3.2.1 Increased Pentose Utilization by Ethanologenic Yeasts by Genetic

Manipulation with Yeast Genes for Xylose Metabolism via Xylitol

It has been known for many years that S. cerevisiae cells can take up xylose from nutrient media; the transport system is one shared by at least 25 sugars (both natural and synthetic), and only major alterations to the pyranose ring structure of hexoses (e. g., in 2-deoxyglucose, a compound that lacks one of the hydroxyl groups of glu­cose) reduce the affinity of the transport system for a carbohydrate.53 Moreover, both D-xylose and L-arabinose can be reduced by S. cerevisiae, the products being the sugar alcohols, xylitol, and L-arabinitol, respectively; three separate genes encode enzymes with overlapping selectivities for xylose and arabinose as substrates.54

Indeed, the wild-type S. cerevisiae genome does contain genes for both xylose reductase and xylitol dehydrogenase, thus being able to isomerize xylulose from xylose, and the resulting xylulose (after phosphorylation catalyzed by a specific xylulokinase) can enter the pentose phosphate pathway (figure 3.2).55,56 Overex­pressing the endogenous yeast genes for xylose catabolism renders the organism capable of growth on xylose in the presence of glucose as cosubstrate under

conditions, although no ethanol is formed; S. cerevisiae may, therefore, have evolved originally to utilize xylose and other pentoses, but this has been muted, possibly because its natural ecological niche altered or the organism changed its range of favored environments.55

Ethanol production from xylose is a rare phenomenon; of 200 species of yeasts tested under controlled conditions in the laboratory, only six accumulated ethanol to more than 1 g/l (0.1% by volume): Pichia stipitis, P. segobiensis, Candida shehatae, C. tenuis, Brettanomyces naardenensis, and Pachysolen tannophilus.56 Even rarer is the ability among yeasts to hydrolyze xylans, only P. stipitis and C. shehatae[24] hav­ing xylanase activity; P. stipitis could, moreover, convert xylan into ethanol at 60% of the theoretical yield as computed from the xylose content of the polymer.58 Three naturally xylose-fermenting yeasts have been used as donors for genes encoding enzymes of xylose utilization for transfer to S. cerevisiae: P. stipitis, C. shehatae, and C. parapsilosis.59-61 These organisms all metabolize xylose by the enzymes of the same low-activity pathway known in S. cerevisiae (figure 3.2), and the relevant enzymes appear to include arabinose as a possible substrate — at least, when the enzymes are assayed in the laboratory.

P. stipitis has been the most widely used donor, probably because it shows relatively little accumulation of xylitol when growing on and fermenting xylose, thus wasting less sugar as xylitol.62 This advantageous property of the yeast does not appear to reside in the enzymes for xylose catabolism but in the occurrence of an alternative respiration pathway (a cyanide-insensitive route widely distributed among yeasts of industrial importance); inhibiting this alternative pathway renders P. stipitis quite capable of accumulating the sugar alcohols xylitol, arabinitol, and ribitol.63 Respiration in P. stipitis is repressed by neither high concentrations of fer­mentable sugars nor by O2 limitation (i. e., the yeast is Crabtree-negative), and as an ethanologen, P. stipitis suffers from the reduction of fermentative ability by aerobic conditions.64

Transferring genetic information from P. stipitis in intact nuclei to S. cerevi­siae produced karyoductants, that is, diploid cells where nuclei from one species have been introduced into protoplasts of another, the two nuclei subsequently fusing, with the ability to grow on both xylose and arabinose; the hybrid organism was, however, inferior to P. stipitis in ethanol production and secreted far more xylitol to the medium than did the donor; its ethanol tolerance was, on the other hand, almost exactly midway between the tolerance ranges of S. cerevisiae and P. stipitis.65 For direct genetic manipulation of S. cerevisiae, however, the most favored strategy (starting in the early 1990s) has been to insert the two genes (xyl1 and xyl2) from P. stipitis coding for xylose reductase (XR) and xylitol dehydrogenase (XDH), respectively.66 Differing ratios of expression of the two foreign genes resulted in smaller or higher amounts of xylitol, glycerol, and acetic acid, and the optimum XR:XDH ratio of 0.06:1 can yield no xylitol, less glycerol and acetic acid, and more ethanol than with other engineered S. cerevisiae strains.67

