3.3.1 Metabolic Routes in Bacteria for Sugar Metabolism and Ethanol Formation

Two principal routes for glucose catabolism are known to classical biochemistry in ethanologenic bacteria: the Embden-Meyerhof-Parnas (EMP) pathway of glycolysis (also present in yeasts, fungi, plants, and animals) and, in a restricted range of bac­teria (but more widely for the catabolism of gluconic acid), the Entner-Doudoroff (ED) pathway (figure 3.4).486152 The two initial steps of the ED pathway resulting in 6-phosphogluconic acid are those of the oxidative pentose phosphate pathway, but Z. mobilis is unique in operating this sequence of reactions under anaerobic condi­tions.153 The EMP and ED pathways converge at pyruvic acid; from pyruvate, a range of fermentative products can be produced, including acids such as lactate and decar — boxylated acids such as 2,3-butanediol (see figure 2.3). In S. cerevisiae and other major ethanol-producing yeasts, ethanol formation requires only two reactions from pyruvate: pyruvate decarboxylase (PDC) catalyzes the formation of acetaldehyde and ADH catalyzes the NADH-oxidizing reduction of acetaldehyde to ethanol; in E. coli, in marked contrast, no PDC naturally exists and pyruvate is catabolized by the PDH reaction under aerobic conditions or by the pyruvate formate lyase (PFL) reaction under anaerobiosis. To maintain the redox balance under fermentative con­ditions, a spectrum of products is generated from glucose in E. coli and many other enteric bacterial species in which ethanol is often a minor component (table 3.5).154

3.3.2 Genetic and Metabolic Engineering of Bacteria for Bioethanol Production

3.3.2.1 Recombinant E. coli: Lineages and Metabolic Capabilities

Attempts (beginning in the 1980s) to genetically “improve” E. coli and other bacteria for efficient ethanol production with recombinant gene technology foundered because they relied on endogenous ADH activities competing with other product pathways.4 Success soon followed when the research group at the University of Florida (Gaines­ville) combined the Z. mobilis genes for PDC and ADH with that bacterium’s homo­ethanol pathway in a single plasmid, the PET operon; in various guises, the PET system has been used to engineer ethanol production in E. coli and other bacteria.155,156 The original PET operon used the P-galactosidase promoter; to construct a vector suitable for directing chromosomal integration by transformation, a PET cartridge was devised to include the PET operon plus chloramphenicol transacetylase for selection, flanked by truncated fragments of the E. coli PFL (pfl) open reading frame (without promoter) to target homologous recombination; upon integration, PET expression was under the control of the chromosomal pfl promoter — although a single-step selection on chlor­amphenicol was required to identify a suitably highly expressing mutant.157

A crucial property of the Z. mobilis PDC is its relatively high affinity (Km) for pyruvate as a substrate: 0.4 mM as compared with 2 mM for PFL, 7 mM for lactate dehydrogenase and 0.4 mM for PDH; in consequence, much reduced amounts of lactic and acetic acids are formed, and the mix of fermentation products is much less toxic to growth and inhibitory to the establishment of dense cell populations.152 Further muta­tions have been utilized to delete genes for the biosynthesis of succinic, acetic, and lac­tic acids to further reduce the waste of sugar carbon to unwanted metabolic acids.158

When a number of the well-characterized E. coli laboratory strains were evaluated for their overall suitability for PET transformation, strain B exhibited the best hardiness to environmental stresses (ethanol tolerance and plasmid stability in nonselective media) and superior ethanol yield on xylose.159,160 This strain was isolated in the 1940s and was widely used for microbiological research in the 1960s; more importantly, it lacks all known genes for pathogenicity.161 A strain B-derived, chromosomally integrated iso­late (KO11), also with a disrupted fumarate reductase gene (for succinate formation), emerged as a leading candidate for industrial ethanol production.162 It shows ethanol production with much-improved selectivity against the mixed-acid range of fermenta­tion products[26] (figure 3.9, cf. table 3.5).157 The capabilities of KO11 have been evaluated and confirmed in laboratories in the United States and Scandinavia.43,163 An encourag­ingly wide range of carbon substrates have been shown to support ethanol production:

• Pine wood acid hydrolysates164

• Sugarcane bagasse and corn stover165

• Corn cobs, hulls, and fibers166,167

• Dilute acid hydrolysate of rice hulls168

• Sweet whey and starch169,170

• Galacturonic acid and other components in orange peel hydrolysates171

• The trisaccharide raffinose (a component of corn steep liquors and molasses)172

