabilized enzymes.114115 With a commercial P-glucosidase from Aspergillus niger, immobilization resulted in two important benefits: greatly improved thermal stability at 65°C and a quite unexpected eightfold increase in maximal enzyme activity at saturating substrate concentration, as well as operational stability dur­ing at least six rounds of lignocellulose hydrolysis.116

Evidence for an unambiguously novel type of cellulose-binding protein in H. jecorina has, however, resulted from the discovery of a family of “swollenins,” proteins that bind to macroscopic cellulose and disrupt the structure of the cellulose fibers without any endoglucanase action.117 This fulfils a prediction made by some of the pioneers of cellulase biotechnology who envisaged that “swelling” factors would be secreted by the fungus to render cellulose more susceptible to cellulase-catalyzed attack.118 Fusing cellulose-disrupting protein domains with cellulase catalytic domains could generate more powerful artificial exo — and endoglucanases. Whether cellulose­binding domains/modules in known cellulases disrupt cellulose structures remains unclear.119 Adding a nonionic detergent to steam-pretreated barley greatly increased total polysaccharide saccharification, and the detergent may have been acting partly as a lignocellulose disrupter.120 An unexpected potential resource for laboratory-based evolution of a new generation of cellulases is the very strong affinity for cellulose exhibited not by a cellulase but by a cellobiose dehydrogenase; nothing has yet been disclosed of attempts to combine this binding activity with cellobiohydrolases.121

The relatively low activities of P-glycosidases in fungal cellulase preparations — possibly an unavoidable consequence of the sophorose induction system—have been considered a barrier to the quantitative saccharification of cellulose. Supplementing H. jecorina cellulase preparations with P-glucosidase reduces the inhibitory effects caused by the accumulation of cellobiose.122 Similarly, supplementing cellulase preparations with P-glucosidase eliminated the measured differences in saccharification rates with solvent-pretreated hardwoods.123 Mixtures of cellulase from different cellulolytic organisms have the advantage of maximally exploiting their native traits.124 Additionally, such a methodologically flexible approach could optimize cellulase saccharification for a single lignocellulosic feedstock with significant seasonal or yearly variation in its composition or to process differing feedstock materials within a short or medium-length time frame.

Finally, novel sources of cellulases have a barely explored serendipitous potential to increase the efficiency of saccharification; for example, cellulases from “nonstandard” fungi (Chaetomium thermophilum, Thielavia terrestris, Thermoascus aurantiacus, Corynascus thermophilus, and Mycellophthora thermophila), all thermophiles with high optimum growth temperatures in the range of 45-60oC, improved the sugar yield from steam-pretreated barley straw incubated with a benchmark cellulase/ P-glucosidase mix; the experimental enzyme preparations all possessed active endo — glucanase activities.125 “Biogeochemistry” aims to explore the natural diversity of coding sequences available in wild-type DNA.126 Forest floors are an obvious source of novel microbes and microbial communities adept at recycling lignocelluloses.

Until the 1980s, the general brewing industry view of yeasts for alcohol production was that most could tolerate only low concentrations (7-8% by volume) of ethanol and, consequently, fermentation media (worts) could be formulated to a maximum of 15-16° (Plato, Brix, or Balling, depending on the industry subsector), equivalent to 15-16% by weight of a sugar solution; the events that radically changed this assessment of yeasts and their physiology were precisely and cogently described by one of the key players:116

• When brewers’ yeasts were grown and measured in the same way as the more ethanol-tolerant distillers’ and sake yeasts, differences in ethanol tol­erance were smaller than previously thought.

• “Stuck” fermentations, that is, ones with little or no active growth in supraopti — mal sugar concentrations, could easily be rescued by avoiding complete anaero — biosis and supplying additional readily utilizable nitrogen for yeast growth.

• By removing insoluble grain residues (to reduce viscosity), recycling clear mashes to prepare more concentrated media from fresh grain, optimizing yeast nutrition in the wort, and increasing cooling capacity, yeast strains with no previous conditioning and genetic manipulation could produce ethanol up to 23.8% by volume.

Very-high-gravity (VHG) technologies have great technical and economic advantages:

1. Water use is greatly reduced.

2. Plant capacity is increased, and fermentor tank volume is more efficiently utilized.

3. Labor productivity is improved.

4. Fewer contamination outbreaks occur.

5. The energy requirements of distillation are reduced because fermented broth is more concentrated (16-23% v/v ethanol).

