Environmental,economic,and geopolitical concerns over continued fossil-fuel dependence have spurred research into the conversion of renewable biomass to ‘‘drop-in’’ fuels. Biofuels derived from hydrocarbon biosynthetic pathways have attracted increasing attention. These fuels have great potential to replace petroleum-based liquid transportation fuels,as they have highenergy content and physicochemical properties comparable to fossil fuels,and hence are compatible with current engines, distribution systems and storage infrastructure. Routes to hydrocarbon biofuels are emerging and mainly fall into two categories based on the metabolic pathways utilized: Fatty acid pathwaybased alkanes/alkenes and isoprenoid biosynthetic pathway based terpenes (Fig. 1). This review focuses on recent progress in the application of hydrocarbon biosynthetic pathways for hydrocarbon biofuel production,together with studies on enzymes, including efforts to engineering them for improved performance.

Fig. 1

Download: JPG larger image

Fig. 1. Overview of the hydrocarbon biofuel biosynthetic pathways as reviewed here. Abbreviations: G3P: glyceraldehyde 3-phosphatel; PEP,phosphaenolpyruvate; ACP,acyl carrier protein; IPP,isoprenyl pyrophosphate; DMAPP,dimethyllallyl pyrophosphate; GPP,geranyl pyrophosphate; FPP,farnesyl pyrophosphate; GGPP, geranylgeranyl pyrophosphate; FS,farnesene synthase; PS,pinene synthase; LS, limonene synthase; BIS,bisabolene synthase.

There are mainly three types of hydrocarbons derived from the fatty acid pathway: Alkanes,α-Olefins and internal olefins. Alkanes are biosynthesized ubiquitously in nature where they function as waterproofing agents in plants’ leaves ^[1] and birds’ feathers ^[2], as pheromones in insects ^[3],and as energy storage molecules in algae ^[4]. Recently,an alkane biosynthesis pathway from cyanobacteria,comprising acyl-coA reductase (cACR) and aldehyde deformylating oxygenase (cADO),was identified through comparative genomics ^[5]. In this pathway,fatty acyl-ACP (CoA) is converted by cACR to fatty aldehydes,which are then converted by cADO to alkanes/alkenes through losing the aldehyde carbon as formate. This heterologous pathway was introduced into E. coli, leading to 300 mg/L alkane production over 40 h ^[5]. A similar hydrocarbon biosynthesis pathway was constructed using carboxylic acid reductase (CAR) from Mycobacterium marinum to convert fatty acids (C8-C16) into the corresponding aldehydes,which were further converted to alkanes (C7-C15) by cADO ^[6].

α-Olefins,hydrocarbons with terminal double bonds,can be blended with diesel fuels. In a recent effort to elucidate their biosynthesis,a terminal olefin-forming fatty acid decarboxylase from the bacteria Jeotgalicoccus was identified and named OleT ^[7]. OleT catalyzes the decarboxylation of free fatty acids to generate α-Olefins and is a cytochrome P450 enzyme. The heterologous expression of OleT enabled E. coli to produce 1-pentadecene and 1,10-heptadecadiene at unreported titers. α-Olefins can also be formed from fatty acids by an elongationdecarboxylation mechanism,similar to that of a polyketide synthase. A gene named ols (olefin synthase) was identified to be involved in the α-Olefin biosynthesis in cyanobacterium Synechococcus sp. PCC 7002 ^[8]. However,no α-Olefin was detected when ols was introduced into E. coli,because it was found that the ols gene was not actively expressed ^[9].

