The depleting petroleum storage and increasing environmental deterioration are threatening the sustainable development of human societies. As such, biofuels and chemical feedstocks generated from renewable sources are becoming increasingly important. Although previous efforts led to great success in bio-ethanol production, higher alcohols, fatty acid derivatives including biodiesels, alkanes, and alkenes offer additional advantages because of their compatibility with existing infrastructure. In addition, some of these compounds are useful chemical feedstocks. Since native organisms do not naturally produce these compounds in high quantities, metabolic engineering becomes essential in constructing producing organisms. In this article, we briefly review the four major metabolic systems, the coenzyme-A mediated pathways, the keto acid pathways, the fatty acid pathway, and the isoprenoid pathways, that allow production of these fuel-grade chemicals.
The depleting petroleum reserve, recurring energy crisis, and global climate change are reigniting the enthusiasm for seeking sustainable technologies for replacing petroleum as a source of fuel and chemicals. In the past few decades, efforts in the development of bio-ethanol as an alternative fuel have led to significant success [14–16, 19]. In 2007, 6.5 billion gallons of bio-ethanol was produced in the United State . However, bio-ethanol exhibits some limitations, such as low energy density, high vapor pressure, and corrosiveness, which prevent its widespread utilization given the existing infrastructure.
Higher alcohols (with more than two carbons), biodiesels, and fatty acid derivatives are thought to be more suitable fuels. Their physicochemical properties are more compatible with gasoline-based fuels and allow direct utilization of existing infrastructure for storage and distribution. Furthermore, some of these fuel molecules also serve as important chemical feedstocks. Although the individual biochemical steps for synthesizing these compounds in microbes have been described previously, efforts in putting together highly productive metabolic systems have only begun recently. In this article, we first summarize the metabolic networks for producing these compounds and then review efforts in engineering the non-native producing organism, Escherichia coli. The metabolic networks discussed include the traditional butanol pathway in Clostridium species, the keto acid pathways for higher alcohols, the isoprenoid pathways, and the fatty acid biosynthesis.
The coenzyme-A-dependent fermentative pathways
For n-butanol biosynthesis, acetoacetate has to go through four steps of NADH-dependent reduction and one step of dehydration. Acetoacetate is first reduced to 3-hydroxybutyryl-CoA by 3-hydroxybutyryl-CoA dehydrogenase (HBD). Then, 3-hydroxybutyryl-CoA is dehydrated to crotonyl-CoA by a crotonase (CRT). Third, a butyryl-CoA dehydrogenase (BCD) catalyzes the reduction of crotonyl-CoA to butyryl-CoA. Finally, an aldehyde/alcohol dehydrogenase (AADH) converts butyryl-CoA to n-butanol through two consecutive reduction reactions.
Isopropanol production in Escherichia coli
The secondary alcohol, isopropanol, is both a desirable fuel and an important chemical feedstock in the petrochemical industry. Its dehydrated product, propylene, serves as the monomer for making polypropylene. In addition, it can be used as an additive to petroleum-based fuels. Replacing methanol with isopropanol in the esterification process of fat and oil could generate crystallization-resistant biodiesels .
As described above, isopropanol is produced by Clostridium species in nature. However, as a native metabolite, it can only be produced in a limited amount for the hosts’ own benefits as a detoxification response to low pH conditions. The maximum titer reported in its native producer, Clostridium, was 1.8 g/l . To improve the production of isopropanol, the fully characterized isopropanol biosynthetic pathway (Fig. 1) was reconstructed in the genetic tractable host E. coli .
Escherichia coli has been reported to produce acetone , the immediate precursor of isopropanol, by expressing the intact pathway from Clostridium acetobutylicum ATCC824 consisting of the acetyl-CoA acyltransferase, ACoAT, ADC encoded by the thl, ctfAB, and adc genes, respectively. The reported titer was around 5.4 g/l, similar to the yield of native host for acetone. Furthermore, with a SADH co-expressed with the acetone pathway in E. coli, the isopropanol production was achieved . The pathway efficiency was tuned by using genes from different organisms, a bio-prospecting approach. Since the genes from Clostridium usually have a low GC content, which may lead to poor expression, the E. coli native genes atoB and atoAD, encoding acetyl-CoA acyltransferase and ACoAT, were also tested as pathway components. Additionally, two genes from C. beijerinckii NRRL B593 and Thermoanaerobacter brockii HTD4, encoding SADHs, were totally synthesized with codon optimization and installed into the pathway to test for production. With these efforts, the strain with a combination of C. acetobutylicum thl, E. coli atoAD, C. acetobutylicum adc, and C. beijerinckii adh achieved the highest titer (~5.0 g/l). The result is promising, since it demonstrates 43.5% (mol/mol) conversion ratio. The theoretical yield is 1 mol isopropanol per mole glucose.