The first patented Saccharomyces strain to coferment xylose and glucose to ethanol (not S. cerevisiae but a fusion between S. diastaticus and S. ovarum able to produce ethanol at 40°C) was constructed by the Laboratory of Renewable

Resources Engineering at Purdue University, with four specific traits tailored for industrial use:68-70

1. To effectively direct carbon flow from xylose to ethanol production rather than to xylitol and other by-products

2. To effectively coferment mixtures of glucose and xylose

3. To easily convert industrial strains of S. cerevisiae to coferment xylose and glucose using plasmids with readily identifiable antibiotic resistance markers controlling gene expression under the direction of promoters of S. cerevisiae glycolytic genes

4. To support rapid bioprocesses with growth on nutritionally rich media

In addition to XR and XDH, the yeast’s own xylulose-phosphorylating xylulokinase (XK, figure 3.2) was also overexpressed via high-copy-number yeast-£. coli shuttle plasmids.68 This extra gene manipulation was crucial because both earlier and contem­porary attempts to transform S. cerevisiae with only genes for XR and XDH produced transformants with slow xylose utilization and poor ethanol production.68 The synthesis of the xylose-metabolizing enzymes not only did not require the presence of xylose, but glucose was incapable of repressing their formation. It was known that S. cerevisiae could consume xylulose anaerobically but only at less than 5% of the rate of glucose utilization; XK activity was very low in unengineered cells, and it was reasoned that providing much higher levels of the enzyme was necessary to metabolize xylose via xylulose because the P. stipitis XDH catalyzed a reversible reaction between xylitol and xylulose, with the equilibrium heavily on the side of xylitol.70-72 Such strains were quickly shown to ferment corn fiber sugars to ethanol and later utilized by the Iogen Corporation in their demonstration process for producing ethanol from wheat straw.7374

The vital importance of increased XK activity in tandem with the XR/XDH pathway for xylose consumption in yeast was demonstrated in S. cerevisiae: not only was xylose consumption increased but xylose as the sole carbon source could be con­verted to ethanol under both aerobic and anaerobic conditions, although ethanol pro­duction was at its most efficient in microaerobiosis (2% O2).75 Large increases in the intracellular concentrations of the xylose-derived metabolites (xylulose 5-phosphate and ribulose 5-phosphate) were demonstrated in the XK-overexpressing strains, but a major drawback was that xylitol formation greatly exceeded ethanol production when O2 levels in the fermentation decreased.

In the intervening years (and subsequently), considerable efforts have been dedicated to achieving higher ethanol productivity with the triple XR/XDH/XK constructs. Apart from continuing attempts to more fully understand the metabolism of xylose by uncon­ventional or little-studied yeast species, two main centers of attention have been evident:

1. Strategies for harmonizing the different cofactor requirements in the path­way, that is, NAPDH-dependent (or NAPDH-preferring) XR and NAD- requiring XDH, and thus reducing xylitol formation

2. Overexpressing a wider array of other pentose-metabolizing enzymes to maximize the rate of xylose use or (broadening the metabolic scope) increas­ing kinetic factors in the central pathways of carbohydrate metabolism