Internal olefins can be generated through the head-to-head condensation of fatty acids catalyzed by a set of enzymes termed OleA,B,C and D. The heterologous expression of a three-gene cluster from Micrococcus luteus (Mlut_13230-13250) in E. coli led to the production of long-chain internal olefins,mainly C27:3 and C29:3 ^[10]. 2.2. Enzymology and engineering of related enzymes 2.2.1. cADO

cADO catalyzes the conversion of fatty aldehydes to alkanes. In contrast to other decarbonylases from insects or plants,which are integral membrane proteins,cADO is soluble and can easily be expressed recombinantly in E. coli,making it more amenable to study. The crystal structure of cADO (pdb 2OC5) fromProchlorococcus marinus shows that cADO is a member of the nonheme di-iron family of oxygenases exemplified by enzymes such as methane monooxygenase (MMO),class I ribonucleotide reductase,and fattyacyl- ACP desaturase. Interestingly,although the deformylation reaction is redox neural,cADO requires molecular oxygen and an external reducing system,either a protein reductive system (NADPH, ferredoxin,and ferredoxin reductase) ^[5] or a chemical reducing system (phenazine methosulfate and NADH) ^[11]. During the reaction,O₂ is completely reduced with one atom of oxygen incorporated into formate and the other into water ^[12]. The mechanism of cADO has been the subject of intense interest,the reader is referred to a recent review for more details ^[13]. However, the activity of cADO in vitro is extremely low. It has been reported that the in vitro activity of cADO is inhibited by hydrogen peroxide (H₂O₂) and interestingly,this inhibition can be relieved by fusing catalase to cADO and convertingH₂O₂ to the cosubstrateO₂ ^[14]. Even so,the highest steady turnover number achieved is only ~1min^-1. 2.2.2. cACR

Despite the fact that it synthesizes highly insoluble products (fatty aldehydes),cACR appears to be a cytosolic enzyme. However, its tendency to form inclusion bodies when expressed in E. coli has hindered mechanistic studies on this enzyme. Recombinant cACR from Synechococcus elongatus has been characterized. The enzyme is specific for NADPH and catalyzes the reduction of fatty acyl-CoA (ACP) to the corresponding aldehydes,rather than to alcohols ^[15]. However,over-expression of cACR in E. coli resulted in the production of both even chain fatty aldehydes and fatty alcohols ^[5]. It is believed that the production of fatty alcohols was due to intrinsic alcohol dehydrogenases in E. coli. The enzyme was shown to function by a well-precedented mechanism involving the formation of an enzyme-thioester intermediate. cACR required divalent metal ions,e.g. Mg²⁺,for activity and was stimulated significantly by K⁺. The enzyme was active toward the reduction of acyl-CoA of chain lengths ranging from 12 to 20 carbon atoms,with the highest enzymatic activity toward stearoyl-CoA. Surprisingly, given the straightforward reduction chemistry involved,cACR exhibited very slow turnovers with k_cat = 0.36 ± 0.023 min^-1. Both cADO and cACR have slow turnover numbers,posing challenges for their use in biofuel application. Furthermore,the toxicity of the aldehyde products,if produced in high concentration in recombinant strains,would need to be considered and resolved. 2.2.3. CAR

CAR catalyzes the reduction of aromatic (including benzoic, vanillic,and ferulic acids) and C4-C18 carboxylic acids to their corresponding aldehydes in a reaction that requires ATP andNADPH ^{[6, 16]}. For activity,CAR requires the prosthetic group 4'-phosphopantetheine, which is covalently bound through a phosphodiester bond to a serine residue ^[17]. This modification is carried out posttranslationally by a separate phosphopantetheinyl transferase enzyme ^[17]. The reaction catalyzed by CAR involves three key steps: (i) adenylylation of the bound fatty acid substrate to form an AMP-fatty acyl complex and pyrophosphate,(ii) transfer of the activated fatty acid to the phosphopantetheine prosthetic group with the formation of a reactive thioester linkage,and (iii) reduction of the thioester intermediate to the aldehyde by NADPH. 2.2.4. OleT