6. The spent yeast can be more readily recycled.

7. The grain solids removed prefermentation can be a valuable coproduct.

With an increased volume of the yeast starter culture added to the wort (higher “pitching rate”) and a prolonged growth phase fueled by adequate O2 and free amino nitrogen (amino acids and peptides), high-gravity worts can be fermented to ethanol concentrations more than 16% v/v even at low temperatures (14°C) within a week and with no evidence of any ethanol “toxicity.”117-120

This is not to say, however, that high ethanol concentrations do not constitute a stress factor. High-alcohol-content worts do still have a tendency to cease fer­mentation, and high ethanol levels are regarded as one of the four major stresses in commercial brewing, the others being high temperature, infection (contamination, sometimes associated with abnormal pH values), and mycotoxins from grain car­rying fungal infections of Aspergillus, Penicillium, Fusarium, Claviceps, or Acre — monium species.121 To some extent, the individual stress factors can be managed and controlled — for example, in extremis, antimicrobial agents that are destroyed during distillation (so that no carryover occurs to the finished products) can be added even in potable alcohol production. It is when the major stresses combine that unique conditions inside a fermentor can be generated. For a potable alcohol producer, these can be disastrous because there is an essential difference between the products of fuel/industrial ethanol and traditional alcoholic beverages: the latter are operated for consistency in flavor and quality of the product; for the beverage producer, flavor and quality outweigh any other consideration — even distinct economic advantages associated with process change and improvement — because of the market risks, especially if a product is to be matured (“aged”) for several years before resale.122 Industrial ethanol is entirely amenable to changes in production practice, strain, trace volatile composition, and even process “excursions” when the stress factors result in out-of-tolerance conditions. Yeast (S. cerevisiae) cells may have the ability to reduce short-term ethanol toxicity by entering a “quiescent” state in their average popula­tion cell cycle, extending a phase of growth-unassociated ethanol production in a laboratory process developed to produce 20% ethanol by volume after 45 hours.123

From the work on VHG fermentations, the realization was gained that typical media were seriously suboptimally supplied with free amino acids and peptides for the crucial early growth phase in the fermentation; increasing the free amino nitrogen content by more than fourfold still resulted in the exhaustion of the extra nitrogen within 48 hours (figure 4.6). With the correct supplements, brewer’s yeast could consume all the fermentable sugars in a concentrated medium (350 g/l) within eight days at 20°C or accumulate 17% (v/v) ethanol within three days.124 Fresh yeast autolysate was another convenient (and cost-effective) means of nitrogen supplemen­tation with an industrial distillery yeast from central Europe — although, with such a strain, while nitrogen additions improved final ethanol concentration and glucose utilization, none of them increased cell viability in the late stages of the fermentation, ethanol yield from sugar, or the maximum rate of ethanol formation.125 Commercial

proteases can liberate free amino acids and peptides from wheat mash, the low — molecular-weight nitrogen sources increasing the maximal growth (cell density) of the yeast cells and reducing the fermentation time in VHG worts from nine to three days — and without (this being an absolute priority) proteolytic degradation of the glucoamylase added previously to the mash to saccharify the wheat starch; rather than adding an extra nitrogen ingredient, a one-time protease digestion could replace medium supplementation.126 Not all amino acids are beneficial: lysine is severely inhibitory to yeast growth if the mash is deficient in freely assimilated nitrogen, but adding extra nitrogen sources such as yeast extract, urea, or ammonium sulfate abol­ishes this effect, promotes uptake of lysine, increases cell viability, and accelerates the fermentation.127

Partial removal of bran from cereal grains (wheat and wheat-rye hybrids) is an effective means of improving the mash in combination with VHG tech­nology with or without nitrogen supplementation (figure 4.7); in a fuel alcohol plant, this would increase plant efficiency and reduce the energy required for heating the fermentation medium and distilling the ethanol produced from the VHG process.128 Conversely, adding particulate materials (wheat bran, wheat mash insolubles, soy or horse gram flour, even alumina) improves sugar utiliza­tion in VHG media: the mechanism may be to offer some (undefined) degree of osmoprotection.129,130

A highly practical goal was in defining optimum conditions for temperature and mash substrate concentration with available yeast strains and fermentation hardware: with a wheat grain-based fermentation, a temperature of 30°C and an initial mash specific gravity of 26% (w/v) gave the best balance of high ethanol productivity, final ethanol concentration, and shortest operating time.131 The conclusions from such investigations are, however, highly dependent on the yeast strain employed and on the type of beer fermentation being optimized: Brazilian investigators working with