The production of isopropanol from glucose is not redox-balanced. Four moles of NADH is produced, while 1 mol of NADPH is consumed per mole of isopropanol. Therefore, an external electron acceptor is required or a byproduct is served as an electron acceptor.
n-Butanol production in E. coli
n-Butanol was proposed to be one of the better substitutes for gasoline-based transportation fuel, because of its high energy density and hydrophobicity. Its energy content (27 MJ/l) is similar to that of gasoline (32 MJ/l). The high hydrophobicity enables its transportation and storage using existing petrochemical infrastructure with minimal modification. In addition, n-butanol has a low vapor pressure of 4 mmHg at 20°C, which allows its mixing with gasoline at any ratio without exceeding air quality specifications.
The microbial production of n-butanol has a history of over 100 years. Traditionally, n-butanol is produced by Clostridium species through the ABE fermentation. However, n-butanol production via this procedure is difficult to control and optimize, particularly because Clostridium exhibits complex physiological features, such as oxygen sensitivity, slow growth rate, and spore-forming life cycles. Thus, it is desirable to create new n-butanol producing organisms using metabolic engineering techniques.
Recently, n-butanol production in a heterologous host, E. coli, using the traditional CoA-dependent pathway originated from C. acetobutylicum (Fig. 1) was reported for the first time . Atsumi et al. created two synthetic operons carrying all the essential genes (thl, hbd, crt, bcd, etfAB, and adhE2) involved in the pathway. Co-expression of the two operons in E. coli led to the initial production of n-butanol at 14 mg/l anaerobically using glucose as sole carbon source. To optimize the pathway, alternative enzymes of different origins were evaluated. More specifically, with E. coli atoB gene in place of C. acetobutylicum thl, a more than threefold increase of n-butanol production was observed. However, replacing the original enzymes for conversion from crotonyl-CoA to butyryl-CoA with homologues and isoenzyme from Megasphaera elsdenii or Streptomyces coelicolor resulted in a much lower yield of n-butanol in E. coli. Nevertheless, this result does not exclude the possibility of the existence of other genes that might improve n-butanol production in E. coli.
Furthermore, n-butanol production does not simply rely on the enzyme activities. The product formation also needs sufficient carbon precursor, acetyl-CoA, and reducing power, NADH. To further improved n-butanol production, the host E. coli strain was engineered by deleting the native pathway competing for both carbon flux and reducing power. The best strain candidate, named JCL88, with the deletion of ldhA, adhE, frdBC, pta, and fnr, allowed a more than twofold increase in n-butanol production, accompanied by the dramatic drop in the formation of lactate, acetate, ethanol, and succinate. The highest titer of 552 mg/l was reported with optimized pathway and improved strain. Although the yield was still low, this work demonstrated the feasibility of heterologous n-butanol production and proposed the principles for further optimization.
The keto acid pathways
Isobutanol production in E. coli
Note that isobutanol is toxic to E. coli at a concentration >10 g/l. However the production of isobutanol occurs mainly in the non-growing phase (Fig. 4) . This result indicates that even though the cells cannot grow at the higher concentration, they nonetheless continue to produce and excrete isobutanol. Thus, even though isobutanol toxicity poses a challenge, the production level can exceed the toxicity level significantly. Mutants with higher isobutanol tolerance have been isolated , which also improves the productivity. Such a high-yield production demonstrates the versatility in exploring the keto acid pathways for biofuel production. The production of isobutanol (<1 g/l) was also reported in a patent application (Donaldson et al., US patent application, US2007/0092957), which is currently under examination. Such results demonstrate that large-scale commercialization of this technology is promising.
n-Butanol and n-propanol production in E. coli
Overexpression of ilvA-leuABCD led to a threefold increase in n-butanol production compared to that without overexpression. Threonine-feeding also caused a dramatic increase in n-butanol production, indicating the limited availability of endogenous threonine for n-butanol production . With the identification of these limiting steps, systematic approaches were taken to further improve the n-butanol and n-propanol co-production in E. coli through deregulation of amino acid biosynthesis and elimination of competing pathways . More specifically, the operon consisting of feedback-resistant thrA and thrBC was overexpressed to relieve the threonine feedback inhibition; the host native genes metA and tdh were disrupted to prevent the carbon flux leakage out of threonine biosynthetic pathway. The genes ilvB and ilvI were also knocked out to avoid the divergence of ketobutyrate to isoleucine, valine, and leucine biosynthesis. With these efforts, a production titer of 2 g/l with a nearly 1:1 ratio of n-butanol and n-propanol was achieved.