OleT is a terminal olefin-forming fatty acid decarboxylase. OleT has 59% amino acid sequence identity to a P450 from Macrococcus caseolyticus and was assigned to the cyp152 family that comprises P450s from various bacteria. Although the function of most P450s in the cyp152 family is unknown,several members of the family,such as Bacillus subtilis P450 (P450_BSb) and Sphingomonas paucimobilis P450 (P450_SPa),have been revealed to hydroxylate fatty acids at either the α- or β-position. Interestingly,it is observed that OleT also catalyzed the α- and β-hydroxylation of fatty acids as side reactions ^{[7, 18]}. Moreover,OleT has a structure highly similar to P450_BSb ^[19]. Therefore,there appears to be a mechanistic link between the ability of a P450 enzyme to hydroxylate fatty acids in the β-position and the ability to oxidatively decarboxylate fatty acids to the terminal olefin. In contrast to other P450s,which require O₂,NADPH,and redox partners,OleT is highly active with hydrogen peroxide as the oxidant and is classified as a peroxygenase. OleT showed activity toward saturated fatty acids ranging from C12 to C20. Recent efforts to engineer a self-sufficient version of OleT by fusing it to the P450 reductase (RhFRED) domain from Rhodococcus sp. NCIMB 9784 resulted in an OleT variant that efficiently decarboxylates long-chain fatty acids in the presence of O₂ and NADPH. When this engineered OleT was expressed in E. coli,the recombinant strain produced 97.6 mg/L α-Olefin ^[18]. 2.2.5. ols

ols,standing for olefin synthase,encodes a large multidomain protein in Synechococcus with homology to type I polyketide synthases,which is capable of converting fatty acyl-ACP into α-Olefins by sequential polyketide synthase chain elongation,keto reduction,sulfonation aided by its sulfotransferase domain (ST), and the subsequent hydrolysis and decarboxylation catalyzed by the thioesterase domain ^[8]. The crystal structure of ST suggested that ST is a member of a distinct protein family relative to other prokaryotic and eukaryotic sulfotransferases,and its substrate selectivity and product formation is affected by a unique dynamic active-site flap ^[20]. 2.2.6. OleABCD

oleABCD,a cluster of four genes,is involved in internal olefin biosynthesis. OleA catalyzes the condensation of two acyl thioesters between the carbonyl of one acyl chain and the alpha carbon of the second to form an aliphatic diketone,which is followed by a series of reductions and dehydrations that is catalyzed by OleD,and possibly OleC and/or OleB that yields the final internal olefin ^{[10, 21]}. The OleABCD protein families were classified to belong to the thiolase,α/βs-hydrolase,AMP dependent ligase/synthase,and short-chain dehydrogenase superfamilies, respectively ^[21]. 3. Isoprenoid pathway based terpenes 3.1. Metabolic pathways and application to biofuel production

Terpenes are a remarkably diverse group of natural products that are derived from isoprene units. They are synthesized by plants,where they play a variety of functions as defense materials, pollinator attractants and allelopathy compounds. Terpenes have traditionally found uses as drugs,fragrances,flavors,nutraceuticals etc.,but more recently have been investigated as advanced biofuel precursors. Short,branched-chain terpenes may be favorable alternatives to gasoline,while longer branched-chain or cyclic ones may be used as diesel and jet fuel substitutes. There are three types of terpenes that are being explored for hydrocarbon fuels: monoterpenes (C10,pinene and limonene),sesquiterpene (C15,bisabolene) and diterpene (C20,taxadinene). These compounds are derived either from the mevalonate (MEV) pathway or the 1-deoxy-D-xylulose-5-phosphate (DXP) pathway,which produce the common five-carbon monomer precursor isoprenyl pyrophosphate (IPP) and its isomer dimethylallyl pyrophosphate (DMAPP) (Fig. 1). Head-to-tail condensation of IPP and DMAPP is carried out by different IPP synthases to generate geranyl pyrophosphate (GPP,C10),farnesyl pyrophosphate (FPP,C15),or geranylgeranyl pyrophosphate (GGPP,C20),which are,in turn, converted by respective terpene synthases to branched or cyclic terpenes.