To increase its hardiness to ethanol and inhibitors present in acid hydrolysates of lignocel — lulosic materials, strain KO11 was adapted to progressively higher ethanol concentrations during a period of months, the culture being reselected for resistance to chlorampheni­col at regular intervals; the resulting strain (LY01) fermented 140 g/l xylose in 96 hr (compared with 120 hr with LO11).173 Increased ethanol tolerance was accompanied (fortuitously) by increased resistance to various growth inhibitors, including aromatic alcohols and acids derived from ligninolysis and various aromatic and nonaromatic alde­hydes (including 4-hydroxybenzaldehyde, syringaldehyde, vanillin, hydroxymethylfur- fural [HMF], and furfural).174-176 Of this multiplicity of inhibitors, the aromatic alcohols proved to be the least toxic to bacterial growth and metabolism, and E. coli strains can be at least as refractory to growth inhibitors as are other microbial ethanologens.

More wide-ranging genetic manipulation of ethanologenic E. coli strains has explored features of the molecular functioning of the recombinant cells as measured by quantitative gene expression and the activities of the heterologous gene prod — ucts.177-180 A long-recognized problem with high-growth-rate bacterial hosts engi­neered to contain and express multiple copies of foreign genes is that of “metabolic burden,” that is, the diversion of nutrients from biosynthesis and cell replication to supporting the expression and copying of the novel gene complement often results in a reduced growth rate in comparison with the host strain. Chromosomal integration of previously plasmid-borne genes does not avoid this metabolic demand, as became evident when attempts were made to substitute a rich laboratory growth medium with possible cheaper industrial media based on ingredients such as corn steep
liquor: uneconomically high concentrations were required to match the productivities observable in laboratory tests, but this could be only partially improved by lavish additions of vitamins, amino acids, and other putative growth-enhancing medium ingredients.177 More immediately influential was increasing PDC activity by insert­ing plasmids with stronger promoters into the chromosomally integrated KO11 strain. Because the laboratory trials in the 1990s achieved often quite low cell densi­ties (3 g dry weight of cells per liter), lower by one or two orders of magnitude than those attainable in industrial fermentations, it is highly probable that some ingenu­ity will be required to develop adequate media for large-scale fermentations while minimizing operating costs.

Genetic manipulation can, however, aid the transition of laboratory strains to commercially relevant media and the physical conditions in high-volume fermen- tors. For example, growth and productivity by the KO11 strain in suboptimal media can be greatly increased by the addition of simple additional carbon sources such as pyruvic acid and acetaldehyde; this has no practical significance because ethanol production cannot necessarily be a biotransformation from more expensive precur­sors but, together with other physiological data, implies that the engineered E. coli cells struggle to adequately partition carbon flow between the demands for growth (amino acids, etc.) and the requirement to reoxidize NADH and produce ethanol as an end product.178 179 Expressing in KO11, a B. subtilis citrate synthase, whose activ­ity is not affected by intracellular NADH concentrations, improves both growth and ethanol yield by more than 50% in a xylose-containing medium; this novel enzyme in a coliform system may act to achieve a better balance and direct more carbon to 2-oxoglutarate and thence to a family of amino acids required for protein and nucleic acid biosynthesis.178 Suppressing acetate formation from pyruvate by delet­ing the endogenous E. coli gene (ackA) for acetate kinase probably has a similar effect by altering carbon flow around the crucial junction represented by pyruvate (see figure 2.3).179

In nutrient-rich media, expressing the Z. mobilis homoethanol pathway genes in E. coli increases growth rate by up to 50% during the anaerobic fermentation of xylose.180 Gene array analysis reveals that, of the nearly 4,300 total open reading frames in the genome, only 8% were expressed at a higher level in KO11 in anaero­bic xylose fermentations when compared with the B strain parent but that nearly 50% of the 30 genes involved in xylose catabolism to pyruvate were expressed at higher levels in the recombinant (figure 3.10). Calculations from bioenergetics show that xylose is a much poorer source of biochemical energy (generating only 33% of the ATP yield per molecule oxidized), and a physiological basis for the changes in growth rate can be deduced from the genomics data in the greatly elevated expres­sions of the genes encoding the initial two enzymes of xylose catabolism, although some of the other changes in the pentose phosphate and glycolytic pathways may be important for intracellular fluxes.180