In addition to the threonine pathway, an alternative pathway was identified in Leptospira interrogans and Methanocaldococcus jannaschii contributing the ketobutyrate formation [8, 13]. The enzyme citramalate synthase (CimA) plays a key role in this pathway (Fig. 5), which directly converts pyruvate to ketobutyrate. This pathway represents a shorter keto acid-mediated pathway to produce n-propanol and n-butanol from glucose. Atsumi and Liao  took advantage of the growth phenotype associated with keto acid deficiency and developed a growth-based screening method to evolve CimA from Methanocaldococcus jannaschii. The best CimA mutant demonstrated both insensitivity to isoleucine feedback inhibition and higher catalytic activity, which enabled 22- and 9-fold increases in n-butanol and n-propanol production compared to wild type CimA. The highest titer of n-propanol and n-butanol reported in this work were around 3.5 and 0.5 g/l, respectively.
2MB and 3MB production in E. coli
The keto aid pathways also enable the biosynthesis of 5-carbon alcohols, including 2MB and 3MB. Their gram-level production in recombinant E. coli was reported recently [7, 10]. Production of 2MB in E. coli harnesses isoleucine biosynthesis. It shares the common intermediate, ketobutyrate, with n-propanol and n-butanol production. Furthermore, 2MB production also shares the ilvIHCD pathway with isobutanol production (Fig. 5). Therefore, to achieve its hyper and selective production is challenging. To shift the carbon flux towards 2MB biosynthesis, the approaches proven to be effective for improving ketobutyrate availability in the n-propanol and n-butanol work were still applicable, which included overexpression of thrABC operon and deletion of metA and tdh. Furthermore, leuABCD operon was disrupted to improve the product specificity. Exploring the biodiversity of enzymes catalyzing the key reactions from threonine to KMV, the authors found that overexpression of Salmonella typhimurium AHAS II (ilvGM) and Corynebacterium glutamicum threonine deaminase (ilvA) was more suitable for 2MB-specific production. Combining these approaches, the engineered strain produced 1.25g/l 2MB in 24 h .
Similarly, leucine biosynthetic pathway leading to 3MB production is an extension of the valine pathway that was engineered for isobutanol production (Fig. 3). 3MB and isobutanol compete for the same substrate KIV. Improving 3MB-specific production in E. coli requiring redistribution of carbon flux between the two branches remained challenging, since the isobutanol production was so efficient already. Overexpressing E. coli native leuABCD operon did not cause much improvement of 3MB-specific production. This was mainly due to the feedback inhibition of free leucine on the activity of 2-isopropylmalate synthase. Thus, leucine-resistant leuA mutant was employed. In addition, the genes tyrB and ilvE were further disrupted in the previously created isobutanol high producer JCL260 to prevent valine and leucine formation. When the JCL260 ΔilvEΔ tyrB strain expressing the mutated leuA along with alsS-ilvCD was examined for 3MB production, a final titer of 1.28 g/l 3MB was obtained in 28 h .
The fatty acid biosynthesis pathway
Fatty acid derivatives: biodiesels and long-chain alkanes/alkenes
As a possible substitute for petroleum-based diesel fuel, biodiesel is made from plant oils through transesterification of triacylglycerols with methanol or ethanol. The generated products are also called fatty acid methyl esters or fatty acid ethyl esters (FAEEs). Although claiming attractive petroleum-diesel-like properties and positive ecological effects, large-scale application of biodiesel seems difficult because of the need to circumvent the geographical and seasonal restrictions, as well as the current costliness of the transesterification procedure . To overcome these drawbacks, Kalscheuer et al., engineered E. coli to produce FAEEs. The traditional ethanol pathway consisting of pyruvate decarboxylase and alcohol dehydrogenase was introduced into E. coli to supply ethanol as building units. Subsequent esterification of ethanol with the acyl moieties of coenzyme-A thioesters of fatty acids was achieved by co-expressing a non-specific acyltransferase from Acinetobacter baylyi ADP1. The metabolically engineered E. coli strain was reported to be able to produce FAEEs at a titer of 1.28g/l, accounting for 26% of cellular dry mass by using glucose and oleic acid as substrates.
