Definitive comparisons of recombinant ethanologenic bacteria in tightly controlled, side-by-side comparisons have not been made public. From data compiled in 2003 from various sources with E. coli, K. oxytoca, and Z. mobilis strains fermenting

Geysers from Yellowstone National Park also yielded one promising isolate.

mixtures of glucose, xylose, and arabinose, conflicting trends are evident.162 For maximum ethanol concentration, the ranking order was

Z. mobilis AX101 >> K. oxytoca P2 = E. coli FBR5,

but for ethanol yield (percentage of maximum possible conversion), E. coli was superior:

E. coli FBR5 > K. oxytoca P2 = Z. mobilis AX101,

and the rankings of ethanol production rate (grams per liter per hour) were again different:

E. coli FBR5 >> Z. mobilis AX101 >> K. oxytoca P2.

All three strains can utilize arabinose, xylose, and glucose, but Z. mobilis AX101 can­not utilize the hemicellulose component hexose sugars galactose or mannose. In tests of E. coli, K. oxytoca, and Erwinia chrysanthemi strains, only E. coli KO11 was able to convert enzyme-degraded polygalacturonic acid (a pectin polymer) to ethanol.221

Enteric bacteria (including E. coli and Klebsiella sp.) have the additional hurdle to overcome of being perceived as potentially injurious to health, and K. oxytoca has been implicated in cases of infectious, hospital-acquired, and antibiotic-associated diseases.238-240 K. oxytoca is well known as a producer of a broad-spectrum P — lactamase, an enzyme capable of inactivating penicillins and other p-lactam antibiotics.241 Immunosuppression of patients under medical supervision or as a result of pathogenesis has led to the identification of infections by hitherto unknown yeast species or by those not considered previously to be pathogenic, including Kluyveromyces marxianus, five species of Candida, and three species of Pichia.17 Biosafety issues and assessments will, therefore, be important where planning (zon­ing) permissions are required to construct bacterial bioethanol facilities.

Because commercially relevant biomass plants for lignocellulosic ethanol only began operating in 2004 (chapter 2, section 2.7), future planned sites in divers parts of the world will inevitably make choices of producing organism that will enor­mously influence the continued development and selection of candidate strains.242 Different substrates and/or producing regions may arrive at different choices for optimized ethanol producer, especially if local enzyme producers influence the choice, if licensing agreements cannot be made on the basis of exclusivity, or if national interests encourage (or dictate) seamless transfer of technologies from labo­ratories to commercial facilities. More than a decade ago, a publication from the National Renewable Energy Laboratory ranked Z. mobilis ahead of (in descending order of suitability): recombinant Saccharomyces, homofermentative Lactobacillus, heterofermentative Lactobacillus, recombinant E. coli, xylose-assimilating yeasts, and clostridia.202 Their list of essential traits included •

• Low fermentation pH (to discourage contaminants)

• High fermentation selectivity

• Broad substrate utilization range

• GRAS status

The secondary list of 19 “desirable” traits included being Crabtree-positive (see above, section 3.1.1), high growth rates, tolerance to high salts, high shear, and elevated tem­perature. No commercialization of the Z. mobilis biocatalyst is yet established, but if bacterial ethanologens remain serious candidates for commercial bioethanol produc­tion, clear evidence for this should appear in the next decade as increasing numbers of scale-up bioethanol facilities are constructed for a variety of biomass feedstocks.

The increasing list of patents issued to companies and institutions for ethanologens testifies to the endeavors in this field of research and development (table 3.6). More problematic is that the field has been gene-led rather than genome — or (more usefully still) metabolome-led, that is, with full cognizance and appreciation of the flexibility and surprises implicit in the biochemical pathway matrices. The “stockpiling” of useful strains, vectors, and genetic manipulation techniques has built a process platform for the commercialization of ethanol production from lignocellulose biomass. The scien­tific community can find grounds for optimism in the new insights in metabolic engi­neering described in this chapter; what is less clear is the precise timescale — 10, 15, or more years — required to translate strain potential into industrial-scale production.243