Pinene dimers have strained ring systems similar to those found in the tactical fuels JP-10 and RJ-5 and thus have high energy densities similar to those of JP-10 and RJ-5 ^[22]. They are formed by pinene dimerization using chemical catalysis. Microbial production of pinene was achieved in E. coli by the introduction of the geranyl pyrophosphate synthase (GPPS) and the pinene synthase (PS) fromdifferent specieswith a titer ranging from1mg/L to 32 mg/L ^{[23, 24, 25]}. Another monoterpene,limonene,was also microbially produced in E. coli at a much higher titer of 400 mg/L through the introduction of a heterologous MEV pathway from Saccharomyces cerevisiae and a limonene synthase ^[26].

Bisabolane,derived from bisabolene by chemical hydrogenation, is regarded as a promising alternative to the D2 diesel fuel ^[27]. It has been demonstrated that bisabolene,the precursor for bisabolane,can be produced in both E. coli and S. cerevisiae with a high titer of > 900 mg/L ^[27]. Included in the engineering efforts undertaken to achieve this titer were enzyme screening for highly efficient sesquiterpene synthases,codon-optimization of the heterologous pathway genes,and promoter optimization. Farnesene is another sesquiterpene that can be hydrogenated into farnesane,which shows a high cetane number of 58. The production of farnesene at a titer up to 23.6 g/L by an engineered strain of E. coli under aerobic,nitrogen limited,fed-batch condition has been reported ^[28]. 3.2. Enzymology and engineering of related enzymes

Terpene synthases are key to the microbial biosynthesis of terpenes as fuel precursors. All terpene synthases from both plants and microorganisms require a divalent metal ion for enzyme activity,normally Mg²⁺ or Mn²⁺,to counteract the negative charge of the diphosphate leaving group in the ionization steps of the reaction. The divalent metal ion is bound by a conserved aspartaterich motif: Asp-Asp-Xaa-Xaa-Asp/Glu (DDXXD/E),which is found in all terpene synthases thus far identified in both plants and microorganisms ^[29]. This DDXXD/E motif is important for the catalytic reaction and might impact the allowable cyclization modes of terpene synthases ^[30]. The gymnosperm monoterpene synthases further depend on a monovalent cation K⁺,while the angiosperm monoterpene synthases do not ^[31]. Crystal structures of several terpene synthases have shown that a similar active site scaffold exits in all terpene synthases,termed the core terpenoid synthase fold,despite the highly diverse nature of the terpene products synthesized. This fold is proposed to have evolved divergently from an ancestor to achieve different biological reactions for diverse natural products ^{[32, 33, 34, 35]}.

Terpene synthases seem to be very promiscuous enzymes; in some cases one terpene synthase is responsible for dozens of products from one substrate via a wide variety of cyclization mechanisms ^{[30, 36]}. This promiscuity makes terpene synthases good targets for protein engineering to improve reaction specificity or to synthesize new products. For example,γ-humulene synthase, which produces 52 different sesquiterpenes from farnesyl diphosphate,was systematically recombined based on a mathematical model to construct seven specific and active terpene synthases,each yielding one or a few sesquiterpene products while largely maintaining the specific activity of the wild-type enzyme ^[37]. 4. Concluding remarks

Hydrocarbon biofuels are excellent alternatives to fossil fuels and can be derived from either the fatty acid biosynthesis pathway or the isoprenoid pathway. Although the production of hydrocarbon biofuels has been successfully demonstrated in recombinant E. coli and S. cerevisiae,the titers are generally low,usually less than 1 g/L,which is far below the requirement for commercial viability. Therefore,further significant efforts are still required to optimize hydrocarbon-producing enzymes,the fatty acid and isoprenoid biosynthetic pathways,together with other aspects of the cellular machinery before the potential of these fuels can be fully realized.