Recently, production of fatty acid derivatives as biofuels has also been reported in patent applications (e.g., Keasling et al., patent application, WO/2007/136762); this process harnessed the existing fatty acid synthesis machinery in E. coli. The E. coli LS9001 was first engineered from membrane-protein-friendly host C41(DE3) by disrupting the fadE gene. The resulting strain was not capable of degrading fatty acids and fatty acyl-CoAs. As control, the constructed host was able to produce fatty alcohol at amounts of only 0.2–0.5 mg/l. When the acyl-CoA reductase gene acrI from A. baylyi ADP1 and acyl-CoA synthetase gene fadD from E. coli were overexpressed, a fivefold increase in fatty alcohol production was observed. An additional increase was achieved by co-expressing E. coli accABCD and tesA, which encode acetyl-CoA carboxylase and thioesterase. To produce wax ester, the wax synthase gene from A. baylyi ADP1 was introduced into constructed fatty alcohol production strain for overexpression. The intracellular wax yield of 10 mg/l was reported with 50-ml shake flask fermentation. More interestingly, selection of various thioesterases and modulation of fatty acid biosynthesis allow for the production of fatty acid derivatives with defined carbon chain length, saturation points, and branch points. Similarly, in the patent application WO/2008/113041, Friedman and Rude identified four genes (oleA, oleB, oleC, and oleD) from Stenotrophomonas maltophilia that encode proteins involved in the biosynthesis of hydrocarbons, such as olefins and hydrocarbon intermediates, such as aliphatic ketones, by shunting fatty acid biosynthesis. Overexpression of oleA, oleC, and oleD in E. coli C41(DE3) resulted in olefin production at 7.5 mg/l with carbon chain length ranging from 27 to 31. The highest titer of 32 mg/l olefins was achieved when the host strain was modified by overexpressing fadD and tesA, and deleting fadE. The patent also disclosed the approaches of cracking olefins and fatty acid esters to short-chain hydrocarbons as fuels and specialty chemicals.
Isoprenoid derivatives as biofuels
Recently, Withers et al. reported two genes in B. subtilis 6051 whose products can convert the prenyl diphosphate precursors to corresponding isoprentenols . On this basis, the patent application US2008/0092829 has been filed by Renninger et al. for the production of unsaturated C5 alcohols, mainly, 3-methyl-3-buten-1-ol, and 3-methyl-2-buten-1-ol. The patent application also disclosed over ten schemes to produce corresponding derivatives using these two alcohols via chemical reactions. The reported titer in the open literature for isoprentenol is relatively low, only 110 mg/l with a production rate of 2.6 mg/l/h . With industrial efforts, the titer reached 1.2 g/l.
Although native organisms may produce a desired compound, they typically regulate the production strictly for their benefit of growth and survival. Therefore, manipulating a native producer for the production purpose may face challenges that limit process improvement. Large-scale transfer and modification of metabolic pathways have become a powerful approach for production of metabolites from renewable sources. As such, metabolic engineers (or synthetic biologists) can readily design metabolic systems that combine genes from several different organisms for the production of fuels in user-friendly organisms. Such practice can both serve as a proof of concept and generate a production host in a short time scale.
The metabolic systems discussed above represent the four routes for production of advanced biofuels. On the basis of published literature to date, the keto acid pathway appears to be the most promising, since the isobutanol production has reached a titer of more than 20 g/l . This pathway branches out from amino acid biosynthesis, which carries significance. The large-scale production of amino acids has enjoyed many decades of commercial success, indicating that the flux through these pathways can be readily manipulated. Because production of isobutanol mainly occurs in the stationary phase, the flux diversion has minimal impact on growth. Since amino acid biosynthesis exists in almost all microorganisms, the keto acid pathway can be implemented in many different organisms for utilizing different raw materials. The keto acid platform technology opens the possibility for producing many higher alcohols, which can be readily dehydrated to yield hydrocarbons. Thus, this platform technology can be used for making bio-gasoline, bio-jet fuel, and biodiesel, as well as chemical feed stocks.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.