Most of the freely available information regarding ethanologenic yeasts has been derived from “laboratory” strains constructed by research groups in academia; some of these strains (or variants thereof) have certainly been applied to industrial-scale fermentations for bioethanol production, but published data mostly refer to strains either grown in chemically defined media or under laboratory conditions (e. g., contin­uous culture) or with strains that can—even under the best available test conditions— accumulate very little ethanol in comparison with modern industrial strains used in potable alcohol or fuel alcohol production (figure 3.7). In addition, strains con­structed with plasmids may not have been tested in nonselective media, and plasmid survival in fermentations is generally speculative although genetic manipulations are routine for constructing “self-selective” plasmid-harboring strains where a chro­mosomal gene in the host is deleted or disrupted and the auxotrophic requirement is supplied as a gene contained on the plasmid.113

Nevertheless, benchmarking studies comparing “laboratory” and “industrial” strains constructed for pentose utilization have appeared, the industrial examples including genetically manipulated polyploid strains typical of S. cerevisiae “work­ing” strains from major brewers or wineries; accounts of engineering such strains for xylose utilization began to be published after 2002.114,115 A comparison of four labo­ratory and five industrial strains surveyed both genetically manipulated and geneti­cally undefined but selected xylose consumers (table 3.4).116 The industrial strains were inferior to the laboratory strains for the yields of both ethanol and xylitol from

xylose in minimal media. Resistance to toxic impurities present in acid hydrolysates of the softwood Norway spruce (Picea abies) was higher with genetically trans­formed industrial strains, but a classically improved industrial strain was no more hardy than the laboratory strains (table 3.4); similarly, while an industrial strain evolved by genetic manipulation and then random mutagenesis had the fastest rate of xylose use, a laboratory strain could accumulate the highest ethanol concentration on minimal medium with 50 g/l of each glucose and xylose. None of the strains had an ideal set of properties for ethanologenesis in xylose-containing media; long-term chemostat cultivation of one industrial strain in microaerobic conditions on xylose as the sole carbon source definitely improved xylose uptake, but neither ethanol nor xylitol yield. Three of the industrial strains could grow in the presence of 10% solu­tions of undetoxified lignocellulose hydrolysate; the most resistant strain grew best (at 4.3 g/l as compared with 3.7-3.8 g/l) but had a marginally low ethanol production (16.8 g/l as compared with 16.9 g/l), perhaps because more of the carbon substrate was used for growth in the absence of any chemical limitation.

The industrial-background strain TMB 3400 (table 3.4) had no obvious meta­bolic advantage in anaerobic batch fermentations with xylose-based media when compared with two laboratory strains, one catabolizing xylose by the XR/XDH/XK pathway, the other by the fungal XI/XK pathway (figure 3.8).101 In the presence of undetoxified lignocellulose hydrolysate, however, only the industrial strain could grow adequately and exhibit good ethanol formation (figure 3.8).

Industrial and laboratory strains engineered for xylose consumption fail to metabolize L-arabinose beyond L-arabitinol.116 With laboratory and industrial strains endowed with recombinant xylose (fungal) and arabinose (bacterial) pathways tested in media containing glucose, xylose, and arabinose, the industrial strain accumulated higher concentrations of ethanol, had a higher conversion efficiency of ethanol per

unit of total pentose utilized, and also converted less xylose to xylitol and less arabi — nose to arabinitol — although 68% of the L-arabinose consumed was still converted only as far as the polyol.117

Interactions between hexose and pentose sugars in the fermentations of lignocel- lulose-derived substrates has often been considered a serious drawback for ethanol production; this is usually phrased as a type of “carbon catabolite repression” by the more readily utilizable hexose carbon source(s), and complex phenotypes can be gen­erated for examination in continuous cultures.118 In batch cultivation, xylose supports slower growth and much delayed entry into ethanol formation in comparison with glucose.119 An ideal ethanologen would co-utilize multiple carbon sources, funneling them all into the central pathways of carbohydrate metabolism — ultimately to

pyruvic acid and thence to acetic acid, acetaldehyde, and ethanol (figures 3.2 and 3.4). No publicly disclosed strain meets these requirements. An emerging major challenge is to achieve the rapid transition from proof-of-concept experiments in synthetic media, using single substrates and in the absence of toxic inhibitors, to demonstra­tions that constructed strains can efficiently convert complex industrial substrates to ethanol.120 In addition to the key criteria of high productivity and tolerance of toxic impurities, process water economy has been emphasized.121

Integration of genes for pentose metabolism is becoming increasingly routine for S. cerevisiae strains intended for industrial use; different constructs have quantitatively variable performance indicators (ethanol production rate, xylose consumption rate, etc.), and this suggests that multiple copies of the heterologous genes must be further optimized as gene dosages may differ for the individual genes packaged into the host strain.122 Such strains can be further improved by a less rigorously defined methodology using “evolutionary engineering,” that is, selection of strains with incremental advan­tages for xylose consumption and ethanol productivity, some of which advantages can be ascribed to increases in measurable enzyme activities for the xylose pathway or the pentose phosphate pathway and with interesting (but not fully interpretable) changes in the pool sizes of the intracellular pathway intermediates.123 124 Efficient utilization of xylose appears to require complex global changes in gene expression, and a reexamina­tion of “natural” S. cerevisiae has revealed that classical selection and strain improve­ment programs can develop yeast cell lines with much shorter doubling times on xylose as the sole carbon source as well as increased XR and XDH activities in a completely nonrecombinant approach.125 This could easily be applied to rationally improve yeast strains with desirable properties that can be isolated in the heavily selective but artifi­cial environment of an industrial fermentation plant — a practice deliberately pursued
for centuries in breweries and wineries but equally applicable to facilities for the fer­mentation of spent sulfite liquor from the pulp and paper industry.126

Defining the capabilities of both industrial and laboratory strains to adapt to the stresses posed by toxic inhibitors in lignocellulose hydrolysates remains a focus of intense activity.127130 Expression of a laccase (from the white rot fungus, Trametes versicolor) offers some promise as a novel means of polymerizing (and precipitat­ing) reactive phenolic aldehydes derived from the hydrolytic breakdown of lignins; S. cerevisiae expressing the laccase could utilize sugars and accumulate ethanol in a medium containing a spruce wood acid hydrolysate at greatly increased rates in comparison with the parental strain.131 S. cerevisiae also contains the gene for phenylacrylic acid decarboxylase, an enzyme catalyzing the degradation of ferulic acid and other phenolic acids; a transformant overexpressing the gene for the decar­boxylase utilized glucose grew up to 25% faster, utilized mannose up to 45% faster, and accumulated ethanol up to 29% more rapidly than did a control transformant not overexpressing the gene.132