Enteric Bacterial Catalysts for Fuel Ethanol Production L. O. Ingram,* H. C. Aldrich, A. C. C. Borges, T. B. Causey, Alfredo Martinez, Fernando Morales, Alif Saleh, S. A. Underwood, L. P. Yomano, S. W. York, Jesus Zaldivar, and Shengde Zhou Department of Microbiology and Cell Science, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, Florida 32611 The technology is available to produce fuel ethanol from renewable lignocellulosic biomass. The current challenge is to assemble the various process options into a commercial venture and begin the task of incremental improvement. Current process designs for lignocellulose are far more complex than grain to ethanol processes. This complexity results in part from the complexity of the substrate and the biological limitations of the catalyst. Our work at the University of Florida has focused primarily on the genetic engineering of Enteric bacteria using genes encoding Zymomonas mobilis pyruvate decarboxylase and alcohol dehydrogenase. These two genes have been assembled into a portable ethanol production cassette, the PET operon, and integrated into the chromosome of Escherichia coli B for use with hemicellulose-derived syrups. The resulting strain, KO11, produces ethanol efficiently from all hexose and pentose sugars present in the polymers of hemicellulose. By using the same approach, we integrated the PET operon into the chromosome of Klebsiella oxytoca to produce strain P2 for use in the simultaneous saccharification and fermentation (SSF) process for cellulose. Strain P2 has the native ability to ferment cellobiose and cellotriose, eliminating the need for one class of cellulase enzymes. Recently, the ability to produce and secrete high levels of endoglucanase has also been added to strain P2, further reducing the requirement for fungal cellulase. The general approach for the genetic engineering of new biocatalysts using the PET operon has been most successful with Enteric bacteria but was also extended to Gram positive bacteria, which have other useful traits for lignocellulose conversion. Many opportunities remain for further improvements in these biocatalysts as we proceed toward the development of single organisms that can be used for the efficient fermentation of both hemicellulosic and cellulosic substrates. Introduction Many environmental and societal benefits would result from the replacement of petroleum-based automotive fuels with renewable fuels from modern biomass (Lynd et al., 1991 Olson and Hahn-Hagerdal, 1996 Wyman, 1995). Each year, the U. S. burns over 120 billion gallons of automotive fuel, roughly equivalent to the total amount of imported petroleum. Development of ethanol as a renewable alternative fuel has the potential to eliminate U.S. dependence on imported oil, to improve the en- vironment, and to provide new employment (Sheehan, 1994). The solution to the problem of imported oil for automo- tive fuel is quite simple, in theory, and can be potentially implemented in many nations around the world. Oil is a fossilized form of biomass, a feedstock which can be chemically transformed into gasoline. Fossil biomass can be replaced as a feedstock by contemporary, renewable, plant materials. Brazil has demonstrated the feasibility of producing ethanol from cane sugar and the use of ethanol as a primary automotive fuel for more than 20 years. Currently, the U.S. produces over 1.2 billion gallons of fuel ethanol each year from corn starch but starch alone cannot supply the 120 billion gallons of fuel needed. Solar energy stored by the conversion of carbon dioxide into green plants represents the most abundant source of renewable energy in the world. The majority of the dry weight of all green plants is lignocellulose, the structural polymers (cellulose, hemicellulose, pectin, and lignin) that comprise the cell wall. The two primary barriers that have prevented large-scale ethanol production from ligno- cellulose are the cost of production relative to current product value and the large capital risks associated with a new technology. Although it can be argued that all of the required components for a biomass to ethanol process are now available (Ingram et al., 1998a), building a cost- effective integrated process remains a challenge. Long term, it will be essential to reduce capital costs through process simplification. For many steps, process complex- ity results from limitations of the biocatalysts. Solving the biological problems that contribute to process complexity represents an opportunity to expand our fundamental understanding of biological systems and * Address all correspondence concerning this manuscript to: L. O. Ingram, Department of Microbiology and Cell Science, Bldg. 981, P.O. Box 110700, University of Florida, Gainesville, FL 32611. Tel: (352) 392-8176. Fax: (352) 392-5922. E-mail: Lingram@ micro.ifas.ufl.edu. 855 Biotechnol. Prog. 1999, 15, 855-866 10.1021/bp9901062 CCC: $15.00 �� 1999 American Chemical Society and American Institute of Chemical Engineers Published on Web 09/11/1999