Because the early attempts to overexpress P. stipitis genes for XR and XDH in S. cerevisiae often resulted in high rates of xylitol formation, if NADPH formed in the oxidative pentose phosphate pathway (figure 3.2) equilibrated with intracellular NAD to form NADH, the reduced availability of NAD could restrict the rate of the XDH reaction in the direction of xylitol oxidation to xylulose; adding external oxidants capable of being reduced by NADH improved ethanol formation and reduced xylitol formation. Two of these were furfural and 5-hydroxyfurfural, known to be sugar deg­radation products present in lignocellulose acid hydrolysates — see chapter 2, sec­tion 2.3.4.78 The adventitious removal of toxic impurities by these reactions probably explained why xylitol accumulation was very low when lignocellulose acid hydroly­sates were used as carbon sources for XR — and XDH-transformed S. cervisiae 7 This line of reasoning does not accord entirely with the high activities of XR measurable in vitro with NADH (63% of the rate with NADPH), but site-specific mutagenesis on the cloned P. stipitis XR gene could alter the activity with NADH to 90% of that with NADPH as cofactor and concomitantly greatly reduce xylitol accumulation although with only a marginally increased xylose utilization rate.69 Further optimization of the xylitol pathway for xylose assimilating was, therefore, entirely possible. Simply coalescing the XR and XDH enzymes into a single fusion protein, with the two active units separated by short peptide linkers, and expressing the chimeric gene in S. cere — visiae resulted in the formation of a bifunctional enzyme; the total activities of XR and XDH were similar to the activities when monomeric enzymes were produced, but the molar yield of xylitol from xylose was reduced, the ethanol yield was higher, and the formation of glycerol was lower, suggesting that the artificially evolved enzyme complex was more selective for NADH in its XR domain as a consequence of the two active sites generating and utilizing NADH being near each other.80

There have also been four direct methodologies tested for altering the preference of XR to use the NADPH cofactor:

1. A mutated gene for a P. stipitis XR with a lower affinity for NADPH replaced the wild-type XR gene and increased the yield of ethanol on xylose while decreasing the xylitol yield but also increasing the acetate and glycerol yields in batch fermentation.81

2. The ammonia-assimilating enzyme glutamate dehydrogenase in S. cerevisiae (and other yeasts) can be either NADPH or NADH specific, setting an artificial transhydrogenase cycle by simultaneously expressing genes for both forms of the enzyme improved xylose utilization rates and ethanol productivity.82

3. Deleting the gene for the NADPH-specific glutamate dehydrogenase aimed to increase the intracellular NADH concentration and the competition between NADH and NADPH for XR but greatly reduced growth rate, ethanol yield, and xylitol yield on a mixture of glucose and xylose; overexpressing the gene for the NADH-specific enzyme in the absence of the NADPH-requir — ing form, however, restored much of the loss in specific growth area and increased both xylose consumption rate when glucose had been exhausted and the ethanol yield while maintaining a low xylitol yield.83

4. Redox (NADPH) regeneration for the XR reaction was approached from a different angle by expressing in a xylose-utilizing S. cerevisiae strain the gene for an NADP-dependent D-glyceraldehyde 3-phosphate dehydrogenase, an enzyme providing precursors for ethanol from either glucose or xylose; the resulting strain fermented xylose to ethanol at a faster rate and with a higher yield.84

Two recent discoveries offer novel biochemical and molecular opportunities: first, the selectivity of P. stipitis XDH has been changed from NAD to NADP by multiple — site-directed mutagenesis of the gene, thereby harmonizing the redox balance with XR; second, an NADH-preferring XR has been demonstrated in the yeast Candida parapsilopsis as a source for a new round of genetic and metabolic engineering.6185

Beyond the initial conversions of xylose and xylitol, pentose metabolism becomes relatively uniform across kingdoms and genera. Most microbial species — and plants, animals, and mammals (including Homo sapiens) — can interconvert some pentose structures via the nonoxidative pentose phosphate pathway (figure 3.2). These reactions are readily reversible, but extended and reorganized, the pathway can function to fully oxidize glucose via the glucose 6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase reactions (both forming NADPH), although the pathway is far more important for the provision of essential biosynthetic intermedi­ates for nucleic acids, amino acids, and cell wall polymers; the reactions can even (when required) run “backward” to generate pentose sugars from triose intermedi­ates of glycolysis.86 Increasing the rate of entry of xylulose into the pentose phosphate pathway by overexpressing endogenous XK activity has been shown to be effec­tive for increasing xylose metabolism to ethanol (while reducing xylitol formation) with XR/XDH transformants of S. cerevisiae.81 In contrast, disrupting the oxidative pentose phosphate pathway genes for either glucose 6-phosphate or 6-phosphoglu — conate oxidation increased ethanol production and decreased xylitol accumulation from xylose by greatly reducing (or eliminating) the main supply route for NADPH; this genetic change also increased the formation of the side products acetic acid and glycerol, and a further deleterious result was a marked decrease in the xylose consumption rate — again, a predictable consequence of low NADPH inside the cells as a coenzyme in the XR reaction.88 Deleting the gene for glucose 6-phosphate dehydrogenase in addition to introducing one for NADP-dependent D-glyceraldehyde 3-phosphate dehydrogenase was an effective means of converting a strain ferment­ing xylose mostly to xylitol and CO2 to an ethanologenic phenotype.84