A further broadening of the substrate range of the KO11 strain was effected by expressing genes (from Klebsiella oxytoca) encoding an uptake mechanism for cello — biose, the disaccharide product of cellulose digestion; an operon was introduced on a plasmid into KO11 containing the two genes for the phosphoenolpyruvate-dependent phosphotransferase transporter for cellobiose (generating phosphorylated cellobiose)

and phospho-P-glucosidase (for hydrolyzing the cellobiose phosphate intracellu — larly).181 The K. oxytoca genes proved to be poorly expressed in the E. coli host, but spontaneous mutants with elevated specific activities for cellobiose metabolism were isolated and shown to have mutations in the plasmid that eliminated the engineered casAB promoter and operator regions; such mutants rapidly fermented cellobiose to ethanol, with an ethanol yield of more than 90% of the theoretical maximum and (with the addition of a commercial cellulase) fermented mixed-waste office paper to ethanol.

A second and distinct major lineage of recombinant E. coli was initiated at the National Center for Agricultural Utilization Research, U. S. Department of Agricul­ture, Peoria, Illinois. The starting point was the incomplete stability of the KO11 strain: phenotypic instability was reported in repeated batch or continuous cultiva­tion, resulting in declining ethanol but increasing lactic acid production.182 183 More­over, the results were different when glucose or xylose provided the carbon supply for continuous culture: on glucose alone, KO11 appeared to be stable, but ethanol productivity declined after five days, and the antibiotic (chloramphenicol) selective marker began to be lost after 30 days.184 185 Novel ethanologenic strains were created by expressing the PET operon on a plasmid in E. coli FMJ39, a strain with deleted genes for lactate dehydrogenase and PFL and, in consequence, incapable of fermen­tative growth on glucose.186 The introduced homoethanol pathway genes comple­mented the mutations and positively selected for plasmid maintenance to enable active growth by fermentation pathways, that is, self-selection under the pressure of fermenting a carbon source.187 The plasmid was accurately maintained by serial culture and transfer under anaerobic conditions with either glucose or xylose as the carbon source and with no selective antibiotic present but quickly disappeared during growth in the aerobic conditions in which the parental strain grew normally.

One of the resulting strains was further adapted for growth on xylose (FBR3); this construct can ferment a 10% (w/v) concentration of glucose, xylose, arabinose, or a mixture of all three sugars at 35°C during a period of 70-80 hr and producing up to 46.6 g/l of ethanol and at up to 91% of the theoretical maximum yield.188 The strains are also able to ferment hydrolysates prepared from corn hulls and germ meal within 60 hr and at a yield of 0.51 g ethanol/g sugar consumed.189 190 Variants of the strains have been constructed that are relatively deficient in glucose uptake because they carry a mutation in the phosphoenolpyruvate-glucose phosphotransferase system. In organisms with this transport mechanism, the presence of glucose represses the uptake of other sugars; obviating this induces the cells to utilize xylose and arabi­nose simultaneously with glucose rather than sequentially after the glucose supply begins to be exhausted and to make ethanol production more rapidly — although the overall productivity (carbon conversion efficiency) is minimally affected.191 The USDA strains will be considered further in section 3.3.2.5 when performance data for the rival bacterial ethanologens are compared.

Molecular evolution of an efficient ethanol pathway in E. coli without resorting to heterologous gene expression was finally accomplished in 2006.192 Work pub­lished over a decade earlier from the University of Sheffield had shown that in the PDH complex, the route of pyruvate oxidation in E. coli under aerobic conditions was encoded by a four-gene operon (pdhR, aceE, aceF, and Ipd), with pyruvate (or a derivative of pyruvate) acting as the inducing agent.193 A mutant of E. coli strain K12 was isolated with the essential genetic mutation occurring in the PDH operon; the phenotype was a novel pathway endowing the capacity to ferment glucose or xylose to ethanol with a yield of 82% under anaerobic conditions, combining a PDC-type enzyme activity with the endogenous ADH activity.192 This may aid the introduc­tion of bacterial ethanol production in regions (including some member states of the European Union) where genetically manipulated organisms are viewed with popular (or populist) concern.

Commercial take-up of recombinant E. coli has proved very slow, but in 2007, dem­onstration facilities using E. coli strains licensed from the University of Florida (table 3.6) have been opened or constructed in the United States and Japan (section 4.8) — in fact, the Japanese site is the world’s first commercial plant to produce ethanol from wood by a bacterial fermentation.