FIGURE 4.7 Grain pearling and very high-gravity grain mash fermentations for fuel ethanol production. (Data from Wang et al.128)

a lager yeast strain found that a lower sugar concentration (20% w/v) and temperature (15°C) were optimal together with a triple supplementation of the wort with yeast extract (as a peptidic nitrogen source), ergosterol (to aid growth), and the surfactant Tween-80 (possibly, to aid O2 transfer in the highly concentrated medium).132

For VHG fermentations, not only is nitrogen nutrition crucial (i. e., the sup­ply of readily utilizable nitrogen-containing nutrients to support growth) but other medium components require optimization: adding 50 mM of a magnesium salt in tandem with a peptone (to supply preformed nitrogen sources) increased ethanol concentrations from 14.2% to 17% within a 48-hr fermentation.133 These results were achieved with a medium based on corn flour (commonly used for ethanol produc­tion in China), the process resembling that in corn ethanol (chapter 1, section 1.4) with starch digestion to glucose with amylase and glucoamylase enzyme treatments. With a range of nutrients tested (glycine, magnesium, yeast extract, peptone, bio­tin, and acetaldehyde), cell densities could dramatically differ: the measured ranges were 74-246 x 106 cells/g mash after 24 hours and 62-392 x 106 cells/g mash after 48 hours. A cocktail of vitamins added at intervals in the first 28-37 hours of the fermentation was another facile strategy for improving final ethanol concentration, average ethanol production rate, specific growth rate, cell yield, and ethanol yield — and a reduced glycerol accumulation.134 Small amounts of acetaldehyde have been claimed to reduce the time required to consume high concentrations of glucose (25% w/v) in VHG fermentations; the mechanism is speculative but could involve increas­ing the intracellular NAD:NADH ratio and accelerating general sugar catabolism by glycolysis (figure 3.1).135 Side effects of acetaldehyde addition included increased accumulation of the higher alcohols 2,3-butanediol and 2-methylpropanol, exem­plifying again how immune fuel ethanol processes are to unwanted “contaminants”
and flavor agents so strictly controlled in potable beverage production. Mutants of brewer’s yeast capable of faster fermentations, more complete utilization of wort carbohydrates (“attenuation”), and higher viability under VHG conditions are eas­ily selected after UV treatment; some of these variants could also exhibit improved fermentation characteristics at low operating temperature (11°C).136

Ethanol diffuses freely across cell membranes, and it seems to be impossible for yeast cells to accumulate ethanol against a concentration gradient.137 This implies that ethanol simply floods out of the cell during the productive phases of alcoholic fermen­tations; the pioneering direct measurement of unidirectional rate constants through the lipid membrane of Z. mobilis confirmed that ethanol transport does not limit etha­nol production and that cytoplasmic ethanol accumulation is highly unlikely to occur during glucose catabolism.138 Nevertheless, even without such an imbalance between internal and external cellular spaces, product inhibition by ethanol is still regarded as an inhibitor of yeast cell growth, if not of product yield, from carbohydrates.139 Yeasts used for the production of sake in Japan are well known as able to accumulate ethanol in primary fermentations to more than 15% (v/v), and both Japanese brewing compa­nies and academic centers have pursued the molecular mechanisms for this:

• With the advent of genomics and the complete sequencing of the S. cerevi — siae genome, whole-genome expression studies of a highly ethanol-tolerant strain showed that ethanol tolerance was heightened in combination with resistance to the stresses imposed by heat, high osmolarity, and oxida­tive conditions, resulting in the accumulation internally of stress protec­tant compounds such as glycerol and trehalose and the overexpression of enzymes, including catalase (catalyzing the degradation of highly reactive hydrogen peroxide).140

• Inositol synthesis as a precursor of inositol-containing glycerophospholip — ids in cellular membranes is a second factor in membrane properties alter­ing (or altered by) ethanol tolerance.141

• Disrupting the FAA1 gene encoding a long-chain fatty acid acyl-CoA syn­thetase and supplying exogenously the long-chain fatty acid palmitic acid were highly effective in stimulating growth of yeast cells in the presence of high ethanol concentrations.142

• Ethanol stress provokes the accumulation of the amino acid L-proline, otherwise recognized as a defense mechanism against osmotic stress; disrupting a gene for proline catabolism increased proline accumulation and ethanol tolerance.143

