In an effort to metabolically engineer microorganisms for efficient isobutanol production, various researchers sought to understand isobutanol production at both the molecular and protein levels. A few years before 2013, a number of studies directed at finding molecular and biochemical bases for isobutanol synthesis have been conducted (Table 7.2). Using a prokaryote (E. coli) as host, Atsumi et al. (2008) demonstrated that KIV can serve as a precursor for efficient isobutanol produc­tion from glucose in a biosynthetic pathway consisting of 2-KDC from L. lactis and ADH2 from S. cerevisiae with broad-range substrate specificity in combination with the expression of the alsS gene (encoding AHAS (acetohydroxy acid synthase)) from B. subtilis and the ilvCD gene from E. coli. A follow-up study on the role of different ADHs on isobutanol production with E. coli showed that chromosomally encoded YqhD is the major isobutyraldehyde-converting enzyme, and ADH2 from S. cerevisiae contributes only to a minor extent to isobutanol production in E. coli (Atsumi et al., 2010). This supposition was made after yqhD gene was deleted from the genome of E. coli; the generated recombinant E. coli strain (yqhD deficient) accumulated isobutyraldehyde during fermen­tation and experienced 80% reduction in isobutanol production. Using an a eukaryote (S. cerevisiae) and work­ing on the assumption that inefficient production of isobu­tanol from glucose may be due to limited supply of KIV, a precursor for the valine biosynthesis pathway (Figure 7.2), Lee et al. (2012) added exogenous KIV (0.5 g/l) into the growth medium. As expected, supplementation of the medium with KIV improved the production of isobutanol, which suggests that the endogenous pathway for produc­ing KIV in S. cerevisiae (and potentially, other producing microorganisms) is the limiting step in the isobutanol pro­duction pathway. Consequently, this finding made KIV biosynthesis a rational target for metabolic engineering toward designing more robust isobutanol-producing strains.

Development of hyper-isobutanol-producing strains has typically followed one of the three approaches: (1) identification of rate-limiting steps in Ehrlich/2-ketoacid biosynthetic pathways and overexpression of KIV biosyn­thetic genes; (2) improvement of heterologous expression of enzymes of the isobutanol pathway by codon optimiza­tion; and (3) removal of feedback inhibition and deletion of other competitive pathways, especially competition for pyruvate. Pursuant to the first strategy, Lee et al. (2012) screened and identified a 2-KDC exhibiting a relatively higher activity on KIV through in vitro activity assays of KDC using crude extracts of transformants overexpress­ing KDCs from various microorganisms. The highest KDC activity with KIV was observed from the transform­ant expressing kivd from L. lactis subsp. lactis KACC13877. Subsequently, Chen et al. (2011) evaluated the effect of overexpressing the genes, ILV2, ILV3, ILV5, ILV6, and BAT2, involved in valine metabolism, in different combi­nations in S. cerevisiae, on isobutanol production. Following cultivation of the ILV2, ILV3, and ILV5 overex­pressing strain (ILV235_XCY561) and the reference strain (CEN. PK113-5D) in mineral glucose medium supple­mented with uracil in fermentors under anaerobic condi­tions, the recombinant strain ILV235_XCY561 produced

0. 97 ± 0.14 mg isobutanol/g glucose, which was sixfold higher than the control strain (Chen et al., 2011), hence attesting to the fact that overexpression of the genes ILV2, ILV3, and ILV5 may have led to a higher concentra­tion of KIV, which resulted in higher isobutanol produc­tion. In parallel, Atsumi et al. (2009) engineered a cyanobacterium, Synechococcus elongatus, by expressing a KDC gene (kivd) from L. lactis in this cyanobacterium using an expression cassette under the control of the iso- propyl-b-D-thiogalactoside-inducible promoter Ptrc and integration into neutral site I (Bustos and Golden, 1992) by homologous recombination (Golden et al., 1987), and strain SA578 was generated. To improve flux toward KIV, alsS gene from B. subtilis and the ilvC and ilvD genes from E. coli were integrated into neutral site II (Andersson et al., 2000) in the genome of strain SA578 to generate strain SA590, which produced high levels of isobutyraldehyde upon cultivation in a Roux culture bottle at 30 °C (Atsumi et al., 2009, Figure 7.2). Given the fact that isobutyralde — hyde can easily undergo a reduction reaction to produce isobutanol, Atsumi et al. (2009) evaluated feasibility of using ADHs (ADH2 from S. cerevisiae, YqhD from E. coli, and AdhA from L. lactis) and Kivd from L. lactis (strain SA590) to achieve this reduction reaction. These genes (ADH2, YqhD, and AdhA) were integrated downstream of kivd individually, hence, strains SA413, SA561 and SA562 were generated, respectively. Whereas YqhD, an NADPH-dependent enzyme, was the most active one in S. elongatus, AdhA and ADH2 were nicotinamide adenine

dinucleotide-dependent, and generated strain SA579, which produced 450 mg/l isobutanol.

The second strategy derives from differences in codon bias among different microorganisms, based on their individual transfer RNA content and requisite expression levels of specific proteins in each microor­ganism (Ikemura, 1985; Percudani et al., 1997). Given the fact that codons at the beginning of an open reading frame play a critical role in protein expression (Vervoort et al., 2000), codon bias influences heterologous expres­sion of foreign proteins to a great extent. For instance, heterologously expressed protein levels in E. coli (Atsumi et al., 2010) and S. cerevisiae (Brat and Boles,

2013) were improved by codon optimization, especially at the 50 end of the coding sequence.

To realize the full potential of heterologously overex­pressed genes in producing microorganisms with respect to efficient isobutanol production, six genes including adhE (ADH), ldhA (lactate dehydrogenase), frd (fumarate reductase), fnr (encodes redox-sensing transcription regulator, which partakes in the regulation of lactate synthesis), pta (phosphate acetyltransferase), and pflB (pyruvate formate lyase) that are involved in byproduct formation in E. coli were deleted following overexpres­sion of AlsS (B. subtilis), IlvC (E. coli), and IlvD (E. coli) (Atsumi et al., 2008). These deletions may have increased the level of pyruvate available for the valine biosynthesis pathway. As a consequence, the isobutanol strain (JCL260/pSA55/pSA69) produced more than 22g/l in 112 h (Atsumi et al., 2008). In a similar study, Kondo
et al. (2012) overexpressed 2-KDC and ADH in S. cerevi- siae to enhance the endogenous activity of the Ehrlich pathway followed by overexpression of Ilv2, which cata­lyzes the first step in the valine synthetic pathway and deletion of the PDC1 gene encoding a major PDC with the intent of reducing ethanol flux via pyruvate. As a result, S. cerevisiae YTD306 was generated. Upon cultiva­tion of S. cerevisiae YTD306 along with modification of culture conditions, a 13-fold increase in isobutanol titer was produced (from 11 mg/l to 143 mg/l) when compared with the control (Table 7.2, Kondo et al., 2012).

The strategy described here, in which amino acid bio­synthetic and 2-ketoacid degradation pathways were exploited for isobutanol production, represents a new paradigm for biofuel production. Indeed, this paradigm employs non-CoA-mediated chemistry and uses only pyruvate as a precursor, unlike ethanol and butanol production by native alcohols producing microorgan­isms that are CoA-dependent (Ezeji et al., 2010; Atsumi et al., 2008).