to reduce the cost of bioconversion. Added process steps that result from limitations of the biocatalysts increase both equipment and operating expenses. Previous techni- cal constraints have included the lack of suitable bio- catalysts for the fermentation of hemicellulose-derived pentose sugars, toxins created during pretreatments and hydrolysis, fermentation rates and yields, ethanol toler- ance, environmental hardiness, the cost of fungal en- zymes for cellulose depolymerization, etc. During the past few years, progress has been made in many of these areas, and at least one lignocellulose to ethanol plant is now under construction in the U.S. (McCoy, 1998). Many aspects of the technology for the bioconversion of ligno- cellulose to fuel ethanol have been demonstrated at pilot scale. These are summarized in excellent books and reviews (du Preez, 1994 Grohmann, 1993 Hahn-Hag- erdal, 1996 Hahn-Hagerdal et al., 1993 Ingram et al., 1997 and 1998a McMillan, 1994a and 1994b Olsson and Hahn-Hagerdal, 1996 Philippidis, 1994 Philippidis et al., 1992 Ramos and Saddler, 1994 von Sivers and Zacchi, 1995). This paper will focus on the development of bacterial biocatalysts in our laboratory and their potential use in a generalized process for the bioconver- sion of lignocellulose. Lignocellulose, a Complex Substrate for Biocon- version. Lignocellulose represents approximately 90% of the dry weight of most plant material (Figure 1). It is a renewable resource in which solar energy is stored in the form of carbohydrate (cellulose, hemicellulose, pectin) and aromatic polymers (lignin). Cellulose, 20%-50% of the dry weight, is a homopolymer of cellobiose, a dimer of glucose. Hemicellulose, 20%-40% of the dry weight, is a complex polymer containing a mixture of pentose (xylose, arabinose) and hexose (glucose, mannose, galac- tose) sugars which contain acetyl and glucuronyl side chains. Pectin, 2%-20% of the dry weight, is a meth- ylated homopolymer of galacturonic acid. These three carbohydrate polymers are potential sources of sugars for bioconversion to ethanol and other chemicals. Lignin, 10%-20% of the dry weight, is a biological thermoplastic composed of aromatic residues that cannot be readily converted to ethanol but can be used as a fuel or biodegradable plastic. Carbohydrate polymers must first be depolymerized into soluble components prior to uptake and metabolism by any microbial biocatalysts. Starch is a relatively simple homopolymer of glucose that can be readily hydrated by heating and rapidly depolymerized with low levels of enzymes from fungi or bacteria. The highly digestible structure of starch is well suited to its biological function, the temporary storage of solar energy in plants. In contrast, lignocellulose has evolved to provide a more permanent structure in plants and to serve as the primary protective barrier that prevents cell destruction by bacteria and fungi. Plant cell walls can be regarded as a giant macro- molecular composite of cellulose fibers embedded in a covalently joined matrix of pectin, lignin, and hemi- cellulose (Brett and Waldron, 1996 Clarke, 1997). The cellulose macrofibers are each composed of crystalline bundles of individual chains of cellulose. The complexity of this lignocellulose has evolved in concert with the saccharification abilities of microorganisms and creates a formidable challenge for bioconversion. In nature, consortia of microorganisms produce a mixture of en- zymes capable of degrading the many different chemical linkages in lignocellulose and these enzymes alone can be used to depolymerize cellulose and hemicellulose. However, in nature this process is relatively slow. Some form of chemical or physical pretreatment appears es- sential to increase the accessibility of hemicellulose and cellulose and thus speed digestion by microbial enzymes (Grohmann, 1993 Grohmann et al., 1985 McMillan, 1994a Ramos and Saddler, 1994 Saddler et al., 1993). Pretreatment methods, such as ammonia freeze explosion (Holtzapple et al., 1991 McMillan, 1994a), carbon dioxide explosion (Zheng et al., 1998), high pressure steam (Nishikawa et al., 1988 Palmqvist et al., 1996 Ramos and Saddler, 1994), and cooking with lime (Gandi et al., 1997) or pulping chemicals, solubilize hemicellulose and part of the lignin and redeposit these polymers in a form that allows cellulose and hemicellulose to be more readily degraded (Holtzapple et al., 1991 Moniruzzaman et al., 1996). Alternatively, hemicellulose can be hydrolyzed by dilute mineral acids at modest temperatures (du Preez, 1994 McMillan, 1994b). This chemical treatment also exposes the cellulose to improve the digestion of cellulose with enzymes. Higher acid concentrations and temper- atures can be used to hydrolyze cellulose directly without the use of cellulase enzymes but provide a lower yield of sugars due to partial chemical destruction (Grohmann, 1994 Philippidis, 1994 von Sivers and Zacchi, 1995). Office paper, a highly purified and refined cellulose, is among the best substrates for enzymatic digestion (Brooks and Ingram, 1995). Lignocellulose pretreatments with dilute mineral acid have an important advantage: the hydrolysis of most linkages in hemicellulose to produce monomeric sugars (Grohmann et al., 1985). Hemicellulose is a branched- chain carbohydrate containing a complex mixture of hexose and pentose sugars, variously substituted with acetyl and uronic acid side chains and covalently bound to lignin (Clarke, 1997). Due in part to the fibrous nature, the cost-effective hydrolysis of hemicellulose with dilute mineral acids to produce a syrup with high concentra- tions of sugar (g100 g/L) is a formidable engineering challenge. With other chemical pretreatments such as ammonia fiber explosion (Holtzapple et al., 1991) or steam-explosion (Palmqvist et al., 1996 Ramos and Saddler, 1994), enzymes can be used for the saccharifi- cation of hemicellulose, but the projected cost for these enzymes remains too high for fuel ethanol production. The abundance of pentose sugars and the diversity of sugar types found in hemicellulose syrups represent a special problem for bioconversion. No yeasts, bacteria, or fungi from nature exhibit the native ability to rapidly and efficiently produce high concentrations of ethanol from both hexose and pentose sugars. Toxins Generated during Pretreatments with Dilute Acid or Steam. In addition to monomeric sugars from hemicellulose, the dilute acid hydrolysis of ligno- Figure 1. Composition of lignocellulose. Approximate composi- tion of lignocellulose is provided as a percentage of total dry weight. Abreviations: ara, arabinose gal, galactose glu, glucose man, mannose and xyl, xylose. 856 Biotechnol. Prog., 1999, Vol. 15, No. 5