• Part of the proline protective effect involves proline accumulation in inter­nal vacuoles — heat shock responses are, however, not changed, and this clearly differentiates cellular and biochemical mechanisms in the various stress reactions.144

Multiple sites for how sake yeasts have adapted (and, presumably, can further adapt) to high ethanol concentrations strongly suggest that continued “blind” selection of mutants that are fitter (in an imposed, Darwinian sense) to function despite the stresses of VHG media might be fruitful in the short to medium term.145 Eventually, however, the need to rationally change multiple sites simultaneously to continue improving the biological properties of yeast ethanologens will require a more proactive use of genomic knowledge.146 The positive properties of sake yeasts can, however, be easily transmitted to other yeast strains to ferment high-gravity worts.147 A compromise between “scien­tific” and traditional methodologies for fuel ethanol production may be to generate fus — ants between recombinant ethanologens and osmo — and ethanol-tolerant sake strains.

A last footnote for sake brewing (but not for bioethanol production) is that the high ethanol concentrations generated during the fermentation extract the antioxi­dant protein thioredoxin from the producing cells so that readily detectable levels of the compound persist in the final sake product.148 In addition to its antioxidant func­tion, thioredoxin is anti-inflammatory for the gastric mucosa and, by cleaving disul­fide bonds in proteins, increases protein digestibility, and sake can be considered as a development stage for “functional foods.”

Xylose reductase, the first step in the pathway of xylose catabolism in most yeast species, functions as an enzyme equally well with L-arabinose as with D-xylose, with a slightly higher affinity (lower Km) and a higher maximal rate (Vmax) for l-arabinose.59 Polyol dehydrogenases, active on xylitol, on the other hand, find either L — or D-arabinose to be a poor substrate.103 Although the XDH activity from P. stipitis was kinetically investigated in 1989, little is known about its func­tional physiology; the catalyzed reaction is reversible but activity is unlikely to be regulated by the NAD/NADH balance inside the cell.104 This yeast also con­tains a second XDH, quite distinct from the well characterized xyl2 gene product, but its role is presently undefined in either xylose catabolism or ethanol produc — tion.105 The NAD-specific XR from S. cerevisiae itself is even less well charac­terized, although the enzyme activity is induced by xylose with the wild-type organism.106

An outline of known enzyme-catalyzed metabolic relationship for pentitols and pentoses is given in figure 3.6; some of these pathways are of increasing contempo­rary interest because either they or their engineered variants could lead to the syn­thesis by whole cells (or in biotransformations with isolated enzymes) of “unnatural” or rare sugars useful for the elaboration of antibiotic or antiviral drugs — this is discussed later in chapter 8 when the Green Chemistry of the biorefinery concept for processing agricultural residues is discussed in depth.

Progress in defining the actual pathways operating in known ethanologenic yeasts was rapid after the year 2000

• gene encoding an L-xylulose reductase (forming xylitol; NADP-depen — dent) was then demonstrated in H. jecorina and overexpressed in S. cere — visiae; the l-arabinose pathway uses as its intermediates l-arabinitol, l-xylulose, xylitol, and (by the action of XDH) D-xylulose; the xylulose reductase exhibited the highest affinity to l-xylulose, but some activity was shown toward d-xylulose, d-fructose, and l-sorbose.109

• In H. jecorina, deletion of the gene for XDH did not abolish growth because ladl-encoded l-arabinitol 4-dehydrogenase compensated for this loss — however, doubly deleting the two dehydrogenase genes abolished the ability to grow on either d-xylose or xylitol.110

With this knowledge, expressing the five genes for L-arabinose catabolism in S. cere — visiae enabled growth on the pentose and, although at a low rate, ethanol production from l-arabinose under anaerobiosis.111 In the same year (2003), the genes of the shorter bacterial pathway for L-arabinose catabolism were inserted into S. cerevi — siae.112 The bacterial pathway (active in, e. g., B. subtilis and E. coli) proceeds via l-ribulose, l-ribulose 5-phosphate, and d-xylulose 5-phosphate (figure 3.6), using the enzymes L-arabinose isomerase, L-ribulokinase, and L-ribulose 5-phosphate epimerase. The coexpression of an arabinose-transporting yeast galactose permease allowed the selection on L-arabinose-containing media of an L-arabinose-utilizing yeast transformant capable of accumulating ethanol at 60% of the theoretical maxi­mum yield from L-arabinose under O2-limiting conditions.112