The first report of overexpression of selected enzymes of the main nonoxidative pentose phosphate pathway (transketolase and transaldolase) in S. cerevisiae harbor­ing the P. stipitis genes for XR and XDH concluded that transaldolase levels found naturally in the yeast were insufficient for efficient metabolism of xylose via the pathway: although xylose could support growth, no ethanol could be produced, and a reduced O2 supply merely impaired growth and increased xylitol accumulation.89 In the most ambitious exercise in metabolic engineering of pentose metabolism by

S. cerevisiae reported to date, high activities of XR and XDH were combined with overexpression of endogenous XK and four enzymes of the nonoxidative pentose phosphate pathway (transketolase, transaldolase, ribulose 5-phosphate epimerase, and ribose 5-phosphate ketolisomerase) and deletion of the endogenous, nonspecific NADPH-dependent aldose reductase (AR), catalyzing the formation of xylitol from xylose.90 In comparison with a strain with lower XR and XDH activities and no other genetic modification other than XK overexpression, fermentation performance on a mixture of glucose (20 g/l) and xylose (50 g/l) was improved, with higher ethanol production, much lower xylitol formation, and faster utilization of xylose; deleting the nonspecific AR had no effect when XR and XDH activities were high, but glyc­erol accumulation was higher (figure 3.5).

To devise strains more suitable for use with industrially relevant mixtures of carbohydrates, the ability of strains to use oligosaccharides remaining undegraded to free hexose and pentose sugars in hydrolysates of cellulose and hemicelluloses is essential. Research groups in South Africa and Japan have explored combinations of heterologous xylanases and P-xylosidases:

• A fusion protein consisting of the xynB P-xylosidase gene from Bacillus pumilus and the S. cerevisiae Mfa1 signal peptide (to ensure the correct posttranslational processing) and the XYN2 P-xylanase gene from Hypo — crea jecorina were separately coexpressed in S. cerevisiae under the con­trol of the glucose-derepressible ADH2 ADH promoter and terminator; coproduction of these xylan-degrading enzymes hydrolyzed birch wood xylan, but no free xylose resulted, probably because of the low affinity of the P-xylosidase for its xylobiose disaccharide substrate.91

• A similar fusion strategy with the xlnD P-xylosidase gene from Aspergillus niger and the XYN2 P-xylanase gene from H. jecorina enabled the yeast to hydrolyze birch wood xylan to free xylose.92

• A xylan-utilizing S. cerevisiae was constructed using cell surface engi­neering based on в-agglutinin (a cell surface glycoprotein involved in cell-cell interactions) to display xylanase II from Hypocrea jecorina and a в-xylosidase from Aspergillus oryzae; with P. stipitis XR and XDH and overexpressed endogenous XK, the strain could generate ethanol directly from birch wood xylan with a conversion efficiency of 0.3 g/g carbohydrate used.93

High XR, XDH, and XK activities combined with the expression of a gene from Aspergillus acleatus for displaying в-glucosidase on the cell surface enabled S. cere­visiae to utilize xylose — and cellulose-derived oligosaccharides from an acid hydro­lysate of wood chips, accumulating 30 g/l of ethanol from a total of 73 g/l of hexose and pentose sugars in 36 hr.94